X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=e8434784518c2d4ffb8797ffaedbc6bf6545b399;hb=e77e2f143f073d08d6764e30771960b2bccde0db;hp=ebe7883b6091d1ee5533814aecd11ee102b63402;hpb=faf82a0b75a45f1e4dbb7ad8cecdfaf9a30a643d;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/regexec.c b/regexec.c
index ebe7883..e843478 100644
--- a/regexec.c
+++ b/regexec.c
@@ -5,6 +5,11 @@
  * "One Ring to rule them all, One Ring to find them..."
  */
 
+/* This file contains functions for executing a regular expression.  See
+ * also regcomp.c which funnily enough, contains functions for compiling
+ * a regular expression.
+ */
+
 /* NOTE: this is derived from Henry Spencer's regexp code, and should not
  * confused with the original package (see point 3 below).  Thanks, Henry!
  */
@@ -67,7 +72,8 @@
  *
  ****    Alterations to Henry's code are...
  ****
- ****    Copyright (c) 1991-2002, Larry Wall
+ ****    Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
+ ****    2000, 2001, 2002, 2003, 2004, by Larry Wall and others
  ****
  ****    You may distribute under the terms of either the GNU General Public
  ****    License or the Artistic License, as specified in the README file.
@@ -86,6 +92,7 @@
 #define RF_warned	2		/* warned about big count? */
 #define RF_evaled	4		/* Did an EVAL with setting? */
 #define RF_utf8		8		/* String contains multibyte chars? */
+#define RF_false	16		/* odd number of nested negatives */
 
 #define UTF ((PL_reg_flags & RF_utf8) != 0)
 
@@ -171,7 +178,7 @@ S_regcppush(pTHX_ I32 parenfloor)
 	Perl_croak(aTHX_ "panic: paren_elems_to_push < 0");
 
 #define REGCP_OTHER_ELEMS 6
-    SSCHECK(paren_elems_to_push + REGCP_OTHER_ELEMS);
+    SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS);
     for (p = PL_regsize; p > parenfloor; p--) {
 /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */
 	SSPUSHINT(PL_regendp[p]);
@@ -544,7 +551,7 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
 	    goto fail_finish;
 	/* we may be pointing at the wrong string */
 	if (s && RX_MATCH_COPIED(prog))
-	    s = prog->subbeg + (s - SvPVX(sv));
+	    s = strbeg + (s - SvPVX(sv));
 	if (data)
 	    *data->scream_olds = s;
     }
@@ -852,10 +859,6 @@ Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos,
 	char *startpos = strbeg;
 
 	t = s;
-	if (prog->reganch & ROPT_UTF8) {	
-	    PL_regdata = prog->data;
-	    PL_bostr = startpos;
-	}
 	cache_re(prog);
 	s = find_byclass(prog, prog->regstclass, s, endpos, startpos, 1);
 	if (!s) {
@@ -955,6 +958,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	I32 doevery = (prog->reganch & ROPT_SKIP) == 0;
 	char *m;
 	STRLEN ln;
+	STRLEN lnc;
+	register STRLEN uskip;
 	unsigned int c1;
 	unsigned int c2;
 	char *e;
@@ -965,7 +970,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	switch (OP(c)) {
 	case ANYOF:
 	    if (do_utf8) {
-		 while (s < strend) {
+		 while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		      if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) ||
 			  !UTF8_IS_INVARIANT((U8)s[0]) ?
 			  reginclass(c, (U8*)s, 0, do_utf8) :
@@ -977,7 +982,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		      }
 		      else 
 			   tmp = 1;
-		      s += UTF8SKIP(s);
+		      s += uskip;
 		 }
 	    }
 	    else {
@@ -1010,10 +1015,12 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	    }
 	    break;
 	case EXACTF:
-	    m = STRING(c);
-	    ln = STR_LEN(c);
+	    m   = STRING(c);
+	    ln  = STR_LEN(c);	/* length to match in octets/bytes */
+	    lnc = (I32) ln;	/* length to match in characters */
 	    if (UTF) {
 	        STRLEN ulen1, ulen2;
+		U8 *sm = (U8 *) m;
 		U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
 		U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
 
