X-Git-Url: http://git.shadowcat.co.uk/gitweb/gitweb.cgi?a=blobdiff_plain;f=regcomp.c;h=b6d8b01c3fef450895223f83b90e6bc5fa576e55;hb=9e5a0491dfd6742194347115bac72d779c95ac13;hp=ae2b02ee75e06108fa57ad3b71378fa62d172d5c;hpb=fbbbcc485c1d03c76a91f998e1e4492c8ad56b38;p=p5sagit%2Fp5-mst-13.2.git

diff --git a/regcomp.c b/regcomp.c
index ae2b02e..b6d8b01 100644
--- a/regcomp.c
+++ b/regcomp.c
@@ -102,7 +102,6 @@
  * Forward declarations for pregcomp()'s friends.
  */
 
-static char* regwhite _((char *, char *));
 #ifndef PERL_OBJECT
 static regnode *reg _((I32, I32 *));
 static regnode *reganode _((U8, U32));
@@ -115,9 +114,10 @@ static regnode *reg_node _((U8));
 static regnode *regpiece _((I32 *));
 static void reginsert _((U8, regnode *));
 static void regoptail _((regnode *, regnode *));
-static void regset _((char *, I32));
 static void regtail _((regnode *, regnode *));
+static char* regwhite _((char *, char *));
 static char* nextchar _((void));
+static void re_croak2 _((const char* pat1,const char* pat2,...)) __attribute__((noreturn));
 #endif
 
 /* Length of a variant. */
@@ -150,8 +150,13 @@ static scan_data_t zero_scan_data = { 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0 };
 #define SF_FIX_BEFORE_EOL	(SF_FIX_BEFORE_SEOL|SF_FIX_BEFORE_MEOL)
 #define SF_FL_BEFORE_EOL	(SF_FL_BEFORE_SEOL|SF_FL_BEFORE_MEOL)
 
-#define SF_FIX_SHIFT_EOL	(+2)
-#define SF_FL_SHIFT_EOL		(+4)
+#ifdef NO_UNARY_PLUS
+#  define SF_FIX_SHIFT_EOL	(0+2)
+#  define SF_FL_SHIFT_EOL		(0+4)
+#else
+#  define SF_FIX_SHIFT_EOL	(+2)
+#  define SF_FL_SHIFT_EOL		(+4)
+#endif
 
 #define SF_FIX_BEFORE_SEOL	(SF_BEFORE_SEOL << SF_FIX_SHIFT_EOL)
 #define SF_FIX_BEFORE_MEOL	(SF_BEFORE_MEOL << SF_FIX_SHIFT_EOL)
@@ -791,6 +796,11 @@ pregcomp(char *exp, char *xend, PMOP *pm)
     r->regstclass = NULL;
     r->naughty = regnaughty >= 10;	/* Probably an expensive pattern. */
     scan = r->program + 1;		/* First BRANCH. */
+
+    /* XXXX To minimize changes to RE engine we always allocate
+       3-units-long substrs field. */
+    Newz(1004, r->substrs, 1, struct reg_substr_data);
+
     if (OP(scan) != BRANCH) {	/* Only one top-level choice. */
 	scan_data_t data;
 	I32 fake;
@@ -885,7 +895,7 @@ pregcomp(char *exp, char *xend, PMOP *pm)
 	    r->float_substr = data.longest_float;
 	    r->float_min_offset = data.offset_float_min;
 	    r->float_max_offset = data.offset_float_max;
-	    fbm_compile(r->float_substr);
+	    fbm_compile(r->float_substr, 0);
 	    BmUSEFUL(r->float_substr) = 100;
 	    if (data.flags & SF_FL_BEFORE_EOL /* Cannot have SEOL and MULTI */
 		&& (!(data.flags & SF_FL_BEFORE_MEOL)
@@ -905,7 +915,7 @@ pregcomp(char *exp, char *xend, PMOP *pm)
 		    || (regflags & PMf_MULTILINE)))) {
 	    r->anchored_substr = data.longest_fixed;
 	    r->anchored_offset = data.offset_fixed;
-	    fbm_compile(r->anchored_substr);
+	    fbm_compile(r->anchored_substr, 0);
 	    BmUSEFUL(r->anchored_substr) = 100;
 	    if (data.flags & SF_FIX_BEFORE_EOL /* Cannot have SEOL and MULTI */
 		&& (!(data.flags & SF_FIX_BEFORE_MEOL)
@@ -1106,8 +1116,11 @@ reg(I32 paren, I32 *flagp)
                 break;
 	    default:
 		--regparse;
-		while (*regparse && strchr("iogcmsx", *regparse))
-		    pmflag(&regflags, *regparse++);
+		while (*regparse && strchr("iogcmsx", *regparse)) {
+		    if (*regparse != 'o')
+			pmflag(&regflags, *regparse);
+		    ++regparse;
+		}
 	      unknown:
 		if (*regparse != ')')
 		    FAIL2("Sequence (?%c...) not recognized", *regparse);
@@ -1784,7 +1797,7 @@ tryagain:
     return(ret);
 }
 
