5.8.4/distrib/regexec.c

110 #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off))
111 #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off))
112 #define HOP(pos,off) (PL_reg_match_utf8 ? reghop((U8*)pos, off) : (U8*)(pos + off))
113 #define HOPMAYBE(pos,off) (PL_reg_match_utf8 ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off))
119 	? reghopmaybe((U8*)pos, -off)	\
121 	? (U8*)(pos - off)		\
122     : (U8*)NULL				\
126 #define reghop3_c(pos,off,lim) ((char*)reghop3((U8*)pos, off, (U8*)lim))
127 #define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim))
128 #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
129 #define HOPMAYBE3(pos,off,lim) (PL_reg_match_utf8 ? reghopmaybe3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off))
133 #define LOAD_UTF8_CHARCLASS(a,b) STMT_START { if (!CAT2(PL_utf8_,a)) { ENTER; save_re_context(); (void)CAT2(is_utf8_, a)((U8*)b); LEAVE; } } STMT_END
140     (PL_regkind[(U8)OP(rn)] == CURLY && ARG1(rn) > 0) \
144     PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \
153 	if (OP(rn) == SUSPEND || PL_regkind[(U8)OP(rn)] == CURLY) \
512 	    I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end;
554 	s = fbm_instr((U8*)(s + start_shift),
555 		      (U8*)(strend - end_shift),
672 	    if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset)
844 	U8* str = (U8*)STRING(prog->regstclass);
845 	int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT
970 			  !UTF8_IS_INVARIANT((U8)s[0]) ?
971 			  reginclass(c, (U8*)s, 0, do_utf8) :
972 			  REGINCLASS(c, (U8*)s)) {
987 		      if (REGINCLASS(c, (U8*)s) ||
1018 		U8 *sm = (U8 *) m;
1019 		U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
1020 		U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
1022 		to_utf8_lower((U8*)m, tmpbuf1, &ulen1);
1023 		to_utf8_upper((U8*)m, tmpbuf2, &ulen2);
1030 		while (sm < ((U8 *) m + ln)) {
1036 		c1 = *(U8*)m;
1044 	    c1 = *(U8*)m;
1065 	        U8 tmpbuf [UTF8_MAXLEN+1];
1066 		U8 foldbuf[UTF8_MAXLEN_FOLD+1];
1073 		        c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &len,
1100 		      c = utf8n_to_uvchr((U8*)s, UTF8_MAXLEN, &len,
1141 			if ( *(U8*)s == c1
1151 			if ( (*(U8*)s == c1 || *(U8*)s == c2)
1169 		    U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1178 				 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1179 				 isALNUM_LC_utf8((U8*)s)))
1212 		    U8 *r = reghop3((U8*)s, -1, (U8*)PL_bostr);
1221 				 swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) :
1222 				 isALNUM_LC_utf8((U8*)s)))
1249 		    if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
1278 		    if (isALNUM_LC_utf8((U8*)s)) {
1307 		    if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) {
1336 		    if (!isALNUM_LC_utf8((U8*)s)) {
1365 		    if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) {
1394 		    if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) {
1423 		    if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) {
1452 		    if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) {
1481 		    if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
1510 		    if (isDIGIT_LC_utf8((U8*)s)) {
1539 		    if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) {
1568 		    if (!isDIGIT_LC_utf8((U8*)s)) {
1715 	   pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60,
1920 	    I32 op = (U8)OP(prog->regstclass);
1934 	      pv_uni_display(dsv0, (U8*)SvPVX(prop), SvCUR(prop), 60,
2298 	    while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len)))
2304 	    while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l)))
2314 		pv_uni_display(dsv0, (U8*)(locinput - pref_len),
2319 		pv_uni_display(dsv1, (U8*)(locinput - pref_len + pref0_len),
2324 		pv_uni_display(dsv2, (U8*)locinput,
2440 			if (NATIVE_TO_UNI(*(U8*)s) !=
2441 			    utf8n_to_uvuni((U8*)l, UTF8_MAXLEN, &ulen,