@@ -1024,6 +1031,11 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 				    0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
 		c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN_UCLC,
 				    0, ckWARN(WARN_UTF8) ? 0 : UTF8_ALLOW_ANY);
+		lnc = 0;
+		while (sm < ((U8 *) m + ln)) {
+		    lnc++;
+		    sm += UTF8SKIP(sm);
+		}
 	    }
 	    else {
 		c1 = *(U8*)m;
@@ -1031,12 +1043,13 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	    }
 	    goto do_exactf;
 	case EXACTFL:
-	    m = STRING(c);
-	    ln = STR_LEN(c);
+	    m   = STRING(c);
+	    ln  = STR_LEN(c);
+	    lnc = (I32) ln;
 	    c1 = *(U8*)m;
 	    c2 = PL_fold_locale[c1];
 	  do_exactf:
-	    e = HOP3c(strend, -(I32)ln, s);
+	    e = HOP3c(strend, -((I32)lnc), s);
 
 	    if (norun && e < s)
 		e = s;			/* Due to minlen logic of intuit() */
@@ -1059,6 +1072,8 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		STRLEN len, foldlen;
 		
 		if (c1 == c2) {
+		    /* Upper and lower of 1st char are equal -
+		     * probably not a "letter". */
 		    while (s <= e) {
 		        c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &len,
 					   ckWARN(WARN_UTF8) ?
@@ -1163,7 +1178,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		tmp = ((OP(c) == BOUND ?
 			isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
 		LOAD_UTF8_CHARCLASS(alnum,"a");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (tmp == !(OP(c) == BOUND ?
 				 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
 				 isALNUM_LC_utf8((U8*)s)))
@@ -1172,7 +1187,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 			if ((norun || regtry(prog, s)))
 			    goto got_it;
 		    }
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1206,14 +1221,14 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		tmp = ((OP(c) == NBOUND ?
 			isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0);
 		LOAD_UTF8_CHARCLASS(alnum,"a");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (tmp == !(OP(c) == NBOUND ?
 				 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
 				 isALNUM_LC_utf8((U8*)s)))
 			tmp = !tmp;
 		    else if ((norun || regtry(prog, s)))
 			goto got_it;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1235,7 +1250,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case ALNUM:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(alnum,"a");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1244,7 +1259,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1264,7 +1279,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case ALNUML:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (isALNUM_LC_utf8((U8*)s)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1273,7 +1288,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1293,7 +1308,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NALNUM:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(alnum,"a");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1302,7 +1317,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1322,7 +1337,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NALNUML:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!isALNUM_LC_utf8((U8*)s)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1331,7 +1346,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1351,7 +1366,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case SPACE:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(space," ");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1360,7 +1375,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1380,7 +1395,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case SPACEL:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1389,7 +1404,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1409,7 +1424,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NSPACE:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(space," ");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1418,7 +1433,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1438,7 +1453,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NSPACEL:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1447,7 +1462,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1467,7 +1482,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case DIGIT:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(digit,"0");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1476,7 +1491,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1496,7 +1511,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case DIGITL:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (isDIGIT_LC_utf8((U8*)s)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1505,7 +1520,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1525,7 +1540,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NDIGIT:
 	    if (do_utf8) {
 		LOAD_UTF8_CHARCLASS(digit,"0");
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1534,7 +1549,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1554,7 +1569,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 	case NDIGITL:
 	    PL_reg_flags |= RF_tainted;
 	    if (do_utf8) {
-		while (s < strend) {
+		while (s + (uskip = UTF8SKIP(s)) <= strend) {
 		    if (!isDIGIT_LC_utf8((U8*)s)) {
 			if (tmp && (norun || regtry(prog, s)))
 			    goto got_it;
@@ -1563,7 +1578,7 @@ S_find_byclass(pTHX_ regexp * prog, regnode *c, char *s, char *strend, char *sta
 		    }
 		    else
 			tmp = 1;
-		    s += UTF8SKIP(s);
+		    s += uskip;
 		}
 	    }
 	    else {
@@ -1866,7 +1881,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
 				  PL_multiline ? FBMrf_MULTILINE : 0))) ) {
 	    /* we may be pointing at the wrong string */
 	    if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog))
-		s = prog->subbeg + (s - SvPVX(sv));
+		s = strbeg + (s - SvPVX(sv));
 	    DEBUG_r( did_match = 1 );
 	    if (HOPc(s, -back_max) > last1) {
 		last1 = HOPc(s, -back_min);
@@ -1955,7 +1970,7 @@ Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *
 		    last = scream_olds; /* Only one occurrence. */
 		/* we may be pointing at the wrong string */
 		else if (RX_MATCH_COPIED(prog))
-		    s = prog->subbeg + (s - SvPVX(sv));
+		    s = strbeg + (s - SvPVX(sv));
 	    }
 	    else {
 		STRLEN len;
@@ -2095,8 +2110,7 @@ S_regtry(pTHX_ regexp *prog, char *startpos)
 	if (PL_reg_sv) {
 	    /* Make $_ available to executed code. */
 	    if (PL_reg_sv != DEFSV) {
-		/* SAVE_DEFSV does *not* suffice here for USE_5005THREADS */
-		SAVESPTR(DEFSV);
+		SAVE_DEFSV;
 		DEFSV = PL_reg_sv;
 	    }
 	
