tahoe-lafs/contrib/fuse/impl_c/blackmatch.py

1516 lines
50 KiB
Python
Raw Normal View History

macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
#!/usr/bin/env python
#-----------------------------------------------------------------------------------------------
from allmydata.uri import CHKFileURI, NewDirectoryURI, LiteralFileURI
from allmydata.scripts.common_http import do_http as do_http_req
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
from allmydata.util.hashutil import tagged_hash
from allmydata.util.assertutil import precondition
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
from allmydata.util import base32, fileutil
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
from allmydata.scripts.common import get_aliases
from twisted.python import usage
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
from twisted.internet.protocol import Factory, Protocol
from twisted.internet import reactor, defer
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
import base64
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
import errno
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
import sha
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
import socket
import stat
import subprocess
import sys
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
import os
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
#import pprint
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
# one needs either python-fuse to have been installed in sys.path, or
# suitable affordances to be made in the build or runtime environment
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
import fuse
import time
import traceback
import simplejson
import urllib
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
VERSIONSTR="0.7"
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
USAGE = 'usage: tahoe fuse [dir_cap_name] [fuse_options] mountpoint'
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
DEFAULT_DIRECTORY_VALIDITY=26
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
if not hasattr(fuse, '__version__'):
raise RuntimeError, \
"your fuse-py doesn't know of fuse.__version__, probably it's too old."
fuse.fuse_python_api = (0, 2)
fuse.feature_assert('stateful_files', 'has_init')
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
class TahoeFuseOptions(usage.Options):
optParameters = [
["node-directory", None, "~/.tahoe",
"Look here to find out which Tahoe node should be used for all "
"operations. The directory should either contain a full Tahoe node, "
"or a file named node.url which points to some other Tahoe node. "
"It should also contain a file named private/aliases which contains "
"the mapping from alias name to root dirnode URI."
],
["node-url", None, None,
"URL of the tahoe node to use, a URL like \"http://127.0.0.1:8123\". "
"This overrides the URL found in the --node-directory ."],
["alias", None, None,
"Which alias should be mounted."],
["root-uri", None, None,
"Which root directory uri should be mounted."],
["cache-timeout", None, 20,
"Time, in seconds, to cache directory data."],
]
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
optFlags = [
['no-split', None,
'run stand-alone; no splitting into client and server'],
['server', None,
'server mode (should not be used by end users)'],
['server-shutdown', None,
'shutdown server (should not be used by end users)'],
]
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
def __init__(self):
usage.Options.__init__(self)
self.fuse_options = []
self.mountpoint = None
def opt_option(self, fuse_option):
"""
Pass mount options directly to fuse. See below.
"""
self.fuse_options.append(fuse_option)
opt_o = opt_option
def parseArgs(self, mountpoint=''):
self.mountpoint = mountpoint
def getSynopsis(self):
return "%s [options] mountpoint" % (os.path.basename(sys.argv[0]),)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
logfile = file('tfuse.log', 'ab')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
def reopen_logfile(fname):
global logfile
log('switching to %s' % (fname,))
logfile.close()
logfile = file(fname, 'ab')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def log(msg):
logfile.write("%s: %s\n" % (time.asctime(), msg))
#time.sleep(0.1)
logfile.flush()
fuse.flog = log
def unicode_to_utf8(u):
precondition(isinstance(u, unicode), repr(u))
return u.encode('utf-8')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def do_http(method, url, body=''):
resp = do_http_req(method, url, body)
log('do_http(%s, %s) -> %s, %s' % (method, url, resp.status, resp.reason))
if resp.status not in (200, 201):
raise RuntimeError('http response (%s, %s)' % (resp.status, resp.reason))
else:
return resp.read()
def flag2mode(flags):
log('flag2mode(%r)' % (flags,))
#md = {os.O_RDONLY: 'r', os.O_WRONLY: 'w', os.O_RDWR: 'w+'}
md = {os.O_RDONLY: 'rb', os.O_WRONLY: 'wb', os.O_RDWR: 'w+b'}
m = md[flags & (os.O_RDONLY | os.O_WRONLY | os.O_RDWR)]
if flags & os.O_APPEND:
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
m = m.replace('w', 'a', 1)
return m
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
class TFSIOError(IOError):
pass
class ENOENT(TFSIOError):
def __init__(self, msg):
TFSIOError.__init__(self, errno.ENOENT, msg)
class EINVAL(TFSIOError):
def __init__(self, msg):
TFSIOError.__init__(self, errno.EINVAL, msg)
class EACCESS(TFSIOError):
def __init__(self, msg):
TFSIOError.__init__(self, errno.EACCESS, msg)
class EEXIST(TFSIOError):
def __init__(self, msg):
TFSIOError.__init__(self, errno.EEXIST, msg)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def logargsretexc(meth):
def inner_logargsretexc(self, *args, **kwargs):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
log("%s(%r, %r)" % (meth, args, kwargs))
try:
ret = meth(self, *args, **kwargs)
except:
log('exception:\n%s' % (traceback.format_exc(),))
raise
log("ret: %r" % (ret, ))
return ret
inner_logargsretexc.__name__ = '<logwrap(%s)>' % (meth,)
return inner_logargsretexc
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def logexc(meth):
def inner_logexc(self, *args, **kwargs):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
try:
ret = meth(self, *args, **kwargs)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
except TFSIOError, tie:
log('error: %s' % (tie,))
raise
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
except:
log('exception:\n%s' % (traceback.format_exc(),))
raise
return ret
inner_logexc.__name__ = '<logwrap(%s)>' % (meth,)
return inner_logexc
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def log_exc():
log('exception:\n%s' % (traceback.format_exc(),))
class TahoeFuseFile(object):
#def __init__(self, path, flags, *mode):
def __init__(self, tfs, path, flags, *mode):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
log("TFF: __init__(%r, %r, %r)" % (path, flags, mode))
self.tfs = tfs
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self._path = path # for tahoe put
try:
self.parent, self.name, self.fnode = self.tfs.get_parent_name_and_child(path)
m = flag2mode(flags)
log('TFF: flags2(mode) -> %s' % (m,))
if m[0] in 'wa':
# write
self.fname = self.tfs.cache.tmp_file(os.urandom(20))
if self.fnode is None:
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
log('TFF: [%s] open() for write: no file node, creating new File %s' % (self.name, self.fname, ))
self.fnode = File(0, 'URI:LIT:')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.fnode.tmp_fname = self.fname # XXX kill this
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.parent.add_child(self.name, self.fnode, {})
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
elif hasattr(self.fnode, 'tmp_fname'):
self.fname = self.fnode.tmp_fname
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
log('TFF: [%s] open() for write: existing file node lists %s' % (self.name, self.fname, ))
else:
log('TFF: [%s] open() for write: existing file node lists no tmp_file, using new %s' % (self.name, self.fname, ))
