Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regexec.c |
2 | */ | |
3 | ||
4 | /* | |
5 | * "One Ring to rule them all, One Ring to find them..." | |
6 | */ | |
7 | ||
61296642 DM |
8 | /* This file contains functions for executing a regular expression. See |
9 | * also regcomp.c which funnily enough, contains functions for compiling | |
166f8a29 | 10 | * a regular expression. |
e4a054ea DM |
11 | * |
12 | * This file is also copied at build time to ext/re/re_exec.c, where | |
13 | * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT. | |
14 | * This causes the main functions to be compiled under new names and with | |
15 | * debugging support added, which makes "use re 'debug'" work. | |
166f8a29 DM |
16 | */ |
17 | ||
a687059c LW |
18 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
19 | * confused with the original package (see point 3 below). Thanks, Henry! | |
20 | */ | |
21 | ||
22 | /* Additional note: this code is very heavily munged from Henry's version | |
23 | * in places. In some spots I've traded clarity for efficiency, so don't | |
24 | * blame Henry for some of the lack of readability. | |
25 | */ | |
26 | ||
e50aee73 AD |
27 | /* The names of the functions have been changed from regcomp and |
28 | * regexec to pregcomp and pregexec in order to avoid conflicts | |
29 | * with the POSIX routines of the same names. | |
30 | */ | |
31 | ||
b9d5759e | 32 | #ifdef PERL_EXT_RE_BUILD |
54df2634 | 33 | #include "re_top.h" |
9041c2e3 | 34 | #endif |
56953603 | 35 | |
a687059c | 36 | /* |
e50aee73 | 37 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c LW |
38 | * |
39 | * Copyright (c) 1986 by University of Toronto. | |
40 | * Written by Henry Spencer. Not derived from licensed software. | |
41 | * | |
42 | * Permission is granted to anyone to use this software for any | |
43 | * purpose on any computer system, and to redistribute it freely, | |
44 | * subject to the following restrictions: | |
45 | * | |
46 | * 1. The author is not responsible for the consequences of use of | |
47 | * this software, no matter how awful, even if they arise | |
48 | * from defects in it. | |
49 | * | |
50 | * 2. The origin of this software must not be misrepresented, either | |
51 | * by explicit claim or by omission. | |
52 | * | |
53 | * 3. Altered versions must be plainly marked as such, and must not | |
54 | * be misrepresented as being the original software. | |
55 | * | |
56 | **** Alterations to Henry's code are... | |
57 | **** | |
4bb101f2 | 58 | **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
fdf8c088 | 59 | **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 by Larry Wall and others |
a687059c | 60 | **** |
9ef589d8 LW |
61 | **** You may distribute under the terms of either the GNU General Public |
62 | **** License or the Artistic License, as specified in the README file. | |
a687059c LW |
63 | * |
64 | * Beware that some of this code is subtly aware of the way operator | |
65 | * precedence is structured in regular expressions. Serious changes in | |
66 | * regular-expression syntax might require a total rethink. | |
67 | */ | |
68 | #include "EXTERN.h" | |
864dbfa3 | 69 | #define PERL_IN_REGEXEC_C |
a687059c | 70 | #include "perl.h" |
0f5d15d6 | 71 | |
54df2634 NC |
72 | #ifdef PERL_IN_XSUB_RE |
73 | # include "re_comp.h" | |
74 | #else | |
75 | # include "regcomp.h" | |
76 | #endif | |
a687059c | 77 | |
c277df42 IZ |
78 | #define RF_tainted 1 /* tainted information used? */ |
79 | #define RF_warned 2 /* warned about big count? */ | |
faec1544 | 80 | |
ab3bbdeb | 81 | #define RF_utf8 8 /* Pattern contains multibyte chars? */ |
a0ed51b3 | 82 | |
eb160463 | 83 | #define UTF ((PL_reg_flags & RF_utf8) != 0) |
ce862d02 IZ |
84 | |
85 | #define RS_init 1 /* eval environment created */ | |
86 | #define RS_set 2 /* replsv value is set */ | |
c277df42 | 87 | |
a687059c LW |
88 | #ifndef STATIC |
89 | #define STATIC static | |
90 | #endif | |
91 | ||
32fc9b6a | 92 | #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c))) |
7d3e948e | 93 | |
c277df42 IZ |
94 | /* |
95 | * Forwards. | |
96 | */ | |
97 | ||
33b8afdf | 98 | #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv)) |
53c4c00c | 99 | #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b) |
a0ed51b3 | 100 | |
3dab1dad YO |
101 | #define HOPc(pos,off) \ |
102 | (char *)(PL_reg_match_utf8 \ | |
52657f30 | 103 | ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \ |
3dab1dad YO |
104 | : (U8*)(pos + off)) |
105 | #define HOPBACKc(pos, off) \ | |
07be1b83 YO |
106 | (char*)(PL_reg_match_utf8\ |
107 | ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \ | |
108 | : (pos - off >= PL_bostr) \ | |
8e11feef | 109 | ? (U8*)pos - off \ |
3dab1dad | 110 | : NULL) |
efb30f32 | 111 | |
e7409c1b | 112 | #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off)) |
1aa99e6b | 113 | #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) |
1aa99e6b | 114 | |
1a4fad37 AL |
115 | #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \ |
116 | if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END | |
117 | #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a") | |
118 | #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0") | |
119 | #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ") | |
120 | #define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86") | |
51371543 | 121 | |
3dab1dad YO |
122 | /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */ |
123 | ||
5f80c4cf | 124 | /* for use after a quantifier and before an EXACT-like node -- japhy */ |
e2d8ce26 JP |
125 | #define JUMPABLE(rn) ( \ |
126 | OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \ | |
cca55fe3 JP |
127 | OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ |
128 | OP(rn) == PLUS || OP(rn) == MINMOD || \ | |
ee9b8eae | 129 | OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \ |
3dab1dad | 130 | (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \ |
e2d8ce26 | 131 | ) |
ee9b8eae | 132 | #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT) |
e2d8ce26 | 133 | |
ee9b8eae YO |
134 | #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF ) |
135 | ||
136 | #if 0 | |
137 | /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so | |
138 | we don't need this definition. */ | |
139 | #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF ) | |
140 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF ) | |
141 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL ) | |
142 | ||
143 | #else | |
144 | /* ... so we use this as its faster. */ | |
145 | #define IS_TEXT(rn) ( OP(rn)==EXACT ) | |
146 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF ) | |
147 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL ) | |
148 | ||
149 | #endif | |
e2d8ce26 | 150 | |
a84d97b6 HS |
151 | /* |
152 | Search for mandatory following text node; for lookahead, the text must | |
153 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
154 | */ | |
cca55fe3 | 155 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
3dab1dad YO |
156 | while (JUMPABLE(rn)) { \ |
157 | const OPCODE type = OP(rn); \ | |
158 | if (type == SUSPEND || PL_regkind[type] == CURLY) \ | |
e2d8ce26 | 159 | rn = NEXTOPER(NEXTOPER(rn)); \ |
3dab1dad | 160 | else if (type == PLUS) \ |
cca55fe3 | 161 | rn = NEXTOPER(rn); \ |
3dab1dad | 162 | else if (type == IFMATCH) \ |
a84d97b6 | 163 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ |
e2d8ce26 | 164 | else rn += NEXT_OFF(rn); \ |
3dab1dad | 165 | } \ |
5f80c4cf | 166 | } STMT_END |
74750237 | 167 | |
c476f425 | 168 | |
acfe0abc | 169 | static void restore_pos(pTHX_ void *arg); |
51371543 | 170 | |
76e3520e | 171 | STATIC CHECKPOINT |
cea2e8a9 | 172 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 173 | { |
97aff369 | 174 | dVAR; |
a3b680e6 | 175 | const int retval = PL_savestack_ix; |
b1ce53c5 | 176 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 177 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 178 | int p; |
40a82448 | 179 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 180 | |
e49a9654 IH |
181 | if (paren_elems_to_push < 0) |
182 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
183 | ||
6bda09f9 | 184 | #define REGCP_OTHER_ELEMS 8 |
4b3c1a47 | 185 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
7f69552c | 186 | |
3280af22 | 187 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 188 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
189 | SSPUSHINT(PL_regendp[p]); |
190 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 191 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 192 | SSPUSHINT(p); |
40a82448 DM |
193 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
194 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", | |
195 | (UV)p, (IV)PL_regstartp[p], | |
196 | (IV)(PL_reg_start_tmp[p] - PL_bostr), | |
197 | (IV)PL_regendp[p] | |
198 | )); | |
a0d0e21e | 199 | } |
b1ce53c5 | 200 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
6bda09f9 YO |
201 | SSPUSHPTR(PL_regstartp); |
202 | SSPUSHPTR(PL_regendp); | |
3280af22 NIS |
203 | SSPUSHINT(PL_regsize); |
204 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 205 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 206 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
207 | #define REGCP_FRAME_ELEMS 2 |
208 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
209 | * are needed for the regexp context stack bookkeeping. */ | |
210 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 211 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 212 | |
a0d0e21e LW |
213 | return retval; |
214 | } | |
215 | ||
c277df42 | 216 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
217 | #define REGCP_SET(cp) \ |
218 | DEBUG_STATE_r( \ | |
ab3bbdeb | 219 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 220 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
221 | (IV)PL_savestack_ix)); \ |
222 | cp = PL_savestack_ix | |
c3464db5 | 223 | |
ab3bbdeb | 224 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 225 | DEBUG_STATE_r( \ |
ab3bbdeb | 226 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
227 | PerlIO_printf(Perl_debug_log, \ |
228 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
229 | (IV)(cp), (IV)PL_savestack_ix)); \ |
230 | regcpblow(cp) | |
c277df42 | 231 | |
76e3520e | 232 | STATIC char * |
097eb12c | 233 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 234 | { |
97aff369 | 235 | dVAR; |
3b6647e0 | 236 | U32 i; |
a0d0e21e | 237 | char *input; |
b1ce53c5 | 238 | |
a3621e74 YO |
239 | GET_RE_DEBUG_FLAGS_DECL; |
240 | ||
b1ce53c5 | 241 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 242 | i = SSPOPINT; |
b1ce53c5 JH |
243 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
244 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 245 | input = (char *) SSPOPPTR; |
a01268b5 | 246 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
247 | *PL_reglastparen = SSPOPINT; |
248 | PL_regsize = SSPOPINT; | |
6bda09f9 YO |
249 | PL_regendp=(I32 *) SSPOPPTR; |
250 | PL_regstartp=(I32 *) SSPOPPTR; | |
b1ce53c5 | 251 | |
6bda09f9 | 252 | |
b1ce53c5 | 253 | /* Now restore the parentheses context. */ |
41123dfd JH |
254 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
255 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 256 | I32 tmps; |
097eb12c | 257 | U32 paren = (U32)SSPOPINT; |
3280af22 | 258 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
259 | PL_regstartp[paren] = SSPOPINT; |
260 | tmps = SSPOPINT; | |
3280af22 NIS |
261 | if (paren <= *PL_reglastparen) |
262 | PL_regendp[paren] = tmps; | |
a3621e74 | 263 | DEBUG_EXECUTE_r( |
c3464db5 | 264 | PerlIO_printf(Perl_debug_log, |
b900a521 | 265 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 266 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 267 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 268 | (IV)PL_regendp[paren], |
3280af22 | 269 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 270 | ); |
a0d0e21e | 271 | } |
a3621e74 | 272 | DEBUG_EXECUTE_r( |
bb7a0f54 | 273 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 274 | PerlIO_printf(Perl_debug_log, |
faccc32b | 275 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 276 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
277 | } |
278 | ); | |
daf18116 | 279 | #if 1 |
dafc8851 JH |
280 | /* It would seem that the similar code in regtry() |
281 | * already takes care of this, and in fact it is in | |
282 | * a better location to since this code can #if 0-ed out | |
283 | * but the code in regtry() is needed or otherwise tests | |
284 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
285 | * (as of patchlevel 7877) will fail. Then again, |
286 | * this code seems to be necessary or otherwise | |
287 | * building DynaLoader will fail: | |
288 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
289 | * --jhi */ | |
3b6647e0 | 290 | for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { |
097eb12c AL |
291 | if (i > PL_regsize) |
292 | PL_regstartp[i] = -1; | |
293 | PL_regendp[i] = -1; | |
a0d0e21e | 294 | } |
dafc8851 | 295 | #endif |
a0d0e21e LW |
296 | return input; |
297 | } | |
298 | ||
02db2b7b | 299 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 300 | |
a687059c | 301 | /* |
e50aee73 | 302 | * pregexec and friends |
a687059c LW |
303 | */ |
304 | ||
76234dfb | 305 | #ifndef PERL_IN_XSUB_RE |
a687059c | 306 | /* |
c277df42 | 307 | - pregexec - match a regexp against a string |
a687059c | 308 | */ |
c277df42 | 309 | I32 |
864dbfa3 | 310 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 311 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
312 | /* strend: pointer to null at end of string */ |
313 | /* strbeg: real beginning of string */ | |
314 | /* minend: end of match must be >=minend after stringarg. */ | |
315 | /* nosave: For optimizations. */ | |
316 | { | |
317 | return | |
9041c2e3 | 318 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
319 | nosave ? 0 : REXEC_COPY_STR); |
320 | } | |
76234dfb | 321 | #endif |
22e551b9 | 322 | |
9041c2e3 | 323 | /* |
cad2e5aa JH |
324 | * Need to implement the following flags for reg_anch: |
325 | * | |
326 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
327 | * USE_INTUIT_ML | |
328 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
329 | * INTUIT_AUTORITATIVE_ML | |
330 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
331 | * INTUIT_ONCE_ML | |
332 | * | |
333 | * Another flag for this function: SECOND_TIME (so that float substrs | |
334 | * with giant delta may be not rechecked). | |
335 | */ | |
336 | ||
337 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
338 | ||
3f7c398e | 339 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
340 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
341 | ||
342 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
343 | ||
6eb5f6b9 JH |
344 | /* XXXX Some places assume that there is a fixed substring. |
345 | An update may be needed if optimizer marks as "INTUITable" | |
346 | RExen without fixed substrings. Similarly, it is assumed that | |
347 | lengths of all the strings are no more than minlen, thus they | |
348 | cannot come from lookahead. | |
40d049e4 YO |
349 | (Or minlen should take into account lookahead.) |
350 | NOTE: Some of this comment is not correct. minlen does now take account | |
351 | of lookahead/behind. Further research is required. -- demerphq | |
352 | ||
353 | */ | |
6eb5f6b9 | 354 | |
2c2d71f5 JH |
355 | /* A failure to find a constant substring means that there is no need to make |
356 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
357 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
358 | regtry() should be needed. |
359 | ||
360 | REx compiler's optimizer found 4 possible hints: | |
361 | a) Anchored substring; | |
362 | b) Fixed substring; | |
363 | c) Whether we are anchored (beginning-of-line or \G); | |
364 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 365 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
366 | string which does not contradict any of them. |
367 | */ | |
2c2d71f5 | 368 | |
6eb5f6b9 JH |
369 | /* Most of decisions we do here should have been done at compile time. |
370 | The nodes of the REx which we used for the search should have been | |
371 | deleted from the finite automaton. */ | |
372 | ||
cad2e5aa JH |
373 | char * |
374 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
375 | char *strend, U32 flags, re_scream_pos_data *data) | |
376 | { | |
97aff369 | 377 | dVAR; |
b7953727 | 378 | register I32 start_shift = 0; |
cad2e5aa | 379 | /* Should be nonnegative! */ |
b7953727 | 380 | register I32 end_shift = 0; |
2c2d71f5 JH |
381 | register char *s; |
382 | register SV *check; | |
a1933d95 | 383 | char *strbeg; |
cad2e5aa | 384 | char *t; |
1de06328 | 385 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 386 | I32 ml_anch; |
bd61b366 SS |
387 | register char *other_last = NULL; /* other substr checked before this */ |
388 | char *check_at = NULL; /* check substr found at this pos */ | |
bbe252da | 389 | const I32 multiline = prog->extflags & RXf_PMf_MULTILINE; |