@@ -2838,6 +2852,7 @@ S_regmatch(pTHX_ regnode *prog)
 	    COP *ocurcop = PL_curcop;
 	    PAD *old_comppad;
 	    SV *ret;
+	    struct regexp *oreg = PL_reg_re;
 	
 	    n = ARG(scan);
 	    PL_op = (OP_4tree*)PL_regdata->data[n];
@@ -2970,8 +2985,10 @@ S_regmatch(pTHX_ regnode *prog)
 		sw = SvTRUE(ret);
 		logical = 0;
 	    }
-	    else
+	    else {
 		sv_setsv(save_scalar(PL_replgv), ret);
+		cache_re(oreg);
+	    }
 	    break;
 	}
 	case OPEN:
@@ -3204,7 +3221,10 @@ S_regmatch(pTHX_ regnode *prog)
 				      "%*s  already tried at this position...\n",
 				      REPORT_CODE_OFF+PL_regindent*2, "")
 			);
-			sayNO_SILENT;
+			if (PL_reg_flags & RF_false)
+			    sayYES;
+			else
+			    sayNO_SILENT;
 		    }
 		    PL_reg_poscache[o] |= (1<<b);
 		}
@@ -3353,7 +3373,7 @@ S_regmatch(pTHX_ regnode *prog)
 	    CHECKPOINT lastcp;
 	
