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, |
b94e2f88 | 59 | **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 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 || \ | |
3dab1dad | 129 | (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \ |
e2d8ce26 JP |
130 | ) |
131 | ||
cca55fe3 | 132 | #define HAS_TEXT(rn) ( \ |
3dab1dad | 133 | PL_regkind[OP(rn)] == EXACT || PL_regkind[OP(rn)] == REF \ |
cca55fe3 | 134 | ) |
e2d8ce26 | 135 | |
a84d97b6 HS |
136 | /* |
137 | Search for mandatory following text node; for lookahead, the text must | |
138 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
139 | */ | |
cca55fe3 | 140 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
3dab1dad YO |
141 | while (JUMPABLE(rn)) { \ |
142 | const OPCODE type = OP(rn); \ | |
143 | if (type == SUSPEND || PL_regkind[type] == CURLY) \ | |
e2d8ce26 | 144 | rn = NEXTOPER(NEXTOPER(rn)); \ |
3dab1dad | 145 | else if (type == PLUS) \ |
cca55fe3 | 146 | rn = NEXTOPER(rn); \ |
3dab1dad | 147 | else if (type == IFMATCH) \ |
a84d97b6 | 148 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ |
e2d8ce26 | 149 | else rn += NEXT_OFF(rn); \ |
3dab1dad | 150 | } \ |
5f80c4cf | 151 | } STMT_END |
74750237 | 152 | |
acfe0abc | 153 | static void restore_pos(pTHX_ void *arg); |
51371543 | 154 | |
76e3520e | 155 | STATIC CHECKPOINT |
cea2e8a9 | 156 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 157 | { |
97aff369 | 158 | dVAR; |
a3b680e6 | 159 | const int retval = PL_savestack_ix; |
b1ce53c5 | 160 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 161 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 162 | int p; |
40a82448 | 163 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 164 | |
e49a9654 IH |
165 | if (paren_elems_to_push < 0) |
166 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
167 | ||
a01268b5 | 168 | #define REGCP_OTHER_ELEMS 6 |
4b3c1a47 | 169 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
3280af22 | 170 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 171 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
172 | SSPUSHINT(PL_regendp[p]); |
173 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 174 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 175 | SSPUSHINT(p); |
40a82448 DM |
176 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
177 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", | |
178 | (UV)p, (IV)PL_regstartp[p], | |
179 | (IV)(PL_reg_start_tmp[p] - PL_bostr), | |
180 | (IV)PL_regendp[p] | |
181 | )); | |
a0d0e21e | 182 | } |
b1ce53c5 | 183 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
3280af22 NIS |
184 | SSPUSHINT(PL_regsize); |
185 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 186 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 187 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
188 | #define REGCP_FRAME_ELEMS 2 |
189 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
190 | * are needed for the regexp context stack bookkeeping. */ | |
191 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 192 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 193 | |
a0d0e21e LW |
194 | return retval; |
195 | } | |
196 | ||
c277df42 | 197 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
198 | #define REGCP_SET(cp) \ |
199 | DEBUG_STATE_r( \ | |
ab3bbdeb | 200 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 201 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
202 | (IV)PL_savestack_ix)); \ |
203 | cp = PL_savestack_ix | |
c3464db5 | 204 | |
ab3bbdeb | 205 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 206 | DEBUG_STATE_r( \ |
ab3bbdeb | 207 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
208 | PerlIO_printf(Perl_debug_log, \ |
209 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
210 | (IV)(cp), (IV)PL_savestack_ix)); \ |
211 | regcpblow(cp) | |
c277df42 | 212 | |
76e3520e | 213 | STATIC char * |
097eb12c | 214 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 215 | { |
97aff369 | 216 | dVAR; |
b1ce53c5 | 217 | I32 i; |
a0d0e21e | 218 | char *input; |
b1ce53c5 | 219 | |
a3621e74 YO |
220 | GET_RE_DEBUG_FLAGS_DECL; |
221 | ||
b1ce53c5 | 222 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 223 | i = SSPOPINT; |
b1ce53c5 JH |
224 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
225 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 226 | input = (char *) SSPOPPTR; |
a01268b5 | 227 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
228 | *PL_reglastparen = SSPOPINT; |
229 | PL_regsize = SSPOPINT; | |
b1ce53c5 JH |
230 | |
231 | /* Now restore the parentheses context. */ | |
41123dfd JH |
232 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
233 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 234 | I32 tmps; |
097eb12c | 235 | U32 paren = (U32)SSPOPINT; |
3280af22 | 236 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
237 | PL_regstartp[paren] = SSPOPINT; |
238 | tmps = SSPOPINT; | |
3280af22 NIS |
239 | if (paren <= *PL_reglastparen) |
240 | PL_regendp[paren] = tmps; | |
a3621e74 | 241 | DEBUG_EXECUTE_r( |
c3464db5 | 242 | PerlIO_printf(Perl_debug_log, |
b900a521 | 243 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 244 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 245 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 246 | (IV)PL_regendp[paren], |
3280af22 | 247 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 248 | ); |
a0d0e21e | 249 | } |
a3621e74 | 250 | DEBUG_EXECUTE_r( |
bb7a0f54 | 251 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 252 | PerlIO_printf(Perl_debug_log, |
faccc32b | 253 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 254 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