self.file = os.fdopen(os.open(self.fname, flags|os.O_CREAT, *mode), m)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.fd = self.file.fileno()
log('TFF: opened(%s) for write' % self.fname)
self.open_for_write = True
else:
# read
assert self.fnode is not None
uri = self.fnode.get_uri()
# XXX make this go away
if hasattr(self.fnode, 'tmp_fname'):
self.fname = self.fnode.tmp_fname
log('TFF: reopening(%s) for reading' % self.fname)
else:
log('TFF: fetching file from cache for reading')
self.fname = self.tfs.cache.get_file(uri)
self.file = os.fdopen(os.open(self.fname, flags, *mode), m)
self.fd = self.file.fileno()
self.open_for_write = False
log('TFF: opened(%s) for read' % self.fname)
except:
log_exc()
raise
def log(self, msg):
log("<TFF(%s:%s)> %s" % (os.path.basename(self.fname), self.name, msg))
@logexc
def read(self, size, offset):
self.log('read(%r, %r)' % (size, offset, ))
self.file.seek(offset)
return self.file.read(size)
@logexc
def write(self, buf, offset):
self.log("write(-%s-, %r)" % (len(buf), offset))
if not self.open_for_write:
return -errno.EACCES
self.file.seek(offset)
self.file.write(buf)
return len(buf)
@logexc
def release(self, flags):
self.log("release(%r)" % (flags,))
self.file.close()
if self.open_for_write:
size = os.path.getsize(self.fname)
self.fnode.size = size
file_cap = self.tfs.upload(self.fname)
self.fnode.ro_uri = file_cap
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
# XXX [ ] TODO: set metadata
# write new uri into parent dir entry
self.parent.add_child(self.name, self.fnode, {})
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.log("uploaded: %s" % (file_cap,))
# dbg
print_tree()
def _fflush(self):
if 'w' in self.file.mode or 'a' in self.file.mode:
self.file.flush()
@logexc
def fsync(self, isfsyncfile):
self.log("fsync(%r)" % (isfsyncfile,))
self._fflush()
if isfsyncfile and hasattr(os, 'fdatasync'):
os.fdatasync(self.fd)
else:
os.fsync(self.fd)
@logexc
def flush(self):
self.log("flush()")
self._fflush()
# cf. xmp_flush() in fusexmp_fh.c
os.close(os.dup(self.fd))
@logexc
def fgetattr(self):
self.log("fgetattr()")
s = os.fstat(self.fd)
self.log("fgetattr() -> %r" % (s,))
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
return stat_to_dict(s)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def ftruncate(self, len):
self.log("ftruncate(%r)" % (len,))
self.file.truncate(len)
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
class TahoeFuseBase(object):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
def __init__(self, tfs):
log("TFB: __init__()")
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.tfs = tfs
self.files = {}
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def log(self, msg):
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
log("<TFB> %s" % (msg, ))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def readlink(self, path):
self.log("readlink(%r)" % (path,))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
node = self.tfs.get_path(path)
if node:
raise EINVAL('Not a symlink') # nothing in tahoe is a symlink
else:
raise ENOENT('Invalid argument')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def unlink(self, path):
self.log("unlink(%r)" % (path,))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.tfs.unlink(path)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def rmdir(self, path):
self.log("rmdir(%r)" % (path,))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.tfs.unlink(path)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def symlink(self, path, path1):
self.log("symlink(%r, %r)" % (path, path1))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.tfs.link(path, path1)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def rename(self, path, path1):
self.log("rename(%r, %r)" % (path, path1))
self.tfs.rename(path, path1)
@logexc
def link(self, path, path1):
self.log("link(%r, %r)" % (path, path1))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.tfs.link(path, path1)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def chmod(self, path, mode):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.log("XX chmod(%r, %r)" % (path, mode))
#return -errno.EOPNOTSUPP
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def chown(self, path, user, group):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.log("XX chown(%r, %r, %r)" % (path, user, group))
#return -errno.EOPNOTSUPP
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def truncate(self, path, len):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.log("XX truncate(%r, %r)" % (path, len))
#return -errno.EOPNOTSUPP
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def utime(self, path, times):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.log("XX utime(%r, %r)" % (path, times))
#return -errno.EOPNOTSUPP
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def statfs(self):
self.log("statfs()")
"""
Should return an object with statvfs attributes (f_bsize, f_frsize...).
Eg., the return value of os.statvfs() is such a thing (since py 2.2).
If you are not reusing an existing statvfs object, start with
fuse.StatVFS(), and define the attributes.
To provide usable information (ie., you want sensible df(1)
output, you are suggested to specify the following attributes:
- f_bsize - preferred size of file blocks, in bytes
- f_frsize - fundamental size of file blcoks, in bytes
[if you have no idea, use the same as blocksize]
- f_blocks - total number of blocks in the filesystem
- f_bfree - number of free blocks
- f_files - total number of file inodes
- f_ffree - nunber of free file inodes
"""
block_size = 4096 # 4k
preferred_block_size = 131072 # 128k, c.f. seg_size
fs_size = 8*2**40 # 8Tb
fs_free = 2*2**40 # 2Tb
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
#s = fuse.StatVfs(f_bsize = preferred_block_size,
s = dict(f_bsize = preferred_block_size,
f_frsize = block_size,
f_blocks = fs_size / block_size,
f_bfree = fs_free / block_size,
f_bavail = fs_free / block_size,
f_files = 2**30, # total files
f_ffree = 2**20, # available files
f_favail = 2**20, # available files (root)
f_flag = 2, # no suid
f_namemax = 255) # max name length
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
#self.log('statfs(): %r' % (s,))
return s
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def fsinit(self):
self.log("fsinit()")
##################################################################
@logexc
def readdir(self, path, offset):
self.log('readdir(%r, %r)' % (path, offset))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
node = self.tfs.get_path(path)
if node is None:
return -errno.ENOENT
dirlist = ['.', '..'] + node.children.keys()
self.log('dirlist = %r' % (dirlist,))
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
#return [fuse.Direntry(d) for d in dirlist]
return dirlist
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def getattr(self, path):
self.log('getattr(%r)' % (path,))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
if path == '/':
# we don't have any metadata for the root (no edge leading to it)
mode = (stat.S_IFDIR | 755)
mtime = self.tfs.root.mtime
s = TStat({}, st_mode=mode, st_nlink=1, st_mtime=mtime)
self.log('getattr(%r) -> %r' % (path, s))
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
#return s
return stat_to_dict(s)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
parent, name, child = self.tfs.get_parent_name_and_child(path)
if not child: # implicitly 'or not parent'
raise ENOENT('No such file or directory')
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
return stat_to_dict(parent.get_stat(name))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
@logexc
def access(self, path, mode):
self.log("access(%r, %r)" % (path, mode))
node = self.tfs.get_path(path)
if not node:
return -errno.ENOENT
accmode = os.O_RDONLY | os.O_WRONLY | os.O_RDWR
if (mode & 0222):
if not node.writable():
log('write access denied for %s (req:%o)' % (path, mode, ))
return -errno.EACCES
#else:
#log('access granted for %s' % (path, ))
@logexc
def mkdir(self, path, mode):