f8fc2ecf | 390 | RXi_GET_DECL(prog,progi); |
30944b6d | 391 | #ifdef DEBUGGING |
890ce7af | 392 | const char * const i_strpos = strpos; |
30944b6d | 393 | #endif |
a3621e74 YO |
394 | |
395 | GET_RE_DEBUG_FLAGS_DECL; | |
396 | ||
a30b2f1f | 397 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 398 | |
bbe252da | 399 | if (prog->extflags & RXf_UTF8) { |
b8d68ded JH |
400 | PL_reg_flags |= RF_utf8; |
401 | } | |
ab3bbdeb YO |
402 | DEBUG_EXECUTE_r( |
403 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
404 | sv ? "Guessing start of match in sv for" |
405 | : "Guessing start of match in string for"); | |
2a782b5b | 406 | ); |
cad2e5aa | 407 | |
c344f387 JH |
408 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
409 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 410 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 411 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 412 | goto fail; |
2c2d71f5 | 413 | } |
1de06328 | 414 | |
a1933d95 | 415 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 416 | PL_regeol = strend; |
33b8afdf JH |
417 | if (do_utf8) { |
418 | if (!prog->check_utf8 && prog->check_substr) | |
419 | to_utf8_substr(prog); | |
420 | check = prog->check_utf8; | |
421 | } else { | |
422 | if (!prog->check_substr && prog->check_utf8) | |
423 | to_byte_substr(prog); | |
424 | check = prog->check_substr; | |
425 | } | |
1de06328 | 426 | if (check == &PL_sv_undef) { |
a3621e74 | 427 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 428 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
429 | goto fail; |
430 | } | |
bbe252da YO |
431 | if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */ |
432 | ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE) | |
433 | || ( (prog->extflags & RXf_ANCH_BOL) | |
7fba1cd6 | 434 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 435 | |
7e25d62c | 436 | if (!ml_anch) { |
bbe252da YO |
437 | if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */ |
438 | && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */ | |
3f7c398e | 439 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
440 | && sv && !SvROK(sv) |
441 | && (strpos != strbeg)) { | |
a3621e74 | 442 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
443 | goto fail; |
444 | } | |
445 | if (prog->check_offset_min == prog->check_offset_max && | |
bbe252da | 446 | !(prog->extflags & RXf_CANY_SEEN)) { |
2c2d71f5 | 447 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
448 | I32 slen; |
449 | ||
1aa99e6b | 450 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 451 | |
653099ff GS |
452 | if (SvTAIL(check)) { |
453 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 454 | |
9041c2e3 | 455 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 456 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 457 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 458 | goto fail_finish; |
cad2e5aa JH |
459 | } |
460 | /* Now should match s[0..slen-2] */ | |
461 | slen--; | |
3f7c398e | 462 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 463 | || (slen > 1 |
3f7c398e | 464 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 465 | report_neq: |
a3621e74 | 466 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
467 | goto fail_finish; |
468 | } | |
cad2e5aa | 469 | } |
3f7c398e | 470 | else if (*SvPVX_const(check) != *s |
653099ff | 471 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 472 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 473 | goto report_neq; |
c315bfe8 | 474 | check_at = s; |
2c2d71f5 | 475 | goto success_at_start; |
7e25d62c | 476 | } |
cad2e5aa | 477 | } |
2c2d71f5 | 478 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 479 | s = strpos; |
2c2d71f5 | 480 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
481 | end_shift = prog->check_end_shift; |
482 | ||
2c2d71f5 | 483 | if (!ml_anch) { |
a3b680e6 | 484 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 485 | - (SvTAIL(check) != 0); |
a3b680e6 | 486 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
487 | |
488 | if (end_shift < eshift) | |
489 | end_shift = eshift; | |
490 | } | |
cad2e5aa | 491 | } |
2c2d71f5 | 492 | else { /* Can match at random position */ |
cad2e5aa JH |
493 | ml_anch = 0; |
494 | s = strpos; | |
1de06328 YO |
495 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
496 | end_shift = prog->check_end_shift; | |
497 | ||
498 | /* end shift should be non negative here */ | |
cad2e5aa JH |
499 | } |
500 | ||
bcdf7404 | 501 | #ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 502 | if (end_shift < 0) |
1de06328 | 503 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 504 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
505 | #endif |
506 | ||
2c2d71f5 JH |
507 | restart: |
508 | /* Find a possible match in the region s..strend by looking for | |
509 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
510 | |
511 | { | |
512 | I32 srch_start_shift = start_shift; | |
513 | I32 srch_end_shift = end_shift; | |
514 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
515 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
516 | srch_start_shift = strbeg - s; | |
517 | } | |
6bda09f9 | 518 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
519 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
520 | (IV)prog->check_offset_min, | |
521 | (IV)srch_start_shift, | |
522 | (IV)srch_end_shift, | |
523 | (IV)prog->check_end_shift); | |
524 | }); | |
525 | ||
cad2e5aa | 526 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 527 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 528 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 529 | |
2c2d71f5 JH |
530 | if (PL_screamfirst[BmRARE(check)] >= 0 |
531 | || ( BmRARE(check) == '\n' | |
85c508c3 | 532 | && (BmPREVIOUS(check) == SvCUR(check) - 1) |
2c2d71f5 | 533 | && SvTAIL(check) )) |
9041c2e3 | 534 | s = screaminstr(sv, check, |
1de06328 | 535 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 536 | else |
2c2d71f5 | 537 | goto fail_finish; |
4addbd3b HS |
538 | /* we may be pointing at the wrong string */ |
539 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 540 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
541 | if (data) |
542 | *data->scream_olds = s; | |
543 | } | |
1de06328 YO |
544 | else { |
545 | U8* start_point; | |
546 | U8* end_point; | |
bbe252da | 547 | if (prog->extflags & RXf_CANY_SEEN) { |
1de06328 YO |
548 | start_point= (U8*)(s + srch_start_shift); |
549 | end_point= (U8*)(strend - srch_end_shift); | |
550 | } else { | |
551 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
552 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
553 | } | |
6bda09f9 | 554 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 555 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 556 | (int)(end_point - start_point), |
fc8cd66c | 557 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
558 | start_point); |
559 | }); | |
560 | ||
561 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 562 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
563 | } |
564 | } | |
cad2e5aa JH |
565 | /* Update the count-of-usability, remove useless subpatterns, |
566 | unshift s. */ | |
2c2d71f5 | 567 | |
ab3bbdeb YO |
568 | DEBUG_EXECUTE_r({ |
569 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
570 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
571 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 572 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
573 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
574 | ? "anchored" : "floating"), | |
575 | quoted, | |
576 | RE_SV_TAIL(check), | |
577 | (s ? " at offset " : "...\n") ); | |
578 | }); | |
2c2d71f5 JH |
579 | |
580 | if (!s) | |
581 | goto fail_finish; | |
2c2d71f5 | 582 | /* Finish the diagnostic message */ |
a3621e74 | 583 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 584 | |
1de06328 YO |
585 | /* XXX dmq: first branch is for positive lookbehind... |
586 | Our check string is offset from the beginning of the pattern. | |
587 | So we need to do any stclass tests offset forward from that | |
588 | point. I think. :-( | |
589 | */ | |
590 | ||
591 | ||
592 | ||
593 | check_at=s; | |
594 | ||
595 | ||
2c2d71f5 JH |
596 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
597 | Start with the other substr. | |
598 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 599 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
600 | *always* match. Probably should be marked during compile... |
601 | Probably it is right to do no SCREAM here... | |
602 | */ | |
603 | ||
1de06328 YO |
604 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
605 | : (prog->float_substr && prog->anchored_substr)) | |
606 | { | |
30944b6d | 607 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
608 | /* XXXX May be hopelessly wrong for UTF... */ |
609 | if (!other_last) | |
6eb5f6b9 | 610 | other_last = strpos; |
33b8afdf | 611 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
612 | do_other_anchored: |
613 | { | |
890ce7af AL |
614 | char * const last = HOP3c(s, -start_shift, strbeg); |
615 | char *last1, *last2; | |
be8e71aa | 616 | char * const saved_s = s; |
33b8afdf | 617 | SV* must; |
2c2d71f5 | 618 | |
2c2d71f5 JH |
619 | t = s - prog->check_offset_max; |
620 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 621 | && (!do_utf8 |
0ce71af7 | 622 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 623 | && t > strpos))) |
6f207bd3 | 624 | NOOP; |
2c2d71f5 JH |
625 | else |
626 | t = strpos; | |
1aa99e6b | 627 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
628 | if (t < other_last) /* These positions already checked */ |
629 | t = other_last; | |
1aa99e6b | 630 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
631 | if (last < last1) |
632 | last1 = last; | |
1de06328 YO |
633 | /* XXXX It is not documented what units *_offsets are in. |
634 | We assume bytes, but this is clearly wrong. | |
635 | Meaning this code needs to be carefully reviewed for errors. | |
636 | dmq. | |
637 | */ | |
638 | ||
2c2d71f5 | 639 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
640 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
641 | if (must == &PL_sv_undef) { | |
642 | s = (char*)NULL; | |
1de06328 | 643 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
644 | } |
645 | else | |
646 | s = fbm_instr( | |
647 | (unsigned char*)t, | |
648 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
649 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
650 | must, | |
7fba1cd6 | 651 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 652 | ); |
ab3bbdeb YO |
653 | DEBUG_EXECUTE_r({ |
654 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
655 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
656 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 657 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
658 | quoted, RE_SV_TAIL(must)); |
659 | }); | |
660 | ||
661 | ||
2c2d71f5 JH |
662 | if (!s) { |
663 | if (last1 >= last2) { | |
a3621e74 | 664 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
665 | ", giving up...\n")); |
666 | goto fail_finish; | |
667 | } | |
a3621e74 | 668 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 669 | ", trying floating at offset %ld...\n", |
be8e71aa | 670 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
671 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
672 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
673 | goto restart; |
674 | } | |
675 | else { | |
a3621e74 | 676 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 677 | (long)(s - i_strpos))); |
1aa99e6b IH |
678 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
679 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 680 | s = saved_s; |
2c2d71f5 JH |
681 | if (t == strpos) |
682 | goto try_at_start; | |
2c2d71f5 JH |
683 | goto try_at_offset; |
684 | } | |
30944b6d | 685 | } |
2c2d71f5 JH |
686 | } |
687 | else { /* Take into account the floating substring. */ | |
33b8afdf | 688 | char *last, *last1; |
be8e71aa | 689 | char * const saved_s = s; |
33b8afdf JH |
690 | SV* must; |
691 | ||
692 | t = HOP3c(s, -start_shift, strbeg); | |
693 | last1 = last = | |
694 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
695 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
696 | last = HOP3c(t, prog->float_max_offset, strend); | |
697 | s = HOP3c(t, prog->float_min_offset, strend); | |
698 | if (s < other_last) | |
699 | s = other_last; | |
2c2d71f5 | 700 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
701 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
702 | /* fbm_instr() takes into account exact value of end-of-str | |
703 | if the check is SvTAIL(ed). Since false positives are OK, | |
704 | and end-of-str is not later than strend we are OK. */ | |
705 | if (must == &PL_sv_undef) { | |
706 | s = (char*)NULL; | |
1de06328 | 707 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
708 | } |
709 | else | |
2c2d71f5 | 710 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
711 | (unsigned char*)last + SvCUR(must) |
712 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 713 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
714 | DEBUG_EXECUTE_r({ |
715 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
716 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
717 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 718 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
719 | quoted, RE_SV_TAIL(must)); |
720 | }); | |
33b8afdf JH |
721 | if (!s) { |
722 | if (last1 == last) { | |
a3621e74 | 723 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
724 | ", giving up...\n")); |
725 | goto fail_finish; | |
2c2d71f5 | 726 | } |
a3621e74 | 727 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 728 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 729 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
730 | other_last = last; |
731 | s = HOP3c(t, 1, strend); | |
732 | goto restart; | |
733 | } | |
734 | else { | |
a3621e74 | 735 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
736 | (long)(s - i_strpos))); |
737 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 738 | s = saved_s; |
33b8afdf JH |
739 | if (t == strpos) |
740 | goto try_at_start; | |
741 | goto try_at_offset; | |
742 | } | |
2c2d71f5 | 743 | } |
cad2e5aa | 744 | } |
2c2d71f5 | 745 | |
1de06328 | 746 | |
9ef43ace | 747 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 748 | |
6bda09f9 | 749 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
750 | PerlIO_printf(Perl_debug_log, |
751 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
752 | (IV)prog->check_offset_min, | |
753 | (IV)prog->check_offset_max, | |
754 | (IV)(s-strpos), | |
755 | (IV)(t-strpos), | |
756 | (IV)(t-s), | |
757 | (IV)(strend-strpos) | |
758 | ) | |
759 | ); | |
760 | ||
2c2d71f5 | 761 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 762 | && (!do_utf8 |
9ef43ace | 763 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
764 | && t > strpos))) |
765 | { | |
2c2d71f5 JH |
766 | /* Fixed substring is found far enough so that the match |
767 | cannot start at strpos. */ | |
768 | try_at_offset: | |
cad2e5aa | 769 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
770 | /* Eventually fbm_*() should handle this, but often |
771 | anchored_offset is not 0, so this check will not be wasted. */ | |
772 | /* XXXX In the code below we prefer to look for "^" even in | |
773 | presence of anchored substrings. And we search even | |
774 | beyond the found float position. These pessimizations | |
775 | are historical artefacts only. */ | |
776 | find_anchor: | |
2c2d71f5 | 777 | while (t < strend - prog->minlen) { |
cad2e5aa | 778 | if (*t == '\n') { |
4ee3650e | 779 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 780 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
781 | /* Since we moved from the found position, |
782 | we definitely contradict the found anchored | |
30944b6d IZ |
783 | substr. Due to the above check we do not |
784 | contradict "check" substr. | |
785 | Thus we can arrive here only if check substr | |
786 | is float. Redo checking for "other"=="fixed". | |
787 | */ | |
9041c2e3 | 788 | strpos = t + 1; |
a3621e74 | 789 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 790 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
791 | goto do_other_anchored; |
792 | } | |
4ee3650e GS |
793 | /* We don't contradict the found floating substring. */ |
794 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 795 | s = t + 1; |
a3621e74 | 796 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 797 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
798 | goto set_useful; |
799 | } | |
4ee3650e GS |
800 | /* Position contradicts check-string */ |
801 | /* XXXX probably better to look for check-string | |
802 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 803 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 804 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 805 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
806 | goto restart; |
807 | } | |
808 | t++; | |
809 | } | |