255 | } |
256 | ); | |
daf18116 | 257 | #if 1 |
dafc8851 JH |
258 | /* It would seem that the similar code in regtry() |
259 | * already takes care of this, and in fact it is in | |
260 | * a better location to since this code can #if 0-ed out | |
261 | * but the code in regtry() is needed or otherwise tests | |
262 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
263 | * (as of patchlevel 7877) will fail. Then again, |
264 | * this code seems to be necessary or otherwise | |
265 | * building DynaLoader will fail: | |
266 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
267 | * --jhi */ | |
bb7a0f54 | 268 | for (i = *PL_reglastparen + 1; (U32)i <= rex->nparens; i++) { |
097eb12c AL |
269 | if (i > PL_regsize) |
270 | PL_regstartp[i] = -1; | |
271 | PL_regendp[i] = -1; | |
a0d0e21e | 272 | } |
dafc8851 | 273 | #endif |
a0d0e21e LW |
274 | return input; |
275 | } | |
276 | ||
02db2b7b | 277 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 278 | |
a687059c | 279 | /* |
e50aee73 | 280 | * pregexec and friends |
a687059c LW |
281 | */ |
282 | ||
76234dfb | 283 | #ifndef PERL_IN_XSUB_RE |
a687059c | 284 | /* |
c277df42 | 285 | - pregexec - match a regexp against a string |
a687059c | 286 | */ |
c277df42 | 287 | I32 |
864dbfa3 | 288 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 289 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
290 | /* strend: pointer to null at end of string */ |
291 | /* strbeg: real beginning of string */ | |
292 | /* minend: end of match must be >=minend after stringarg. */ | |
293 | /* nosave: For optimizations. */ | |
294 | { | |
295 | return | |
9041c2e3 | 296 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
297 | nosave ? 0 : REXEC_COPY_STR); |
298 | } | |
76234dfb | 299 | #endif |
22e551b9 | 300 | |
9041c2e3 | 301 | /* |
cad2e5aa JH |
302 | * Need to implement the following flags for reg_anch: |
303 | * | |
304 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
305 | * USE_INTUIT_ML | |
306 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
307 | * INTUIT_AUTORITATIVE_ML | |
308 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
309 | * INTUIT_ONCE_ML | |
310 | * | |
311 | * Another flag for this function: SECOND_TIME (so that float substrs | |
312 | * with giant delta may be not rechecked). | |
313 | */ | |
314 | ||
315 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
316 | ||
3f7c398e | 317 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
318 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
319 | ||
320 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
321 | ||
6eb5f6b9 JH |
322 | /* XXXX Some places assume that there is a fixed substring. |
323 | An update may be needed if optimizer marks as "INTUITable" | |
324 | RExen without fixed substrings. Similarly, it is assumed that | |
325 | lengths of all the strings are no more than minlen, thus they | |
326 | cannot come from lookahead. | |
327 | (Or minlen should take into account lookahead.) */ | |
328 | ||
2c2d71f5 JH |
329 | /* A failure to find a constant substring means that there is no need to make |
330 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
331 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
332 | regtry() should be needed. |
333 | ||
334 | REx compiler's optimizer found 4 possible hints: | |
335 | a) Anchored substring; | |
336 | b) Fixed substring; | |
337 | c) Whether we are anchored (beginning-of-line or \G); | |
338 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 339 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
340 | string which does not contradict any of them. |
341 | */ | |
2c2d71f5 | 342 | |
6eb5f6b9 JH |
343 | /* Most of decisions we do here should have been done at compile time. |
344 | The nodes of the REx which we used for the search should have been | |
345 | deleted from the finite automaton. */ | |
346 | ||
cad2e5aa JH |
347 | char * |
348 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
349 | char *strend, U32 flags, re_scream_pos_data *data) | |
350 | { | |
97aff369 | 351 | dVAR; |
b7953727 | 352 | register I32 start_shift = 0; |
cad2e5aa | 353 | /* Should be nonnegative! */ |
b7953727 | 354 | register I32 end_shift = 0; |
2c2d71f5 JH |
355 | register char *s; |
356 | register SV *check; | |
a1933d95 | 357 | char *strbeg; |
cad2e5aa | 358 | char *t; |
1de06328 | 359 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 360 | I32 ml_anch; |
bd61b366 SS |
361 | register char *other_last = NULL; /* other substr checked before this */ |
362 | char *check_at = NULL; /* check substr found at this pos */ | |
1df70142 | 363 | const I32 multiline = prog->reganch & PMf_MULTILINE; |
30944b6d | 364 | #ifdef DEBUGGING |
890ce7af | 365 | const char * const i_strpos = strpos; |
30944b6d | 366 | #endif |
a3621e74 YO |
367 | |
368 | GET_RE_DEBUG_FLAGS_DECL; | |
369 | ||
a30b2f1f | 370 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 371 | |
b8d68ded | 372 | if (prog->reganch & ROPT_UTF8) { |
b8d68ded JH |
373 | PL_reg_flags |= RF_utf8; |
374 | } | |
ab3bbdeb YO |
375 | DEBUG_EXECUTE_r( |
376 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
377 | sv ? "Guessing start of match in sv for" |
378 | : "Guessing start of match in string for"); | |
2a782b5b | 379 | ); |
cad2e5aa | 380 | |
c344f387 JH |
381 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
382 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 383 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 384 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 385 | goto fail; |
2c2d71f5 | 386 | } |
1de06328 | 387 | |
a1933d95 | 388 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 389 | PL_regeol = strend; |
33b8afdf JH |