self.log("mkdir(%r, %r)" % (path, mode))
self.tfs.mkdir(path)
##################################################################
# file methods
def open(self, path, flags):
self.log('open(%r, %r)' % (path, flags, ))
if path in self.files:
# XXX todo [ ] should consider concurrent open files of differing modes
return
else:
tffobj = TahoeFuseFile(self.tfs, path, flags)
self.files[path] = tffobj
def create(self, path, flags, mode):
self.log('create(%r, %r, %r)' % (path, flags, mode))
if path in self.files:
# XXX todo [ ] should consider concurrent open files of differing modes
return
else:
tffobj = TahoeFuseFile(self.tfs, path, flags, mode)
self.files[path] = tffobj
def _get_file(self, path):
if not path in self.files:
raise ENOENT('No such file or directory: %s' % (path,))
return self.files[path]
##
def read(self, path, size, offset):
self.log('read(%r, %r, %r)' % (path, size, offset, ))
return self._get_file(path).read(size, offset)
@logexc
def write(self, path, buf, offset):
self.log("write(%r, -%s-, %r)" % (path, len(buf), offset))
return self._get_file(path).write(buf, offset)
@logexc
def release(self, path, flags):
self.log("release(%r, %r)" % (path, flags,))
self._get_file(path).release(flags)
del self.files[path]
@logexc
def fsync(self, path, isfsyncfile):
self.log("fsync(%r, %r)" % (path, isfsyncfile,))
return self._get_file(path).fsync(isfsyncfile)
@logexc
def flush(self, path):
self.log("flush(%r)" % (path,))
return self._get_file(path).flush()
@logexc
def fgetattr(self, path):
self.log("fgetattr(%r)" % (path,))
return self._get_file(path).fgetattr()
@logexc
def ftruncate(self, path, len):
self.log("ftruncate(%r, %r)" % (path, len,))
return self._get_file(path).ftruncate(len)
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
class TahoeFuseLocal(TahoeFuseBase, fuse.Fuse):
def __init__(self, tfs, *args, **kw):
log("TFL: __init__(%r, %r)" % (args, kw))
TahoeFuseBase.__init__(self, tfs)
fuse.Fuse.__init__(self, *args, **kw)
def log(self, msg):
log("<TFL> %s" % (msg, ))
def main(self, *a, **kw):
self.log("main(%r, %r)" % (a, kw))
return fuse.Fuse.main(self, *a, **kw)
# overrides for those methods which return objects not marshalled
def fgetattr(self, path):
return TStat({}, **(TahoeFuseBase.fgetattr(self, path)))
def getattr(self, path):
return TStat({}, **(TahoeFuseBase.getattr(self, path)))
def statfs(self):
return fuse.StatVfs(**(TahoeFuseBase.statfs(self)))
#self.log('statfs()')
#ret = fuse.StatVfs(**(TahoeFuseBase.statfs(self)))
#self.log('statfs(): %r' % (ret,))
#return ret
@logexc
def readdir(self, path, offset):
return [ fuse.Direntry(d) for d in TahoeFuseBase.readdir(self, path, offset) ]
class TahoeFuseShim(fuse.Fuse):
def __init__(self, trpc, *args, **kw):
log("TF: __init__(%r, %r)" % (args, kw))
self.trpc = trpc
fuse.Fuse.__init__(self, *args, **kw)
def log(self, msg):
log("<TFs> %s" % (msg, ))
@logexc
def readlink(self, path):
self.log("readlink(%r)" % (path,))
return self.trpc.call('readlink', path)
@logexc
def unlink(self, path):
self.log("unlink(%r)" % (path,))
return self.trpc.call('unlink', path)
@logexc
def rmdir(self, path):
self.log("rmdir(%r)" % (path,))
return self.trpc.call('unlink', path)
@logexc
def symlink(self, path, path1):
self.log("symlink(%r, %r)" % (path, path1))
return self.trpc.call('link', path, path1)
@logexc
def rename(self, path, path1):
self.log("rename(%r, %r)" % (path, path1))
return self.trpc.call('rename', path, path1)
@logexc
def link(self, path, path1):
self.log("link(%r, %r)" % (path, path1))
return self.trpc.call('link', path, path1)
@logexc
def chmod(self, path, mode):
self.log("XX chmod(%r, %r)" % (path, mode))
return self.trpc.call('chmod', path, mode)
@logexc
def chown(self, path, user, group):
self.log("XX chown(%r, %r, %r)" % (path, user, group))
return self.trpc.call('chown', path, user, group)
@logexc
def truncate(self, path, len):
self.log("XX truncate(%r, %r)" % (path, len))
return self.trpc.call('truncate', path, len)
@logexc
def utime(self, path, times):
self.log("XX utime(%r, %r)" % (path, times))
return self.trpc.call('utime', path, times)
@logexc
def statfs(self):
self.log("statfs()")
response = self.trpc.call('statfs')
#self.log("statfs(): %r" % (response,))
kwargs = dict([ (str(k),v) for k,v in response.items() ])
return fuse.StatVfs(**kwargs)
def fsinit(self):
self.log("fsinit()")
def main(self, *a, **kw):
self.log("main(%r, %r)" % (a, kw))
return fuse.Fuse.main(self, *a, **kw)
##################################################################
@logexc
def readdir(self, path, offset):
self.log('readdir(%r, %r)' % (path, offset))
return [ fuse.Direntry(d) for d in self.trpc.call('readdir', path, offset) ]
@logexc
def getattr(self, path):
self.log('getattr(%r)' % (path,))
response = self.trpc.call('getattr', path)
kwargs = dict([ (str(k),v) for k,v in response.items() ])
s = TStat({}, **kwargs)
self.log('getattr(%r) -> %r' % (path, s))
return s
@logexc
def access(self, path, mode):
self.log("access(%r, %r)" % (path, mode))
return self.trpc.call('access', path, mode)
@logexc
def mkdir(self, path, mode):
self.log("mkdir(%r, %r)" % (path, mode))
return self.trpc.call('mkdir', path, mode)
##################################################################
# file methods
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
def open(self, path, flags):
self.log('open(%r, %r)' % (path, flags, ))
return self.trpc.call('open', path, flags)
def create(self, path, flags, mode):
self.log('create(%r, %r, %r)' % (path, flags, mode))
return self.trpc.call('create', path, flags, mode)
##
def read(self, path, size, offset):
self.log('read(%r, %r, %r)' % (path, size, offset, ))
return self.trpc.call('read', path, size, offset)
@logexc
def write(self, path, buf, offset):
self.log("write(%r, -%s-, %r)" % (path, len(buf), offset))
return self.trpc.call('write', path, buf, offset)
@logexc
def release(self, path, flags):
self.log("release(%r, %r)" % (path, flags,))
return self.trpc.call('release', path, flags)
@logexc
def fsync(self, path, isfsyncfile):
self.log("fsync(%r, %r)" % (path, isfsyncfile,))
return self.trpc.call('fsync', path, isfsyncfile)
@logexc
def flush(self, path):
self.log("flush(%r)" % (path,))
return self.trpc.call('flush', path)
@logexc
def fgetattr(self, path):
self.log("fgetattr(%r)" % (path,))
#return self.trpc.call('fgetattr', path)
response = self.trpc.call('fgetattr', path)
kwargs = dict([ (str(k),v) for k,v in response.items() ])
s = TStat({}, **kwargs)
self.log('getattr(%r) -> %r' % (path, s))
return s
@logexc
def ftruncate(self, path, len):
self.log("ftruncate(%r, %r)" % (path, len,))
return self.trpc.call('ftruncate', path, len)
def launch_tahoe_fuse(tf_class, tobj, argv):
sys.argv = ['tahoe fuse'] + list(argv)
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
log('setting sys.argv=%r' % (sys.argv,))
config = TahoeFuseOptions()
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
version = "%prog " +VERSIONSTR+", fuse "+ fuse.__version__
server = tf_class(tobj, version=version, usage=config.getSynopsis(), dash_s_do='setsingle')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
server.parse(errex=1)
server.main()
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
def getnodeurl(nodedir):
f = file(os.path.expanduser(os.path.join(nodedir, "node.url")), 'rb')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
nu = f.read().strip()
f.close()
if nu[-1] != "/":
nu += "/"
return nu
def fingerprint(uri):
if uri is None:
return None
return base64.b32encode(sha.new(uri).digest()).lower()[:6]
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
stat_fields = [ 'st_mode', 'st_ino', 'st_dev', 'st_nlink', 'st_uid', 'st_gid', 'st_size',
'st_atime', 'st_mtime', 'st_ctime', ]
def stat_to_dict(statobj, fields=None):
if fields is None:
fields = stat_fields
d = {}
for f in fields:
d[f] = getattr(statobj, f, None)
return d
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
class TStat(fuse.Stat):
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
