X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regexec.c;h=747b61be3bfb6383d24d20a75045700907255474;hb=ce5e94717f361c3fd6b9b0fb704412d30f3ccf66;hp=890736cc132485d7108f9ce0a39ea7828321b318;hpb=dad790286e318c5c7f4b6ccd52b4fd512c87c763;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/regexec.c b/regexec.c
index 890736c..747b61b 100644
--- a/regexec.c
+++ b/regexec.c
@@ -31,28 +31,7 @@
 */
 
 #ifdef PERL_EXT_RE_BUILD
-/* need to replace pregcomp et al, so enable that */
-#  ifndef PERL_IN_XSUB_RE
-#    define PERL_IN_XSUB_RE
-#  endif
-/* need access to debugger hooks */
-#  if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING)
-#    define DEBUGGING
-#  endif
-#endif
-
-#ifdef PERL_IN_XSUB_RE
-/* We *really* need to overwrite these symbols: */
-#  define Perl_regexec_flags my_regexec
-#  define Perl_regdump my_regdump
-#  define Perl_regprop my_regprop
-#  define Perl_re_intuit_start my_re_intuit_start
-/* *These* symbols are masked to allow static link. */
-#  define Perl_pregexec my_pregexec
-#  define Perl_reginitcolors my_reginitcolors
-#  define Perl_regclass_swash my_regclass_swash
-
-#  define PERL_NO_GET_CONTEXT
+#include "re_top.h"
 #endif
 
 /*
@@ -91,7 +70,11 @@
 #define PERL_IN_REGEXEC_C
 #include "perl.h"
 
-#include "regcomp.h"
+#ifdef PERL_IN_XSUB_RE
+#  include "re_comp.h"
+#else
+#  include "regcomp.h"
+#endif
 
 #define RF_tainted	1		/* tainted information used? */
 #define RF_warned	2		/* warned about big count? */
@@ -299,6 +282,7 @@ S_regcppop(pTHX_ const regexp *rex)
  * pregexec and friends
  */
 
+#ifndef PERL_IN_XSUB_RE
 /*
  - pregexec - match a regexp against a string
  */
@@ -314,7 +298,7 @@ Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *stren
 	regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL,
 		      nosave ? 0 : REXEC_COPY_STR);
 }
-
+#endif
 
 /*
  * Need to implement the following flags for reg_anch:
@@ -2298,6 +2282,9 @@ typedef union re_unwind_t {
 /* Make sure there is a test for this +1 options in re_tests */
 #define TRIE_INITAL_ACCEPT_BUFFLEN 4;
 
+/* this value indiciates that the c1/c2 "next char" test should be skipped */
+#define CHRTEST_VOID -1000
+
 #define SLAB_FIRST(s) (&(s)->states[0])
 #define SLAB_LAST(s)  (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1])
 
