Fix loop for classes containing \p or \P and just one ascii character.

git-svn-id: svn://vcs.exim.org/pcre/code/trunk@223 2f5784b3-3f2a-0410-8824-cb99058d5e15

7 files changed, 120 insertions, 12 deletions

diff --git a/ChangeLog b/ChangeLog
index 031ec34..94b478f 100644
--- a/ChangeLog
+++ b/ChangeLog
@@ -1,7 +1,7 @@
 ChangeLog for PCRE
 ------------------
 
-Version 7.3 17-Aug-07
+Version 7.3 20-Aug-07
 ---------------------
 
  1. In the rejigging of the build system that eventually resulted in 7.1, the
@@ -145,6 +145,9 @@ Version 7.3 17-Aug-07
 
 25. Using pcregrep in multiline, inverted mode (-Mv) caused it to loop.
 
+26. Patterns such as [\P{Yi}A] which include \p or \P and just one other 
+    character were causing crashes (broken optimization).
+
 
 Version 7.2 19-Jun-07
 ---------------------
diff --git a/configure.ac b/configure.ac
index cb534b4..2379045 100644
--- a/configure.ac
+++ b/configure.ac
@@ -8,8 +8,8 @@ dnl empty.
 
 m4_define(pcre_major, [7])
 m4_define(pcre_minor, [3])
-m4_define(pcre_prerelease, [-RC7])
-m4_define(pcre_date, [2007-08-17])
+m4_define(pcre_prerelease, [-RC8])
+m4_define(pcre_date, [2007-08-20])
 
 # Libtool shared library interface versions (current:revision:age)
 m4_define(libpcre_version, [0:1:0])
diff --git a/pcre_compile.c b/pcre_compile.c
index 05f370a..a1370d0 100644
--- a/pcre_compile.c
+++ b/pcre_compile.c
@@ -3176,11 +3176,14 @@ for (;; ptr++)
       }
 
     /* If class_charcount is 1, we saw precisely one character whose value is
-    less than 256. In non-UTF-8 mode we can always optimize. In UTF-8 mode, we
-    can optimize the negative case only if there were no characters >= 128
-    because OP_NOT and the related opcodes like OP_NOTSTAR operate on
-    single-bytes only. This is an historical hangover. Maybe one day we can
-    tidy these opcodes to handle multi-byte characters.
+    less than 256. As long as there were no characters >= 128 and there was no 
+    use of \p or \P, in other words, no use of any XCLASS features, we can 
+    optimize. 
+ 
+    In UTF-8 mode, we can optimize the negative case only if there were no
+    characters >= 128 because OP_NOT and the related opcodes like OP_NOTSTAR
+    operate on single-bytes only. This is an historical hangover. Maybe one day
+    we can tidy these opcodes to handle multi-byte characters.
 
     The optimization throws away the bit map. We turn the item into a
     1-character OP_CHAR[NC] if it's positive, or OP_NOT if it's negative. Note
@@ -3190,10 +3193,8 @@ for (;; ptr++)
     reqbyte, save the previous value for reinstating. */
 
 #ifdef SUPPORT_UTF8
-    if (class_charcount == 1 &&
-          (!utf8 ||
-          (!class_utf8 && (!negate_class || class_lastchar < 128))))
-
+    if (class_charcount == 1 && !class_utf8 && 
+      (!utf8 || !negate_class || class_lastchar < 128))
 #else
     if (class_charcount == 1)
 #endif
diff --git a/testdata/testinput10 b/testdata/testinput10
index 28fef9d..726a389 100644
--- a/testdata/testinput10
+++ b/testdata/testinput10
@@ -105,4 +105,20 @@ are all themselves checked in other tests. --/
 
 /(  (?(1)0|)*   )/xBM
 
+/[a]/BM
+
+/[a]/8BM
+
+/[\xaa]/BM
+
+/[\xaa]/8BM
+
+/[^a]/BM
+
+/[^a]/8BM
+
+/[^\xaa]/BM
+
+/[^\xaa]/8BM
+
 / End of testinput10 /
diff --git a/testdata/testinput6 b/testdata/testinput6
index 400b14f..14b0645 100644
--- a/testdata/testinput6
+++ b/testdata/testinput6
@@ -806,4 +806,12 @@ was broken in all cases./
 
 /(\P{Yi}{2}\277)?/
 
+/[\P{Yi}A]/
+
+/[\P{Yi}\P{Yi}\P{Yi}A]/
+
+/[^\P{Yi}A]/
+
+/[^\P{Yi}\P{Yi}\P{Yi}A]/
+
 / End of testinput6 /
diff --git a/testdata/testoutput10 b/testdata/testoutput10
index ffc84fa..dbd5924 100644
--- a/testdata/testoutput10
+++ b/testdata/testoutput10
@@ -594,4 +594,76 @@ Memory allocation (code space): 30
  29     End
 ------------------------------------------------------------------
 
+/[a]/BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     a
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[a]/8BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     a
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[\xaa]/BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     \xaa
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[\xaa]/8BM
+Memory allocation (code space): 10
+------------------------------------------------------------------
+  0   6 Bra
+  3     \x{aa}
+  6   6 Ket
+  9     End
+------------------------------------------------------------------
+
+/[^a]/BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     [^a]
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[^a]/8BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     [^a]
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[^\xaa]/BM
+Memory allocation (code space): 9
+------------------------------------------------------------------
+  0   5 Bra
+  3     [^\xaa]
+  5   5 Ket
+  8     End
+------------------------------------------------------------------
+
+/[^\xaa]/8BM
+Memory allocation (code space): 40
+------------------------------------------------------------------
+  0  36 Bra
+  3     [\x00-\xa9\xab-\xff] (neg)
+ 36  36 Ket
+ 39     End
+------------------------------------------------------------------
+
 / End of testinput10 /
diff --git a/testdata/testoutput6 b/testdata/testoutput6
index 2fc2db0..a39cf0e 100644
--- a/testdata/testoutput6
+++ b/testdata/testoutput6
@@ -1496,4 +1496,12 @@ No match
 
 /(\P{Yi}{2}\277)?/
 
+/[\P{Yi}A]/
+
+/[\P{Yi}\P{Yi}\P{Yi}A]/
+
+/[^\P{Yi}A]/
+
+/[^\P{Yi}\P{Yi}\P{Yi}A]/
+
 / End of testinput6 /