-static char *
+STATIC char *
 regwhite(char *p, char *e)
 {
     while (p < e) {
@@ -1801,15 +1814,6 @@ regwhite(char *p, char *e)
     return p;
 }
 
-STATIC void
-regset(char *opnd, register I32 c)
-{
-    if (SIZE_ONLY)
-	return;
-    c &= 0xFF;
-    opnd[1 + (c >> 3)] |= (1 << (c & 7));
-}
-
 STATIC regnode *
 regclass(void)
 {
@@ -1873,63 +1877,67 @@ regclass(void)
 	    Class = UCHARAT(regparse++);
 	    switch (Class) {
 	    case 'w':
-		if (regflags & PMf_LOCALE) {
-		    if (!SIZE_ONLY)
+		if (!SIZE_ONLY) {
+		    if (regflags & PMf_LOCALE)
 			*opnd |= ANYOF_ALNUML;
-		}
-		else {
-		    for (Class = 0; Class < 256; Class++)
-			if (isALNUM(Class))
-			    regset(opnd, Class);
+		    else {
+			for (Class = 0; Class < 256; Class++)
+			    if (isALNUM(Class))
+				ANYOF_SET(opnd, Class);
+		    }
 		}
 		lastclass = 1234;
 		continue;
 	    case 'W':
-		if (regflags & PMf_LOCALE) {
-		    if (!SIZE_ONLY)
+		if (!SIZE_ONLY) {
+		    if (regflags & PMf_LOCALE)
 			*opnd |= ANYOF_NALNUML;
-		}
-		else {
-		    for (Class = 0; Class < 256; Class++)
-			if (!isALNUM(Class))
-			    regset(opnd, Class);
+		    else {
+			for (Class = 0; Class < 256; Class++)
+			    if (!isALNUM(Class))
+				ANYOF_SET(opnd, Class);
+		    }
 		}
 		lastclass = 1234;
 		continue;
 	    case 's':
-		if (regflags & PMf_LOCALE) {
-		    if (!SIZE_ONLY)
+		if (!SIZE_ONLY) {
+		    if (regflags & PMf_LOCALE)
 			*opnd |= ANYOF_SPACEL;
-		}
-		else {
-		    for (Class = 0; Class < 256; Class++)
-			if (isSPACE(Class))
-			    regset(opnd, Class);
+		    else {
+			for (Class = 0; Class < 256; Class++)
+			    if (isSPACE(Class))
+				ANYOF_SET(opnd, Class);
+		    }
 		}
 		lastclass = 1234;
 		continue;
 	    case 'S':
-		if (regflags & PMf_LOCALE) {
-		    if (!SIZE_ONLY)
+		if (!SIZE_ONLY) {
+		    if (regflags & PMf_LOCALE)
 			*opnd |= ANYOF_NSPACEL;
-		}
-		else {
-		    for (Class = 0; Class < 256; Class++)
-			if (!isSPACE(Class))
-			    regset(opnd, Class);
+		    else {
+			for (Class = 0; Class < 256; Class++)
+			    if (!isSPACE(Class))
+				ANYOF_SET(opnd, Class);
+		    }
 		}
 		lastclass = 1234;
 		continue;
 	    case 'd':
-		for (Class = '0'; Class <= '9'; Class++)
-		    regset(opnd, Class);
+		if (!SIZE_ONLY) {
+		    for (Class = '0'; Class <= '9'; Class++)
+			ANYOF_SET(opnd, Class);
+		}
 		lastclass = 1234;
 		continue;
 	    case 'D':
-		for (Class = 0; Class < '0'; Class++)
-		    regset(opnd, Class);
-		for (Class = '9' + 1; Class < 256; Class++)
-		    regset(opnd, Class);
+		if (!SIZE_ONLY) {
+		    for (Class = 0; Class < '0'; Class++)
+			ANYOF_SET(opnd, Class);
+		    for (Class = '9' + 1; Class < 256; Class++)
+			ANYOF_SET(opnd, Class);
+		}
 		lastclass = 1234;
 		continue;
 	    case 'n':
@@ -1982,13 +1990,31 @@ regclass(void)
 		continue;	/* do it next time */
 	    }
 	}
-	for ( ; lastclass <= Class; lastclass++)
-	    regset(opnd, lastclass);
+	if (!SIZE_ONLY) {
+	    for ( ; lastclass <= Class; lastclass++)
+		ANYOF_SET(opnd, lastclass);
+	}
 	lastclass = Class;
     }
     if (*regparse != ']')
 	FAIL("unmatched [] in regexp");
     nextchar();
+    /* optimize case-insensitive simple patterns (e.g. /[a-z]/i) */
+    if (!SIZE_ONLY && (*opnd & (0xFF ^ ANYOF_INVERT)) == ANYOF_FOLD) {
+	for (Class = 0; Class < 256; ++Class) {
+	    if (ANYOF_TEST(opnd, Class)) {
+		I32 cf = fold[Class];
+		ANYOF_SET(opnd, cf);
+	    }
+	}
+	*opnd &= ~ANYOF_FOLD;
+    }
+    /* optimize inverted simple patterns (e.g. [^a-z]) */
+    if (!SIZE_ONLY && (*opnd & 0xFF) == ANYOF_INVERT) {
+	for (Class = 0; Class < 32; ++Class)
+	    opnd[1 + Class] ^= 0xFF;
+	*opnd = 0;
+    }
     return ret;
 }
 