@@ -2306,6 +2293,9 @@ typedef union re_unwind_t {
 STATIC regmatch_state *
 S_push_slab(pTHX)
 {
+#if PERL_VERSION < 9
+    dMY_CXT;
+#endif
     regmatch_slab *s = PL_regmatch_slab->next;
     if (!s) {
 	Newx(s, 1, regmatch_slab);
@@ -2473,6 +2463,9 @@ S_push_slab(pTHX)
 STATIC I32			/* 0 failure, 1 success */
 S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 {
+#if PERL_VERSION < 9
+    dMY_CXT;
+#endif
     dVAR;
     register const bool do_utf8 = PL_reg_match_utf8;
     const U32 uniflags = UTF8_ALLOW_DEFAULT;
@@ -3872,94 +3865,109 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		scan += NEXT_OFF(scan); /* Skip former OPEN. */
 	    PL_reginput = locinput;
 	    st->u.curlym.maxwanted = st->minmod ? st->ln : n;
-	    if (st->u.curlym.maxwanted) {
-		while (PL_reginput < PL_regeol && st->u.curlym.matches < st->u.curlym.maxwanted) {
-		    /* resume to current state on success */
-		    st->u.yes.prev_yes_state = yes_state;
-		    yes_state = st;
-		    REGMATCH(scan, CURLYM1);
-		    yes_state = st->u.yes.prev_yes_state;
-		    /*** all unsaved local vars undefined at this point */
-		    if (!result)
-			break;
-		    /* on first match, determine length, u.curlym.l */
-		    if (!st->u.curlym.matches++) {
-			if (PL_reg_match_utf8) {
-			    char *s = locinput;
-			    while (s < PL_reginput) {
-				st->u.curlym.l++;
-				s += UTF8SKIP(s);
-			    }
-			}
-			else {
-			    st->u.curlym.l = PL_reginput - locinput;
-			}
-			if (st->u.curlym.l == 0) {
-			    st->u.curlym.matches = st->u.curlym.maxwanted;
-			    break;
+	    while (PL_reginput < PL_regeol && st->u.curlym.matches < st->u.curlym.maxwanted) {
+		/* resume to current state on success */
+		st->u.yes.prev_yes_state = yes_state;
+		yes_state = st;
+		REGMATCH(scan, CURLYM1);
+		yes_state = st->u.yes.prev_yes_state;
+		/*** all unsaved local vars undefined at this point */
+		if (!result)
+		    break;
+		/* on first match, determine length, u.curlym.l */
+		if (!st->u.curlym.matches++) {
+		    if (PL_reg_match_utf8) {
+			char *s = locinput;
+			while (s < PL_reginput) {
+			    st->u.curlym.l++;
+			    s += UTF8SKIP(s);
 			}
 		    }
-		    locinput = PL_reginput;
+		    else {
+			st->u.curlym.l = PL_reginput - locinput;
+		    }
+		    if (st->u.curlym.l == 0) {
+			st->u.curlym.matches = st->u.curlym.maxwanted;
+			break;
+		    }
 		}
+		locinput = PL_reginput;
 	    }
 
 	    PL_reginput = locinput;
-
-	    if (st->minmod) {
+	    if (st->u.curlym.matches < st->ln) {
 		st->minmod = 0;
-		if (st->ln && st->u.curlym.matches < st->ln)
-		    sayNO;
-		if (HAS_TEXT(next) || JUMPABLE(next)) {
-		    regnode *text_node = next;
+		sayNO;
+	    }
 
-		    if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
+	    DEBUG_EXECUTE_r(
+		PerlIO_printf(Perl_debug_log,
+			  "%*s  matched %"IVdf" times, len=%"IVdf"...\n",
+			  (int)(REPORT_CODE_OFF+PL_regindent*2), "",
+			  (IV) st->u.curlym.matches, (IV)st->u.curlym.l)
+	    );
+
+	    /* calculate c1 and c1 for possible match of 1st char
+	     * following curly */
+	    st->u.curlym.c1 = st->u.curlym.c2 = CHRTEST_VOID;
+	    if (HAS_TEXT(next) || JUMPABLE(next)) {
+		regnode *text_node = next;
+		if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
+		if (HAS_TEXT(text_node)
+		    && PL_regkind[(U8)OP(text_node)] != REF)
+		{
+		    st->u.curlym.c1 = (U8)*STRING(text_node);
+		    st->u.curlym.c2 =
+			(OP(text_node) == EXACTF || OP(text_node) == REFF)
+			? PL_fold[st->u.curlym.c1]
+			: (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
+			    ? PL_fold_locale[st->u.curlym.c1]
+			    : st->u.curlym.c1;
+		}
+	    }
 