2454 			if (NATIVE_TO_UNI(*((U8*)l)) !=
2455 			    utf8n_to_uvuni((U8*)s, UTF8_MAXLEN, &ulen,
2500 			   (U8)l[0] == 0xC3 &&
2502 			   (U8)l[1] == 0x9F))
2530 	        if (!reginclass(scan, (U8*)locinput, &inclasslen, do_utf8))
2541 		if (!REGINCLASS(scan, (U8*)locinput))
2568 		      ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
2569 		      : isALNUM_LC_utf8((U8*)locinput)))
2591 		    ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8)
2592 		    : isALNUM_LC_utf8((U8*)locinput))
2616 		    U8 *r = reghop3((U8*)locinput, -1, (U8*)PL_bostr);
2623 		    n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8);
2627 		    n = isALNUM_LC_utf8((U8*)locinput);
2656 			  ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
2657 			  : isSPACE_LC_utf8((U8*)locinput)))
2686 		    ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8)
2687 		    : isSPACE_LC_utf8((U8*)locinput))
2709 		      ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
2710 		      : isDIGIT_LC_utf8((U8*)locinput)))
2732 		    ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8)
2733 		    : isDIGIT_LC_utf8((U8*)locinput))
2751 		if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
2755 		       swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8))
2788 		    U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
2789 		    U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
2793 			toLOWER_utf8((U8*)s, tmpbuf1, &ulen1);
2794 			toLOWER_utf8((U8*)l, tmpbuf2, &ulen2);
3394 			if (PL_regkind[(U8)OP(text_node)] == REF) {
3398 			else { c1 = (U8)*STRING(text_node); }
3465 			    if (PL_regkind[(U8)OP(text_node)] == REF) {
3469 			    else { c1 = (U8)*STRING(text_node); }
3556 		U8 *s;
3563 		    if (PL_regkind[(U8)OP(text_node)] == REF) {
3567 		    else { s = (U8*)STRING(text_node); }
3579 			     U8 tmpbuf1[UTF8_MAXLEN_UCLC+1];
3580 			     U8 tmpbuf2[UTF8_MAXLEN_UCLC+1];
3582 			     to_utf8_lower((U8*)s, tmpbuf1, &ulen1);
3583 			     to_utf8_upper((U8*)s, tmpbuf2, &ulen2);
3619 			    while (UTF8_IS_CONTINUATION(*(U8*)e))
3654 				       utf8n_to_uvchr((U8*)locinput,
3665 				    UV c = utf8n_to_uvchr((U8*)locinput,
3702 			    c = utf8n_to_uvchr((U8*)PL_reginput,
3735 		if (ln < n && PL_regkind[(U8)OP(next)] == EOL &&
3752 				c = utf8n_to_uvchr((U8*)PL_reginput,
3775 				c = utf8n_to_uvchr((U8*)PL_reginput,
4056 	c = (U8)*STRING(p);
4061 	c = (U8)*STRING(p);
4068 	c = (U8)*STRING(p);
4077 		   reginclass(p, (U8*)scan, 0, do_utf8)) {
4082 	    while (scan < loceol && REGINCLASS(p, (U8*)scan))
4091 		   swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
4105 		   isALNUM_LC_utf8((U8*)scan)) {
4119 		   !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) {
4133 		   !isALNUM_LC_utf8((U8*)scan)) {
4148 		    swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
4162 		   (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
4177 		     swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) {
4191 		   !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) {
4205 		   swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
4219 		   !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) {
4360 S_reginclass(pTHX_ register regnode *n, register U8* p, STRLEN* lenp, register bool do_utf8)
4406 		        U8 tmpbuf[UTF8_MAXLEN_FOLD+1];
4423 	    U8 f;
4478 STATIC U8 *
4479 S_reghop(pTHX_ U8 *s, I32 off)
4481     return S_reghop3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr));
4484 STATIC U8 *
4485 S_reghop3(pTHX_ U8 *s, I32 off, U8* lim)
4498 		    while (s > (U8*)lim && UTF8_IS_CONTINUATION(*s))
4508 STATIC U8 *
4509 S_reghopmaybe(pTHX_ U8 *s, I32 off)
4511     return S_reghopmaybe3(aTHX_ s, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr));
4514 STATIC U8 *
4515 S_reghopmaybe3(pTHX_ U8* s, I32 off, U8* lim)
4530 		    while (s > (U8*)lim && UTF8_IS_CONTINUATION(*s))