390 | if (do_utf8) { |
391 | if (!prog->check_utf8 && prog->check_substr) | |
392 | to_utf8_substr(prog); | |
393 | check = prog->check_utf8; | |
394 | } else { | |
395 | if (!prog->check_substr && prog->check_utf8) | |
396 | to_byte_substr(prog); | |
397 | check = prog->check_substr; | |
398 | } | |
1de06328 | 399 | if (check == &PL_sv_undef) { |
a3621e74 | 400 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 401 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
402 | goto fail; |
403 | } | |
2c2d71f5 | 404 | if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ |
cad2e5aa JH |
405 | ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) |
406 | || ( (prog->reganch & ROPT_ANCH_BOL) | |
7fba1cd6 | 407 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 408 | |
7e25d62c JH |
409 | if (!ml_anch) { |
410 | if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ | |
411 | | ROPT_IMPLICIT)) /* not a real BOL */ | |
3f7c398e | 412 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
413 | && sv && !SvROK(sv) |
414 | && (strpos != strbeg)) { | |
a3621e74 | 415 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
416 | goto fail; |
417 | } | |
418 | if (prog->check_offset_min == prog->check_offset_max && | |
cce850e4 | 419 | !(prog->reganch & ROPT_CANY_SEEN)) { |
2c2d71f5 | 420 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
421 | I32 slen; |
422 | ||
1aa99e6b | 423 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 424 | |
653099ff GS |
425 | if (SvTAIL(check)) { |
426 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 427 | |
9041c2e3 | 428 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 429 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 430 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 431 | goto fail_finish; |
cad2e5aa JH |
432 | } |
433 | /* Now should match s[0..slen-2] */ | |
434 | slen--; | |
3f7c398e | 435 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 436 | || (slen > 1 |
3f7c398e | 437 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 438 | report_neq: |
a3621e74 | 439 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
440 | goto fail_finish; |
441 | } | |
cad2e5aa | 442 | } |
3f7c398e | 443 | else if (*SvPVX_const(check) != *s |
653099ff | 444 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 445 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 446 | goto report_neq; |
c315bfe8 | 447 | check_at = s; |
2c2d71f5 | 448 | goto success_at_start; |
7e25d62c | 449 | } |
cad2e5aa | 450 | } |
2c2d71f5 | 451 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 452 | s = strpos; |
2c2d71f5 | 453 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
454 | end_shift = prog->check_end_shift; |
455 | ||
2c2d71f5 | 456 | if (!ml_anch) { |
a3b680e6 | 457 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 458 | - (SvTAIL(check) != 0); |
a3b680e6 | 459 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
460 | |
461 | if (end_shift < eshift) | |
462 | end_shift = eshift; | |
463 | } | |
cad2e5aa | 464 | } |
2c2d71f5 | 465 | else { /* Can match at random position */ |
cad2e5aa JH |
466 | ml_anch = 0; |
467 | s = strpos; | |
1de06328 YO |
468 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
469 | end_shift = prog->check_end_shift; | |
470 | ||
471 | /* end shift should be non negative here */ | |
cad2e5aa JH |
472 | } |
473 | ||
2c2d71f5 | 474 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 475 | if (end_shift < 0) |
1de06328 | 476 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 477 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
478 | #endif |
479 | ||
2c2d71f5 JH |
480 | restart: |
481 | /* Find a possible match in the region s..strend by looking for | |
482 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
483 | |
484 | { | |
485 | I32 srch_start_shift = start_shift; | |
486 | I32 srch_end_shift = end_shift; | |
487 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
488 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
489 | srch_start_shift = strbeg - s; | |
490 | } | |
491 | DEBUG_OPTIMISE_r({ | |
492 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", | |
493 | (IV)prog->check_offset_min, | |
494 | (IV)srch_start_shift, | |
495 | (IV)srch_end_shift, | |
496 | (IV)prog->check_end_shift); | |
497 | }); | |
498 | ||
cad2e5aa | 499 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 500 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 501 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 502 | |
2c2d71f5 JH |
503 | if (PL_screamfirst[BmRARE(check)] >= 0 |
504 | || ( BmRARE(check) == '\n' | |
505 | && (BmPREVIOUS(check) == SvCUR(check) - 1) | |
506 | && SvTAIL(check) )) | |
9041c2e3 | 507 | s = screaminstr(sv, check, |
1de06328 | 508 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 509 | else |
2c2d71f5 | 510 | goto fail_finish; |
4addbd3b HS |
511 | /* we may be pointing at the wrong string */ |
512 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 513 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
514 | if (data) |
515 | *data->scream_olds = s; | |
516 | } | |
1de06328 YO |
517 | else { |
518 | U8* start_point; | |
519 | U8* end_point; | |
520 | if (prog->reganch & ROPT_CANY_SEEN) { | |
521 | start_point= (U8*)(s + srch_start_shift); | |
522 | end_point= (U8*)(strend - srch_end_shift); | |
523 | } else { | |
524 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
525 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
526 | } | |
527 | DEBUG_OPTIMISE_r({ | |
56570a2c | 528 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 529 | (int)(end_point - start_point), |
fc8cd66c | 530 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
531 | start_point); |
532 | }); | |