@@ -2205,11 +2231,11 @@ regcurly(register char *s)
     return TRUE;
 }
 
-#ifdef DEBUGGING
 
 STATIC regnode *
 dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
 {
+#ifdef DEBUGGING
     register char op = EXACT;	/* Arbitrary non-END op. */
     register regnode *next, *onode;
 
@@ -2266,6 +2292,7 @@ dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
 	else if (op == WHILEM)
 	    l--;
     }
+#endif	/* DEBUGGING */
     return node;
 }
 
@@ -2275,6 +2302,7 @@ dumpuntil(regnode *start, regnode *node, regnode *last, SV* sv, I32 l)
 void
 regdump(regexp *r)
 {
+#ifdef DEBUGGING
     SV *sv = sv_newmortal();
 
     (void)dumpuntil(r->program, r->program + 1, NULL, sv, 0);
@@ -2327,6 +2355,7 @@ regdump(regexp *r)
 	PerlIO_printf(Perl_debug_log, "implicit ");
     PerlIO_printf(Perl_debug_log, "minlen %ld ", (long) r->minlen);
     PerlIO_printf(Perl_debug_log, "\n");
+#endif	/* DEBUGGING */
 }
 
 /*
@@ -2335,6 +2364,7 @@ regdump(regexp *r)
 void
 regprop(SV *sv, regnode *o)
 {
+#ifdef DEBUGGING
     register char *p = 0;
 
     sv_setpv(sv, ":");
@@ -2532,8 +2562,8 @@ regprop(SV *sv, regnode *o)
     }
     if (p)
 	sv_catpv(sv, p);
+#endif	/* DEBUGGING */
 }
-#endif /* DEBUGGING */
 
 void
 pregfree(struct regexp *r)
@@ -2544,10 +2574,13 @@ pregfree(struct regexp *r)
 	Safefree(r->precomp);
     if (r->subbase)
 	Safefree(r->subbase);
-    if (r->anchored_substr)
-	SvREFCNT_dec(r->anchored_substr);
-    if (r->float_substr)
-	SvREFCNT_dec(r->float_substr);
+    if (r->substrs) {
+	if (r->anchored_substr)
+	    SvREFCNT_dec(r->anchored_substr);
+	if (r->float_substr)
+	    SvREFCNT_dec(r->float_substr);
+	Safefree(r->substrs);
+    }
     if (r->data) {
 	int n = r->data->count;
 	while (--n >= 0) {
@@ -2600,17 +2633,8 @@ regnext(register regnode *p)
 #endif
 }
 
-#ifdef I_STDARG
-void	
+STATIC void	
 re_croak2(const char* pat1,const char* pat2,...)
-#else
-/*VARARGS0*/
-void	
-re_croak2(const char* pat1,const char* pat2, va_alist)
-    const char* pat1;
-    const char* pat2;
-    va_dcl
-#endif 
 {
     va_list args;
     STRLEN l1 = strlen(pat1);
@@ -2626,11 +2650,7 @@ re_croak2(const char* pat1,const char* pat2, va_alist)
     Copy(pat2, buf + l1, l2 , char);
     buf[l1 + l2 + 1] = '\n';
     buf[l1 + l2 + 2] = '\0';
-#ifdef I_STDARG
     va_start(args, pat2);
-#else
-    va_start(args);
-#endif
     message = mess(buf, &args);
     va_end(args);
     l1 = strlen(message);