-		    if (! HAS_TEXT(text_node)) st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-		    else {
-			if (PL_regkind[(U8)OP(text_node)] == REF) {
-			    st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-			    goto assume_ok_MM;
+	    REGCP_SET(st->u.curlym.lastcp);
+
+	    st->u.curlym.minmod = st->minmod;
+	    st->minmod = 0;
+	    while (st->u.curlym.matches >= st->ln
+		&& (st->u.curlym.matches <= n
+		    /* for REG_INFTY, ln could overflow to negative */
+		    || (n == REG_INFTY && st->u.curlym.matches >= 0)))
+	    { 
+		/* If it could work, try it. */
+		if (st->u.curlym.c1 == CHRTEST_VOID ||
+		    UCHARAT(PL_reginput) == st->u.curlym.c1 ||
+		    UCHARAT(PL_reginput) == st->u.curlym.c2)
+		{
+		    DEBUG_EXECUTE_r(
+			PerlIO_printf(Perl_debug_log,
+			    "%*s  trying tail with matches=%"IVdf"...\n",
+			    (int)(REPORT_CODE_OFF+PL_regindent*2),
+			    "", (IV)st->u.curlym.matches)
+			);
+		    if (st->u.curlym.paren) {
+			if (st->u.curlym.matches) {
+			    PL_regstartp[st->u.curlym.paren]
+				= HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
+			    PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
 			}
-			else { st->u.curlym.c1 = (U8)*STRING(text_node); }
-			if (OP(text_node) == EXACTF || OP(text_node) == REFF)
-			    st->u.curlym.c2 = PL_fold[st->u.curlym.c1];
-			else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
-			    st->u.curlym.c2 = PL_fold_locale[st->u.curlym.c1];
 			else
-			    st->u.curlym.c2 = st->u.curlym.c1;
+			    PL_regendp[st->u.curlym.paren] = -1;
 		    }
+		    /* resume to current state on success */
+		    st->u.yes.prev_yes_state = yes_state;
+		    yes_state = st;
+		    REGMATCH(next, CURLYM2);
+		    yes_state = st->u.yes.prev_yes_state;
+		    /*** all unsaved local vars undefined at this point */
+		    if (result)
+			/* XXX tmp sayYES; */
+			sayYES_FINAL;
+		    REGCP_UNWIND(st->u.curlym.lastcp);
 		}
-		else
-		    st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-	    assume_ok_MM:
-		REGCP_SET(st->u.curlym.lastcp);
-		while (n >= st->ln || (n == REG_INFTY && st->ln > 0)) { /* ln overflow ? */
-		    /* If it could work, try it. */
-		    if (st->u.curlym.c1 == -1000 ||
-			UCHARAT(PL_reginput) == st->u.curlym.c1 ||
-			UCHARAT(PL_reginput) == st->u.curlym.c2)
-		    {
-			if (st->u.curlym.paren) {
-			    if (st->ln) {
-				PL_regstartp[st->u.curlym.paren] =
-				    HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
-				PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
-			    }
-			    else
-				PL_regendp[st->u.curlym.paren] = -1;
-			}
-			/* resume to current state on success */
-			st->u.yes.prev_yes_state = yes_state;
-			yes_state = st;
-			REGMATCH(next, CURLYM2);
-			yes_state = st->u.yes.prev_yes_state;
-			/*** all unsaved local vars undefined at this point */
-			if (result)
-			    /* XXX tmp sayYES; */
-			    sayYES_FINAL;
-			REGCP_UNWIND(st->u.curlym.lastcp);
-		    }
-		    /* Couldn't or didn't -- move forward. */
+		/* Couldn't or didn't -- move forward/backward. */