 	    /* We suppose that the next guy does not need
-	       backtracking: in particular, it is of constant length,
+	       backtracking: in particular, it is of constant non-zero length,
 	       and has no parenths to influence future backrefs. */
 	    ln = ARG1(scan);  /* min to match */
 	    n  = ARG2(scan);  /* max to match */
@@ -3372,15 +3392,6 @@ S_regmatch(pTHX_ regnode *prog)
 		minmod = 0;
 		if (ln && regrepeat_hard(scan, ln, &l) < ln)
 		    sayNO;
-		/* if we matched something zero-length we don't need to
-		   backtrack - capturing parens are already defined, so
-		   the caveat in the maximal case doesn't apply
-
-		   XXXX if ln == 0, we can redo this check first time
-		   through the following loop
-		*/
-		if (ln && l == 0)
-		    n = ln;	/* don't backtrack */
 		locinput = PL_reginput;
 		if (HAS_TEXT(next) || JUMPABLE(next)) {
 		    regnode *text_node = next;
@@ -3390,19 +3401,8 @@ S_regmatch(pTHX_ regnode *prog)
 		    if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
 		    else {
 			if (PL_regkind[(U8)OP(text_node)] == REF) {
-			    I32 n, ln;
-			    n = ARG(text_node);  /* which paren pair */
-			    ln = PL_regstartp[n];
-			    /* assume yes if we haven't seen CLOSEn */
-			    if (
-				(I32)*PL_reglastparen < n ||
-				ln == -1 ||
-				ln == PL_regendp[n]
-			    ) {
-				c1 = c2 = -1000;
-				goto assume_ok_MM;
-			    }
-			    c1 = *(PL_bostr + ln);
+			    c1 = c2 = -1000;
+			    goto assume_ok_MM;
 			}
 			else { c1 = (U8)*STRING(text_node); }
 			if (OP(text_node) == EXACTF || OP(text_node) == REFF)
@@ -3417,8 +3417,7 @@ S_regmatch(pTHX_ regnode *prog)
 		    c1 = c2 = -1000;
 	    assume_ok_MM:
 		REGCP_SET(lastcp);
-		/* This may be improved if l == 0.  */
-		while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */
+		while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */
 		    /* If it could work, try it. */
 		    if (c1 == -1000 ||
 			UCHARAT(PL_reginput) == c1 ||
@@ -3449,13 +3448,6 @@ S_regmatch(pTHX_ regnode *prog)
 	    }
 	    else {
 		n = regrepeat_hard(scan, n, &l);
-		/* if we matched something zero-length we don't need to
-		   backtrack, unless the minimum count is zero and we
-		   are capturing the result - in that case the capture
-		   being defined or not may affect later execution
-		*/
-		if (n != 0 && l == 0 && !(paren && ln == 0))
-		    ln = n;	/* don't backtrack */
 		locinput = PL_reginput;
 		DEBUG_r(
 		    PerlIO_printf(Perl_debug_log,
@@ -3472,19 +3464,8 @@ S_regmatch(pTHX_ regnode *prog)
 			if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
 			else {
 			    if (PL_regkind[(U8)OP(text_node)] == REF) {
-				I32 n, ln;
-				n = ARG(text_node);  /* which paren pair */
-				ln = PL_regstartp[n];
-				/* assume yes if we haven't seen CLOSEn */
-				if (
-				    (I32)*PL_reglastparen < n ||
-				    ln == -1 ||
-				    ln == PL_regendp[n]
-				) {
-				    c1 = c2 = -1000;
-				    goto assume_ok_REG;
-				}
-				c1 = *(PL_bostr + ln);
+				c1 = c2 = -1000;
+				goto assume_ok_REG;
 			    }
 			    else { c1 = (U8)*STRING(text_node); }
 
@@ -3581,19 +3562,8 @@ S_regmatch(pTHX_ regnode *prog)
 		if (! HAS_TEXT(text_node)) c1 = c2 = -1000;
 		else {
 		    if (PL_regkind[(U8)OP(text_node)] == REF) {
-			I32 n, ln;
-			n = ARG(text_node);  /* which paren pair */
-			ln = PL_regstartp[n];
-			/* assume yes if we haven't seen CLOSEn */
-			if (
-			    (I32)*PL_reglastparen < n ||
-			    ln == -1 ||
-			    ln == PL_regendp[n]
-			) {
-			    c1 = c2 = -1000;
-			    goto assume_ok_easy;
-			}
-			s = (U8*)PL_bostr + ln;
+			c1 = c2 = -1000;
+			goto assume_ok_easy;
 		    }
 		    else { s = (U8*)STRING(text_node); }
 
@@ -3887,6 +3857,7 @@ S_regmatch(pTHX_ regnode *prog)
 	    }
 	    else
 		PL_reginput = locinput;
+	    PL_reg_flags ^= RF_false;
 	    goto do_ifmatch;
 	case IFMATCH:
 	    n = 1;
@@ -3902,6 +3873,8 @@ S_regmatch(pTHX_ regnode *prog)
 	  do_ifmatch:
 	    inner = NEXTOPER(NEXTOPER(scan));
 	    if (regmatch(inner) != n) {
+		if (n == 0)
+		    PL_reg_flags ^= RF_false;
 	      say_no:
 		if (logical) {
 		    logical = 0;
@@ -3911,6 +3884,8 @@ S_regmatch(pTHX_ regnode *prog)
 		else
 		    sayNO;
 	    }
+	    if (n == 0)
+		PL_reg_flags ^= RF_false;
 	  say_yes:
 	    if (logical) {
 		logical = 0;
@@ -4277,7 +4252,7 @@ S_regrepeat(pTHX_ regnode *p, I32 max)
 /*
  - regrepeat_hard - repeatedly match something, report total lenth and length
  *
- * The repeater is supposed to have constant length.
+ * The repeater is supposed to have constant non-zero length.
  */
 
 STATIC I32