X-Git-Url: http://erislabs.net/gitweb/?a=blobdiff_plain;f=regex.c;h=5b1b391dab7dff072152bd3e7cf7ddb94bdf6b28;hb=69a9b8f8d7acc8677b5657adb83a8ba448737e0c;hp=022ca13eba6f13fca636e2e136bedd49f8e30254;hpb=972e3e06142a2a253a022b8ad3d8c4e0baf80c8b;p=gnulib.git

diff --git a/regex.c b/regex.c
index 022ca13eb..5b1b391da 100644
--- a/regex.c
+++ b/regex.c
@@ -33,9 +33,6 @@
   #pragma alloca
 #endif
 
-#undef	_GNU_SOURCE
-#define _GNU_SOURCE
-
 #ifdef HAVE_CONFIG_H
 # include <config.h>
 #endif
@@ -160,8 +157,9 @@
        {						    		\
 	 re_char *dtemp = (p) == (str2) ? (end1) : (p);		    	\
 	 re_char *dlimit = ((p) > (str2) && (p) <= (end2)) ? (str2) : (str1); \
-	 while (dtemp-- > dlimit && !CHAR_HEAD_P (*dtemp));		\
-	 c = STRING_CHAR (dtemp, (p) - dtemp);				\
+	 re_char *d0 = dtemp;						\
+	 PREV_CHAR_BOUNDARY (d0, dlimit);				\
+	 c = STRING_CHAR (d0, dtemp - d0);				\
        }						    		\
      else						    		\
        (c = ((p) == (str2) ? (end1) : (p))[-1]);			\
@@ -238,6 +236,7 @@ enum syntaxcode { Swhitespace = 0, Sword = 1 };
 # define SINGLE_BYTE_CHAR_P(c) (1)
 # define SAME_CHARSET_P(c1, c2) (1)
 # define MULTIBYTE_FORM_LENGTH(p, s) (1)
+# define PREV_CHAR_BOUNDARY(p, limit) ((p)--)
 # define STRING_CHAR(p, s) (*(p))
 # define RE_STRING_CHAR STRING_CHAR
 # define CHAR_STRING(c, s) (*(s) = (c), 1)
@@ -2066,7 +2065,7 @@ re_wctype_to_bit (cc)
 static void
 extend_range_table_work_area (work_area)
      struct range_table_work_area *work_area;
-{									
+{
   work_area->allocated += 16 * sizeof (int);
   if (work_area->table)
     work_area->table
@@ -2105,7 +2104,7 @@ set_image_of_range_1 (work_area, start, end, translate)
 
      `strange' indicates a character that has more than one
      case-equivalent.  */
-     
+
   enum case_type {one_case, two_case, strange};
 
   /* Describe the run that is in progress,
@@ -2175,7 +2174,7 @@ set_image_of_range_1 (work_area, start, end, translate)
 	    }
 	  run_type = strange;
 	}
-	      
+
       if (this_type == strange)
 	{
 	  /* For a strange character, add each of its equivalents, one
@@ -2625,7 +2624,7 @@ regex_compile (pattern, size, syntax, bufp)
 		      (simple || !analyse_first (laststart, b, NULL, 0))
 		      ? on_failure_jump : on_failure_jump_loop;
 		    assert (skip_one_char (laststart) <= b);
-		    
+
 		    if (!zero_times_ok && simple)
 		      { /* Since simple * loops can be made faster by using
 		    	   on_failure_keep_string_jump, we turn simple P+
@@ -3942,7 +3941,7 @@ analyse_first (p, pend, fastmap, multibyte)
 	     case has already been handled, so we only need to look at the
 	     fallthrough case.  */
 	  continue;
-	  
+
 	case succeed_n:
 	  /* If N == 0, it should be an on_failure_jump_loop instead.  */
 	  DEBUG_STATEMENT (EXTRACT_NUMBER (j, p + 2); assert (j > 0));
@@ -4067,6 +4066,10 @@ re_search (bufp, string, size, startpos, range, regs)
 }
 WEAK_ALIAS (__re_search, re_search)
 
+/* Head address of virtual concatenation of string.  */
+#define HEAD_ADDR_VSTRING(P)		\
+  (((P) >= size1 ? string2 : string1))
+
 /* End address of virtual concatenation of string.  */
 #define STOP_ADDR_VSTRING(P)				\
   (((P) >= size1 ? string2 + size2 : string1 + size1))
@@ -4302,26 +4305,17 @@ re_search_2 (bufp, str1, size1, str2, size2, startpos, range, regs, stop)
 	  /* Update STARTPOS to the previous character boundary.  */
 	  if (multibyte)
 	    {
-	      re_char *p = POS_ADDR_VSTRING (startpos);
-	      int len = 0;
+	      re_char *p = POS_ADDR_VSTRING (startpos) + 1;
+	      re_char *p0 = p;
+	      re_char *phead = HEAD_ADDR_VSTRING (startpos);
 
 	      /* Find the head of multibyte form.  */
-	      while (!CHAR_HEAD_P (*p))
-		p--, len++;
-
-	      /* Adjust it. */
-#if 0				/* XXX */
-	      if (MULTIBYTE_FORM_LENGTH (p, len + 1) != (len + 1))
-		;
-	      else
-#endif
-		{
-		  range += len;
-		  if (range > 0)
-		    break;
+	      PREV_CHAR_BOUNDARY (p, phead);
+	      range += p0 - 1 - p;
+	      if (range > 0)
+		break;
 
-		  startpos -= len;
-		}
+	      startpos -= p0 - 1 - p;
 	    }
 	}
     }
@@ -4430,7 +4424,7 @@ skip_one_char (p)
     {
     case anychar:
       break;
-      
+
     case exactn:
       p += *p + 1;
       break;
@@ -4447,7 +4441,7 @@ skip_one_char (p)
       else
 	p += 1 + CHARSET_BITMAP_SIZE (p - 1);
       break;
-      
+
     case syntaxspec:
     case notsyntaxspec:
 #ifdef emacs
@@ -4530,7 +4524,7 @@ mutually_exclusive_p (bufp, p1, p2)
 	  return 1;
 	}
       break;
-      
+
     case endline:
     case exactn:
       {
@@ -4640,7 +4634,7 @@ mutually_exclusive_p (bufp, p1, p2)
 	  }
       }
       break;
-      
+
     case charset_not:
       switch (SWITCH_ENUM_CAST (*p1))
 	{
@@ -5324,7 +5318,7 @@ re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
 
 	  assert (!REG_UNSET (regstart[*p]));
 	  /* Strictly speaking, there should be code such as:
-	     
+
 		assert (REG_UNSET (regend[*p]));
 		PUSH_FAILURE_REGSTOP ((unsigned int)*p);
 
@@ -5741,7 +5735,7 @@ re_match_2_internal (bufp, string1, size1, string2, size2, pos, regs, stop)
 	      PREFETCH ();
 	      c2 = RE_STRING_CHAR (d, dend - d);
 	      s2 = SYNTAX (c2);
-	
+
 	      /* Case 2: S2 is not Sword. */
 	      if (s2 != Sword)
 		goto fail;