+		if (st->u.curlym.minmod) {
 		    PL_reginput = locinput;
 		    /* resume to current state on success */
 		    st->u.yes.prev_yes_state = yes_state;
@@ -3968,80 +3976,13 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		    yes_state = st->u.yes.prev_yes_state;
 		    /*** all unsaved local vars undefined at this point */
 		    if (result) {
-			st->ln++;
+			st->u.curlym.matches++;
 			locinput = PL_reginput;
 		    }
 		    else
 			sayNO;
 		}
-	    }
-	    else {
-		DEBUG_EXECUTE_r(
-		    PerlIO_printf(Perl_debug_log,
-			      "%*s  matched %"IVdf" times, len=%"IVdf"...\n",
-			      (int)(REPORT_CODE_OFF+PL_regindent*2), "",
-			      (IV) st->u.curlym.matches, (IV)st->u.curlym.l)
-		    );
-		if (st->u.curlym.matches >= st->ln) {
-		    if (HAS_TEXT(next) || JUMPABLE(next)) {
-			regnode *text_node = next;
-
-			if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
-
-			if (! HAS_TEXT(text_node)) st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-			else {
-			    if (PL_regkind[(U8)OP(text_node)] == REF) {
-				st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-				goto assume_ok_REG;
-			    }
-			    else { st->u.curlym.c1 = (U8)*STRING(text_node); }
-
-			    if (OP(text_node) == EXACTF || OP(text_node) == REFF)
-				st->u.curlym.c2 = PL_fold[st->u.curlym.c1];
-			    else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL)
-				st->u.curlym.c2 = PL_fold_locale[st->u.curlym.c1];
-			    else
-				st->u.curlym.c2 = st->u.curlym.c1;
-			}
-		    }
-		    else
-			st->u.curlym.c1 = st->u.curlym.c2 = -1000;
-		}
-	    assume_ok_REG:
-		REGCP_SET(st->u.curlym.lastcp);
-		while (st->u.curlym.matches >= st->ln) {
-		    /* If it could work, try it. */
-		    if (st->u.curlym.c1 == -1000 ||
-			UCHARAT(PL_reginput) == st->u.curlym.c1 ||
-			UCHARAT(PL_reginput) == st->u.curlym.c2)
-		    {
-			DEBUG_EXECUTE_r(
-			    PerlIO_printf(Perl_debug_log,
-				"%*s  trying tail with matches=%"IVdf"...\n",
-				(int)(REPORT_CODE_OFF+PL_regindent*2),
-				"", (IV)st->u.curlym.matches)
-			    );
-			if (st->u.curlym.paren) {
-			    if (st->u.curlym.matches) {
-				PL_regstartp[st->u.curlym.paren]
-				    = HOPc(PL_reginput, -st->u.curlym.l) - PL_bostr;
-				PL_regendp[st->u.curlym.paren] = PL_reginput - PL_bostr;
-			    }
-			    else
-				PL_regendp[st->u.curlym.paren] = -1;
-			}
-			/* resume to current state on success */
-			st->u.yes.prev_yes_state = yes_state;
-			yes_state = st;
-			REGMATCH(next, CURLYM4);
-			yes_state = st->u.yes.prev_yes_state;
-			/*** all unsaved local vars undefined at this point */
-			if (result)
-			    /* XXX tmp sayYES; */
-			    sayYES_FINAL;
-			REGCP_UNWIND(st->u.curlym.lastcp);
-		    }
-		    /* Couldn't or didn't -- back up. */
+		else {
 		    st->u.curlym.matches--;
 		    locinput = HOPc(locinput, -st->u.curlym.l);
 		    PL_reginput = locinput;
@@ -4096,10 +4037,11 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 
 		if (! HAS_TEXT(text_node)) FIND_NEXT_IMPT(text_node);
 
