[glib] Update the included copy of PCRE

[Matthias Clasen <matthiasc src gnome org>]

commit c03dc6bf7dfc6e326e7249e9e377676db885d99e
Author: Matthias Clasen <mclasen redhat com>
Date:   Fri Jan 21 23:10:01 2011 -0500

    Update the included copy of PCRE
    
    Update PCRE to version 8.12.
    At the same time, also add Unicode 6.0 script support.

 glib/gregex.c      |   16 ++++++++++++++++
 glib/tests/regex.c |    5 ++---
 2 files changed, 18 insertions(+), 3 deletions(-)
---
diff --git a/glib/gregex.c b/glib/gregex.c
index b62bda7..a1d2315 100644
--- a/glib/gregex.c
+++ b/glib/gregex.c
@@ -230,6 +230,10 @@ match_error (gint errcode)
       return _("workspace limit for empty substrings reached");
     case PCRE_ERROR_BADNEWLINE:
       return _("invalid combination of newline flags");
+    case PCRE_ERROR_BADOFFSET:
+      return _("bad offset");
+    case PCRE_ERROR_SHORTUTF8:
+      return _("short utf8");
     default:
       break;
     }
@@ -565,6 +569,14 @@ g_match_info_next (GMatchInfo  *match_info,
   prev_match_start = match_info->offsets[0];
   prev_match_end = match_info->offsets[1];
 
+  if (match_info->pos > match_info->string_len)
+    {
+      /* we have reached the end of the string */
+      match_info->pos = -1;
+      match_info->matches = PCRE_ERROR_NOMATCH;
+      return FALSE;
+    }
+
   match_info->matches = pcre_exec (match_info->regex->pcre_re,
                                    match_info->regex->extra,
                                    match_info->string,
@@ -1197,6 +1209,8 @@ g_regex_new (const gchar         *pattern,
       compile_options |= PCRE_NEWLINE_ANY;
     }
 
+  compile_options |= PCRE_UCP;
+
   /* compile the pattern */
   re = pcre_compile2 (pattern, compile_options, &errcode,
                       &errmsg, &erroffset, NULL);
@@ -1792,6 +1806,7 @@ g_regex_split_simple (const gchar        *pattern,
   regex = g_regex_new (pattern, compile_options, 0, NULL);
   if (!regex)
     return NULL;
+
   result = g_regex_split_full (regex, string, -1, 0, match_options, 0, NULL);
   g_regex_unref (regex);
   return result;
@@ -1924,6 +1939,7 @@ g_regex_split_full (const GRegex      *regex,
 
   match_ok = g_regex_match_full (regex, string, string_len, start_position,
                                  match_options, &match_info, &tmp_error);
+
   while (tmp_error == NULL)
     {
       if (match_ok)
diff --git a/glib/tests/regex.c b/glib/tests/regex.c
index 6e015e9..ef3ac68 100644
--- a/glib/tests/regex.c
+++ b/glib/tests/regex.c
@@ -1342,6 +1342,7 @@ test_match_all (gconstpointer d)
 #define PCRE_UTF8               0x00000800
 #define PCRE_NO_UTF8_CHECK      0x00002000
 #define PCRE_NEWLINE_ANY        0x00400000
+#define PCRE_UCP                0x20000000
 
 static void
 test_basic (void)
@@ -1353,7 +1354,7 @@ test_basic (void)
   regex = g_regex_new ("[A-Z]+", cflags, mflags, NULL);
 
   g_assert (regex != NULL);
-  g_assert_cmpint (g_regex_get_compile_flags (regex), ==, cflags|PCRE_UTF8|PCRE_NO_UTF8_CHECK|PCRE_NEWLINE_ANY );
+  g_assert_cmpint (g_regex_get_compile_flags (regex), ==, cflags|PCRE_UTF8|PCRE_NO_UTF8_CHECK|PCRE_NEWLINE_ANY|PCRE_UCP );
   g_assert_cmpint (g_regex_get_match_flags (regex), ==, mflags|PCRE_NO_UTF8_CHECK);
 
   g_regex_unref (regex);
@@ -2063,8 +2064,6 @@ main (int argc, char *argv[])
 {
   setlocale (LC_ALL, "");
 
-  g_setenv ("G_DEBUG", "fatal_warnings", TRUE);
-
   g_test_init (&argc, &argv, NULL);
 
   g_test_add_func ("/regex/basic", test_basic);