a3621e74 | 810 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 811 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 812 | goto fail_finish; |
cad2e5aa | 813 | } |
f5952150 | 814 | else { |
a3621e74 | 815 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 816 | PL_colors[0], PL_colors[1])); |
f5952150 | 817 | } |
cad2e5aa JH |
818 | s = t; |
819 | set_useful: | |
33b8afdf | 820 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
821 | } |
822 | else { | |
f5952150 | 823 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 824 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
825 | unless it was an MBOL and we are not after MBOL, |
826 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
827 | try_at_start: |
828 | /* Even in this situation we may use MBOL flag if strpos is offset | |
829 | wrt the start of the string. */ | |
05b4157f | 830 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 831 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d | 832 | /* May be due to an implicit anchor of m{.*foo} */ |
bbe252da | 833 | && !(prog->intflags & PREGf_IMPLICIT)) |
d506a20d | 834 | { |
cad2e5aa JH |
835 | t = strpos; |
836 | goto find_anchor; | |
837 | } | |
a3621e74 | 838 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 839 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
70685ca0 | 840 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 841 | ); |
2c2d71f5 | 842 | success_at_start: |
bbe252da | 843 | if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
844 | && (do_utf8 ? ( |
845 | prog->check_utf8 /* Could be deleted already */ | |
846 | && --BmUSEFUL(prog->check_utf8) < 0 | |
847 | && (prog->check_utf8 == prog->float_utf8) | |
848 | ) : ( | |
849 | prog->check_substr /* Could be deleted already */ | |
850 | && --BmUSEFUL(prog->check_substr) < 0 | |
851 | && (prog->check_substr == prog->float_substr) | |
852 | ))) | |
66e933ab | 853 | { |
cad2e5aa | 854 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 855 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
856 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
857 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
858 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
859 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
860 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
861 | check = NULL; /* abort */ | |
cad2e5aa | 862 | s = strpos; |
3cf5c195 IZ |
863 | /* XXXX This is a remnant of the old implementation. It |
864 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 865 | other heuristics. */ |
bbe252da | 866 | prog->extflags &= ~RXf_USE_INTUIT; |
cad2e5aa JH |
867 | } |
868 | else | |
869 | s = strpos; | |
870 | } | |
871 | ||
6eb5f6b9 JH |
872 | /* Last resort... */ |
873 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
874 | /* trie stclasses are too expensive to use here, we are better off to |
875 | leave it to regmatch itself */ | |
f8fc2ecf | 876 | if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
877 | /* minlen == 0 is possible if regstclass is \b or \B, |
878 | and the fixed substr is ''$. | |
879 | Since minlen is already taken into account, s+1 is before strend; | |
880 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
881 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
882 | regstclass does not come from lookahead... */ | |
883 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
884 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
f8fc2ecf YO |
885 | const U8* const str = (U8*)STRING(progi->regstclass); |
886 | const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT | |
887 | ? CHR_DIST(str+STR_LEN(progi->regstclass), str) | |
66e933ab | 888 | : 1); |
1de06328 YO |
889 | char * endpos; |
890 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
891 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
892 | else if (prog->float_substr || prog->float_utf8) | |
893 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
894 | else | |
895 | endpos= strend; | |
896 | ||
70685ca0 JH |
897 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n", |
898 | (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg))); | |
1de06328 | 899 | |
6eb5f6b9 | 900 | t = s; |
f8fc2ecf | 901 | s = find_byclass(prog, progi->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
902 | if (!s) { |
903 | #ifdef DEBUGGING | |
cbbf8932 | 904 | const char *what = NULL; |
6eb5f6b9 JH |
905 | #endif |
906 | if (endpos == strend) { | |
a3621e74 | 907 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
908 | "Could not match STCLASS...\n") ); |
909 | goto fail; | |
910 | } | |
a3621e74 | 911 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 912 | "This position contradicts STCLASS...\n") ); |
bbe252da | 913 | if ((prog->extflags & RXf_ANCH) && !ml_anch) |
653099ff | 914 | goto fail; |
6eb5f6b9 | 915 | /* Contradict one of substrings */ |
33b8afdf JH |
916 | if (prog->anchored_substr || prog->anchored_utf8) { |
917 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 918 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 919 | hop_and_restart: |
1aa99e6b | 920 | s = HOP3c(t, 1, strend); |
66e933ab GS |
921 | if (s + start_shift + end_shift > strend) { |
922 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 923 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
924 | "Could not match STCLASS...\n") ); |
925 | goto fail; | |
926 | } | |
5e39e1e5 HS |
927 | if (!check) |
928 | goto giveup; | |
a3621e74 | 929 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 930 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
931 | what, (long)(s + start_shift - i_strpos)) ); |
932 | goto restart; | |
933 | } | |
66e933ab | 934 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
935 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
936 | goto retry_floating_check; | |
937 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 938 | s = check_at; |
5e39e1e5 HS |
939 | if (!check) |
940 | goto giveup; | |
a3621e74 | 941 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 942 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
943 | (long)(other_last - i_strpos)) ); |
944 | goto do_other_anchored; | |
945 | } | |
60e71179 GS |
946 | /* Another way we could have checked stclass at the |
947 | current position only: */ | |
948 | if (ml_anch) { | |
949 | s = t = t + 1; | |
5e39e1e5 HS |
950 | if (!check) |
951 | goto giveup; | |
a3621e74 | 952 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 953 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 954 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 955 | goto try_at_offset; |
66e933ab | 956 | } |
33b8afdf | 957 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 958 | goto fail; |
6eb5f6b9 JH |
959 | /* Check is floating subtring. */ |
960 | retry_floating_check: | |
961 | t = check_at - start_shift; | |
a3621e74 | 962 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
963 | goto hop_and_restart; |
964 | } | |
b7953727 | 965 | if (t != s) { |
a3621e74 | 966 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 967 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
968 | (long)(t - i_strpos), (long)(s - i_strpos)) |
969 | ); | |
970 | } | |
971 | else { | |
a3621e74 | 972 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
973 | "Does not contradict STCLASS...\n"); |
974 | ); | |
975 | } | |
6eb5f6b9 | 976 | } |
5e39e1e5 | 977 | giveup: |
a3621e74 | 978 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
979 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
980 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 981 | return s; |
2c2d71f5 JH |
982 | |
983 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
984 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
985 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 986 | fail: |
a3621e74 | 987 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 988 | PL_colors[4], PL_colors[5])); |
bd61b366 | 989 | return NULL; |
cad2e5aa | 990 | } |
9661b544 | 991 | |
786e8c11 | 992 | |
3b0527fe | 993 | |
55eed653 NC |
994 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ |
995 | uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ | |
4cadc6a9 YO |
996 | switch (trie_type) { \ |
997 | case trie_utf8_fold: \ | |
998 | if ( foldlen>0 ) { \ | |
999 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
1000 | foldlen -= len; \ | |
1001 | uscan += len; \ | |
1002 | len=0; \ | |
1003 | } else { \ | |
1004 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
1005 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
1006 | foldlen -= UNISKIP( uvc ); \ | |
1007 | uscan = foldbuf + UNISKIP( uvc ); \ | |
1008 | } \ | |
1009 | break; \ | |
1010 | case trie_utf8: \ | |
1011 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
1012 | break; \ | |
1013 | case trie_plain: \ | |
1014 | uvc = (UV)*uc; \ | |
1015 | len = 1; \ | |
1016 | } \ | |
1017 | \ | |
1018 | if (uvc < 256) { \ | |
1019 | charid = trie->charmap[ uvc ]; \ | |
1020 | } \ | |
1021 | else { \ | |
1022 | charid = 0; \ | |
55eed653 NC |
1023 | if (widecharmap) { \ |
1024 | SV** const svpp = hv_fetch(widecharmap, \ | |
4cadc6a9 YO |
1025 | (char*)&uvc, sizeof(UV), 0); \ |
1026 | if (svpp) \ | |
1027 | charid = (U16)SvIV(*svpp); \ | |
1028 | } \ | |
1029 | } \ | |
1030 | } STMT_END | |
1031 | ||
1032 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1033 | if ( (CoNd) \ | |
1034 | && (ln == len || \ | |
1035 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1036 | m, NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1037 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1038 | goto got_it; \ |
1039 | else { \ | |
1040 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1041 | uvchr_to_utf8(tmpbuf, c); \ | |
1042 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1043 | if ( f != c \ | |
1044 | && (f == c1 || f == c2) \ | |
1045 | && (ln == foldlen || \ | |
1046 | !ibcmp_utf8((char *) foldbuf, \ | |
1047 | NULL, foldlen, do_utf8, \ | |
1048 | m, \ | |
1049 | NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1050 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1051 | goto got_it; \ |
1052 | } \ | |
1053 | s += len | |
1054 | ||
1055 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1056 | STMT_START { \ | |
1057 | while (s <= e) { \ | |
1058 | if ( (CoNd) \ | |
1059 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1060 | ? ibcmp(s, m, ln) \ | |
1061 | : ibcmp_locale(s, m, ln))) \ | |
24b23f37 | 1062 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1063 | goto got_it; \ |
1064 | s++; \ | |
1065 | } \ | |
1066 | } STMT_END | |
1067 | ||
1068 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1069 | STMT_START { \ | |
1070 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1071 | CoDe \ | |
1072 | s += uskip; \ | |
1073 | } \ | |
1074 | } STMT_END | |
1075 | ||
1076 | #define REXEC_FBC_SCAN(CoDe) \ | |
1077 | STMT_START { \ | |
1078 | while (s < strend) { \ | |
1079 | CoDe \ | |
1080 | s++; \ | |
1081 | } \ | |
1082 | } STMT_END | |
1083 | ||
1084 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1085 | REXEC_FBC_UTF8_SCAN( \ | |
1086 | if (CoNd) { \ | |
24b23f37 | 1087 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1088 | goto got_it; \ |
1089 | else \ | |
1090 | tmp = doevery; \ | |
1091 | } \ | |
1092 | else \ | |
1093 | tmp = 1; \ | |
1094 | ) | |
1095 | ||
1096 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1097 | REXEC_FBC_SCAN( \ | |
1098 | if (CoNd) { \ | |
24b23f37 | 1099 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1100 | goto got_it; \ |
1101 | else \ | |
1102 | tmp = doevery; \ | |
1103 | } \ | |
1104 | else \ | |
1105 | tmp = 1; \ | |
1106 | ) | |
1107 | ||
1108 | #define REXEC_FBC_TRYIT \ | |
24b23f37 | 1109 | if ((!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1110 | goto got_it |
1111 | ||
1112 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ | |
1113 | if (do_utf8) { \ | |
1114 | UtFpReLoAd; \ | |
1115 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1116 | } \ | |
1117 | else { \ | |
1118 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1119 | } \ | |
1120 | break | |
1121 | ||
1122 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1123 | PL_reg_flags |= RF_tainted; \ | |
1124 | if (do_utf8) { \ | |
1125 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1126 | } \ | |
1127 | else { \ | |
1128 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1129 | } \ | |
1130 | break | |
1131 | ||
786e8c11 YO |
1132 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1133 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1134 | ||
1135 | /* We know what class REx starts with. Try to find this position... */ | |
1136 | /* if reginfo is NULL, its a dryrun */ | |
1137 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1138 | in regmatch. /grrr */ | |
1139 | ||
3c3eec57 | 1140 | STATIC char * |
07be1b83 | 1141 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
24b23f37 | 1142 | const char *strend, regmatch_info *reginfo) |
a687059c | 1143 | { |
27da23d5 | 1144 | dVAR; |
bbe252da | 1145 | const I32 doevery = (prog->intflags & PREGf_SKIP) == 0; |
6eb5f6b9 | 1146 | char *m; |
d8093b23 | 1147 | STRLEN ln; |
5dab1207 | 1148 | STRLEN lnc; |
078c425b | 1149 | register STRLEN uskip; |
d8093b23 G |
1150 | unsigned int c1; |
1151 | unsigned int c2; | |
6eb5f6b9 JH |
1152 | char *e; |
1153 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1154 | register const bool do_utf8 = PL_reg_match_utf8; |
f8fc2ecf YO |
1155 | RXi_GET_DECL(prog,progi); |
1156 | ||
6eb5f6b9 JH |
1157 | /* We know what class it must start with. */ |
1158 | switch (OP(c)) { | |
6eb5f6b9 | 1159 | case ANYOF: |
388cc4de | 1160 | if (do_utf8) { |
4cadc6a9 | 1161 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1162 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1163 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1164 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1165 | } |
1166 | else { | |
1167 | while (s < strend) { | |
1168 | STRLEN skip = 1; | |
1169 | ||
32fc9b6a | 1170 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1171 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1172 | /* The assignment of 2 is intentional: | |
1173 | * for the folded sharp s, the skip is 2. */ | |
1174 | (skip = SHARP_S_SKIP))) { | |
24b23f37 | 1175 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
388cc4de HS |
1176 | goto got_it; |
1177 | else | |
1178 | tmp = doevery; | |
1179 | } | |
1180 | else | |
1181 | tmp = 1; | |
1182 | s += skip; | |
1183 | } | |
a0d0e21e | 1184 | } |
6eb5f6b9 | 1185 | break; |
f33976b4 | 1186 | case CANY: |
4cadc6a9 | 1187 | REXEC_FBC_SCAN( |
24b23f37 | 1188 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
f33976b4 DB |
1189 | goto got_it; |
1190 | else | |
1191 | tmp = doevery; | |
4cadc6a9 | 1192 | ); |
f33976b4 | 1193 | break; |
6eb5f6b9 | 1194 | case EXACTF: |
5dab1207 NIS |
1195 | m = STRING(c); |
1196 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1197 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1198 | if (UTF) { |
a2a2844f | 1199 | STRLEN ulen1, ulen2; |
5dab1207 | 1200 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1201 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1202 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1203 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1204 | |
1205 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1206 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1207 | ||
89ebb4a3 | 1208 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1209 | 0, uniflags); |
89ebb4a3 | 1210 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1211 | 0, uniflags); |
5dab1207 NIS |
1212 | lnc = 0; |
1213 | while (sm < ((U8 *) m + ln)) { | |
1214 | lnc++; | |
1215 | sm += UTF8SKIP(sm); | |
1216 | } | |
1aa99e6b IH |
1217 | } |
1218 | else { | |
1219 | c1 = *(U8*)m; | |
1220 | c2 = PL_fold[c1]; | |
1221 | } | |
6eb5f6b9 JH |
1222 | goto do_exactf; |
1223 | case EXACTFL: | |
5dab1207 NIS |
1224 | m = STRING(c); |
1225 | ln = STR_LEN(c); | |
1226 | lnc = (I32) ln; | |
d8093b23 | 1227 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1228 | c2 = PL_fold_locale[c1]; |
1229 | do_exactf: | |
db12adc6 | 1230 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1231 | |
3b0527fe | 1232 | if (!reginfo && e < s) |
6eb5f6b9 | 1233 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1234 | |
60a8b682 JH |
1235 | /* The idea in the EXACTF* cases is to first find the |
1236 | * first character of the EXACTF* node and then, if | |
1237 | * necessary, case-insensitively compare the full | |
1238 | * text of the node. The c1 and c2 are the first | |
1239 | * characters (though in Unicode it gets a bit | |
1240 | * more complicated because there are more cases | |
7f16dd3d JH |
1241 | * than just upper and lower: one needs to use |
1242 | * the so-called folding case for case-insensitive | |
1243 | * matching (called "loose matching" in Unicode). | |
1244 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1245 | |
1aa99e6b | 1246 | if (do_utf8) { |
575cac57 | 1247 | UV c, f; |
