Don't return duplicate matches when matching empty strings. (#515944) Add

2008-02-17  Marco Barisione  <marco@barisione.org>

	* glib/gregex.c: (match_info_new), (g_match_info_next): Don't return
	duplicate matches when matching empty strings.  (#515944)
	* tests/regex-test.c: Add tests.

svn path=/trunk/; revision=6527

ChangeLog

@@ -1,3 +1,9 @@
+2008-02-17  Marco Barisione  <marco@barisione.org>
+
+	* glib/gregex.c: (match_info_new), (g_match_info_next): Don't return
+	duplicate matches when matching empty strings.  (#515944)
+	* tests/regex-test.c: Add tests.
+
 2008-02-17  Hans Breuer  <hans@breuer.org>
 
 	* glib/gutils.c : define CSIDL_MYPICTURES if not available

glib/gregex.c

@@ -411,7 +411,11 @@ match_info_new (const GRegex *regex,
                      PCRE_INFO_CAPTURECOUNT, &capture_count);
       match_info->n_offsets = (capture_count + 1) * 3;
     }
+
   match_info->offsets = g_new0 (gint, match_info->n_offsets);
+  /* Set an invalid position for the previous match. */
+  match_info->offsets[0] = -1;
+  match_info->offsets[1] = -1;
 
   return match_info;
 }
@@ -495,6 +499,8 @@ g_match_info_next (GMatchInfo  *match_info,
 		   GError     **error)
 {
   gint opts;
+  gint prev_match_start;
+  gint prev_match_end;
 
   g_return_val_if_fail (match_info != NULL, FALSE);
   g_return_val_if_fail (error == NULL || *error == NULL, FALSE);
@@ -502,6 +508,9 @@ g_match_info_next (GMatchInfo  *match_info,
 
   opts = match_info->regex->match_opts | match_info->match_opts;
  
+  prev_match_start = match_info->offsets[0];
+  prev_match_end = match_info->offsets[1];
+
   match_info->matches = pcre_exec (match_info->regex->pcre_re,
 				   match_info->regex->extra,
 				   match_info->string,
@@ -540,6 +549,25 @@ g_match_info_next (GMatchInfo  *match_info,
       match_info->pos = match_info->offsets[1];
     }
 
+  /* it's possibile to get two identical matches when we are matching
+   * empty strings, for instance if the pattern is "(?=[A-Z0-9])" and
+   * the string is "RegExTest" we have:
+   *  - search at position 0: match from 0 to 0
+   *  - search at position 1: match from 3 to 3
+   *  - search at position 3: match from 3 to 3 (duplicate)
+   *  - search at position 4: match from 5 to 5
+   *  - search at position 5: match from 5 to 5 (duplicate)
+   *  - search at position 6: no match -> stop
+   * so we have to ignore the duplicates.
+   * see bug #515944: http://bugzilla.gnome.org/show_bug.cgi?id=515944 */
+  if (match_info->matches >= 0 &&
+      prev_match_start == match_info->offsets[0] &&
+      prev_match_end == match_info->offsets[1])
+    {
+      /* ignore this match and search the next one */
+      return g_match_info_next (match_info, error);
+    }
+
   return match_info->matches >= 0;
 }
 

tests/regex-test.c

@@ -1786,6 +1786,7 @@ main (int argc, char *argv[])
   TEST_MATCH_NEXT3("a", "aaxa", -1, 0, "a", 0, 1, "a", 1, 2, "a", 3, 4);
   TEST_MATCH_NEXT3("a", "aa" OGRAVE "a", -1, 0, "a", 0, 1, "a", 1, 2, "a", 4, 5);
   TEST_MATCH_NEXT3("a*", "aax", -1, 0, "aa", 0, 2, "", 2, 2, "", 3, 3);
+  TEST_MATCH_NEXT3("(?=[A-Z0-9])", "RegExTest", -1, 0, "", 0, 0, "", 3, 3, "", 5, 5);
   TEST_MATCH_NEXT4("a*", "aaxa", -1, 0, "aa", 0, 2, "", 2, 2, "a", 3, 4, "", 4, 4);
 
   /* TEST_MATCH_COUNT(pattern, string, start_position, match_opts, expected_count) */
@@ -2047,6 +2048,8 @@ main (int argc, char *argv[])
   TEST_REPLACE_LIT("[^-]", "-" EURO "-" AGRAVE "-" HSTROKE, 0, "a", "-a-a-a");
   TEST_REPLACE_LIT("[^-]", "-" EURO "-" AGRAVE, 0, "a\\g<0>a", "-a\\g<0>a-a\\g<0>a");
   TEST_REPLACE_LIT("-", "-" EURO "-" AGRAVE "-" HSTROKE, 0, "", EURO AGRAVE HSTROKE);
+  TEST_REPLACE_LIT("(?=[A-Z0-9])", "RegExTest", 0, "_", "_Reg_Ex_Test");
+  TEST_REPLACE_LIT("(?=[A-Z0-9])", "RegExTest", 1, "_", "Reg_Ex_Test");
 
   /* TEST_GET_STRING_NUMBER(pattern, name, expected_num) */
   TEST_GET_STRING_NUMBER("", "A", -1);