From 7fcbf9d85ceef2c5c292e0f8873a4afea3633b4b Mon Sep 17 00:00:00 2001
From: Joel Dice <git@seibutsu.mailsnare.net>
Date: Fri, 14 Aug 2009 08:52:31 -0600
Subject: [PATCH] fix reading 2-byte UTF-8 constants

---
 src/machine.cpp   | 35 +++++++++++++++++++++--------------
 test/Strings.java | 29 +++++++++++++++++++++++++++++
 2 files changed, 50 insertions(+), 14 deletions(-)
 create mode 100644 test/Strings.java

diff --git a/src/machine.cpp b/src/machine.cpp
index 9905f63ddf..53a134393c 100644
--- a/src/machine.cpp
+++ b/src/machine.cpp
@@ -19,6 +19,8 @@ using namespace vm;
 
 namespace {
 
+const unsigned NoByte = 0xFFFF;
+
 bool
 find(Thread* t, Thread* o)
 {
@@ -540,9 +542,21 @@ makeByteArray(Thread* t, const char* format, va_list a)
   return s;
 }
 
+unsigned
+readByte(Stream& s, unsigned* value)
+{
+  if (*value == NoByte) {
+    return s.read1();
+  } else {
+    unsigned r = *value;
+    *value = NoByte;
+    return r;
+  }
+}
+
 object
 parseUtf8NonAscii(Thread* t, Stream& s, object bytesSoFar, unsigned byteCount,
-                  unsigned sourceIndex, unsigned lastByteRead)
+                  unsigned sourceIndex, unsigned byteA, unsigned byteB)
 {
   PROTECT(t, bytesSoFar);
   
@@ -554,15 +568,14 @@ parseUtf8NonAscii(Thread* t, Stream& s, object bytesSoFar, unsigned byteCount,
     charArrayBody(t, value, vi) = byteArrayBody(t, bytesSoFar, vi);
   }
 
-  unsigned a = lastByteRead;
-  unsigned si = sourceIndex;
-  while (true) {
+  for (unsigned si = sourceIndex; si < length; ++si) {
+    unsigned a = readByte(s, &byteA);
     if (a & 0x80) {
       if (a & 0x20) {
 	// 3 bytes
 	si += 2;
 	assert(t, si < length);
-	unsigned b = s.read1();
+        unsigned b = readByte(s, &byteB);
 	unsigned c = s.read1();
 	charArrayBody(t, value, vi++)
           = ((a & 0xf) << 12) | ((b & 0x3f) << 6) | (c & 0x3f);
@@ -570,7 +583,7 @@ parseUtf8NonAscii(Thread* t, Stream& s, object bytesSoFar, unsigned byteCount,
 	// 2 bytes
 	++ si;
 	assert(t, si < length);
-	unsigned b = s.read1();
+        unsigned b = readByte(s, &byteB);
 
 	if (a == 0xC0 and b == 0x80) {
 	  charArrayBody(t, value, vi++) = 0;
@@ -580,12 +593,6 @@ parseUtf8NonAscii(Thread* t, Stream& s, object bytesSoFar, unsigned byteCount,
       }
     } else {
       charArrayBody(t, value, vi++) = a;
-    }    
-    
-    if (++si < length) {
-      a = s.read1();
-    } else {
-      break;
     }
   }
 
@@ -611,7 +618,7 @@ parseUtf8(Thread* t, Stream& s, unsigned length)
     if (a & 0x80) {
       if (a & 0x20) {
 	// 3 bytes
-        return parseUtf8NonAscii(t, s, value, vi, si, a);
+        return parseUtf8NonAscii(t, s, value, vi, si, a, NoByte);
       } else {
 	// 2 bytes
 	unsigned b = s.read1();
@@ -621,7 +628,7 @@ parseUtf8(Thread* t, Stream& s, unsigned length)
           assert(t, si < length);
 	  byteArrayBody(t, value, vi++) = 0;
 	} else {
-          return parseUtf8NonAscii(t, s, value, vi, si, a);
+          return parseUtf8NonAscii(t, s, value, vi, si, a, b);
 	}
       }
     } else {
diff --git a/test/Strings.java b/test/Strings.java
new file mode 100644
index 0000000000..c71fffa1aa
--- /dev/null
+++ b/test/Strings.java
@@ -0,0 +1,29 @@
+public class Strings {
+  private static void expect(boolean v) {
+    if (! v) throw new RuntimeException();
+  }
+
+  public static void main(String[] args) {
+    expect(new String(new byte[] { 99, 111, 109, 46, 101, 99, 111, 118, 97,
+                                   116, 101, 46, 110, 97, 116, 46, 98, 117,
+                                   115, 46, 83, 121, 109, 98, 111, 108 })
+      .equals("com.ecovate.nat.bus.Symbol"));
+    
+    final String months = "Jan\u00aeFeb\u00aeMar\u00ae";
+
+    System.out.println(months.split("\u00ae")[0]);
+    System.out.println(months.length());
+    System.out.println(months);
+    for (int i = 0; i < months.length(); ++i) {
+      System.out.print(Integer.toHexString(months.charAt(i)) + " ");
+    }
+    System.out.println();
+
+    expect(months.split("\u00ae").length == 3);
+
+    StringBuilder sb = new StringBuilder();
+    sb.append('$');
+    sb.append('2');
+    expect(sb.substring(1).equals("2"));
+  }
+}