Changeset 15455 in webkit for trunk/JavaScriptCore

Timestamp:

Jul 15, 2006, 8:30:03 AM (19 years ago)

Author:

darin

Message:

JavaScriptCore:

Reviewed by Geoff.

• fix ​http://bugzilla.opendarwin.org/show_bug.cgi?id=8395 <rdar://problem/4613467> REGRESSION: RegEx seems broken for hex escaped non breaking space

Test: fast/js/regexp-extended-characters-more.html

• pcre/pcre_exec.c: (match): Got rid of utf16Length local variable to guarantee there's no extra stack usage in recursive calls. Fixed two places in the PCRE_UTF16 code that were using the length variable, which is the UTF-8 length of a character in the pattern, to move in the UTF-16 subject string. Instead they hardcode lengths of 1 and 2 since the code already handles BMP characters and surrogate pairs separately. Also fixed some DPRINTF so I could compile with DEBUG on. (pcre_exec): Changed a place that was checking for multibyte characters in the subject string to use ISMIDCHAR. Instead it was using hardcoded logic that was right for UTF-8 but wrong for UTF-16.

• pcre/pcre_compile.c: (pcre_compile2): Fixed a DPRINTF so I could compile with DEBUG on.

LayoutTests:

Reviewed by Geoff.

• test for ​http://bugzilla.opendarwin.org/show_bug.cgi?id=8395 <rdar://problem/4613467> REGRESSION: RegEx seems broken for hex escaped non breaking space

• fast/js/regexp-extended-characters-more-expected.txt: Added.
• fast/js/regexp-extended-characters-more.html: Added.
• fast/js/resources/regexp-extended-characters-more.js: Added.

Location:

trunk/JavaScriptCore

Files:

• ChangeLog (modified) (1 diff)
• pcre/pcre_compile.c (modified) (1 diff)
• pcre/pcre_exec.c (modified) (6 diffs)

trunk/JavaScriptCore/ChangeLog

@@ +1 @@
+2006-07-15  Darin Adler  <[email protected]>
+
+        Reviewed by Geoff.
+
+        - fix http://bugzilla.opendarwin.org/show_bug.cgi?id=8395
+          <rdar://problem/4613467>
+          REGRESSION: RegEx seems broken for hex escaped non breaking space
+
+        Test: fast/js/regexp-extended-characters-more.html
+
+        * pcre/pcre_exec.c:
+        (match): Got rid of utf16Length local variable to guarantee there's no
+        extra stack usage in recursive calls. Fixed two places in the PCRE_UTF16
+        code that were using the length variable, which is the UTF-8 length of
+        a character in the pattern, to move in the UTF-16 subject string. Instead
+        they hardcode lengths of 1 and 2 since the code already handles BMP
+        characters and surrogate pairs separately. Also fixed some DPRINTF so
+        I could compile with DEBUG on.
+        (pcre_exec): Changed a place that was checking for multibyte characters
+        in the subject string to use ISMIDCHAR. Instead it was using hardcoded
+        logic that was right for UTF-8 but wrong for UTF-16.
+
+        * pcre/pcre_compile.c: (pcre_compile2): Fixed a DPRINTF so I could compile
+        with DEBUG on.
+
 2006-07-14  Geoffrey Garen  <[email protected]>
 

trunk/JavaScriptCore/pcre/pcre_compile.c

@@ -3996 +3996 @@
 
 DPRINTF(("------------------------------------------------------------------\n"));
+#if !PCRE_UTF16
 DPRINTF(("%s\n", pattern));
+#endif
 
 /* The first thing to do is to make a pass over the pattern to compute the

trunk/JavaScriptCore/pcre/pcre_exec.c

@@ -1952 +1952 @@
       GETUTF8CHARLEN(fc, ecode, length);
       {
-      int utf16Length; /* don't initialize on this line as workaround for Win32 compile problem */
-      utf16Length = fc > 0xFFFF ? 2 : 1;
-      if (min * utf16Length > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
+      if (min * (fc > 0xFFFF ? 2 : 1) > md->end_subject - eptr) RRETURN(MATCH_NOMATCH);
       ecode += length;
 
-      if (utf16Length == 1)
+      if (fc <= 0xFFFF)
         {
 #ifdef SUPPORT_UCP
@@ -1991 +1989 @@
           for (i = min; i < max; i++)
             {
-            if (eptr > md->end_subject - length) break;
+            if (eptr >= md->end_subject) break;
             if (*eptr != fc && *eptr != othercase) break;
             ++eptr;
@@ -2039 +2037 @@
             {
             int nc;
-            if (eptr > md->end_subject - length) break;
+            if (eptr > md->end_subject - 2) break;
             GETCHAR(nc, eptr);
             if (*eptr != fc) break;
@@ -2162 +2160 @@
     maximizing, find the maximum number of characters and work backwards. */
 
+#if PCRE_UTF16
+    DPRINTF(("matching %c{%d,%d}\n", fc, min, max));
+#else
     DPRINTF(("matching %c{%d,%d} against subject %.*s\n", fc, min, max,
       max, eptr));
+#endif
 
     if ((ims & PCRE_CASELESS) != 0)
@@ -2307 +2309 @@
     characters and work backwards. */
 
+#if PCRE_UTF16
+    DPRINTF(("negative matching %c{%d,%d}\n", fc, min, max));
+#else
     DPRINTF(("negative matching %c{%d,%d} against subject %.*s\n", fc, min, max,
       max, eptr));
+#endif
 
     if ((ims & PCRE_CASELESS) != 0)
@@ -3733 +3739 @@
 #ifdef SUPPORT_UTF8
     if (match_block.utf8)
-      while(start_match < end_subject && (*start_match & 0xc0) == 0x80)
+      while(start_match < end_subject && ISMIDCHAR(*start_match))
         start_match++;
 #endif