533 | ||
534 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 535 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
536 | } |
537 | } | |
cad2e5aa JH |
538 | /* Update the count-of-usability, remove useless subpatterns, |
539 | unshift s. */ | |
2c2d71f5 | 540 | |
ab3bbdeb YO |
541 | DEBUG_EXECUTE_r({ |
542 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
543 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
544 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 545 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
546 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
547 | ? "anchored" : "floating"), | |
548 | quoted, | |
549 | RE_SV_TAIL(check), | |
550 | (s ? " at offset " : "...\n") ); | |
551 | }); | |
2c2d71f5 JH |
552 | |
553 | if (!s) | |
554 | goto fail_finish; | |
2c2d71f5 | 555 | /* Finish the diagnostic message */ |
a3621e74 | 556 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 557 | |
1de06328 YO |
558 | /* XXX dmq: first branch is for positive lookbehind... |
559 | Our check string is offset from the beginning of the pattern. | |
560 | So we need to do any stclass tests offset forward from that | |
561 | point. I think. :-( | |
562 | */ | |
563 | ||
564 | ||
565 | ||
566 | check_at=s; | |
567 | ||
568 | ||
2c2d71f5 JH |
569 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
570 | Start with the other substr. | |
571 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 572 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
573 | *always* match. Probably should be marked during compile... |
574 | Probably it is right to do no SCREAM here... | |
575 | */ | |
576 | ||
1de06328 YO |
577 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
578 | : (prog->float_substr && prog->anchored_substr)) | |
579 | { | |
30944b6d | 580 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
581 | /* XXXX May be hopelessly wrong for UTF... */ |
582 | if (!other_last) | |
6eb5f6b9 | 583 | other_last = strpos; |
33b8afdf | 584 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
585 | do_other_anchored: |
586 | { | |
890ce7af AL |
587 | char * const last = HOP3c(s, -start_shift, strbeg); |
588 | char *last1, *last2; | |
be8e71aa | 589 | char * const saved_s = s; |
33b8afdf | 590 | SV* must; |
2c2d71f5 | 591 | |
2c2d71f5 JH |
592 | t = s - prog->check_offset_max; |
593 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 594 | && (!do_utf8 |
0ce71af7 | 595 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 596 | && t > strpos))) |
6f207bd3 | 597 | NOOP; |
2c2d71f5 JH |
598 | else |
599 | t = strpos; | |
1aa99e6b | 600 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
601 | if (t < other_last) /* These positions already checked */ |
602 | t = other_last; | |
1aa99e6b | 603 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
604 | if (last < last1) |
605 | last1 = last; | |
1de06328 YO |
606 | /* XXXX It is not documented what units *_offsets are in. |
607 | We assume bytes, but this is clearly wrong. | |
608 | Meaning this code needs to be carefully reviewed for errors. | |
609 | dmq. | |
610 | */ | |
611 | ||
2c2d71f5 | 612 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
613 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
614 | if (must == &PL_sv_undef) { | |
615 | s = (char*)NULL; | |
1de06328 | 616 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
617 | } |
618 | else | |
619 | s = fbm_instr( | |
620 | (unsigned char*)t, | |
621 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
622 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
623 | must, | |
7fba1cd6 | 624 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 625 | ); |
ab3bbdeb YO |
626 | DEBUG_EXECUTE_r({ |
627 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
628 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
629 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 630 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
631 | quoted, RE_SV_TAIL(must)); |
632 | }); | |
633 | ||
634 | ||
2c2d71f5 JH |
635 | if (!s) { |
636 | if (last1 >= last2) { | |
a3621e74 | 637 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
638 | ", giving up...\n")); |
639 | goto fail_finish; | |
640 | } | |
a3621e74 | 641 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 642 | ", trying floating at offset %ld...\n", |
be8e71aa | 643 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
644 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
645 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
646 | goto restart; |
647 | } | |
648 | else { | |
a3621e74 | 649 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 650 | (long)(s - i_strpos))); |
1aa99e6b IH |
651 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
652 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 653 | s = saved_s; |
2c2d71f5 JH |
654 | if (t == strpos) |
655 | goto try_at_start; | |
2c2d71f5 JH |
656 | goto try_at_offset; |
657 | } | |
30944b6d | 658 | } |
2c2d71f5 JH |
659 | } |
660 | else { /* Take into account the floating substring. */ | |
33b8afdf | 661 | char *last, *last1; |
be8e71aa | 662 | char * const saved_s = s; |
33b8afdf JH |
663 | SV* must; |
664 | ||
665 | t = HOP3c(s, -start_shift, strbeg); | |
666 | last1 = last = | |
667 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
668 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
669 | last = HOP3c(t, prog->float_max_offset, strend); | |
670 | s = HOP3c(t, prog->float_min_offset, strend); | |
671 | if (s < other_last) | |
672 | s = other_last; | |
2c2d71f5 | 673 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
674 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
