From 6138bb612cd23c0ca37faadbe3fc9944be275228 Mon Sep 17 00:00:00 2001
From: Patrik Nyblom <pan@erlang.org>
Date: Mon, 29 Jul 2013 10:54:20 +0200
Subject: Integrate patch for PCRE bug id 1370

---
 erts/emulator/pcre/pcre-8.33_1370.diff | 60 ++++++++++++++++++++++++++++++++++
 erts/emulator/pcre/pcre_exec.c         | 38 ++++++++++++++++-----
 2 files changed, 90 insertions(+), 8 deletions(-)
 create mode 100644 erts/emulator/pcre/pcre-8.33_1370.diff

(limited to 'erts')

diff --git a/erts/emulator/pcre/pcre-8.33_1370.diff b/erts/emulator/pcre/pcre-8.33_1370.diff
new file mode 100644
index 0000000000..d62398985d
--- /dev/null
+++ b/erts/emulator/pcre/pcre-8.33_1370.diff
@@ -0,0 +1,60 @@
+--- code/trunk/pcre_exec.c	2013/07/02 18:37:36	1346
++++ code/trunk/pcre_exec.c	2013/07/26 10:03:38	1350
+@@ -5637,7 +5637,7 @@
+           }
+         }
+ 
+-      /* Match extended Unicode sequences. We will get here only if the
++      /* Match extended Unicode grapheme clusters. We will get here only if the
+       support is in the binary; otherwise a compile-time error occurs. */
+ 
+       else if (ctype == OP_EXTUNI)
+@@ -5670,21 +5670,41 @@
+         /* eptr is now past the end of the maximum run */
+ 
+         if (possessive) continue;    /* No backtracking */
++         
+         for(;;)
+           {
+-          if (eptr == pp) goto TAIL_RECURSE;
++          int lgb, rgb; 
++          PCRE_PUCHAR fptr;
++            
++          if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
+           RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
+           if (rrc != MATCH_NOMATCH) RRETURN(rrc);
++
++          /* Backtracking over an extended grapheme cluster involves inspecting
++          the previous two characters (if present) to see if a break is
++          permitted between them. */
++ 
+           eptr--;
+-          for (;;)                        /* Move back over one extended */
++          if (!utf) c = *eptr; else
++            {
++            BACKCHAR(eptr);
++            GETCHAR(c, eptr);
++            }
++          rgb = UCD_GRAPHBREAK(c);
++
++          for (;;)
+             {
+-            if (!utf) c = *eptr; else
++            if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
++            fptr = eptr - 1;
++            if (!utf) c = *fptr; else
+               {
+-              BACKCHAR(eptr);
+-              GETCHAR(c, eptr);
++              BACKCHAR(fptr);
++              GETCHAR(c, fptr);
+               }
+-            if (UCD_CATEGORY(c) != ucp_M) break;
+-            eptr--;
++            lgb = UCD_GRAPHBREAK(c);        
++            if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
++            eptr = fptr;
++            rgb = lgb;
+             }
+           }
+         }
diff --git a/erts/emulator/pcre/pcre_exec.c b/erts/emulator/pcre/pcre_exec.c
index 0371f9efa8..1cab78cdd8 100644
--- a/erts/emulator/pcre/pcre_exec.c
+++ b/erts/emulator/pcre/pcre_exec.c
@@ -5818,7 +5818,7 @@ for (;;)
           }
         }
 
-      /* Match extended Unicode sequences. We will get here only if the
+      /* Match extended Unicode grapheme clusters. We will get here only if the
       support is in the binary; otherwise a compile-time error occurs. */
 
       else if (ctype == OP_EXTUNI)
@@ -5855,21 +5855,43 @@ for (;;)
         /* eptr is now past the end of the maximum run */
 
         if (possessive) continue;    /* No backtracking */
+         
         for(;;) /* LOOP_COUNT: Ok */
           {
-          if (eptr == pp) goto TAIL_RECURSE;
+#ifndef ERLANG_INTEGRATION
+          int lgb, rgb; 
+#endif
+          PCRE_PUCHAR fptr;
+            
+          if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
           RMATCH(eptr, ecode, offset_top, md, eptrb, RM45);
           if (rrc != MATCH_NOMATCH) RRETURN(rrc);
+
+          /* Backtracking over an extended grapheme cluster involves inspecting
+          the previous two characters (if present) to see if a break is
+          permitted between them. */
+ 
           eptr--;
-          for (;;)                        /* Move back over one extended */  /* LOOP_COUNT: COST */
+          if (!utf) c = *eptr; else
             {
-            if (!utf) c = *eptr; else
+            BACKCHAR(eptr);
+            GETCHAR(c, eptr);
+            }
+          rgb = UCD_GRAPHBREAK(c);
+
+          for (;;) /* LOOP_COUNT: COST */
+            {
+            if (eptr == pp) goto TAIL_RECURSE;   /* At start of char run */
+            fptr = eptr - 1;
+            if (!utf) c = *fptr; else
               {
-              BACKCHAR(eptr);
-              GETCHAR(c, eptr);
+              BACKCHAR(fptr);
+              GETCHAR(c, fptr);
               }
-            if (UCD_CATEGORY(c) != ucp_M) break;
-            eptr--;
+            lgb = UCD_GRAPHBREAK(c);        
+            if ((PRIV(ucp_gbtable)[lgb] & (1 << rgb)) == 0) break;
+            eptr = fptr;
+            rgb = lgb;
 	    COST(1);
             }
           }
-- 
cgit v1.2.3