89ebb4a3 | 1248 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1249 | STRLEN len, foldlen; |
4ad0818d | 1250 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1251 | if (c1 == c2) { |
5dab1207 NIS |
1252 | /* Upper and lower of 1st char are equal - |
1253 | * probably not a "letter". */ | |
1aa99e6b | 1254 | while (s <= e) { |
89ebb4a3 | 1255 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1256 | uniflags); |
4cadc6a9 | 1257 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1258 | } |
09091399 JH |
1259 | } |
1260 | else { | |
1aa99e6b | 1261 | while (s <= e) { |
89ebb4a3 | 1262 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1263 | uniflags); |
80aecb99 | 1264 | |
60a8b682 | 1265 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1266 | * Note that not all the possible combinations |
1267 | * are handled here: some of them are handled | |
1268 | * by the standard folding rules, and some of | |
1269 | * them (the character class or ANYOF cases) | |
1270 | * are handled during compiletime in | |
1271 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1272 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1273 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1274 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1275 | |
4cadc6a9 | 1276 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1277 | } |
09091399 | 1278 | } |
1aa99e6b IH |
1279 | } |
1280 | else { | |
1281 | if (c1 == c2) | |
4cadc6a9 | 1282 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1283 | else |
4cadc6a9 | 1284 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1285 | } |
1286 | break; | |
bbce6d69 | 1287 | case BOUNDL: |
3280af22 | 1288 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1289 | /* FALL THROUGH */ |
a0d0e21e | 1290 | case BOUND: |
ffc61ed2 | 1291 | if (do_utf8) { |
12d33761 | 1292 | if (s == PL_bostr) |
ffc61ed2 JH |
1293 | tmp = '\n'; |
1294 | else { | |
6136c704 | 1295 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1296 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1297 | } |
1298 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1299 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1300 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1301 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1302 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1303 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1304 | isALNUM_LC_utf8((U8*)s))) |
1305 | { | |
1306 | tmp = !tmp; | |
4cadc6a9 | 1307 | REXEC_FBC_TRYIT; |
a687059c | 1308 | } |
4cadc6a9 | 1309 | ); |
a0d0e21e | 1310 | } |
667bb95a | 1311 | else { |
12d33761 | 1312 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1313 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1314 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1315 | if (tmp == |
1316 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1317 | tmp = !tmp; | |
4cadc6a9 | 1318 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1319 | } |
4cadc6a9 | 1320 | ); |
a0ed51b3 | 1321 | } |
24b23f37 | 1322 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1323 | goto got_it; |
1324 | break; | |
bbce6d69 | 1325 | case NBOUNDL: |
3280af22 | 1326 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1327 | /* FALL THROUGH */ |
a0d0e21e | 1328 | case NBOUND: |
ffc61ed2 | 1329 | if (do_utf8) { |
12d33761 | 1330 | if (s == PL_bostr) |
ffc61ed2 JH |
1331 | tmp = '\n'; |
1332 | else { | |
6136c704 | 1333 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1334 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1335 | } |
1336 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1337 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1338 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1339 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1340 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1341 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1342 | isALNUM_LC_utf8((U8*)s))) |
1343 | tmp = !tmp; | |
4cadc6a9 YO |
1344 | else REXEC_FBC_TRYIT; |
1345 | ); | |
a0d0e21e | 1346 | } |
667bb95a | 1347 | else { |
12d33761 | 1348 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1349 | tmp = ((OP(c) == NBOUND ? |
1350 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1351 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1352 | if (tmp == |
1353 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1354 | tmp = !tmp; | |
4cadc6a9 YO |
1355 | else REXEC_FBC_TRYIT; |
1356 | ); | |
a0ed51b3 | 1357 | } |
24b23f37 | 1358 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1359 | goto got_it; |
1360 | break; | |
a0d0e21e | 1361 | case ALNUM: |
4cadc6a9 YO |
1362 | REXEC_FBC_CSCAN_PRELOAD( |
1363 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1364 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1365 | isALNUM(*s) | |
1366 | ); | |
bbce6d69 | 1367 | case ALNUML: |
4cadc6a9 YO |
1368 | REXEC_FBC_CSCAN_TAINT( |
1369 | isALNUM_LC_utf8((U8*)s), | |
1370 | isALNUM_LC(*s) | |
1371 | ); | |
a0d0e21e | 1372 | case NALNUM: |
4cadc6a9 YO |
1373 | REXEC_FBC_CSCAN_PRELOAD( |
1374 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1375 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1376 | !isALNUM(*s) | |
1377 | ); | |
bbce6d69 | 1378 | case NALNUML: |
4cadc6a9 YO |
1379 | REXEC_FBC_CSCAN_TAINT( |
1380 | !isALNUM_LC_utf8((U8*)s), | |
1381 | !isALNUM_LC(*s) | |
1382 | ); | |
a0d0e21e | 1383 | case SPACE: |
4cadc6a9 YO |
1384 | REXEC_FBC_CSCAN_PRELOAD( |
1385 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1386 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1387 | isSPACE(*s) | |
1388 | ); | |
bbce6d69 | 1389 | case SPACEL: |
4cadc6a9 YO |
1390 | REXEC_FBC_CSCAN_TAINT( |
1391 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1392 | isSPACE_LC(*s) | |
1393 | ); | |
a0d0e21e | 1394 | case NSPACE: |
4cadc6a9 YO |
1395 | REXEC_FBC_CSCAN_PRELOAD( |
1396 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1397 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1398 | !isSPACE(*s) | |
1399 | ); | |
bbce6d69 | 1400 | case NSPACEL: |
4cadc6a9 YO |
1401 | REXEC_FBC_CSCAN_TAINT( |
1402 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1403 | !isSPACE_LC(*s) | |
1404 | ); | |
a0d0e21e | 1405 | case DIGIT: |
4cadc6a9 YO |
1406 | REXEC_FBC_CSCAN_PRELOAD( |
1407 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1408 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1409 | isDIGIT(*s) | |
1410 | ); | |
b8c5462f | 1411 | case DIGITL: |
4cadc6a9 YO |
1412 | REXEC_FBC_CSCAN_TAINT( |
1413 | isDIGIT_LC_utf8((U8*)s), | |
1414 | isDIGIT_LC(*s) | |
1415 | ); | |
a0d0e21e | 1416 | case NDIGIT: |
4cadc6a9 YO |
1417 | REXEC_FBC_CSCAN_PRELOAD( |
1418 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1419 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1420 | !isDIGIT(*s) | |
1421 | ); | |
b8c5462f | 1422 | case NDIGITL: |
4cadc6a9 YO |
1423 | REXEC_FBC_CSCAN_TAINT( |
1424 | !isDIGIT_LC_utf8((U8*)s), | |
1425 | !isDIGIT_LC(*s) | |
1426 | ); | |
1de06328 YO |
1427 | case AHOCORASICKC: |
1428 | case AHOCORASICK: | |
07be1b83 YO |
1429 | { |
1430 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1431 | trie_type = do_utf8 ? | |
1432 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1433 | : trie_plain; | |
1434 | /* what trie are we using right now */ | |
1435 | reg_ac_data *aho | |
f8fc2ecf | 1436 | = (reg_ac_data*)progi->data->data[ ARG( c ) ]; |
3251b653 NC |
1437 | reg_trie_data *trie |
1438 | = (reg_trie_data*)progi->data->data[ aho->trie ]; | |
55eed653 | 1439 | HV *widecharmap = (HV*) progi->data->data[ aho->trie + 1 ]; |
07be1b83 YO |
1440 | |
1441 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1442 | #ifdef DEBUGGING |
07be1b83 | 1443 | const char *real_start = s; |
6148ee25 | 1444 | #endif |
07be1b83 | 1445 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1446 | SV *sv_points; |
1447 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1448 | when reading a given char. For ASCII this |
be8e71aa YO |
1449 | is unnecessary overhead as the relationship |
1450 | is always 1:1, but for unicode, especially | |
1451 | case folded unicode this is not true. */ | |
f9e705e8 | 1452 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1453 | U8 *bitmap=NULL; |
1454 | ||
07be1b83 YO |
1455 | |
1456 | GET_RE_DEBUG_FLAGS_DECL; | |
1457 | ||
be8e71aa YO |
1458 | /* We can't just allocate points here. We need to wrap it in |
1459 | * an SV so it gets freed properly if there is a croak while | |
1460 | * running the match */ | |
1461 | ENTER; | |
1462 | SAVETMPS; | |
1463 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1464 | SvCUR_set(sv_points, | |
1465 | maxlen * sizeof(U8 *)); | |
1466 | SvPOK_on(sv_points); | |
1467 | sv_2mortal(sv_points); | |
1468 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1469 | if ( trie_type != trie_utf8_fold |
1470 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1471 | { | |
786e8c11 YO |
1472 | if (trie->bitmap) |
1473 | bitmap=(U8*)trie->bitmap; | |
1474 | else | |
1475 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1476 | } |
786e8c11 YO |
1477 | /* this is the Aho-Corasick algorithm modified a touch |
1478 | to include special handling for long "unknown char" | |
1479 | sequences. The basic idea being that we use AC as long | |
1480 | as we are dealing with a possible matching char, when | |
1481 | we encounter an unknown char (and we have not encountered | |
1482 | an accepting state) we scan forward until we find a legal | |
1483 | starting char. | |
1484 | AC matching is basically that of trie matching, except | |
1485 | that when we encounter a failing transition, we fall back | |
1486 | to the current states "fail state", and try the current char | |
1487 | again, a process we repeat until we reach the root state, | |
1488 | state 1, or a legal transition. If we fail on the root state | |
1489 | then we can either terminate if we have reached an accepting | |
1490 | state previously, or restart the entire process from the beginning | |
1491 | if we have not. | |
1492 | ||
1493 | */ | |
07be1b83 YO |
1494 | while (s <= last_start) { |
1495 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1496 | U8 *uc = (U8*)s; | |
1497 | U16 charid = 0; | |
1498 | U32 base = 1; | |
1499 | U32 state = 1; | |
1500 | UV uvc = 0; | |
1501 | STRLEN len = 0; | |
1502 | STRLEN foldlen = 0; | |
1503 | U8 *uscan = (U8*)NULL; | |
1504 | U8 *leftmost = NULL; | |
786e8c11 YO |
1505 | #ifdef DEBUGGING |
1506 | U32 accepted_word= 0; | |
1507 | #endif | |
07be1b83 YO |
1508 | U32 pointpos = 0; |
1509 | ||
1510 | while ( state && uc <= (U8*)strend ) { | |
1511 | int failed=0; | |
786e8c11 YO |
1512 | U32 word = aho->states[ state ].wordnum; |
1513 | ||
1de06328 YO |
1514 | if( state==1 ) { |
1515 | if ( bitmap ) { | |
1516 | DEBUG_TRIE_EXECUTE_r( | |
1517 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1518 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1519 | (char *)uc, do_utf8 ); | |
1520 | PerlIO_printf( Perl_debug_log, | |
1521 | " Scanning for legal start char...\n"); | |
1522 | } | |
1523 | ); | |
1524 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1525 | uc++; | |
786e8c11 | 1526 | } |
1de06328 | 1527 | s= (char *)uc; |
786e8c11 | 1528 | } |
786e8c11 YO |
1529 | if (uc >(U8*)last_start) break; |
1530 | } | |
1531 | ||
1532 | if ( word ) { | |
1533 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1534 | if (!leftmost || lpos < leftmost) { | |
1535 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1536 | leftmost= lpos; |
786e8c11 | 1537 | } |
07be1b83 | 1538 | if (base==0) break; |
786e8c11 | 1539 | |
07be1b83 YO |
1540 | } |
1541 | points[pointpos++ % maxlen]= uc; | |
55eed653 NC |
1542 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
1543 | uscan, len, uvc, charid, foldlen, | |
1544 | foldbuf, uniflags); | |
786e8c11 YO |
1545 | DEBUG_TRIE_EXECUTE_r({ |
1546 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1547 | s, do_utf8 ); | |
07be1b83 | 1548 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1549 | " Charid:%3u CP:%4"UVxf" ", |
1550 | charid, uvc); | |
1551 | }); | |
07be1b83 YO |
1552 | |
1553 | do { | |
6148ee25 | 1554 | #ifdef DEBUGGING |
786e8c11 | 1555 | word = aho->states[ state ].wordnum; |
6148ee25 | 1556 | #endif |
07be1b83 YO |
1557 | base = aho->states[ state ].trans.base; |
1558 | ||
786e8c11 YO |
1559 | DEBUG_TRIE_EXECUTE_r({ |
1560 | if (failed) | |
1561 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1562 | s, do_utf8 ); | |
07be1b83 | 1563 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1564 | "%sState: %4"UVxf", word=%"UVxf, |
1565 | failed ? " Fail transition to " : "", | |
1566 | (UV)state, (UV)word); | |
1567 | }); | |
07be1b83 YO |
1568 | if ( base ) { |
1569 | U32 tmp; | |
1570 | if (charid && | |
1571 | (base + charid > trie->uniquecharcount ) | |
1572 | && (base + charid - 1 - trie->uniquecharcount | |
1573 | < trie->lasttrans) | |
1574 | && trie->trans[base + charid - 1 - | |
1575 | trie->uniquecharcount].check == state | |
1576 | && (tmp=trie->trans[base + charid - 1 - | |
1577 | trie->uniquecharcount ].next)) | |
1578 | { | |
786e8c11 YO |
1579 | DEBUG_TRIE_EXECUTE_r( |
1580 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1581 | state = tmp; |
1582 | break; | |
1583 | } | |
1584 | else { | |
786e8c11 YO |
1585 | DEBUG_TRIE_EXECUTE_r( |
1586 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1587 | failed = 1; | |
1588 | state = aho->fail[state]; | |
07be1b83 YO |
1589 | } |
1590 | } | |
1591 | else { | |
1592 | /* we must be accepting here */ | |
786e8c11 YO |
1593 | DEBUG_TRIE_EXECUTE_r( |
1594 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1595 | failed = 1; | |
07be1b83 YO |
1596 | break; |
1597 | } | |
1598 | } while(state); | |
786e8c11 | 1599 | uc += len; |
07be1b83 YO |
1600 | if (failed) { |
1601 | if (leftmost) | |
1602 | break; | |
786e8c11 | 1603 | if (!state) state = 1; |
07be1b83 YO |
1604 | } |
1605 | } | |
1606 | if ( aho->states[ state ].wordnum ) { | |
1607 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1608 | if (!leftmost || lpos < leftmost) { |
1609 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1610 | leftmost = lpos; |
786e8c11 | 1611 | } |
07be1b83 | 1612 | } |
07be1b83 YO |
1613 | if (leftmost) { |
1614 | s = (char*)leftmost; | |
786e8c11 YO |
1615 | DEBUG_TRIE_EXECUTE_r({ |
1616 | PerlIO_printf( | |
70685ca0 JH |
1617 | Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n", |
1618 | (UV)accepted_word, (IV)(s - real_start) | |
786e8c11 YO |
1619 | ); |
1620 | }); | |
24b23f37 | 1621 | if (!reginfo || regtry(reginfo, &s)) { |
be8e71aa YO |
1622 | FREETMPS; |
1623 | LEAVE; | |
07be1b83 | 1624 | goto got_it; |
be8e71aa | 1625 | } |
07be1b83 | 1626 | s = HOPc(s,1); |
786e8c11 YO |
1627 | DEBUG_TRIE_EXECUTE_r({ |
1628 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1629 | }); | |
07be1b83 | 1630 | } else { |
786e8c11 YO |
1631 | DEBUG_TRIE_EXECUTE_r( |
1632 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1633 | break; |
1634 | } | |
1635 | } | |
be8e71aa YO |
1636 | FREETMPS; |
1637 | LEAVE; | |
07be1b83 YO |
1638 | } |
1639 | break; | |
b3c9acc1 | 1640 | default: |
3c3eec57 GS |
1641 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1642 | break; | |
d6a28714 | 1643 | } |
6eb5f6b9 JH |
1644 | return 0; |
1645 | got_it: | |
1646 | return s; | |
1647 | } | |
1648 | ||
1649 | /* | |
1650 | - regexec_flags - match a regexp against a string | |
1651 | */ | |
1652 | I32 | |
1653 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1654 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1655 | /* strend: pointer to null at end of string */ | |
1656 | /* strbeg: real beginning of string */ | |
1657 | /* minend: end of match must be >=minend after stringarg. */ | |
58e23c8d YO |
1658 | /* data: May be used for some additional optimizations. |
1659 | Currently its only used, with a U32 cast, for transmitting | |
1660 | the ganch offset when doing a /g match. This will change */ | |
6eb5f6b9 JH |
1661 | /* nosave: For optimizations. */ |
1662 | { | |
97aff369 | 1663 | dVAR; |
24b23f37 | 1664 | /*register*/ char *s; |
6eb5f6b9 | 1665 | register regnode *c; |
24b23f37 | 1666 | /*register*/ char *startpos = stringarg; |
6eb5f6b9 JH |
1667 | I32 minlen; /* must match at least this many chars */ |
1668 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1669 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1670 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1671 | char *scream_olds = NULL; |
3dab1dad | 1672 | SV* const oreplsv = GvSV(PL_replgv); |
f9f4320a | 1673 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1674 | I32 multiline; |
f8fc2ecf | 1675 | RXi_GET_DECL(prog,progi); |
3b0527fe | 1676 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1677 | |
1678 | GET_RE_DEBUG_FLAGS_DECL; | |
1679 | ||
9d4ba2ae | 1680 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1681 | |
1682 | /* Be paranoid... */ | |
1683 | if (prog == NULL || startpos == NULL) { | |
1684 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1685 | return 0; | |
1686 | } | |
1687 | ||
bbe252da | 1688 | multiline = prog->extflags & RXf_PMf_MULTILINE; |