# in fuse 0.2, these are set by fuse.Stat.__init__
# in fuse 0.2-pre3 (hardy) they are not. badness unsues if they're missing
st_mode = None
st_ino = 0
st_dev = 0
st_nlink = None
st_uid = 0
st_gid = 0
st_size = 0
st_atime = 0
st_mtime = 0
st_ctime = 0
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
fields = [ 'st_mode', 'st_ino', 'st_dev', 'st_nlink', 'st_uid', 'st_gid', 'st_size',
'st_atime', 'st_mtime', 'st_ctime', ]
def __init__(self, metadata, **kwargs):
# first load any stat fields present in 'metadata'
for st in [ 'mtime', 'ctime' ]:
if st in metadata:
setattr(self, "st_%s" % st, metadata[st])
for st in self.fields:
if st in metadata:
setattr(self, st, metadata[st])
# then set any values passed in as kwargs
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
fuse.Stat.__init__(self, **kwargs)
def __repr__(self):
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
return "<Stat%r>" % (stat_to_dict(self),)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
class Directory(object):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def __init__(self, tfs, ro_uri, rw_uri):
self.tfs = tfs
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.ro_uri = ro_uri
self.rw_uri = rw_uri
assert (rw_uri or ro_uri)
self.children = {}
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.last_load = None
self.last_data = None
self.mtime = 0
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def __repr__(self):
return "<Directory %s>" % (fingerprint(self.get_uri()),)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def maybe_refresh(self, name=None):
"""
if the previously cached data was retrieved within the cache
validity period, does nothing. otherwise refetches the data
for this directory and reloads itself
"""
now = time.time()
if self.last_load is None or (now - self.last_load) > self.tfs.cache_validity:
self.load(name)
def load(self, name=None):
now = time.time()
log('%s.loading(%s)' % (self, name))
url = self.tfs.compose_url("uri/%s?t=json", self.get_uri())
data = urllib.urlopen(url).read()
h = tagged_hash('cache_hash', data)
if h == self.last_data:
self.last_load = now
log('%s.load() : no change h(data)=%s' % (self, base32.b2a(h), ))
return
try:
parsed = simplejson.loads(data)
except ValueError:
log('%s.load(): unable to parse json data for dir:\n%r' % (self, data))
return
nodetype, d = parsed
assert nodetype == 'dirnode'
self.children.clear()
for cname,details in d['children'].items():
cname = unicode_to_utf8(cname)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
ctype, cattrs = details
metadata = cattrs.get('metadata', {})
if ctype == 'dirnode':
cobj = self.tfs.dir_for(cname, cattrs.get('ro_uri'), cattrs.get('rw_uri'))
else:
assert ctype == "filenode"
cobj = File(cattrs.get('size'), cattrs.get('ro_uri'))
self.children[cname] = cobj, metadata
self.last_load = now
self.last_data = h
self.mtime = now
log('%s.load() loaded: \n%s' % (self, self.pprint(),))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def get_children(self):
return self.children.keys()
def get_child(self, name):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
return self.children[name][0]
def add_child(self, name, child, metadata):
log('%s.add_child(%r, %r, %r)' % (self, name, child, metadata, ))
self.children[name] = child, metadata
url = self.tfs.compose_url("uri/%s/%s?t=uri", self.get_uri(), name)
child_cap = do_http('PUT', url, child.get_uri())
# XXX [ ] TODO: push metadata to tahoe node
assert child_cap == child.get_uri()
self.mtime = time.time()
log('added child %r with %r to %r' % (name, child_cap, self))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def remove_child(self, name):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
log('%s.remove_child(%r)' % (self, name, ))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
del self.children[name]
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
url = self.tfs.compose_url("uri/%s/%s", self.get_uri(), name)
resp = do_http('DELETE', url)
self.mtime = time.time()
log('child (%s) removal yielded %r' % (name, resp,))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def get_uri(self):
return self.rw_uri or self.ro_uri
def writable(self):
return self.rw_uri and self.rw_uri != self.ro_uri
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def pprint(self, prefix='', printed=None, suffix=''):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
ret = []
if printed is None:
printed = set()
writable = self.writable() and '+' or ' '
if self in printed:
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
ret.append(" %s/%s ... <%s> : %s" % (prefix, writable, fingerprint(self.get_uri()), suffix, ))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
else:
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
ret.append("[%s] %s/%s : %s" % (fingerprint(self.get_uri()), prefix, writable, suffix, ))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
printed.add(self)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
for name,(child,metadata) in sorted(self.children.items()):
ret.append(child.pprint(' ' * (len(prefix)+1)+name, printed, repr(metadata)))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
return '\n'.join(ret)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def get_metadata(self, name):
return self.children[name][1]
def get_stat(self, name):
child,metadata = self.children[name]
log("%s.get_stat(%s) md: %r" % (self, name, metadata))
if isinstance(child, Directory):
child.maybe_refresh(name)
mode = metadata.get('st_mode') or (stat.S_IFDIR | 0755)
s = TStat(metadata, st_mode=mode, st_nlink=1, st_mtime=child.mtime)
else:
if hasattr(child, 'tmp_fname'):
s = os.stat(child.tmp_fname)
log("%s.get_stat(%s) returning local stat of tmp file" % (self, name, ))
else:
s = TStat(metadata,
st_nlink = 1,
st_size = child.size,
st_mode = metadata.get('st_mode') or (stat.S_IFREG | 0444),
st_mtime = metadata.get('mtime') or self.mtime,
)
return s
log("%s.get_stat(%s)->%s" % (self, name, s))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
return s
class File(object):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def __init__(self, size, ro_uri):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.size = size
if ro_uri:
ro_uri = str(ro_uri)
self.ro_uri = ro_uri
def __repr__(self):
return "<File %s>" % (fingerprint(self.ro_uri) or [self.tmp_fname],)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def pprint(self, prefix='', printed=None, suffix=''):
return " %s (%s) : %s" % (prefix, self.size, suffix, )
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def get_uri(self):
return self.ro_uri
def writable(self):
return True
class TFS(object):
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
def __init__(self, nodedir, nodeurl, root_uri,
cache_validity_period=DEFAULT_DIRECTORY_VALIDITY):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.cache_validity = cache_validity_period
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.nodeurl = nodeurl
self.root_uri = root_uri
self.dirs = {}
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
cachedir = os.path.expanduser(os.path.join(nodedir, '_cache'))
self.cache = FileCache(nodeurl, cachedir)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
ro_uri = NewDirectoryURI.init_from_string(self.root_uri).get_readonly()
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.root = Directory(self, ro_uri, self.root_uri)
self.root.maybe_refresh('<root>')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def log(self, msg):
log("<TFS> %s" % (msg, ))
def pprint(self):
return self.root.pprint()
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def compose_url(self, fmt, *args):
return self.nodeurl + (fmt % tuple(map(urllib.quote, args)))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def get_parent_name_and_child(self, path):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
"""
find the parent dir node, name of child relative to that parent, and
child node within the TFS object space.