675 | /* fbm_instr() takes into account exact value of end-of-str | |
676 | if the check is SvTAIL(ed). Since false positives are OK, | |
677 | and end-of-str is not later than strend we are OK. */ | |
678 | if (must == &PL_sv_undef) { | |
679 | s = (char*)NULL; | |
1de06328 | 680 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
681 | } |
682 | else | |
2c2d71f5 | 683 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
684 | (unsigned char*)last + SvCUR(must) |
685 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 686 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
687 | DEBUG_EXECUTE_r({ |
688 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
689 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
690 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 691 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
692 | quoted, RE_SV_TAIL(must)); |
693 | }); | |
33b8afdf JH |
694 | if (!s) { |
695 | if (last1 == last) { | |
a3621e74 | 696 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
697 | ", giving up...\n")); |
698 | goto fail_finish; | |
2c2d71f5 | 699 | } |
a3621e74 | 700 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 701 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 702 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
703 | other_last = last; |
704 | s = HOP3c(t, 1, strend); | |
705 | goto restart; | |
706 | } | |
707 | else { | |
a3621e74 | 708 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
709 | (long)(s - i_strpos))); |
710 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 711 | s = saved_s; |
33b8afdf JH |
712 | if (t == strpos) |
713 | goto try_at_start; | |
714 | goto try_at_offset; | |
715 | } | |
2c2d71f5 | 716 | } |
cad2e5aa | 717 | } |
2c2d71f5 | 718 | |
1de06328 | 719 | |
9ef43ace | 720 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 YO |
721 | |
722 | DEBUG_OPTIMISE_r( | |
723 | PerlIO_printf(Perl_debug_log, | |
724 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
725 | (IV)prog->check_offset_min, | |
726 | (IV)prog->check_offset_max, | |
727 | (IV)(s-strpos), | |
728 | (IV)(t-strpos), | |
729 | (IV)(t-s), | |
730 | (IV)(strend-strpos) | |
731 | ) | |
732 | ); | |
733 | ||
2c2d71f5 | 734 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 735 | && (!do_utf8 |
9ef43ace | 736 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
737 | && t > strpos))) |
738 | { | |
2c2d71f5 JH |
739 | /* Fixed substring is found far enough so that the match |
740 | cannot start at strpos. */ | |
741 | try_at_offset: | |
cad2e5aa | 742 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
743 | /* Eventually fbm_*() should handle this, but often |
744 | anchored_offset is not 0, so this check will not be wasted. */ | |
745 | /* XXXX In the code below we prefer to look for "^" even in | |
746 | presence of anchored substrings. And we search even | |
747 | beyond the found float position. These pessimizations | |
748 | are historical artefacts only. */ | |
749 | find_anchor: | |
2c2d71f5 | 750 | while (t < strend - prog->minlen) { |
cad2e5aa | 751 | if (*t == '\n') { |
4ee3650e | 752 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 753 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
754 | /* Since we moved from the found position, |
755 | we definitely contradict the found anchored | |
30944b6d IZ |
756 | substr. Due to the above check we do not |
757 | contradict "check" substr. | |
758 | Thus we can arrive here only if check substr | |
759 | is float. Redo checking for "other"=="fixed". | |
760 | */ | |
9041c2e3 | 761 | strpos = t + 1; |
a3621e74 | 762 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 763 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
764 | goto do_other_anchored; |
765 | } | |
4ee3650e GS |
766 | /* We don't contradict the found floating substring. */ |
767 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 768 | s = t + 1; |
a3621e74 | 769 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 770 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
771 | goto set_useful; |
772 | } | |
4ee3650e GS |
773 | /* Position contradicts check-string */ |
774 | /* XXXX probably better to look for check-string | |
775 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 776 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 777 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 778 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
779 | goto restart; |
780 | } | |
781 | t++; | |
782 | } | |
a3621e74 | 783 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 784 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 785 | goto fail_finish; |
cad2e5aa | 786 | } |
f5952150 | 787 | else { |
a3621e74 | 788 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 789 | PL_colors[0], PL_colors[1])); |
f5952150 | 790 | } |
cad2e5aa JH |
791 | s = t; |
792 | set_useful: | |
33b8afdf | 793 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
794 | } |
795 | else { | |
f5952150 | 796 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 797 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
798 | unless it was an MBOL and we are not after MBOL, |
799 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
800 | try_at_start: |
801 | /* Even in this situation we may use MBOL flag if strpos is offset | |
802 | wrt the start of the string. */ | |
05b4157f | 803 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 804 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d IZ |
805 | /* May be due to an implicit anchor of m{.*foo} */ |
806 | && !(prog->reganch & ROPT_IMPLICIT)) | |
807 | { | |
cad2e5aa JH |
808 | t = strpos; |
809 | goto find_anchor; | |
810 | } | |