3b0527fe | 1689 | reginfo.prog = prog; |
2757e526 | 1690 | |
bac06658 | 1691 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1692 | DEBUG_EXECUTE_r( |
1693 | debug_start_match(prog, do_utf8, startpos, strend, | |
1694 | "Matching"); | |
1695 | ); | |
bac06658 | 1696 | |
6eb5f6b9 | 1697 | minlen = prog->minlen; |
1de06328 YO |
1698 | |
1699 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1700 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1701 | "String too short [regexec_flags]...\n")); |
1702 | goto phooey; | |
1aa99e6b | 1703 | } |
6eb5f6b9 | 1704 | |
1de06328 | 1705 | |
6eb5f6b9 | 1706 | /* Check validity of program. */ |
f8fc2ecf | 1707 | if (UCHARAT(progi->program) != REG_MAGIC) { |
6eb5f6b9 JH |
1708 | Perl_croak(aTHX_ "corrupted regexp program"); |
1709 | } | |
1710 | ||
1711 | PL_reg_flags = 0; | |
1712 | PL_reg_eval_set = 0; | |
1713 | PL_reg_maxiter = 0; | |
1714 | ||
bbe252da | 1715 | if (prog->extflags & RXf_UTF8) |
6eb5f6b9 JH |
1716 | PL_reg_flags |= RF_utf8; |
1717 | ||
1718 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1719 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1720 | PL_bostr = strbeg; |
3b0527fe | 1721 | reginfo.sv = sv; |
6eb5f6b9 JH |
1722 | |
1723 | /* Mark end of line for $ (and such) */ | |
1724 | PL_regeol = strend; | |
1725 | ||
1726 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1727 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1728 | |
6eb5f6b9 JH |
1729 | /* If there is a "must appear" string, look for it. */ |
1730 | s = startpos; | |
1731 | ||
bbe252da | 1732 | if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1733 | MAGIC *mg; |
1734 | ||
1735 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
58e23c8d | 1736 | reginfo.ganch = startpos + prog->gofs; |
6eb5f6b9 JH |
1737 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1738 | && SvMAGIC(sv) | |
14befaf4 DM |
1739 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1740 | && mg->mg_len >= 0) { | |
3b0527fe | 1741 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
bbe252da | 1742 | if (prog->extflags & RXf_ANCH_GPOS) { |
3b0527fe | 1743 | if (s > reginfo.ganch) |
6eb5f6b9 | 1744 | goto phooey; |
58e23c8d | 1745 | s = reginfo.ganch - prog->gofs; |
6eb5f6b9 JH |
1746 | } |
1747 | } | |
58e23c8d | 1748 | else if (data) { |
70685ca0 | 1749 | reginfo.ganch = strbeg + PTR2UV(data); |
58e23c8d | 1750 | } else /* pos() not defined */ |
3b0527fe | 1751 | reginfo.ganch = strbeg; |
6eb5f6b9 | 1752 | } |
c74340f9 YO |
1753 | if (PL_curpm && (PM_GETRE(PL_curpm) == prog)) { |
1754 | I32 *t; | |
f8fc2ecf | 1755 | if (!progi->swap) { |
c74340f9 YO |
1756 | /* We have to be careful. If the previous successful match |
1757 | was from this regex we don't want a subsequent paritally | |
1758 | successful match to clobber the old results. | |
1759 | So when we detect this possibility we add a swap buffer | |
1760 | to the re, and switch the buffer each match. If we fail | |
1761 | we switch it back, otherwise we leave it swapped. | |
1762 | */ | |
f8fc2ecf | 1763 | Newxz(progi->swap, 1, regexp_paren_ofs); |
c74340f9 | 1764 | /* no need to copy these */ |
f8fc2ecf YO |
1765 | Newxz(progi->swap->startp, prog->nparens + 1, I32); |
1766 | Newxz(progi->swap->endp, prog->nparens + 1, I32); | |
c74340f9 | 1767 | } |
f8fc2ecf YO |
1768 | t = progi->swap->startp; |
1769 | progi->swap->startp = prog->startp; | |
c74340f9 | 1770 | prog->startp = t; |
f8fc2ecf YO |
1771 | t = progi->swap->endp; |
1772 | progi->swap->endp = prog->endp; | |
c74340f9 YO |
1773 | prog->endp = t; |
1774 | } | |
a0714e2c | 1775 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1776 | re_scream_pos_data d; |
1777 | ||
1778 | d.scream_olds = &scream_olds; | |
1779 | d.scream_pos = &scream_pos; | |
1780 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1781 | if (!s) { |
a3621e74 | 1782 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1783 | goto phooey; /* not present */ |
3fa9c3d7 | 1784 | } |
6eb5f6b9 JH |
1785 | } |
1786 | ||
1de06328 | 1787 | |
6eb5f6b9 JH |
1788 | |
1789 | /* Simplest case: anchored match need be tried only once. */ | |
1790 | /* [unless only anchor is BOL and multiline is set] */ | |
bbe252da | 1791 | if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) { |
24b23f37 | 1792 | if (s == startpos && regtry(®info, &startpos)) |
6eb5f6b9 | 1793 | goto got_it; |
bbe252da YO |
1794 | else if (multiline || (prog->intflags & PREGf_IMPLICIT) |
1795 | || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */ | |
6eb5f6b9 JH |
1796 | { |
1797 | char *end; | |
1798 | ||
1799 | if (minlen) | |
1800 | dontbother = minlen - 1; | |
1aa99e6b | 1801 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1802 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1803 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1804 | if (s == startpos) |
1805 | goto after_try; | |
1806 | while (1) { | |
24b23f37 | 1807 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1808 | goto got_it; |
1809 | after_try: | |
1810 | if (s >= end) | |
1811 | goto phooey; | |
bbe252da | 1812 | if (prog->extflags & RXf_USE_INTUIT) { |
6eb5f6b9 JH |
1813 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); |
1814 | if (!s) | |
1815 | goto phooey; | |
1816 | } | |
1817 | else | |
1818 | s++; | |
1819 | } | |
1820 | } else { | |
1821 | if (s > startpos) | |
1822 | s--; | |
1823 | while (s < end) { | |
1824 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
24b23f37 | 1825 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1826 | goto got_it; |
1827 | } | |
1828 | } | |
1829 | } | |
1830 | } | |
1831 | goto phooey; | |
bbe252da | 1832 | } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK)) |
f9f4320a YO |
1833 | { |
1834 | /* the warning about reginfo.ganch being used without intialization | |
bbe252da | 1835 | is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN |
f9f4320a | 1836 | and we only enter this block when the same bit is set. */ |
58e23c8d YO |
1837 | char *tmp_s = reginfo.ganch - prog->gofs; |
1838 | if (regtry(®info, &tmp_s)) | |
6eb5f6b9 JH |
1839 | goto got_it; |
1840 | goto phooey; | |
1841 | } | |
1842 | ||
1843 | /* Messy cases: unanchored match. */ | |
bbe252da | 1844 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) { |
6eb5f6b9 JH |
1845 | /* we have /x+whatever/ */ |
1846 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1847 | char ch; |
bf93d4cc GS |
1848 | #ifdef DEBUGGING |
1849 | int did_match = 0; | |
1850 | #endif | |
33b8afdf JH |
1851 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1852 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1853 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1854 | |
1aa99e6b | 1855 | if (do_utf8) { |
4cadc6a9 | 1856 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1857 | if (*s == ch) { |
a3621e74 | 1858 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1859 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1860 | s += UTF8SKIP(s); |
1861 | while (s < strend && *s == ch) | |
1862 | s += UTF8SKIP(s); | |
1863 | } | |
4cadc6a9 | 1864 | ); |
6eb5f6b9 JH |
1865 | } |
1866 | else { | |
4cadc6a9 | 1867 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1868 | if (*s == ch) { |
a3621e74 | 1869 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1870 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1871 | s++; |
1872 | while (s < strend && *s == ch) | |
1873 | s++; | |
1874 | } | |
4cadc6a9 | 1875 | ); |
6eb5f6b9 | 1876 | } |
a3621e74 | 1877 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1878 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1879 | "Did not find anchored character...\n") |
1880 | ); | |
6eb5f6b9 | 1881 | } |
a0714e2c SS |
1882 | else if (prog->anchored_substr != NULL |
1883 | || prog->anchored_utf8 != NULL | |
1884 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1885 | && prog->float_max_offset < strend - s)) { |
1886 | SV *must; | |
1887 | I32 back_max; | |
1888 | I32 back_min; | |
1889 | char *last; | |
6eb5f6b9 | 1890 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1891 | #ifdef DEBUGGING |
1892 | int did_match = 0; | |
1893 | #endif | |
33b8afdf JH |
1894 | if (prog->anchored_substr || prog->anchored_utf8) { |
1895 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1896 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1897 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1898 | back_max = back_min = prog->anchored_offset; | |
1899 | } else { | |
1900 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1901 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1902 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1903 | back_max = prog->float_max_offset; | |
1904 | back_min = prog->float_min_offset; | |
1905 | } | |
1de06328 YO |
1906 | |
1907 | ||
33b8afdf JH |
1908 | if (must == &PL_sv_undef) |
1909 | /* could not downgrade utf8 check substring, so must fail */ | |
1910 | goto phooey; | |
1911 | ||
1de06328 YO |
1912 | if (back_min<0) { |
1913 | last = strend; | |
1914 | } else { | |
1915 | last = HOP3c(strend, /* Cannot start after this */ | |
1916 | -(I32)(CHR_SVLEN(must) | |
1917 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1918 | } | |
6eb5f6b9 JH |
1919 | if (s > PL_bostr) |
1920 | last1 = HOPc(s, -1); | |
1921 | else | |
1922 | last1 = s - 1; /* bogus */ | |
1923 | ||
a0288114 | 1924 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1925 | check_substr==must. */ |
1926 | scream_pos = -1; | |
1927 | dontbother = end_shift; | |
1928 | strend = HOPc(strend, -dontbother); | |
1929 | while ( (s <= last) && | |
9041c2e3 | 1930 | ((flags & REXEC_SCREAM) |
1de06328 | 1931 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1932 | end_shift, &scream_pos, 0)) |
1de06328 | 1933 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1934 | (unsigned char*)strend, must, |
7fba1cd6 | 1935 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1936 | /* we may be pointing at the wrong string */ |
1937 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1938 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1939 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1940 | if (HOPc(s, -back_max) > last1) { |
1941 | last1 = HOPc(s, -back_min); | |
1942 | s = HOPc(s, -back_max); | |
1943 | } | |
1944 | else { | |
52657f30 | 1945 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1946 | |
1947 | last1 = HOPc(s, -back_min); | |
52657f30 | 1948 | s = t; |
6eb5f6b9 | 1949 | } |
1aa99e6b | 1950 | if (do_utf8) { |
6eb5f6b9 | 1951 | while (s <= last1) { |
24b23f37 | 1952 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1953 | goto got_it; |
1954 | s += UTF8SKIP(s); | |
1955 | } | |
1956 | } | |
1957 | else { | |
1958 | while (s <= last1) { | |
24b23f37 | 1959 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1960 | goto got_it; |
1961 | s++; | |
1962 | } | |
1963 | } | |
1964 | } | |
ab3bbdeb YO |
1965 | DEBUG_EXECUTE_r(if (!did_match) { |
1966 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
1967 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
1968 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 1969 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 1970 | ? "anchored" : "floating"), |
ab3bbdeb YO |
1971 | quoted, RE_SV_TAIL(must)); |
1972 | }); | |
6eb5f6b9 JH |
1973 | goto phooey; |
1974 | } | |
f8fc2ecf | 1975 | else if ( (c = progi->regstclass) ) { |
f14c76ed | 1976 | if (minlen) { |
f8fc2ecf | 1977 | const OPCODE op = OP(progi->regstclass); |
66e933ab | 1978 | /* don't bother with what can't match */ |
786e8c11 | 1979 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
1980 | strend = HOPc(strend, -(minlen - 1)); |
1981 | } | |
a3621e74 | 1982 | DEBUG_EXECUTE_r({ |
be8e71aa | 1983 | SV * const prop = sv_newmortal(); |
32fc9b6a | 1984 | regprop(prog, prop, c); |
0df25f3d | 1985 | { |
ab3bbdeb YO |
1986 | RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1), |
1987 | s,strend-s,60); | |
0df25f3d | 1988 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 1989 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 1990 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 1991 | quoted, (int)(strend - s)); |
0df25f3d | 1992 | } |
ffc61ed2 | 1993 | }); |
3b0527fe | 1994 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1995 | goto got_it; |
07be1b83 | 1996 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
1997 | } |
1998 | else { | |
1999 | dontbother = 0; | |
a0714e2c | 2000 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 2001 | /* Trim the end. */ |
d6a28714 | 2002 | char *last; |
33b8afdf JH |
2003 | SV* float_real; |
2004 | ||
2005 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
2006 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
2007 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
2008 | |
2009 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 2010 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
2011 | end_shift, &scream_pos, 1); /* last one */ |
2012 | if (!last) | |
ffc61ed2 | 2013 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
2014 | /* we may be pointing at the wrong string */ |
2015 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 2016 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 2017 | } |
d6a28714 JH |
2018 | else { |
2019 | STRLEN len; | |
cfd0369c | 2020 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 2021 | |
33b8afdf | 2022 | if (SvTAIL(float_real)) { |
d6a28714 JH |
2023 | if (memEQ(strend - len + 1, little, len - 1)) |
2024 | last = strend - len + 1; | |
7fba1cd6 | 2025 | else if (!multiline) |
9041c2e3 | 2026 | last = memEQ(strend - len, little, len) |
bd61b366 | 2027 | ? strend - len : NULL; |
b8c5462f | 2028 | else |
d6a28714 JH |
2029 | goto find_last; |
2030 | } else { | |
2031 | find_last: | |
9041c2e3 | 2032 | if (len) |
d6a28714 | 2033 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 2034 | else |
a0288114 | 2035 | last = strend; /* matching "$" */ |
b8c5462f | 2036 | } |
b8c5462f | 2037 | } |
bf93d4cc | 2038 | if (last == NULL) { |
6bda09f9 YO |
2039 | DEBUG_EXECUTE_r( |
2040 | PerlIO_printf(Perl_debug_log, | |
2041 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
2042 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
2043 | goto phooey; /* Should not happen! */ |
2044 | } | |
d6a28714 JH |
2045 | dontbother = strend - last + prog->float_min_offset; |
2046 | } | |
2047 | if (minlen && (dontbother < minlen)) | |
2048 | dontbother = minlen - 1; | |
2049 | strend -= dontbother; /* this one's always in bytes! */ | |
2050 | /* We don't know much -- general case. */ | |
1aa99e6b | 2051 | if (do_utf8) { |
d6a28714 | 2052 | for (;;) { |
24b23f37 | 2053 | if (regtry(®info, &s)) |
d6a28714 JH |
2054 | goto got_it; |
2055 | if (s >= strend) | |
2056 | break; | |
b8c5462f | 2057 | s += UTF8SKIP(s); |
d6a28714 JH |
2058 | }; |
2059 | } | |
2060 | else { | |
2061 | do { | |
24b23f37 | 2062 | if (regtry(®info, &s)) |
d6a28714 JH |
2063 | goto got_it; |
2064 | } while (s++ < strend); | |
2065 | } | |
2066 | } | |
2067 | ||
2068 | /* Failure. */ | |
2069 | goto phooey; | |
2070 | ||
2071 | got_it: | |
2072 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2073 | ||
2074 | if (PL_reg_eval_set) { | |
2075 | /* Preserve the current value of $^R */ | |
2076 | if (oreplsv != GvSV(PL_replgv)) | |
2077 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2078 | restored, the value remains | |
2079 | the same. */ | |
4f639d21 | 2080 | restore_pos(aTHX_ prog); |
d6a28714 | 2081 | } |
81714fb9 YO |
2082 | if (prog->paren_names) |
2083 | (void)hv_iterinit(prog->paren_names); | |
d6a28714 JH |
2084 | |
2085 | /* make sure $`, $&, $', and $digit will work later */ | |
2086 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2087 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2088 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2089 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2090 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2091 | if ((SvIsCOW(sv) |
2092 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2093 | if (DEBUG_C_TEST) { | |
2094 | PerlIO_printf(Perl_debug_log, | |
2095 | "Copy on write: regexp capture, type %d\n", | |
2096 | (int) SvTYPE(sv)); | |
2097 | } | |
2098 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2099 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2100 | assert (SvPOKp(prog->saved_copy)); |
2101 | } else | |
2102 | #endif | |
2103 | { | |
2104 | RX_MATCH_COPIED_on(prog); | |
2105 | s = savepvn(strbeg, i); | |
2106 | prog->subbeg = s; | |
2107 | } | |
d6a28714 | 2108 | prog->sublen = i; |
d6a28714 JH |
2109 | } |
2110 | else { | |
2111 | prog->subbeg = strbeg; | |
2112 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2113 | } | |
2114 | } | |
9041c2e3 | 2115 | |
d6a28714 JH |
2116 | return 1; |
2117 | ||
2118 | phooey: | |
a3621e74 | 2119 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2120 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2121 | if (PL_reg_eval_set) |
4f639d21 | 2122 | restore_pos(aTHX_ prog); |
f8fc2ecf | 2123 | if (progi->swap) { |
c74340f9 YO |
2124 | /* we failed :-( roll it back */ |