@returns: (parent, name, child) if the child is found
(parent, name, None) if the child is missing from the parent
(None, name, None) if the parent is not found
"""
if path == '/':
return
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
dirname, name = os.path.split(path)
parent = self.get_path(dirname)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
if parent:
try:
child = parent.get_child(name)
return parent, name, child
except KeyError:
return parent, name, None
else:
return None, name, None
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def get_path(self, path):
comps = path.strip('/').split('/')
if comps == ['']:
comps = []
cursor = self.root
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
c_name = '<root>'
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
for comp in comps:
if not isinstance(cursor, Directory):
self.log('path "%s" is not a dir' % (path,))
return None
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
cursor.maybe_refresh(c_name)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
try:
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
cursor = cursor.get_child(comp)
c_name = comp
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
except KeyError:
self.log('path "%s" not found' % (path,))
return None
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
if isinstance(cursor, Directory):
cursor.maybe_refresh(c_name)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
return cursor
def dir_for(self, name, ro_uri, rw_uri):
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
#self.log('dir_for(%s) [%s/%s]' % (name, fingerprint(ro_uri), fingerprint(rw_uri)))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
if ro_uri:
ro_uri = str(ro_uri)
if rw_uri:
rw_uri = str(rw_uri)
uri = rw_uri or ro_uri
assert uri
dirobj = self.dirs.get(uri)
if not dirobj:
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
self.log('dir_for(%s) creating new Directory' % (name, ))
dirobj = Directory(self, ro_uri, rw_uri)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
self.dirs[uri] = dirobj
return dirobj
def upload(self, fname):
self.log('upload(%r)' % (fname,))
fh = file(fname, 'rb')
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
url = self.compose_url("uri")
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
file_cap = do_http('PUT', url, fh)
self.log('uploaded to: %r' % (file_cap,))
return file_cap
def mkdir(self, path):
self.log('mkdir(%r)' % (path,))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
parent, name, child = self.get_parent_name_and_child(path)
if child:
raise EEXIST('File exists: %s' % (name,))
if not parent:
raise ENOENT('No such file or directory: %s' % (path,))
url = self.compose_url("uri?t=mkdir")
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
new_dir_cap = do_http('PUT', url)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
ro_uri = NewDirectoryURI.init_from_string(new_dir_cap).get_readonly()
child = Directory(self, ro_uri, new_dir_cap)
parent.add_child(name, child, {})
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def rename(self, path, path1):
self.log('rename(%s, %s)' % (path, path1))
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
src_parent, src_name, src_child = self.get_parent_name_and_child(path)
dst_parent, dst_name, dst_child = self.get_parent_name_and_child(path1)
if not src_child or not dst_parent:
raise ENOENT('No such file or directory')
dst_parent.add_child(dst_name, src_child, {})
src_parent.remove_child(src_name)
def unlink(self, path):
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
parent, name, child = self.get_parent_name_and_child(path)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
if child is None: # parent or child is missing
raise ENOENT('No such file or directory')
if not parent.writable():
raise EACCESS('Permission denied')
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
parent.remove_child(name)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
def link(self, path, path1):
src = self.get_path(path)
dst_parent, dst_name, dst_child = self.get_parent_name_and_child(path1)
if not src:
raise ENOENT('No such file or directory')
if dst_parent is None:
raise ENOENT('No such file or directory')
if not dst_parent.writable():
raise EACCESS('Permission denied')
dst_parent.add_child(dst_name, src, {})
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
class FileCache(object):
def __init__(self, nodeurl, cachedir):
self.nodeurl = nodeurl
self.cachedir = cachedir
if not os.path.exists(self.cachedir):
os.makedirs(self.cachedir)
self.tmpdir = os.path.join(self.cachedir, 'tmp')
if not os.path.exists(self.tmpdir):
os.makedirs(self.tmpdir)
def log(self, msg):
log("<FC> %s" % (msg, ))
def get_file(self, uri):
self.log('get_file(%s)' % (uri,))
if uri.startswith("URI:LIT"):
return self.get_literal(uri)
else:
return self.get_chk(uri)
def get_literal(self, uri):
h = sha.new(uri).digest()
u = LiteralFileURI.init_from_string(uri)
fname = os.path.join(self.cachedir, '__'+base64.b32encode(h).lower())
size = len(u.data)
self.log('writing literal file %s (%s)' % (fname, size, ))
fh = open(fname, 'wb')
fh.write(u.data)
fh.close()
return fname
def get_chk(self, uri):
u = CHKFileURI.init_from_string(str(uri))
storage_index = u.storage_index
size = u.size
fname = os.path.join(self.cachedir, base64.b32encode(storage_index).lower())
if os.path.exists(fname):
fsize = os.path.getsize(fname)
if fsize == size:
return fname
else:
self.log('warning file "%s" is too short %s < %s' % (fname, fsize, size))
self.log('downloading file %s (%s)' % (fname, size, ))
fh = open(fname, 'wb')
url = "%suri/%s" % (self.nodeurl, uri)
download = urllib.urlopen(''.join([ self.nodeurl, "uri/", uri ]))
while True:
chunk = download.read(4096)
if not chunk:
break
fh.write(chunk)
fh.close()
return fname
def tmp_file(self, id):
fname = os.path.join(self.tmpdir, base64.b32encode(id).lower())
return fname
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
_tfs = None # to appease pyflakes; is set in main()
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
def print_tree():
log('tree:\n' + _tfs.pprint())
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
def unmarshal(obj):
if obj is None or isinstance(obj, int) or isinstance(obj, long) or isinstance(obj, float):
