Commit Graph

2 Commits

Author SHA1 Message Date
Johannes Schindelin
e2105670a0 Regex compiler: fall back to TrivialPattern when possible
While at it, let's get rid of the unescaping in TrivialPattern which was
buggy anyway: special operators such as \b were misinterpreted as trivial
patterns.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2013-12-03 12:28:10 -06:00
Johannes Schindelin
04d8955f98 Regex: Implement compiler for regular expression patterns
Originally, this developer wanted to (ab)use the PikeVM with a
hand-crafted program and an added "callback" opcode to parse the regular
expressions.

However, this turned out to be completely unnecessary: there are no
ambiguities in regular expression patterns, so there is no need to do
anything else than parse the pattern, one character at a time, into a
nested expression that then knows how to write itself into a program for
the PikeVM.

For the moment, we still hardcode the program for the regular expression
pattern demonstrating the challenge with the prioritized threads because
the compiler cannot yet parse reluctant operators.

Signed-off-by: Johannes Schindelin <johannes.schindelin@gmx.de>
2013-12-03 12:28:10 -06:00