2125 | I32 *t; | |
f8fc2ecf YO |
2126 | t = progi->swap->startp; |
2127 | progi->swap->startp = prog->startp; | |
c74340f9 | 2128 | prog->startp = t; |
f8fc2ecf YO |
2129 | t = progi->swap->endp; |
2130 | progi->swap->endp = prog->endp; | |
c74340f9 YO |
2131 | prog->endp = t; |
2132 | } | |
d6a28714 JH |
2133 | return 0; |
2134 | } | |
2135 | ||
6bda09f9 | 2136 | |
ded05c2a YO |
2137 | |
2138 | ||
d6a28714 JH |
2139 | /* |
2140 | - regtry - try match at specific point | |
2141 | */ | |
2142 | STATIC I32 /* 0 failure, 1 success */ | |
24b23f37 | 2143 | S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) |
d6a28714 | 2144 | { |
97aff369 | 2145 | dVAR; |
d6a28714 JH |
2146 | register I32 *sp; |
2147 | register I32 *ep; | |
2148 | CHECKPOINT lastcp; | |
3b0527fe | 2149 | regexp *prog = reginfo->prog; |
f8fc2ecf | 2150 | RXi_GET_DECL(prog,progi); |
a3621e74 | 2151 | GET_RE_DEBUG_FLAGS_DECL; |
24b23f37 | 2152 | reginfo->cutpoint=NULL; |
d6a28714 | 2153 | |
bbe252da | 2154 | if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) { |
d6a28714 JH |
2155 | MAGIC *mg; |
2156 | ||
2157 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2158 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2159 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2160 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2161 | )); |
ea8d6ae1 | 2162 | SAVESTACK_CXPOS(); |
d6a28714 JH |
2163 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2164 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2165 | SAVETMPS; | |
2166 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2167 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2168 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2169 | ||
3b0527fe | 2170 | if (reginfo->sv) { |
d6a28714 | 2171 | /* Make $_ available to executed code. */ |
3b0527fe | 2172 | if (reginfo->sv != DEFSV) { |
59f00321 | 2173 | SAVE_DEFSV; |
3b0527fe | 2174 | DEFSV = reginfo->sv; |
b8c5462f | 2175 | } |
d6a28714 | 2176 | |
3b0527fe DM |
2177 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2178 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2179 | /* prepare for quick setting of pos */ |
d300d9fa | 2180 | #ifdef PERL_OLD_COPY_ON_WRITE |
51a9ea20 NC |
2181 | if (SvIsCOW(reginfo->sv)) |
2182 | sv_force_normal_flags(reginfo->sv, 0); | |
d300d9fa | 2183 | #endif |
3dab1dad | 2184 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2185 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2186 | mg->mg_len = -1; |
b8c5462f | 2187 | } |
d6a28714 JH |
2188 | PL_reg_magic = mg; |
2189 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2190 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2191 | } |
09687e5a | 2192 | if (!PL_reg_curpm) { |
a02a5408 | 2193 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2194 | #ifdef USE_ITHREADS |
2195 | { | |
be8e71aa | 2196 | SV* const repointer = newSViv(0); |
577e12cc | 2197 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2198 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2199 | av_push(PL_regex_padav,repointer); |
2200 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2201 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2202 | } | |
2203 | #endif | |
2204 | } | |
aaa362c4 | 2205 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2206 | PL_reg_oldcurpm = PL_curpm; |
2207 | PL_curpm = PL_reg_curpm; | |
2208 | if (RX_MATCH_COPIED(prog)) { | |
2209 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2210 | since it may be needed if this match fails. Thus | |
2211 | $` inside (?{}) could fail... */ | |
2212 | PL_reg_oldsaved = prog->subbeg; | |
2213 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2214 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2215 | PL_nrs = prog->saved_copy; |
2216 | #endif | |
d6a28714 JH |
2217 | RX_MATCH_COPIED_off(prog); |
2218 | } | |
2219 | else | |
bd61b366 | 2220 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2221 | prog->subbeg = PL_bostr; |
2222 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2223 | } | |
24b23f37 YO |
2224 | DEBUG_EXECUTE_r(PL_reg_starttry = *startpos); |
2225 | prog->startp[0] = *startpos - PL_bostr; | |
2226 | PL_reginput = *startpos; | |
d6a28714 | 2227 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2228 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2229 | prog->lastparen = 0; |
03994de8 | 2230 | prog->lastcloseparen = 0; |
d6a28714 | 2231 | PL_regsize = 0; |
6bda09f9 YO |
2232 | PL_regstartp = prog->startp; |
2233 | PL_regendp = prog->endp; | |
d6a28714 JH |
2234 | if (PL_reg_start_tmpl <= prog->nparens) { |
2235 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2236 | if(PL_reg_start_tmp) | |
2237 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2238 | else | |
a02a5408 | 2239 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2240 | } |
2241 | ||
2242 | /* XXXX What this code is doing here?!!! There should be no need | |
2243 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2244 | this! --ilya*/ |
dafc8851 JH |
2245 | |
2246 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2247 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2248 | * PL_reglastparen), is not needed at all by the test suite |
2249 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2250 | * enough, for building DynaLoader, or otherwise this | |
2251 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2252 | * will happen. Meanwhile, this code *is* needed for the | |
2253 | * above-mentioned test suite tests to succeed. The common theme | |
2254 | * on those tests seems to be returning null fields from matches. | |
2255 | * --jhi */ | |
dafc8851 | 2256 | #if 1 |
c74340f9 YO |
2257 | sp = PL_regstartp; |
2258 | ep = PL_regendp; | |
d6a28714 | 2259 | if (prog->nparens) { |
097eb12c | 2260 | register I32 i; |
eb160463 | 2261 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2262 | *++sp = -1; |
2263 | *++ep = -1; | |
2264 | } | |
2265 | } | |
dafc8851 | 2266 | #endif |
02db2b7b | 2267 | REGCP_SET(lastcp); |
f8fc2ecf | 2268 | if (regmatch(reginfo, progi->program + 1)) { |
c74340f9 | 2269 | PL_regendp[0] = PL_reginput - PL_bostr; |
d6a28714 JH |
2270 | return 1; |
2271 | } | |
24b23f37 YO |
2272 | if (reginfo->cutpoint) |
2273 | *startpos= reginfo->cutpoint; | |
02db2b7b | 2274 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2275 | return 0; |
2276 | } | |
2277 | ||
02db2b7b | 2278 | |
8ba1375e MJD |
2279 | #define sayYES goto yes |
2280 | #define sayNO goto no | |
262b90c4 | 2281 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2282 | |
f9f4320a YO |
2283 | /* we dont use STMT_START/END here because it leads to |
2284 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2285 | #define CACHEsayNO \ | |
c476f425 DM |
2286 | if (ST.cache_mask) \ |
2287 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2288 | sayNO |
3298f257 | 2289 | |
a3621e74 | 2290 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2291 | 'failed...' are printed. It should be set such that messages |
2292 | are inline with the regop output that created them. | |
a3621e74 | 2293 | */ |
265c4333 | 2294 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2295 | |
2296 | ||
2297 | /* Make sure there is a test for this +1 options in re_tests */ | |
2298 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2299 | ||
40a82448 DM |
2300 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2301 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2302 | |
86545054 DM |
2303 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2304 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2305 | ||
5d9a96ca DM |
2306 | /* grab a new slab and return the first slot in it */ |
2307 | ||
2308 | STATIC regmatch_state * | |
2309 | S_push_slab(pTHX) | |
2310 | { | |
54df2634 NC |
2311 | #if PERL_VERSION < 9 |
2312 | dMY_CXT; | |
2313 | #endif | |
5d9a96ca DM |
2314 | regmatch_slab *s = PL_regmatch_slab->next; |
2315 | if (!s) { | |
2316 | Newx(s, 1, regmatch_slab); | |
2317 | s->prev = PL_regmatch_slab; | |
2318 | s->next = NULL; | |
2319 | PL_regmatch_slab->next = s; | |
2320 | } | |
2321 | PL_regmatch_slab = s; | |
86545054 | 2322 | return SLAB_FIRST(s); |
5d9a96ca | 2323 | } |
5b47454d | 2324 | |
95b24440 | 2325 | |
40a82448 DM |
2326 | /* push a new state then goto it */ |
2327 | ||
2328 | #define PUSH_STATE_GOTO(state, node) \ | |
2329 | scan = node; \ | |
2330 | st->resume_state = state; \ | |
2331 | goto push_state; | |
2332 | ||
2333 | /* push a new state with success backtracking, then goto it */ | |
2334 | ||
2335 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2336 | scan = node; \ | |
2337 | st->resume_state = state; \ | |
2338 | goto push_yes_state; | |
2339 | ||
aa283a38 | 2340 | |
aa283a38 | 2341 | |
d6a28714 | 2342 | /* |
95b24440 | 2343 | |
bf1f174e DM |
2344 | regmatch() - main matching routine |
2345 | ||
2346 | This is basically one big switch statement in a loop. We execute an op, | |
2347 | set 'next' to point the next op, and continue. If we come to a point which | |
2348 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2349 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2350 | state, and re-enter the loop at the state indicated. If there are no more | |
2351 | states to pop, we return failure. | |
2352 | ||
2353 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2354 | after successfully matching one A, we need to go back and try to | |
2355 | match another one; similarly for lookahead assertions: if the assertion | |
2356 | completes successfully, we backtrack to the state just before the assertion | |
2357 | and then carry on. In these cases, the pushed state is marked as | |
2358 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2359 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2360 | the nearest yes state, discarding any intermediate failure-only states. | |
2361 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2362 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2363 | it to free the inner regex. | |
2364 | ||
2365 | Note that failure backtracking rewinds the cursor position, while | |
2366 | success backtracking leaves it alone. | |
2367 | ||
2368 | A pattern is complete when the END op is executed, while a subpattern | |
2369 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2370 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2371 | behaviour. | |
2372 | ||
2373 | A common convention in this function is to use A and B to refer to the two | |
2374 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2375 | the subpattern to be matched possibly multiple times, while B is the entire | |
2376 | rest of the pattern. Variable and state names reflect this convention. | |
2377 | ||
2378 | The states in the main switch are the union of ops and failure/success of | |
2379 | substates associated with with that op. For example, IFMATCH is the op | |
2380 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2381 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2382 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2383 | just failed to match A. Resume states always come in pairs. The backtrack | |
2384 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2385 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2386 | on success or failure. | |
2387 | ||
2388 | The struct that holds a backtracking state is actually a big union, with | |
2389 | one variant for each major type of op. The variable st points to the | |
2390 | top-most backtrack struct. To make the code clearer, within each | |
2391 | block of code we #define ST to alias the relevant union. | |
2392 | ||
2393 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2394 | implementation: | |
2395 | ||
2396 | switch (state) { | |
2397 | .... | |
2398 | ||
2399 | #define ST st->u.ifmatch | |
2400 | ||
2401 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2402 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2403 | ... |
bf1f174e DM |
2404 | // push a yes backtrack state with a resume value of |
2405 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2406 | // first node of A: | |
2407 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2408 | // NOTREACHED | |
2409 | ||
2410 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2411 | next = B; | |
2412 | bar = ST.foo; // do something with the preserved value | |
2413 | break; | |
2414 | ||
2415 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2416 | ...; // do some housekeeping, then ... | |
2417 | sayNO; // propagate the failure | |
2418 | ||
2419 | #undef ST | |
95b24440 | 2420 | |
bf1f174e DM |
2421 | ... |
2422 | } | |
95b24440 | 2423 | |
bf1f174e DM |
2424 | For any old-timers reading this who are familiar with the old recursive |
2425 | approach, the code above is equivalent to: | |
95b24440 | 2426 | |
bf1f174e DM |
2427 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2428 | { | |
2429 | int foo = ... | |
95b24440 | 2430 | ... |
bf1f174e DM |
2431 | if (regmatch(A)) { |
2432 | next = B; | |
2433 | bar = foo; | |
2434 | break; | |
95b24440 | 2435 | } |
bf1f174e DM |
2436 | ...; // do some housekeeping, then ... |
2437 | sayNO; // propagate the failure | |
95b24440 | 2438 | } |
bf1f174e DM |
2439 | |
2440 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2441 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2442 | save, then do one of | |
2443 | ||
2444 | PUSH_STATE_GOTO(resume_state, node); | |
2445 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2446 | ||
2447 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2448 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2449 | On backtracking, the free slot is popped, and the saved state becomes the | |
2450 | new free state. An ST.foo field in this new top state can be temporarily | |
2451 | accessed to retrieve values, but once the main loop is re-entered, it | |
2452 | becomes available for reuse. | |
2453 | ||
2454 | Note that the depth of the backtrack stack constantly increases during the | |
2455 | left-to-right execution of the pattern, rather than going up and down with | |
2456 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2457 | end of the pattern, rather than at X in the following: | |
2458 | ||
2459 | /(((X)+)+)+....(Y)+....Z/ | |
2460 | ||
2461 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2462 | (?>A), which pop all the backtrack states associated with A before | |
2463 | continuing. | |
2464 | ||
2465 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2466 | PL_regmatch_state and st always point to the currently active state, | |
2467 | and PL_regmatch_slab points to the slab currently containing | |
2468 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2469 | allocated, and is never freed until interpreter destruction. When the slab | |
2470 | is full, a new one is allocated and chained to the end. At exit from | |
2471 | regmatch(), slabs allocated since entry are freed. | |
2472 | ||
2473 | */ | |
95b24440 | 2474 | |
40a82448 | 2475 | |
5bc10b2c | 2476 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2477 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2478 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2479 | PerlIO_printf(Perl_debug_log, \ | |
5d458dd8 | 2480 | " %*s"pp" %s%s%s%s%s\n", \ |
5bc10b2c | 2481 | depth*2, "", \ |
5d458dd8 YO |
2482 | reg_name[st->resume_state], \ |
2483 | ((st==yes_state||st==mark_state) ? "[" : ""), \ | |
2484 | ((st==yes_state) ? "Y" : ""), \ | |
2485 | ((st==mark_state) ? "M" : ""), \ | |
2486 | ((st==yes_state||st==mark_state) ? "]" : "") \ | |
2487 | ); \ | |
265c4333 | 2488 | }); |
5bc10b2c | 2489 | |
40a82448 | 2490 | |
3dab1dad | 2491 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2492 | |
3df15adc | 2493 | #ifdef DEBUGGING |
5bc10b2c | 2494 | |
ab3bbdeb YO |
2495 | STATIC void |
2496 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2497 | const char *start, const char *end, const char *blurb) | |
2498 | { | |
bbe252da | 2499 | const bool utf8_pat= prog->extflags & RXf_UTF8 ? 1 : 0; |
ab3bbdeb YO |
2500 | if (!PL_colorset) |
2501 | reginitcolors(); | |
2502 | { | |
2503 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2504 | prog->precomp, prog->prelen, 60); | |
2505 | ||
2506 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2507 | start, end - start, 60); | |
2508 | ||
2509 | PerlIO_printf(Perl_debug_log, | |
2510 | "%s%s REx%s %s against %s\n", | |
2511 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2512 | ||
2513 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2514 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2515 | utf8_pat ? "pattern" : "", | |
2516 | utf8_pat && do_utf8 ? " and " : "", | |
2517 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2518 | ); |
2519 | } | |
2520 | } | |
3df15adc YO |
2521 | |
2522 | STATIC void | |
786e8c11 YO |
2523 | S_dump_exec_pos(pTHX_ const char *locinput, |
2524 | const regnode *scan, | |
2525 | const char *loc_regeol, | |
2526 | const char *loc_bostr, | |
2527 | const char *loc_reg_starttry, | |
2528 | const bool do_utf8) | |
07be1b83 | 2529 | { |