return obj
elif isinstance(obj, unicode):
#log('unmarshal(%r)' % (obj,))
return base64.b64decode(obj)
elif isinstance(obj, list):
return map(unmarshal, obj)
elif isinstance(obj, dict):
return dict([ (k,unmarshal(v)) for k,v in obj.items() ])
else:
raise ValueError('object type not int,str,list,dict,none (%s) (%r)' % (type(obj), obj))
def marshal(obj):
if obj is None or isinstance(obj, int) or isinstance(obj, long) or isinstance(obj, float):
return obj
elif isinstance(obj, str):
return base64.b64encode(obj)
elif isinstance(obj, list) or isinstance(obj, tuple):
return map(marshal, obj)
elif isinstance(obj, dict):
return dict([ (k,marshal(v)) for k,v in obj.items() ])
else:
raise ValueError('object type not int,str,list,dict,none (%s)' % type(obj))
class TRPCProtocol(Protocol):
compute_response_sha1 = True
log_all_requests = False
def connectionMade(self):
self.buf = []
def dataReceived(self, data):
if data == 'keepalive\n':
log('keepalive connection on %r' % (self.transport,))
self.keepalive = True
return
if not data.endswith('\n'):
self.buf.append(data)
return
if self.buf:
self.buf.append(data)
reqstr = ''.join(self.buf)
self.buf = []
self.dispatch_request(reqstr)
else:
self.dispatch_request(data)
def dispatch_request(self, reqstr):
try:
req = simplejson.loads(reqstr)
except ValueError, ve:
log(ve)
return
d = defer.maybeDeferred(self.handle_request, req)
d.addCallback(self.send_response)
d.addErrback(self.send_error)
def send_error(self, failure):
log('failure: %s' % (failure,))
if failure.check(TFSIOError):
e = failure.value
self.send_response(['error', 'errno', e.args[0], e.args[1]])
else:
self.send_response(['error', 'failure', str(failure)])
def send_response(self, result):
response = simplejson.dumps(result)
header = { 'len': len(response), }
if self.compute_response_sha1:
header['sha1'] = base64.b64encode(sha.new(response).digest())
hdr = simplejson.dumps(header)
self.transport.write(hdr)
self.transport.write('\n')
self.transport.write(response)
self.transport.loseConnection()
def connectionLost(self, reason):
if hasattr(self, 'keepalive'):
log('keepalive connection %r lost, shutting down' % (self.transport,))
reactor.callLater(0, reactor.stop)
def handle_request(self, req):
if type(req) is not list or not req or len(req) < 1:
return ['error', 'malformed request']
if req[0] == 'call':
if len(req) < 3:
return ['error', 'malformed request']
methname = req[1]
try:
args = unmarshal(req[2])
except ValueError, ve:
return ['error', 'malformed arguments', str(ve)]
try:
meth = getattr(self.factory.server, methname)
except AttributeError, ae:
return ['error', 'no such method', str(ae)]
if self.log_all_requests:
log('call %s(%s)' % (methname, ', '.join(map(repr, args))))
try:
result = meth(*args)
except TFSIOError, e:
log('errno: %s; %s' % e.args)
return ['error', 'errno', e.args[0], e.args[1]]
except Exception, e:
log('exception: ' + traceback.format_exc())
return ['error', 'exception', str(e)]
d = defer.succeed(None)
d.addCallback(lambda junk: result) # result may be Deferred
d.addCallback(lambda res: ['result', marshal(res)]) # only applies if not errback
return d
class TFSServer(object):
def __init__(self, socket_path, server=None):
self.socket_path = socket_path
log('TFSServer init socket: %s' % (socket_path,))
self.factory = Factory()
self.factory.protocol = TRPCProtocol
if server:
self.factory.server = server
else:
self.factory.server = self
def get_service(self):
if not hasattr(self, 'svc'):
from twisted.application import strports
self.svc = strports.service('unix:'+self.socket_path, self.factory)
return self.svc
def run(self):
svc = self.get_service()
def ss():
try:
svc.startService()
except:
reactor.callLater(0, reactor.stop)
raise
reactor.callLater(0, ss)
reactor.run()
def hello(self):
return 'pleased to meet you'
def echo(self, arg):
return arg
def failex(self):
raise ValueError('expected')
def fail(self):
return defer.maybeDeferred(self.failex)
class RPCError(RuntimeError):
pass
class TRPC(object):
def __init__(self, socket_fname):
self.socket_fname = socket_fname
self.keepalive = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
self.keepalive.connect(self.socket_fname)
self.keepalive.send('keepalive\n')
log('requested keepalive on %s' % (self.keepalive,))
def req(self, req):
# open conenction to trpc server
s = socket.socket(socket.AF_UNIX, socket.SOCK_STREAM)
s.connect(self.socket_fname)
# send request
s.send(simplejson.dumps(req))
s.send('\n')
# read response header
hdr_data = s.recv(8192)
first_newline = hdr_data.index('\n')
header = hdr_data[:first_newline]
data = hdr_data[first_newline+1:]
hdr = simplejson.loads(header)
hdr_len = hdr['len']
if hdr.has_key('sha1'):
hdr_sha1 = base64.b64decode(hdr['sha1'])
spool = [data]
spool_sha = sha.new(data)
# spool response
while True:
data = s.recv(8192)
if data:
spool.append(data)
spool_sha.update(data)
else:
break
else:
spool = [data]
# spool response
while True:
data = s.recv(8192)
if data:
spool.append(data)
else:
break
s.close()
# decode response
resp = ''.join(spool)
spool = None
assert hdr_len == len(resp), str((hdr_len, len(resp), repr(resp)))
if hdr.has_key('sha1'):
data_sha1 = spool_sha.digest()
spool = spool_sha = None
assert hdr_sha1 == data_sha1, str((base32.b2a(hdr_sha1), base32.b2a(data_sha1)))
#else:
#print 'warning, server provided no sha1 to check'
return resp
def call(self, methodname, *args):
res = self.req(['call', methodname, marshal(args)])
result = simplejson.loads(res)
if not result or len(result) < 2:
raise TypeError('malformed response %r' % (result,))
if result[0] == 'error':
if result[1] == 'errno':
raise TFSIOError(result[2], result[3])
else:
raise RPCError(*(result[1:])) # error, exception / error, failure
elif result[0] == 'result':
return unmarshal(result[1])
else:
raise TypeError('unknown response type %r' % (result[0],))
def shutdown(self):
log('shutdown() closing keepalive %s' % (self.keepalive,))
self.keepalive.close()
# (cut-n-pasted here due to an ImportError / some py2app linkage issues)
#from twisted.scripts._twistd_unix import daemonize
def daemonize():
# See http://www.erlenstar.demon.co.uk/unix/faq_toc.html#TOC16
if os.fork(): # launch child and...
os._exit(0) # kill off parent
os.setsid()
if os.fork(): # launch child and...
os._exit(0) # kill off parent again.