a3621e74 | 811 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 812 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
e4584336 | 813 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 814 | ); |
2c2d71f5 | 815 | success_at_start: |
30944b6d | 816 | if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
817 | && (do_utf8 ? ( |
818 | prog->check_utf8 /* Could be deleted already */ | |
819 | && --BmUSEFUL(prog->check_utf8) < 0 | |
820 | && (prog->check_utf8 == prog->float_utf8) | |
821 | ) : ( | |
822 | prog->check_substr /* Could be deleted already */ | |
823 | && --BmUSEFUL(prog->check_substr) < 0 | |
824 | && (prog->check_substr == prog->float_substr) | |
825 | ))) | |
66e933ab | 826 | { |
cad2e5aa | 827 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 828 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
829 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
830 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
831 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
832 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
833 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
834 | check = NULL; /* abort */ | |
cad2e5aa | 835 | s = strpos; |
3cf5c195 IZ |
836 | /* XXXX This is a remnant of the old implementation. It |
837 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 838 | other heuristics. */ |
cad2e5aa JH |
839 | prog->reganch &= ~RE_USE_INTUIT; |
840 | } | |
841 | else | |
842 | s = strpos; | |
843 | } | |
844 | ||
6eb5f6b9 JH |
845 | /* Last resort... */ |
846 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
847 | /* trie stclasses are too expensive to use here, we are better off to |
848 | leave it to regmatch itself */ | |
786e8c11 | 849 | if (prog->regstclass && PL_regkind[OP(prog->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
850 | /* minlen == 0 is possible if regstclass is \b or \B, |
851 | and the fixed substr is ''$. | |
852 | Since minlen is already taken into account, s+1 is before strend; | |
853 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
854 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
855 | regstclass does not come from lookahead... */ | |
856 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
857 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
890ce7af | 858 | const U8* const str = (U8*)STRING(prog->regstclass); |
3dab1dad | 859 | const int cl_l = (PL_regkind[OP(prog->regstclass)] == EXACT |
1aa99e6b | 860 | ? CHR_DIST(str+STR_LEN(prog->regstclass), str) |
66e933ab | 861 | : 1); |
1de06328 YO |
862 | char * endpos; |
863 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
864 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
865 | else if (prog->float_substr || prog->float_utf8) | |
866 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
867 | else | |
868 | endpos= strend; | |
869 | ||
56570a2c | 870 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %d s: %d endpos: %d\n", |
b57a0404 | 871 | (IV)start_shift, check_at - strbeg, s - strbeg, endpos - strbeg)); |
1de06328 | 872 | |
6eb5f6b9 | 873 | t = s; |
3b0527fe | 874 | s = find_byclass(prog, prog->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
875 | if (!s) { |
876 | #ifdef DEBUGGING | |
cbbf8932 | 877 | const char *what = NULL; |
6eb5f6b9 JH |
878 | #endif |
879 | if (endpos == strend) { | |
a3621e74 | 880 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
881 | "Could not match STCLASS...\n") ); |
882 | goto fail; | |
883 | } | |
a3621e74 | 884 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 885 | "This position contradicts STCLASS...\n") ); |
653099ff GS |
886 | if ((prog->reganch & ROPT_ANCH) && !ml_anch) |
887 | goto fail; | |
6eb5f6b9 | 888 | /* Contradict one of substrings */ |
33b8afdf JH |
889 | if (prog->anchored_substr || prog->anchored_utf8) { |
890 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 891 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 892 | hop_and_restart: |
1aa99e6b | 893 | s = HOP3c(t, 1, strend); |
66e933ab GS |
894 | if (s + start_shift + end_shift > strend) { |
895 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 896 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
897 | "Could not match STCLASS...\n") ); |
898 | goto fail; | |
899 | } | |
5e39e1e5 HS |
900 | if (!check) |
901 | goto giveup; | |
a3621e74 | 902 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 903 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
904 | what, (long)(s + start_shift - i_strpos)) ); |
905 | goto restart; | |
906 | } | |
66e933ab | 907 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
908 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
909 | goto retry_floating_check; | |
910 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 911 | s = check_at; |
5e39e1e5 HS |
912 | if (!check) |
913 | goto giveup; | |
a3621e74 | 914 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 915 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
916 | (long)(other_last - i_strpos)) ); |
917 | goto do_other_anchored; | |
918 | } | |
60e71179 GS |
919 | /* Another way we could have checked stclass at the |
920 | current position only: */ | |
921 | if (ml_anch) { | |
922 | s = t = t + 1; | |
5e39e1e5 HS |
923 | if (!check) |
924 | goto giveup; | |
a3621e74 | 925 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 926 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 927 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 928 | goto try_at_offset; |
66e933ab | 929 | } |
33b8afdf | 930 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 931 | goto fail; |
6eb5f6b9 JH |
932 | /* Check is floating subtring. */ |
933 | retry_floating_check: | |
934 | t = check_at - start_shift; | |
a3621e74 | 935 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