786e8c11 | 2530 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2531 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2532 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2533 | /* The part of the string before starttry has one color |
2534 | (pref0_len chars), between starttry and current | |
2535 | position another one (pref_len - pref0_len chars), | |
2536 | after the current position the third one. | |
2537 | We assume that pref0_len <= pref_len, otherwise we | |
2538 | decrease pref0_len. */ | |
786e8c11 YO |
2539 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2540 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2541 | int pref0_len; |
2542 | ||
2543 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2544 | pref_len++; | |
786e8c11 YO |
2545 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2546 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2547 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2548 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2549 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2550 | l--; | |
2551 | if (pref0_len < 0) | |
2552 | pref0_len = 0; | |
2553 | if (pref0_len > pref_len) | |
2554 | pref0_len = pref_len; | |
2555 | { | |
3df15adc | 2556 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2557 | |
ab3bbdeb | 2558 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2559 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2560 | |
ab3bbdeb | 2561 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2562 | (locinput - pref_len + pref0_len), |
1de06328 | 2563 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2564 | |
ab3bbdeb | 2565 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2566 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2567 | |
1de06328 | 2568 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2569 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2570 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2571 | (IV)(locinput - loc_bostr), |
07be1b83 | 2572 | len0, s0, |
07be1b83 | 2573 | len1, s1, |
07be1b83 | 2574 | (docolor ? "" : "> <"), |
07be1b83 | 2575 | len2, s2, |
f9f4320a | 2576 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2577 | ""); |
2578 | } | |
2579 | } | |
3df15adc | 2580 | |
07be1b83 YO |
2581 | #endif |
2582 | ||
0a4db386 YO |
2583 | /* reg_check_named_buff_matched() |
2584 | * Checks to see if a named buffer has matched. The data array of | |
2585 | * buffer numbers corresponding to the buffer is expected to reside | |
2586 | * in the regexp->data->data array in the slot stored in the ARG() of | |
2587 | * node involved. Note that this routine doesn't actually care about the | |
2588 | * name, that information is not preserved from compilation to execution. | |
2589 | * Returns the index of the leftmost defined buffer with the given name | |
2590 | * or 0 if non of the buffers matched. | |
2591 | */ | |
2592 | STATIC I32 | |
2593 | S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) { | |
2594 | I32 n; | |
f8fc2ecf YO |
2595 | RXi_GET_DECL(rex,rexi); |
2596 | SV *sv_dat=(SV*)rexi->data->data[ ARG( scan ) ]; | |
0a4db386 YO |
2597 | I32 *nums=(I32*)SvPVX(sv_dat); |
2598 | for ( n=0; n<SvIVX(sv_dat); n++ ) { | |
2599 | if ((I32)*PL_reglastparen >= nums[n] && | |
2600 | PL_regendp[nums[n]] != -1) | |
2601 | { | |
2602 | return nums[n]; | |
2603 | } | |
2604 | } | |
2605 | return 0; | |
2606 | } | |
2607 | ||
d6a28714 | 2608 | STATIC I32 /* 0 failure, 1 success */ |
24b23f37 | 2609 | S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2610 | { |
54df2634 NC |
2611 | #if PERL_VERSION < 9 |
2612 | dMY_CXT; | |
2613 | #endif | |
27da23d5 | 2614 | dVAR; |
95b24440 | 2615 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2616 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2617 | |
3b0527fe | 2618 | regexp *rex = reginfo->prog; |
f8fc2ecf YO |
2619 | RXi_GET_DECL(rex,rexi); |
2620 | ||
5d9a96ca DM |
2621 | regmatch_slab *orig_slab; |
2622 | regmatch_state *orig_state; | |
a3621e74 | 2623 | |
5d9a96ca DM |
2624 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2625 | register regmatch_state *st; | |
95b24440 | 2626 | |
5d9a96ca DM |
2627 | /* cache heavy used fields of st in registers */ |
2628 | register regnode *scan; | |
2629 | register regnode *next; | |
438e9bae | 2630 | register U32 n = 0; /* general value; init to avoid compiler warning */ |
24d3c4a9 | 2631 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ |
5d9a96ca | 2632 | register char *locinput = PL_reginput; |
5d9a96ca | 2633 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2634 | |
b69b0499 | 2635 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2636 | int depth = 0; /* depth of backtrack stack */ |
1a147d38 | 2637 | int nochange_depth = 0; /* depth of GOSUB recursion with nochange*/ |
77cb431f DM |
2638 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2639 | subpattern */ | |
e2e6a0f1 YO |
2640 | /* mark_state piggy backs on the yes_state logic so that when we unwind |
2641 | the stack on success we can update the mark_state as we go */ | |
2642 | regmatch_state *mark_state = NULL; /* last mark state we have seen */ | |
5d458dd8 | 2643 | |
faec1544 | 2644 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2645 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2646 | U32 state_num; |
5d458dd8 YO |
2647 | bool no_final = 0; /* prevent failure from backtracking? */ |
2648 | bool do_cutgroup = 0; /* no_final only until next branch/trie entry */ | |
e2e6a0f1 | 2649 | char *startpoint = PL_reginput; |
5d458dd8 YO |
2650 | SV *popmark = NULL; /* are we looking for a mark? */ |
2651 | SV *sv_commit = NULL; /* last mark name seen in failure */ | |
2652 | SV *sv_yes_mark = NULL; /* last mark name we have seen | |
2653 | during a successfull match */ | |
2654 | U32 lastopen = 0; /* last open we saw */ | |
2655 | bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0; | |
58e23c8d | 2656 | |
5d458dd8 | 2657 | |
24d3c4a9 DM |
2658 | /* these three flags are set by various ops to signal information to |
2659 | * the very next op. They have a useful lifetime of exactly one loop | |
2660 | * iteration, and are not preserved or restored by state pushes/pops | |
2661 | */ | |
2662 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2663 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2664 | int logical = 0; /* the following EVAL is: | |
2665 | 0: (?{...}) | |
2666 | 1: (?(?{...})X|Y) | |
2667 | 2: (??{...}) | |
2668 | or the following IFMATCH/UNLESSM is: | |
2669 | false: plain (?=foo) | |
2670 | true: used as a condition: (?(?=foo)) | |
2671 | */ | |
2672 | ||
95b24440 | 2673 | #ifdef DEBUGGING |
e68ec53f | 2674 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2675 | #endif |
2676 | ||
58e23c8d | 2677 | DEBUG_OPTIMISE_r( { |
24b23f37 YO |
2678 | PerlIO_printf(Perl_debug_log,"regmatch start\n"); |
2679 | }); | |
5d9a96ca DM |
2680 | /* on first ever call to regmatch, allocate first slab */ |
2681 | if (!PL_regmatch_slab) { | |
2682 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2683 | PL_regmatch_slab->prev = NULL; | |
2684 | PL_regmatch_slab->next = NULL; | |
86545054 | 2685 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2686 | } |
2687 | ||
2688 | /* remember current high-water mark for exit */ | |
2689 | /* XXX this should be done with SAVE* instead */ | |
2690 | orig_slab = PL_regmatch_slab; | |
2691 | orig_state = PL_regmatch_state; | |
2692 | ||
2693 | /* grab next free state slot */ | |
2694 | st = ++PL_regmatch_state; | |
86545054 | 2695 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2696 | st = PL_regmatch_state = S_push_slab(aTHX); |
2697 | ||
d6a28714 JH |
2698 | /* Note that nextchr is a byte even in UTF */ |
2699 | nextchr = UCHARAT(locinput); | |
2700 | scan = prog; | |
2701 | while (scan != NULL) { | |
8ba1375e | 2702 | |
a3621e74 | 2703 | DEBUG_EXECUTE_r( { |
6136c704 | 2704 | SV * const prop = sv_newmortal(); |
1de06328 | 2705 | regnode *rnext=regnext(scan); |
786e8c11 | 2706 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2707 | regprop(rex, prop, scan); |
07be1b83 YO |
2708 | |
2709 | PerlIO_printf(Perl_debug_log, | |
2710 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
f8fc2ecf | 2711 | (IV)(scan - rexi->program), depth*2, "", |
07be1b83 | 2712 | SvPVX_const(prop), |
1de06328 | 2713 | (PL_regkind[OP(scan)] == END || !rnext) ? |
f8fc2ecf | 2714 | 0 : (IV)(rnext - rexi->program)); |
2a782b5b | 2715 | }); |
d6a28714 JH |
2716 | |
2717 | next = scan + NEXT_OFF(scan); | |
2718 | if (next == scan) | |
2719 | next = NULL; | |
40a82448 | 2720 | state_num = OP(scan); |
d6a28714 | 2721 | |
40a82448 DM |
2722 | reenter_switch: |
2723 | switch (state_num) { | |
d6a28714 | 2724 | case BOL: |
7fba1cd6 | 2725 | if (locinput == PL_bostr) |
d6a28714 | 2726 | { |
3b0527fe | 2727 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2728 | break; |
2729 | } | |
d6a28714 JH |
2730 | sayNO; |
2731 | case MBOL: | |
12d33761 HS |
2732 | if (locinput == PL_bostr || |
2733 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2734 | { |
b8c5462f JH |
2735 | break; |
2736 | } | |
d6a28714 JH |
2737 | sayNO; |
2738 | case SBOL: | |
c2a73568 | 2739 | if (locinput == PL_bostr) |
b8c5462f | 2740 | break; |
d6a28714 JH |
2741 | sayNO; |
2742 | case GPOS: | |
3b0527fe | 2743 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2744 | break; |
2745 | sayNO; | |
ee9b8eae YO |
2746 | |
2747 | case KEEPS: | |
2748 | /* update the startpoint */ | |
2749 | st->u.keeper.val = PL_regstartp[0]; | |
2750 | PL_reginput = locinput; | |
2751 | PL_regstartp[0] = locinput - PL_bostr; | |
2752 | PUSH_STATE_GOTO(KEEPS_next, next); | |
2753 | /*NOT-REACHED*/ | |
2754 | case KEEPS_next_fail: | |
2755 | /* rollback the start point change */ | |
2756 | PL_regstartp[0] = st->u.keeper.val; | |
2757 | sayNO_SILENT; | |
2758 | /*NOT-REACHED*/ | |
d6a28714 | 2759 | case EOL: |
d6a28714 JH |
2760 | goto seol; |
2761 | case MEOL: | |
d6a28714 | 2762 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2763 | sayNO; |
b8c5462f | 2764 | break; |
d6a28714 JH |
2765 | case SEOL: |
2766 | seol: | |
2767 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2768 | sayNO; |
d6a28714 | 2769 | if (PL_regeol - locinput > 1) |
b8c5462f | 2770 | sayNO; |
b8c5462f | 2771 | break; |
d6a28714 JH |
2772 | case EOS: |
2773 | if (PL_regeol != locinput) | |
b8c5462f | 2774 | sayNO; |
d6a28714 | 2775 | break; |
ffc61ed2 | 2776 | case SANY: |
d6a28714 | 2777 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2778 | sayNO; |
f33976b4 DB |
2779 | if (do_utf8) { |
2780 | locinput += PL_utf8skip[nextchr]; | |
2781 | if (locinput > PL_regeol) | |
2782 | sayNO; | |
2783 | nextchr = UCHARAT(locinput); | |
2784 | } | |
2785 | else | |
2786 | nextchr = UCHARAT(++locinput); | |
2787 | break; | |
2788 | case CANY: | |
2789 | if (!nextchr && locinput >= PL_regeol) | |
2790 | sayNO; | |
b8c5462f | 2791 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2792 | break; |
ffc61ed2 | 2793 | case REG_ANY: |
1aa99e6b IH |
2794 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2795 | sayNO; | |
2796 | if (do_utf8) { | |
b8c5462f | 2797 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2798 | if (locinput > PL_regeol) |
2799 | sayNO; | |
a0ed51b3 | 2800 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2801 | } |
1aa99e6b IH |
2802 | else |
2803 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2804 | break; |
166ba7cd DM |
2805 | |
2806 | #undef ST | |
2807 | #define ST st->u.trie | |
786e8c11 YO |
2808 | case TRIEC: |
2809 | /* In this case the charclass data is available inline so | |
2810 | we can fail fast without a lot of extra overhead. | |
2811 | */ | |
2812 | if (scan->flags == EXACT || !do_utf8) { | |
2813 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2814 | DEBUG_EXECUTE_r( | |
2815 | PerlIO_printf(Perl_debug_log, | |
2816 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2817 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2818 | ); |
2819 | sayNO_SILENT; | |
2820 | /* NOTREACHED */ | |
2821 | } | |
2822 | } | |
2823 | /* FALL THROUGH */ | |
5b47454d | 2824 | case TRIE: |
3dab1dad | 2825 | { |
07be1b83 | 2826 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2827 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2828 | trie_type = do_utf8 ? | |
2829 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2830 | : trie_plain; | |
2831 | ||
2832 | /* what trie are we using right now */ | |
be8e71aa | 2833 | reg_trie_data * const trie |
f8fc2ecf | 2834 | = (reg_trie_data*)rexi->data->data[ ARG( scan ) ]; |
55eed653 | 2835 | HV * widecharmap = (HV *)rexi->data->data[ ARG( scan ) + 1 ]; |
3dab1dad | 2836 | U32 state = trie->startstate; |
166ba7cd | 2837 | |
3dab1dad YO |
2838 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2839 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2840 | ) { | |
2841 | if (trie->states[ state ].wordnum) { | |
2842 | DEBUG_EXECUTE_r( | |
2843 | PerlIO_printf(Perl_debug_log, | |
2844 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2845 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2846 | ); |
2847 | break; | |
2848 | } else { | |
2849 | DEBUG_EXECUTE_r( | |
2850 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2851 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2852 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2853 | ); |
2854 | sayNO_SILENT; | |
2855 | } | |
2856 | } | |
166ba7cd | 2857 | |
786e8c11 YO |
2858 | { |
2859 | U8 *uc = ( U8* )locinput; | |
2860 | ||
2861 | STRLEN len = 0; | |
2862 | STRLEN foldlen = 0; | |
2863 | U8 *uscan = (U8*)NULL; | |
2864 | STRLEN bufflen=0; | |
2865 | SV *sv_accept_buff = NULL; | |
2866 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2867 | ||
2868 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2869 | ST.B = next; | |
2870 | ST.jump = trie->jump; | |
786e8c11 | 2871 | ST.me = scan; |
786e8c11 | 2872 | |
07be1b83 YO |
2873 | /* |
2874 | traverse the TRIE keeping track of all accepting states | |
2875 | we transition through until we get to a failing node. | |
2876 | */ | |
2877 | ||
a3621e74 | 2878 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 2879 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 2880 | UV uvc = 0; |
786e8c11 YO |
2881 | U16 charid; |
2882 | /* We use charid to hold the wordnum as we don't use it | |
2883 | for charid until after we have done the wordnum logic. | |
2884 | We define an alias just so that the wordnum logic reads | |
2885 | more naturally. */ | |
2886 | ||
2887 | #define got_wordnum charid | |
2888 | got_wordnum = trie->states[ state ].wordnum; | |
2889 | ||
2890 | if ( got_wordnum ) { | |
2891 | if ( ! ST.accepted ) { | |
5b47454d DM |
2892 | ENTER; |
2893 | SAVETMPS; | |
2894 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2895 | sv_accept_buff=newSV(bufflen * | |
2896 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2897 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2898 | SvPOK_on(sv_accept_buff); |
2899 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2900 | SAVETMPS; |
2901 | ST.accept_buff = | |
5b47454d DM |
2902 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2903 | } | |
786e8c11 | 2904 | do { |
166ba7cd | 2905 | if (ST.accepted >= bufflen) { |
5b47454d | 2906 | bufflen *= 2; |
166ba7cd | 2907 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2908 | SvGROW(sv_accept_buff, |
2909 | bufflen * sizeof(reg_trie_accepted)); | |
2910 | } | |
2911 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2912 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2913 | |
786e8c11 YO |
2914 | |
2915 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2916 | ST.accept_buff[ST.accepted].endpos = uc; | |
2917 | ++ST.accepted; | |
2918 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2919 | } | |
2920 | #undef got_wordnum | |
a3621e74 | 2921 | |
07be1b83 | 2922 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2923 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2924 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2925 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2926 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2927 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2928 | }); |
a3621e74 YO |
2929 | |
2930 | if ( base ) { | |
55eed653 NC |
2931 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
2932 | uscan, len, uvc, charid, foldlen, | |
2933 | foldbuf, uniflags); | |
a3621e74 | 2934 | |
5b47454d DM |
2935 | if (charid && |
2936 | (base + charid > trie->uniquecharcount ) | |
2937 | && (base + charid - 1 - trie->uniquecharcount | |
2938 | < trie->lasttrans) | |
2939 | && trie->trans[base + charid - 1 - | |
2940 | trie->uniquecharcount].check == state) | |
2941 | { | |
2942 | state = trie->trans[base + charid - 1 - | |
2943 | trie->uniquecharcount ].next; | |
2944 | } | |
2945 | else { | |
2946 | state = 0; | |
2947 | } | |
2948 | uc += len; | |
2949 | ||
2950 | } | |
2951 | else { | |
a3621e74 YO |
2952 | state = 0; |
2953 | } | |
2954 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 2955 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2956 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 2957 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