-		if (! HAS_TEXT(text_node)) st->u.plus.c1 = st->u.plus.c2 = -1000;
+		if (! HAS_TEXT(text_node))
+		    st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
 		else {
 		    if (PL_regkind[(U8)OP(text_node)] == REF) {
-			st->u.plus.c1 = st->u.plus.c2 = -1000;
+			st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
 			goto assume_ok_easy;
 		    }
 		    else { s = (U8*)STRING(text_node); }
@@ -4133,7 +4075,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		}
 	    }
 	    else
-		st->u.plus.c1 = st->u.plus.c2 = -1000;
+		st->u.plus.c1 = st->u.plus.c2 = CHRTEST_VOID;
 	assume_ok_easy:
 	    PL_reginput = locinput;
 	    if (st->minmod) {
@@ -4142,7 +4084,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		    sayNO;
 		locinput = PL_reginput;
 		REGCP_SET(st->u.plus.lastcp);
-		if (st->u.plus.c1 != -1000) {
+		if (st->u.plus.c1 != CHRTEST_VOID) {
 		    st->u.plus.old = locinput;
 		    st->u.plus.count = 0;
 
@@ -4191,11 +4133,11 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 				    st->u.plus.count++;
 				}
 			    } else {
-				STRLEN len;
 				/* count initialised to
 				 * utf8_distance(old, locinput) */
 				while (locinput <= st->u.plus.e) {
-				    UV c = utf8n_to_uvchr((U8*)locinput,
+				    STRLEN len;
+				    const UV c = utf8n_to_uvchr((U8*)locinput,
 							  UTF8_MAXBYTES, &len,
 							  uniflags);
 				    if (c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
@@ -4230,7 +4172,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		else
 		while (n >= st->ln || (n == REG_INFTY && st->ln > 0)) { /* ln overflow ? */
 		    UV c;
-		    if (st->u.plus.c1 != -1000) {
+		    if (st->u.plus.c1 != CHRTEST_VOID) {
 			if (do_utf8)
 			    c = utf8n_to_uvchr((U8*)PL_reginput,
 					       UTF8_MAXBYTES, 0,
@@ -4246,7 +4188,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		        }
 		    }
 		    /* If it could work, try it. */
-		    else if (st->u.plus.c1 == -1000)
+		    else if (st->u.plus.c1 == CHRTEST_VOID)
 		    {
 			TRYPAREN(st->u.plus.paren, st->ln, PL_reginput, PLUS3);
 			/*** all unsaved local vars undefined at this point */
@@ -4280,7 +4222,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 		{
 		    UV c = 0;
 		    while (n >= st->ln) {
-			if (st->u.plus.c1 != -1000) {
+			if (st->u.plus.c1 != CHRTEST_VOID) {
 			    if (do_utf8)
 				c = utf8n_to_uvchr((U8*)PL_reginput,
 						   UTF8_MAXBYTES, 0,
@@ -4289,7 +4231,7 @@ S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog)
 				c = UCHARAT(PL_reginput);
 			}
 			/* If it could work, try it. */
-			if (st->u.plus.c1 == -1000 || c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
+			if (st->u.plus.c1 == CHRTEST_VOID || c == (UV)st->u.plus.c1 || c == (UV)st->u.plus.c2)
 			    {
 				TRYPAREN(st->u.plus.paren, n, PL_reginput, PLUS4);
 				/*** all unsaved local vars undefined at this point */
@@ -4433,8 +4375,9 @@ yes_final:
 	 * pop to the state marked by yes_state and continue from there */
 
 	/*XXX tmp for CURLYM*/
-	regmatch_slab *oslab = PL_regmatch_slab;
-	regmatch_state *ost = st, *oys=yes_state;
+	regmatch_slab * const oslab = PL_regmatch_slab;
+	regmatch_state * const ost = st;
+	regmatch_state * const oys = yes_state;
 	int odepth = depth;
 
 	assert(st != yes_state);
@@ -4463,7 +4406,7 @@ yes_final:
 	    /* Restore parens of the caller without popping the
 	     * savestack */
 	    {
-		I32 tmp = PL_savestack_ix;
+		const I32 tmp = PL_savestack_ix;
 		PL_savestack_ix = st->u.eval.lastcp;
 		regcppop(rex);
 		PL_savestack_ix = tmp;
@@ -4496,7 +4439,6 @@ yes_final:
 	case resume_CURLYM1:
 	case resume_CURLYM2:
 	case resume_CURLYM3:
-	case resume_CURLYM4:
 	    PL_regmatch_slab =oslab;
 	    st = ost;
 	    PL_regmatch_state = st;
@@ -4550,8 +4492,6 @@ yes:
 	    goto resume_point_CURLYM2;
 	case resume_CURLYM3:
 	    goto resume_point_CURLYM3;
-	case resume_CURLYM4:
-	    goto resume_point_CURLYM4;
 	case resume_PLUS1:
 	    goto resume_point_PLUS1;
 	case resume_PLUS2:
@@ -4678,8 +4618,6 @@ do_no:
 	    goto resume_point_CURLYM2;
 	case resume_CURLYM3:
 	    goto resume_point_CURLYM3;
-	case resume_CURLYM4:
-	    goto resume_point_CURLYM4;
 	case resume_IFMATCH:
 	    yes_state = st->u.yes.prev_yes_state;
 	    if (st->logical) {
@@ -4718,10 +4656,10 @@ final_exit:
 
     /* free all slabs above current one */
     if (orig_slab->next) {
-	regmatch_slab *osl, *sl = orig_slab->next;
+	regmatch_slab *sl = orig_slab->next;
 	orig_slab->next = NULL;
 	while (sl) {
-	    osl = sl;
+	    regmatch_slab * const osl = sl;
 	    sl = sl->next;
 	    Safefree(osl);
 	}
@@ -4980,6 +4918,7 @@ S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max)
 }
 
 
+#ifndef PERL_IN_XSUB_RE
 /*
 - regclass_swash - prepare the utf8 swash
 */
@@ -5027,6 +4966,7 @@ Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool
 
     return sw;
 }
+#endif
 
 /*
  - reginclass - determine if a character falls into a character class