936 | goto hop_and_restart; |
937 | } | |
b7953727 | 938 | if (t != s) { |
a3621e74 | 939 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 940 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
941 | (long)(t - i_strpos), (long)(s - i_strpos)) |
942 | ); | |
943 | } | |
944 | else { | |
a3621e74 | 945 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
946 | "Does not contradict STCLASS...\n"); |
947 | ); | |
948 | } | |
6eb5f6b9 | 949 | } |
5e39e1e5 | 950 | giveup: |
a3621e74 | 951 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
952 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
953 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 954 | return s; |
2c2d71f5 JH |
955 | |
956 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
957 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
958 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 959 | fail: |
a3621e74 | 960 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 961 | PL_colors[4], PL_colors[5])); |
bd61b366 | 962 | return NULL; |
cad2e5aa | 963 | } |
9661b544 | 964 | |
786e8c11 | 965 | |
3b0527fe | 966 | |
4cadc6a9 YO |
967 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, uvc, charid, \ |
968 | foldlen, foldbuf, uniflags) STMT_START { \ | |
969 | switch (trie_type) { \ | |
970 | case trie_utf8_fold: \ | |
971 | if ( foldlen>0 ) { \ | |
972 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
973 | foldlen -= len; \ | |
974 | uscan += len; \ | |
975 | len=0; \ | |
976 | } else { \ | |
977 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
978 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
979 | foldlen -= UNISKIP( uvc ); \ | |
980 | uscan = foldbuf + UNISKIP( uvc ); \ | |
981 | } \ | |
982 | break; \ | |
983 | case trie_utf8: \ | |
984 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
985 | break; \ | |
986 | case trie_plain: \ | |
987 | uvc = (UV)*uc; \ | |
988 | len = 1; \ | |
989 | } \ | |
990 | \ | |
991 | if (uvc < 256) { \ | |
992 | charid = trie->charmap[ uvc ]; \ | |
993 | } \ | |
994 | else { \ | |
995 | charid = 0; \ | |
996 | if (trie->widecharmap) { \ | |
997 | SV** const svpp = hv_fetch(trie->widecharmap, \ | |
998 | (char*)&uvc, sizeof(UV), 0); \ | |
999 | if (svpp) \ | |
1000 | charid = (U16)SvIV(*svpp); \ | |
1001 | } \ | |
1002 | } \ | |
1003 | } STMT_END | |
1004 | ||
1005 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1006 | if ( (CoNd) \ | |
1007 | && (ln == len || \ | |
1008 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1009 | m, NULL, ln, (bool)UTF)) \ | |
1010 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1011 | goto got_it; \ | |
1012 | else { \ | |
1013 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1014 | uvchr_to_utf8(tmpbuf, c); \ | |
1015 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1016 | if ( f != c \ | |
1017 | && (f == c1 || f == c2) \ | |
1018 | && (ln == foldlen || \ | |
1019 | !ibcmp_utf8((char *) foldbuf, \ | |
1020 | NULL, foldlen, do_utf8, \ | |
1021 | m, \ | |
1022 | NULL, ln, (bool)UTF)) \ | |
1023 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1024 | goto got_it; \ | |
1025 | } \ | |
1026 | s += len | |
1027 | ||
1028 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1029 | STMT_START { \ | |
1030 | while (s <= e) { \ | |
1031 | if ( (CoNd) \ | |
1032 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1033 | ? ibcmp(s, m, ln) \ | |
1034 | : ibcmp_locale(s, m, ln))) \ | |
1035 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1036 | goto got_it; \ | |
1037 | s++; \ | |
1038 | } \ | |
1039 | } STMT_END | |
1040 | ||
1041 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1042 | STMT_START { \ | |
1043 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1044 | CoDe \ | |
1045 | s += uskip; \ | |
1046 | } \ | |
1047 | } STMT_END | |
1048 | ||
1049 | #define REXEC_FBC_SCAN(CoDe) \ | |
1050 | STMT_START { \ | |
1051 | while (s < strend) { \ | |
1052 | CoDe \ | |
1053 | s++; \ | |
1054 | } \ | |
1055 | } STMT_END | |
1056 | ||
1057 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1058 | REXEC_FBC_UTF8_SCAN( \ | |
1059 | if (CoNd) { \ | |
1060 | if (tmp && (!reginfo || regtry(reginfo, s))) \ | |
1061 | goto got_it; \ | |
1062 | else \ | |
1063 | tmp = doevery; \ | |
1064 | } \ | |
1065 | else \ | |
1066 | tmp = 1; \ | |
1067 | ) | |
1068 | ||
1069 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1070 | REXEC_FBC_SCAN( \ | |
1071 | if (CoNd) { \ | |
1072 | if (tmp && (!reginfo || regtry(reginfo, s))) \ | |
1073 | goto got_it; \ | |
1074 | else \ | |
1075 | tmp = doevery; \ | |
1076 | } \ | |
1077 | else \ | |
1078 | tmp = 1; \ | |
1079 | ) | |
1080 | ||
1081 | #define REXEC_FBC_TRYIT \ | |
1082 | if ((!reginfo || regtry(reginfo, s))) \ | |
1083 | goto got_it | |
1084 | ||
1085 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ | |
1086 | if (do_utf8) { \ | |
1087 | UtFpReLoAd; \ | |
1088 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1089 | } \ | |
1090 | else { \ | |
1091 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1092 | } \ | |
1093 | break | |
1094 | ||
1095 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1096 | PL_reg_flags |= RF_tainted; \ | |
1097 | if (do_utf8) { \ | |
1098 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1099 | } \ | |
1100 | else { \ | |
1101 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1102 | } \ | |
1103 | break | |
1104 | ||
786e8c11 YO |
1105 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1106 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1107 | ||
1108 | /* We know what class REx starts with. Try to find this position... */ | |
1109 | /* if reginfo is NULL, its a dryrun */ | |