2958 | ); |
2959 | } | |
166ba7cd | 2960 | if (!ST.accepted ) |
a3621e74 | 2961 | sayNO; |
a3621e74 | 2962 | |
166ba7cd DM |
2963 | DEBUG_EXECUTE_r( |
2964 | PerlIO_printf( Perl_debug_log, | |
2965 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 2966 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
2967 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
2968 | ); | |
786e8c11 | 2969 | }} |
166ba7cd DM |
2970 | |
2971 | /* FALL THROUGH */ | |
166ba7cd | 2972 | case TRIE_next_fail: /* we failed - try next alterative */ |
5d458dd8 YO |
2973 | if (do_cutgroup) { |
2974 | do_cutgroup = 0; | |
2975 | no_final = 0; | |
2976 | } | |
166ba7cd DM |
2977 | if ( ST.accepted == 1 ) { |
2978 | /* only one choice left - just continue */ | |
2979 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
2980 | AV *const trie_words |
2981 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
2982 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 2983 | ST.accept_buff[ 0 ].wordnum-1, 0 ); |
de734bd5 A |
2984 | SV *sv= tmp ? sv_newmortal() : NULL; |
2985 | ||
166ba7cd DM |
2986 | PerlIO_printf( Perl_debug_log, |
2987 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 2988 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 2989 | ST.accept_buff[ 0 ].wordnum, |
de734bd5 A |
2990 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
2991 | PL_colors[0], PL_colors[1], | |
2992 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
2993 | ) | |
2994 | : "not compiled under -Dr", | |
166ba7cd DM |
2995 | PL_colors[5] ); |
2996 | }); | |
2997 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
2998 | /* in this case we free tmps/leave before we call regmatch | |
2999 | as we wont be using accept_buff again. */ | |
5d458dd8 | 3000 | |
166ba7cd DM |
3001 | locinput = PL_reginput; |
3002 | nextchr = UCHARAT(locinput); | |
5d458dd8 YO |
3003 | if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum]) |
3004 | scan = ST.B; | |
3005 | else | |
3006 | scan = ST.me + ST.jump[ST.accept_buff[0].wordnum]; | |
3007 | if (!has_cutgroup) { | |
3008 | FREETMPS; | |
3009 | LEAVE; | |
3010 | } else { | |
3011 | ST.accepted--; | |
3012 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
3013 | } | |
786e8c11 | 3014 | |
166ba7cd DM |
3015 | continue; /* execute rest of RE */ |
3016 | } | |
3017 | ||
3018 | if (!ST.accepted-- ) { | |
5d458dd8 YO |
3019 | DEBUG_EXECUTE_r({ |
3020 | PerlIO_printf( Perl_debug_log, | |
3021 | "%*s %sTRIE failed...%s\n", | |
3022 | REPORT_CODE_OFF+depth*2, "", | |
3023 | PL_colors[4], | |
3024 | PL_colors[5] ); | |
3025 | }); | |
166ba7cd DM |
3026 | FREETMPS; |
3027 | LEAVE; | |
5d458dd8 | 3028 | sayNO_SILENT; |
166ba7cd DM |
3029 | } |
3030 | ||
a3621e74 | 3031 | /* |
166ba7cd DM |
3032 | There are at least two accepting states left. Presumably |
3033 | the number of accepting states is going to be low, | |
3034 | typically two. So we simply scan through to find the one | |
3035 | with lowest wordnum. Once we find it, we swap the last | |
3036 | state into its place and decrement the size. We then try to | |
3037 | match the rest of the pattern at the point where the word | |
3038 | ends. If we succeed, control just continues along the | |
3039 | regex; if we fail we return here to try the next accepting | |
3040 | state | |
3041 | */ | |
a3621e74 | 3042 | |
166ba7cd DM |
3043 | { |
3044 | U32 best = 0; | |
3045 | U32 cur; | |
3046 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
3047 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 3048 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 3049 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 3050 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
3051 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
3052 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
3053 | ); | |
3054 | ||
3055 | if (ST.accept_buff[cur].wordnum < | |
3056 | ST.accept_buff[best].wordnum) | |
3057 | best = cur; | |
a3621e74 | 3058 | } |
166ba7cd DM |
3059 | |
3060 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
3061 | AV *const trie_words |
3062 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
3063 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 3064 | ST.accept_buff[ best ].wordnum - 1, 0 ); |
7f69552c | 3065 | regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ? |
265c4333 | 3066 | ST.B : |
7f69552c | 3067 | ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
de734bd5 A |
3068 | SV *sv= tmp ? sv_newmortal() : NULL; |
3069 | ||
265c4333 YO |
3070 | PerlIO_printf( Perl_debug_log, |
3071 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 3072 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3073 | ST.accept_buff[best].wordnum, |
de734bd5 A |
3074 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3075 | PL_colors[0], PL_colors[1], | |
3076 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3077 | ) : "not compiled under -Dr", | |
265c4333 | 3078 | REG_NODE_NUM(nextop), |
166ba7cd DM |
3079 | PL_colors[5] ); |
3080 | }); | |
3081 | ||
3082 | if ( best<ST.accepted ) { | |
3083 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
3084 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
3085 | ST.accept_buff[ ST.accepted ] = tmp; | |
3086 | best = ST.accepted; | |
a3621e74 | 3087 | } |
166ba7cd | 3088 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
7f69552c | 3089 | if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) { |
5d458dd8 | 3090 | scan = ST.B; |
786e8c11 YO |
3091 | /* NOTREACHED */ |
3092 | } else { | |
5d458dd8 | 3093 | scan = ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
786e8c11 YO |
3094 | /* NOTREACHED */ |
3095 | } | |
5d458dd8 YO |
3096 | if (has_cutgroup) { |
3097 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
3098 | /* NOTREACHED */ | |
3099 | } else { | |
3100 | PUSH_STATE_GOTO(TRIE_next, scan); | |
3101 | /* NOTREACHED */ | |
3102 | } | |
786e8c11 | 3103 | /* NOTREACHED */ |
166ba7cd | 3104 | } |
166ba7cd | 3105 | /* NOTREACHED */ |
5d458dd8 YO |
3106 | case TRIE_next: |
3107 | FREETMPS; | |
3108 | LEAVE; | |
3109 | sayYES; | |
166ba7cd DM |
3110 | #undef ST |
3111 | ||
95b24440 DM |
3112 | case EXACT: { |
3113 | char *s = STRING(scan); | |
24d3c4a9 | 3114 | ln = STR_LEN(scan); |
eb160463 | 3115 | if (do_utf8 != UTF) { |
bc517b45 | 3116 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 3117 | char *l = locinput; |
24d3c4a9 | 3118 | const char * const e = s + ln; |
a72c7584 | 3119 | |
5ff6fc6d JH |
3120 | if (do_utf8) { |
3121 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 3122 | while (s < e) { |
a3b680e6 | 3123 | STRLEN ulen; |
1aa99e6b | 3124 | if (l >= PL_regeol) |
5ff6fc6d JH |
3125 | sayNO; |
3126 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 3127 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 3128 | uniflags)) |
5ff6fc6d | 3129 | sayNO; |
bc517b45 | 3130 | l += ulen; |
5ff6fc6d | 3131 | s ++; |
1aa99e6b | 3132 | } |
5ff6fc6d JH |
3133 | } |
3134 | else { | |
3135 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 3136 | while (s < e) { |
a3b680e6 | 3137 | STRLEN ulen; |
1aa99e6b IH |
3138 | if (l >= PL_regeol) |
3139 | sayNO; | |
5ff6fc6d | 3140 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 3141 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 3142 | uniflags)) |
1aa99e6b | 3143 | sayNO; |
bc517b45 | 3144 | s += ulen; |
a72c7584 | 3145 | l ++; |
1aa99e6b | 3146 | } |
5ff6fc6d | 3147 | } |
1aa99e6b IH |
3148 | locinput = l; |
3149 | nextchr = UCHARAT(locinput); | |
3150 | break; | |
3151 | } | |
bc517b45 | 3152 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
3153 | /* Inline the first character, for speed. */ |
3154 | if (UCHARAT(s) != nextchr) | |
3155 | sayNO; | |
24d3c4a9 | 3156 | if (PL_regeol - locinput < ln) |
d6a28714 | 3157 | sayNO; |
24d3c4a9 | 3158 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3159 | sayNO; |
24d3c4a9 | 3160 | locinput += ln; |
d6a28714 JH |
3161 | nextchr = UCHARAT(locinput); |
3162 | break; | |
95b24440 | 3163 | } |
d6a28714 | 3164 | case EXACTFL: |
b8c5462f JH |
3165 | PL_reg_flags |= RF_tainted; |
3166 | /* FALL THROUGH */ | |
95b24440 | 3167 | case EXACTF: { |
be8e71aa | 3168 | char * const s = STRING(scan); |
24d3c4a9 | 3169 | ln = STR_LEN(scan); |
d6a28714 | 3170 | |
d07ddd77 JH |
3171 | if (do_utf8 || UTF) { |
3172 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3173 | const char * const l = locinput; |
d07ddd77 | 3174 | char *e = PL_regeol; |
bc517b45 | 3175 | |
24d3c4a9 | 3176 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3177 | l, &e, 0, do_utf8)) { |
5486206c JH |
3178 | /* One more case for the sharp s: |
3179 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3180 | * the 0xC3 0x9F are the UTF-8 | |
3181 | * byte sequence for the U+00DF. */ | |
3182 | if (!(do_utf8 && | |
3183 | toLOWER(s[0]) == 's' && | |
24d3c4a9 | 3184 | ln >= 2 && |
5486206c JH |
3185 | toLOWER(s[1]) == 's' && |
3186 | (U8)l[0] == 0xC3 && | |
3187 | e - l >= 2 && | |
3188 | (U8)l[1] == 0x9F)) | |
3189 | sayNO; | |
3190 | } | |
d07ddd77 JH |
3191 | locinput = e; |
3192 | nextchr = UCHARAT(locinput); | |
3193 | break; | |
a0ed51b3 | 3194 | } |
d6a28714 | 3195 | |
bc517b45 JH |
3196 | /* Neither the target and the pattern are utf8. */ |
3197 | ||
d6a28714 JH |
3198 | /* Inline the first character, for speed. */ |
3199 | if (UCHARAT(s) != nextchr && | |
3200 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3201 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3202 | sayNO; |
24d3c4a9 | 3203 | if (PL_regeol - locinput < ln) |
b8c5462f | 3204 | sayNO; |
24d3c4a9 DM |
3205 | if (ln > 1 && (OP(scan) == EXACTF |
3206 | ? ibcmp(s, locinput, ln) | |
3207 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3208 | sayNO; |
24d3c4a9 | 3209 | locinput += ln; |
d6a28714 | 3210 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3211 | break; |
95b24440 | 3212 | } |
d6a28714 | 3213 | case ANYOF: |
ffc61ed2 | 3214 | if (do_utf8) { |
9e55ce06 JH |
3215 | STRLEN inclasslen = PL_regeol - locinput; |
3216 | ||
32fc9b6a | 3217 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3218 | goto anyof_fail; |
ffc61ed2 JH |
3219 | if (locinput >= PL_regeol) |
3220 | sayNO; | |
0f0076b4 | 3221 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3222 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3223 | break; |
ffc61ed2 JH |
3224 | } |
3225 | else { | |
3226 | if (nextchr < 0) | |
3227 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3228 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3229 | goto anyof_fail; |
ffc61ed2 JH |
3230 | if (!nextchr && locinput >= PL_regeol) |
3231 | sayNO; | |
3232 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3233 | break; |
3234 | } | |
262b90c4 | 3235 | anyof_fail: |
e0f9d4a8 JH |
3236 | /* If we might have the case of the German sharp s |
3237 | * in a casefolding Unicode character class. */ | |
3238 | ||
ebc501f0 JH |
3239 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3240 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3241 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3242 | } |
e0f9d4a8 JH |
3243 | else |
3244 | sayNO; | |
b8c5462f | 3245 | break; |
d6a28714 | 3246 | case ALNUML: |
b8c5462f JH |
3247 | PL_reg_flags |= RF_tainted; |
3248 | /* FALL THROUGH */ | |
d6a28714 | 3249 | case ALNUM: |
b8c5462f | 3250 | if (!nextchr) |
4633a7c4 | 3251 | sayNO; |
ffc61ed2 | 3252 | if (do_utf8) { |
1a4fad37 | 3253 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3254 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3255 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3256 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3257 | { |
3258 | sayNO; | |
a0ed51b3 | 3259 | } |
b8c5462f | 3260 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3261 | nextchr = UCHARAT(locinput); |
3262 | break; | |
3263 | } | |
ffc61ed2 | 3264 | if (!(OP(scan) == ALNUM |
d6a28714 | 3265 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3266 | sayNO; |
b8c5462f | 3267 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3268 | break; |
d6a28714 | 3269 | case NALNUML: |
b8c5462f JH |
3270 | PL_reg_flags |= RF_tainted; |
3271 | /* FALL THROUGH */ | |
d6a28714 JH |
3272 | case NALNUM: |
3273 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3274 | sayNO; |
ffc61ed2 | 3275 | if (do_utf8) { |
1a4fad37 | 3276 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3277 | if (OP(scan) == NALNUM |
bb7a0f54 | 3278 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3279 | : isALNUM_LC_utf8((U8*)locinput)) |
3280 | { | |
b8c5462f | 3281 | sayNO; |
d6a28714 | 3282 | } |
b8c5462f JH |
3283 | locinput += PL_utf8skip[nextchr]; |
3284 | nextchr = UCHARAT(locinput); | |
3285 | break; | |
3286 | } | |
ffc61ed2 | 3287 | if (OP(scan) == NALNUM |
d6a28714 | 3288 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3289 | sayNO; |
76e3520e | 3290 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3291 | break; |
d6a28714 JH |
3292 | case BOUNDL: |
3293 | case NBOUNDL: | |
3280af22 | 3294 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3295 | /* FALL THROUGH */ |
d6a28714 JH |
3296 | case BOUND: |
3297 | case NBOUND: | |
3298 | /* was last char in word? */ | |
ffc61ed2 | 3299 | if (do_utf8) { |
12d33761 | 3300 | if (locinput == PL_bostr) |
24d3c4a9 | 3301 | ln = '\n'; |
ffc61ed2 | 3302 | else { |
a3b680e6 | 3303 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3304 | |
24d3c4a9 | 3305 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3306 | } |
3307 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3308 | ln = isALNUM_uni(ln); |
1a4fad37 | 3309 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3310 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3311 | } |
3312 | else { | |
24d3c4a9 | 3313 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3314 | n = isALNUM_LC_utf8((U8*)locinput); |
3315 | } | |
a0ed51b3 | 3316 | } |
d6a28714 | 3317 | else { |
24d3c4a9 | 3318 | ln = (locinput != PL_bostr) ? |
12d33761 | 3319 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3320 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3321 | ln = isALNUM(ln); |
ffc61ed2 JH |
3322 | n = isALNUM(nextchr); |
3323 | } | |
3324 | else { | |
24d3c4a9 | 3325 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3326 | n = isALNUM_LC(nextchr); |
3327 | } | |
d6a28714 | 3328 | } |
24d3c4a9 | 3329 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3330 | OP(scan) == BOUNDL)) |
3331 | sayNO; | |
a0ed51b3 | 3332 | break; |
d6a28714 | 3333 | case SPACEL: |
3280af22 | 3334 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3335 | /* FALL THROUGH */ |
d6a28714 | 3336 | case SPACE: |
9442cb0e | 3337 | if (!nextchr) |
4633a7c4 | 3338 | sayNO; |
1aa99e6b | 3339 | if (do_utf8) { |
fd400ab9 | 3340 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3341 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3342 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3343 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3344 | : isSPACE_LC_utf8((U8*)locinput))) |
3345 | { | |
3346 | sayNO; | |
3347 | } | |
3348 | locinput += PL_utf8skip[nextchr]; | |
3349 | nextchr = UCHARAT(locinput); | |
3350 | break; | |
d6a28714 | 3351 | } |
ffc61ed2 JH |
3352 | if (!(OP(scan) == SPACE |
3353 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3354 | sayNO; | |
3355 | nextchr = UCHARAT(++locinput); | |
3356 | } | |
3357 | else { | |
3358 | if (!(OP(scan) == SPACE | |
3359 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3360 | sayNO; | |
3361 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3362 | } |
a0ed51b3 | 3363 | break; |
d6a28714 | 3364 | case NSPACEL: |
3280af22 | 3365 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3366 | /* FALL THROUGH */ |
d6a28714 | 3367 | case NSPACE: |
9442cb0e | 3368 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3369 | sayNO; |
1aa99e6b | 3370 | if (do_utf8) { |
1a4fad37 | 3371 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3372 | if (OP(scan) == NSPACE |
bb7a0f54 | 3373 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3374 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3375 | { |
3376 | sayNO; | |
3377 | } | |
3378 | locinput += PL_utf8skip[nextchr]; | |
3379 | nextchr = UCHARAT(locinput); | |
3380 | break; | |
a0ed51b3 | 3381 | } |
ffc61ed2 | 3382 | if (OP(scan) == NSPACE |
d6a28714 | 3383 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3384 | sayNO; |
76e3520e | 3385 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3386 | break; |
d6a28714 | 3387 | case DIGITL: |
a0ed51b3 LW |
3388 | PL_reg_flags |= RF_tainted; |
3389 | /* FALL THROUGH */ | |
d6a28714 | 3390 | case DIGIT: |
9442cb0e | 3391 | if (!nextchr) |
a0ed51b3 | 3392 | sayNO; |
1aa99e6b | 3393 | if (do_utf8) { |
1a4fad37 | 3394 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3395 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3396 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3397 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3398 | { |
a0ed51b3 | 3399 | sayNO; |
dfe13c55 | 3400 | } |