Upgrade to Unicode 3.1 beta 2001-03-01.

[p5sagit/p5-mst-13.2.git] / regexec.c

diff --git a/regexec.c b/regexec.c
index eef5f59..5d9e8ac 100644 (file)
--- a/regexec.c
+++ b/regexec.c
@@ -147,8 +147,12 @@ S_regcppush(pTHX_ I32 parenfloor)
     SSPUSHINT(PL_regsize);
     SSPUSHINT(*PL_reglastparen);
     SSPUSHPTR(PL_reginput);
-    SSPUSHINT(paren_elems_to_push + (REGCP_PAREN_ELEMS - 1));
+#define REGCP_FRAME_ELEMS 2
+/* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and
+ * are needed for the regexp context stack bookkeeping. */
+    SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
     SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */
+
     return retval;
 }
 
@@ -179,7 +183,8 @@ S_regcppop(pTHX)
     PL_regsize = SSPOPINT;
 
     /* Now restore the parentheses context. */
-    for (i -= (REGCP_PAREN_ELEMS - 1); i > 0; i -= REGCP_PAREN_ELEMS) {
+    for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS);
+        i > 0; i -= REGCP_PAREN_ELEMS) {
        paren = (U32)SSPOPINT;
        PL_reg_start_tmp[paren] = (char *) SSPOPPTR;
        PL_regstartp[paren] = SSPOPINT;
@@ -393,7 +398,8 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
              DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n"));
              goto fail;
          }
-         if (prog->check_offset_min == prog->check_offset_max) {
+         if (prog->check_offset_min == prog->check_offset_max &&
+             !(prog->reganch & ROPT_SANY_SEEN)) {
            /* Substring at constant offset from beg-of-str... */
            I32 slen;
 
@@ -469,6 +475,10 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
        if (data)
            *data->scream_olds = s;
     }
+    else if (prog->reganch & ROPT_SANY_SEEN)
+       s = fbm_instr((U8*)(s + start_shift),
+                     (U8*)(strend - end_shift),
+                     check, PL_multiline ? FBMrf_MULTILINE : 0);
     else
        s = fbm_instr(HOP3(s, start_shift, strend),
                      HOP3(strend, -end_shift, strbeg),
@@ -1402,7 +1412,8 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
 
     minlen = prog->minlen;
     if (do_utf8) {
-      if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey;
+      if (!(prog->reganch & ROPT_SANY_SEEN))
+        if (utf8_distance((U8*)strend, (U8*)startpos) < minlen) goto phooey;
     }
     else {
       if (strend - startpos < minlen) goto phooey;
@@ -2070,13 +2081,6 @@ S_regmatch(pTHX_ regnode *prog)
                sayNO;
            break;
        case SANY:
-           if (do_utf8) {
-               locinput += PL_utf8skip[nextchr];
-               if (locinput > PL_regeol)
-                   sayNO;
-               nextchr = UCHARAT(locinput);
-               break;
-           }
            if (!nextchr && locinput >= PL_regeol)
                sayNO;
            nextchr = UCHARAT(++locinput);
@@ -3558,15 +3562,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max)
        }
        break;
     case SANY:
-       if (do_utf8) {
-           loceol = PL_regeol;
-           while (hardcount < max && scan < loceol) {
-               scan += UTF8SKIP(scan);
-               hardcount++;
-           }
-       } else {
-           scan = loceol;
-       }
+       scan = loceol;
        break;
     case EXACT:                /* length of string is 1 */
        c = (U8)*STRING(p);