os.umask(077)
null=os.open('/dev/null', os.O_RDWR)
for i in range(3):
try:
os.dup2(null, i)
except OSError, e:
if e.errno != errno.EBADF:
raise
os.close(null)
def main(argv):
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
log("main(%s)" % (argv,))
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
# check for version or help options (no args == help)
if not argv:
argv = ['--help']
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
if len(argv) == 1 and argv[0] in ['-h', '--help']:
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
config = TahoeFuseOptions()
print >> sys.stderr, config
print >> sys.stderr, 'fuse usage follows:'
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
if len(argv) == 1 and argv[0] in ['-h', '--help', '--version']:
launch_tahoe_fuse(TahoeFuseLocal, None, argv)
return -2
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
# parse command line options
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
config = TahoeFuseOptions()
try:
#print 'parsing', argv
config.parseOptions(argv)
except usage.error, e:
print config
print e
return -1
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
# check for which alias or uri is specified
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
if config['alias']:
alias = config['alias']
#print 'looking for aliases in', config['node-directory']
aliases = get_aliases(os.path.expanduser(config['node-directory']))
if alias not in aliases:
raise usage.error('Alias %r not found' % (alias,))
root_uri = aliases[alias]
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
root_name = alias
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
elif config['root-uri']:
root_uri = config['root-uri']
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
root_name = 'uri_' + base32.b2a(tagged_hash('root_name', root_uri))[:12]
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
# test the uri for structural validity:
try:
NewDirectoryURI.init_from_string(root_uri)
except:
raise usage.error('root-uri must be a valid directory uri (not %r)' % (root_uri,))
else:
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
raise usage.error('At least one of --alias or --root-uri must be specified')
nodedir = config['node-directory']
nodeurl = config['node-url']
if not nodeurl:
nodeurl = getnodeurl(nodedir)
macfuse: slew of updates various updates to improve the functionality of the mac fuse plugin 1. caching previously, the experimental tahoefuse plugin pre-loaded the whole structure of the specified mount into memory at launch time. changes which were made through that fuse plugin would be remembered, but any changes made through other tahoe clients would not be reflected. now directory contents are only loaded when needed, and the data is cached for a limited time. any use of Directory objects should first call maybe_refresh() which will check the time since the cache was last loaded, and if the data is older than some validity period (currently 26s) then the directory's contents will be refetched and reloaded. this replaces the 'load_dir()' method of TFS whenever a local change is made to a Directory object, or when the aforementioned cache reloading notices a change in directory data, the mtime of the directory is automatically updated. 2. stat / metadata the retrieval of 'stat' information for getattr(), and the way that metadata is handled, has been refactored to better reflect the fact that metadata in tahoe is only represented by 'edges' (i.e entries in directories) not on 'nodes' (files or directories themselves) hence a stat lookup should be a query to the parent directory (specifically the parent specified by the path being queried in the case that a node has multiple parents) for details known by that directory for the given child, rather than a query to the child itself. the TStat utility class for returning stat information to the python- fuse layer has been extended to accept a 'metadata' argument in its constructor. any fields found in the metadata dict which match the names of the stat attributes are loaded into the TStat object. the 'ctime' and 'mtime' attributes are translated to st_ctime and st_mtime to mesh with the existing timestamp handling code. any fields specified by kwargs to the constructor override things that might be loaded from the metadata dict. Directory objects now track their children as a dict mapping name to (child_obj, metadata) tuples. This is because the metadata in tahoe will be stored exclusively on the edges of the graph. each Directory maintains its own mtime however, and get_stat() calls will report the mtime of a directory based on the last modification of the Directory object, not based on any mtime records from the parent directory's metadata for that child. This addresses the fact that since directories may be shared, a given parent may or may not reflect the latest changes, however one of the Finder's behaviours is to examine the stat of a directory, and not to bother doing a readdir() if the stat is unchanged. i.e. unless directories report their changes in their stat info, the Finder will not show changes within that directory. 3. refactoring reporting of many error codes has been refactored to raise IOError subclasses with the appropriate errno. this exploits python-fuse's built-in mechanism for catching IOError and reporting the errno embedded within it automatically, while simplifying the code within the plugin. the add_child() method on TFS was removed in favour of simply having an add_child() method on Directory objects. this provides a more OO approach in that Directory is responsible for maintaining its own in memory state and also writing changes back to the node. similarly for remove_child() these changes, along with the new tfs.compose_url() method, significantly simplify and improve readability of mkdir, rename methods along with the newer link and unlink. these also get improved error reporting. various operations (chmod, chown, truncate, utime) are now ignored. previously they would report an unsupported operation (EOPNOTSUPP) but now are simply logged and ignored. this surpresses errors caused by some client programs which try to use these operations, but at the moment those operations are meaningless to the tahoe filesystem anyway. 4. link / unlink / rmdir link, symlink calls are now supported, though with semantics differing from posix, both equivalent. unlink, rmdir calls are now supported, also equivalent. link or symlink calls duplicate the uri of the named source and adds it as a child of another directory according to the destination path. for directories, this creates a 'hard' link, i.e. the same directory will appear in multiple locations within the filesystem, and changes in any place will be reflected everywhere. for files, by contrast, since the uri being duplicated is an immutable CHK uri, link/symlink for files is equivalent to a copy - though significantly cheaper. (a file copy with the fuse plugin is likely to cause a new file to be written and uploaded, the link command simply adds an entry referring to an existing uri) in testing, the 'ln' command is unable to make hard links (i.e. call link()) for directories, though symlink ('ln -s') is supported. either forms works equivalently for files. unlink and rmdir both remove the specified entry from its parent directory. 5. logging the 'tfuse.log' file now only reports launches of the fuse plugin. once the plugin has parsed the options, it reopens the log file with the name of the mount, e.g. tfuse.root_dir.log, so that multiple instances running concurrently will not interfere with each others' logging. 6. bug fixes the tmp_file in the cache dir backing files opened for write was intermittently failing to open the file. added O_CREAT to the os.open call so that files will be created if missing, not throw errors. a failure to correctly parse arguments if no mount (dir_cap) name was given but also no fuse options were given has been fixed. now the command 'tahoe fuse mountpoint' will correctly default to root_dir also when running from source, arguments to tahoefuse were not handled to correctly match the 'tahoe fuse ...' behaviour.
2008-03-01 03:12:41 +00:00
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
# allocate socket
socket_dir = os.path.join(os.path.expanduser(nodedir), "tfuse.sockets")
socket_path = os.path.join(socket_dir, root_name)
if len(socket_path) > 103:
# try googling AF_UNIX and sun_len for some taste of why this oddity exists.
raise OSError(errno.ENAMETOOLONG, 'socket path too long (%s)' % (socket_path,))
fileutil.make_dirs(socket_dir, 0700)
if os.path.exists(socket_path):
log('socket exists')
if config['server-shutdown']:
log('calling shutdown')
trpc = TRPC(socket_path)
result = trpc.shutdown()
log('result: %r' % (result,))
log('called shutdown')
return
else:
raise OSError(errno.EEXIST, 'fuse already running (%r exists)' % (socket_path,))
elif config['server-shutdown']:
raise OSError(errno.ENOTCONN, '--server-shutdown specified, but server not running')
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
if not os.path.exists(config.mountpoint):
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
raise OSError(errno.ENOENT, 'No such file or directory: "%s"' % (config.mountpoint,))
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
global _tfs
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
#
# Standalone ("no-split")
#
if config['no-split']:
reopen_logfile('tfuse.%s.unsplit.log' % (root_name,))
log('\n'+(24*'_')+'init (unsplit)'+(24*'_')+'\n')
cache_timeout = float(config['cache-timeout'])
tfs = TFS(nodedir, nodeurl, root_uri, cache_timeout)
#print tfs.pprint()
# make tfs instance accesible to print_tree() for dbg
_tfs = tfs
args = [ '-o'+opt for opt in config.fuse_options ] + [config.mountpoint]
launch_tahoe_fuse(TahoeFuseLocal, tfs, args)
#
# Server
#
elif config['server']:
reopen_logfile('tfuse.%s.server.log' % (root_name,))
log('\n'+(24*'_')+'init (server)'+(24*'_')+'\n')
log('daemonizing')
daemonize()
cache_timeout = float(config['cache-timeout'])
tfs = TFS(nodedir, nodeurl, root_uri, cache_timeout)
#print tfs.pprint()
# make tfs instance accesible to print_tree() for dbg
_tfs = tfs
log('launching tfs server')
tfuse = TahoeFuseBase(tfs)
tfs_server = TFSServer(socket_path, tfuse)
tfs_server.run()
log('tfs server ran, exiting')
#
# Client
#
else:
reopen_logfile('tfuse.%s.client.log' % (root_name,))
log('\n'+(24*'_')+'init (client)'+(24*'_')+'\n')
server_args = [sys.executable, sys.argv[0], '--server'] + argv
if 'Allmydata.app/Contents/MacOS' in sys.executable:
# in this case blackmatch is the 'fuse' subcommand of the 'tahoe' executable
# otherwise we assume blackmatch is being run from source
server_args.insert(2, 'fuse')
#print 'launching server:', server_args
server = subprocess.Popen(server_args)
waiting_since = time.time()
wait_at_most = 8
while not os.path.exists(socket_path):
log('waiting for appearance of %r' % (socket_path,))
time.sleep(1)
if time.time() - waiting_since > wait_at_most:
log('%r did not appear within %ss' % (socket_path, wait_at_most))
raise IOError(2, 'no socket %s' % (socket_path,))
#print 'launched server'
trpc = TRPC(socket_path)
args = [ '-o'+opt for opt in config.fuse_options ] + [config.mountpoint]
launch_tahoe_fuse(TahoeFuseShim, trpc, args)
macfuse: another tahoe fuse implementation This is the result of various experimentation done into using python-fuse to provide access to tahoe on the mac. It's rough in quite a few places, and is really the result of investigation more than a thorough implemenation of the fuse api. upon launch, it looks for the users root_dir by opening ~/.tahoe/node.url and ~/.tahoe/private/root_dir.cap it then proceeds to cache the directory structure found by walking the users tahoe drive (safely in the face of directory loops) into memory and then mounts that filesystem. when a file is read, it calls the tahoe node to first download the file into a cache directory (~/.tahoe/_cache) and then serves up the file from there. when a file is written, a temporary file is allocated within the tmp dir of the cache, and upon close() (specifically upon release()) the file is uploaded to the tahoe node, and the new directory entry written. note that while the durectory structure is cached into memory only when the filesystem is mounted, that it is 'write through' i.e. changes made via fuse are reflected into the underlying tahoe fs, even though changes made to the tahoe fs otherwise show up only upon restart. in addition to opening files for read and write, the mkdir() and rename() calls are supported. most other file system operations are not yet supported. notably stat() metadata is not currently tracked by tahoe, and is variably reported by this fs depending on write cache files. also note that this version does not fully support Finder. access through normal unix commands such as cat, cp, mv, ls etc works fine, and read access to file from within finder (including preview images and double- click to open) work ok. but copies to the tahoe drive from within finder may or may not succeed, but will always report an error. This is still under investigation. also note that this does not include any build integration. the included _fusemodule.so was built on mac os 10.4 against macfuse 1.3.0, and is known to not work against 10.5-1.3.1 it's possible it may also contain dependencies upon parts of macports used to build the python that it was built against. this will be cleaned up later. usage: python tahoefuse.py /Path/to/choice/of/mountpoint or optionally python tahoefuse.py -ovolicon=/Path/to/icon.icns /Path/to/mountpoint upon startup, tahoefuse will walk the tahoe directory, then print a summary of files and folders found, and then daemonise itself. to exit, either eject the 'drive' (note: 10.5 doesn't show it as a drive, since it considers fuse to be a connected server instead) or unmount it via umount /Path/to/mountpoint etc.
2008-02-15 01:35:10 +00:00
fuse/blackmatch: split into client/server (twisted server) This implements a client/server split for blackmatch, where the client implements the fuse_main bindings and a simple blocking rpc client mechanism. The server implements the other half of that rpc mechanism, and contains all the actual logic for interpreting fuse requests in the context of the on disk cache and requests to the tahoe node. The server is based on a twisted reactor. The rpc mechanism implements a simple method dispatch including marshalling, using json, of basic inert data types, in a flat namespace (no objects). The client side is written in a blocking idiom, to interface with the threading model used by the fuse_main bindings, whereas the server side is written for a twisted reactor-based environment, intended to facilitate implementing more sophisticated logic in that paradigm. The two communicate over a unix domain socket, allocated within the nodedir. Command line usage is unchanged; the server is launched automatically by the client. The server daemonizes itself, to avoid preventing the original parent process (e.g. 'runtests') from waiting upon the server exiting. The client keeps open a 'keepalive' connection to the server; upon loss thereof the server will exit. This addresses the fact that the python-fuse bindings provide no notification of exit of the client process upon unmount. The client thus provides a relatively thin 'shim' proxying requests from the fuse_main bindings across the rpc to the server process, which handles the logic behind each request. For the time being, a '--no-split' option is provided to surpress the splitting into client/server, yielding the prior behaviour. Once the server logic gets more complex and more entrenched in a twisted idiom, this might be removed. The 'runtests' test harness currently tests both modes, as 'impl_c' and 'impl_c_no_split'
2008-10-16 15:08:46 +00:00
if __name__ == '__main__':
fuse/impl_c: reworking of mac/tahoefuse, command line options, test integration a handful of changes to the tahoefuse implementation used by the mac build, to make command line option parsing more flexible and robust, and moreover to facilitate integration of this implementation with the 'runtests' test harness used to test the other two implementations. this patch includes; - improvements to command line option parsing [ see below ] - support for 'aliases' akin to other tahoe tools - tweaks to support linux (ubuntu hardy) the linux support tweaks are, or at least seem to be, a result of the fact that hardy ships with fuse 0.2pre3, as opposed to the fuse0.2 that macfuse is based upon. at least the versions I was working with have discrepencies in their interfaces, but on reflection this is probably a 'python-fuse' version issue rather than fuse per se. At any rate, the fixes to handling the Stat objects should be safe against either version, it's just that the bindings on hardy lacked code that was in the 'fuse' python module on the mac... command line options: the need for more flexible invocation in support of the runtests harness led me to rework the argument parsing from some simple positional hacks with a pass-through of the remainder to the fuse binding's 'fuse_main' to a system using twisted.usage to parse arguments, and having just one option '-o' being explicitly a pass-through for -o options to fuse_main. the options are now: --node-directory NODEDIR : this is used to look up the node-url to connect to if that's not specified concretely on the command line, and also used to determine the location of the cache directory used by the implementation, specifically '_cache' within the nodedir. default value: ~/.tahoe --node-url NODEURL : specify a node-url taking precendence over that found in the node.url file within the nodedir --alias ALIAS : specifies the named alias should be mounted. a lookup is performed in the alias table within 'nodedir' to find the root dir cap the named alias must exist in the alias table of the specified nodedir --root-uri ROOTURI : specifies that the given directory uri should be mounted at least one of --alias and --root-uri must be given (which directory to mount must be specified somehow) if both are given --alias takes precedence. --cache-timeout TIMEOUTSECS : specifies the number of seconds that cached directory data should be considered valid for. this tahoefuse implementation implements directory caching for a limited time; largely because the mac (i.e. the Finder in particular) tends to make a large number of requests in quick successsion when browsing the filesystem. on the flip side, the 'runtests' unit tests fail in the face of such caching because the changes made to the underlying tahoe directories are not reflected in the fuse presentation. by specifying a cache-timeout of 0 seconds, runtests can force the fuse layer into refetching directory data upon each request. any number of -oname=value options may be specified on the command line, and they will all be passed into the underlying fuse_main call. a single non-optional argument, the mountpoint, must also be given.
2008-09-25 00:15:35 +00:00
sys.exit(main(sys.argv[1:]))