1110 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1111 | in regmatch. /grrr */ | |
1112 | ||
3c3eec57 | 1113 | STATIC char * |
07be1b83 YO |
1114 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
1115 | const char *strend, const regmatch_info *reginfo) | |
a687059c | 1116 | { |
27da23d5 | 1117 | dVAR; |
1df70142 | 1118 | const I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
6eb5f6b9 | 1119 | char *m; |
d8093b23 | 1120 | STRLEN ln; |
5dab1207 | 1121 | STRLEN lnc; |
078c425b | 1122 | register STRLEN uskip; |
d8093b23 G |
1123 | unsigned int c1; |
1124 | unsigned int c2; | |
6eb5f6b9 JH |
1125 | char *e; |
1126 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1127 | register const bool do_utf8 = PL_reg_match_utf8; |
cad2e5aa | 1128 | |
6eb5f6b9 JH |
1129 | /* We know what class it must start with. */ |
1130 | switch (OP(c)) { | |
6eb5f6b9 | 1131 | case ANYOF: |
388cc4de | 1132 | if (do_utf8) { |
4cadc6a9 | 1133 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1134 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1135 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1136 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1137 | } |
1138 | else { | |
1139 | while (s < strend) { | |
1140 | STRLEN skip = 1; | |
1141 | ||
32fc9b6a | 1142 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1143 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1144 | /* The assignment of 2 is intentional: | |
1145 | * for the folded sharp s, the skip is 2. */ | |
1146 | (skip = SHARP_S_SKIP))) { | |
3b0527fe | 1147 | if (tmp && (!reginfo || regtry(reginfo, s))) |
388cc4de HS |
1148 | goto got_it; |
1149 | else | |
1150 | tmp = doevery; | |
1151 | } | |
1152 | else | |
1153 | tmp = 1; | |
1154 | s += skip; | |
1155 | } | |
a0d0e21e | 1156 | } |
6eb5f6b9 | 1157 | break; |
f33976b4 | 1158 | case CANY: |
4cadc6a9 | 1159 | REXEC_FBC_SCAN( |
3b0527fe | 1160 | if (tmp && (!reginfo || regtry(reginfo, s))) |
f33976b4 DB |
1161 | goto got_it; |
1162 | else | |
1163 | tmp = doevery; | |
4cadc6a9 | 1164 | ); |
f33976b4 | 1165 | break; |
6eb5f6b9 | 1166 | case EXACTF: |
5dab1207 NIS |
1167 | m = STRING(c); |
1168 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1169 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1170 | if (UTF) { |
a2a2844f | 1171 | STRLEN ulen1, ulen2; |
5dab1207 | 1172 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1173 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1174 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1175 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1176 | |
1177 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1178 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1179 | ||
89ebb4a3 | 1180 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1181 | 0, uniflags); |
89ebb4a3 | 1182 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1183 | 0, uniflags); |
5dab1207 NIS |
1184 | lnc = 0; |
1185 | while (sm < ((U8 *) m + ln)) { | |
1186 | lnc++; | |
1187 | sm += UTF8SKIP(sm); | |
1188 | } | |
1aa99e6b IH |
1189 | } |
1190 | else { | |
1191 | c1 = *(U8*)m; | |
1192 | c2 = PL_fold[c1]; | |
1193 | } | |
6eb5f6b9 JH |
1194 | goto do_exactf; |
1195 | case EXACTFL: | |
5dab1207 NIS |
1196 | m = STRING(c); |
1197 | ln = STR_LEN(c); | |
1198 | lnc = (I32) ln; | |
d8093b23 | 1199 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1200 | c2 = PL_fold_locale[c1]; |
1201 | do_exactf: | |
db12adc6 | 1202 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1203 | |
3b0527fe | 1204 | if (!reginfo && e < s) |
6eb5f6b9 | 1205 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1206 | |
60a8b682 JH |
1207 | /* The idea in the EXACTF* cases is to first find the |
1208 | * first character of the EXACTF* node and then, if | |
1209 | * necessary, case-insensitively compare the full | |
1210 | * text of the node. The c1 and c2 are the first | |
1211 | * characters (though in Unicode it gets a bit | |
1212 | * more complicated because there are more cases | |
7f16dd3d JH |
1213 | * than just upper and lower: one needs to use |
1214 | * the so-called folding case for case-insensitive | |
1215 | * matching (called "loose matching" in Unicode). | |
1216 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1217 | |
1aa99e6b | 1218 | if (do_utf8) { |
575cac57 | 1219 | UV c, f; |
89ebb4a3 | 1220 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1221 | STRLEN len, foldlen; |
4ad0818d | 1222 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1223 | if (c1 == c2) { |
5dab1207 NIS |
1224 | /* Upper and lower of 1st char are equal - |
1225 | * probably not a "letter". */ | |
1aa99e6b | 1226 | while (s <= e) { |
89ebb4a3 | 1227 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1228 | uniflags); |
4cadc6a9 | 1229 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1230 | } |
09091399 JH |
1231 | } |
1232 | else { | |
1aa99e6b | 1233 | while (s <= e) { |
89ebb4a3 | 1234 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1235 | uniflags); |
80aecb99 | 1236 | |
60a8b682 | 1237 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1238 | * Note that not all the possible combinations |
1239 | * are handled here: some of them are handled | |
1240 | * by the standard folding rules, and some of | |
1241 | * them (the character class or ANYOF cases) | |
1242 | * are handled during compiletime in | |
1243 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1244 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1245 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1246 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1247 | |
4cadc6a9 | 1248 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1249 | } |
09091399 | 1250 | } |
1aa99e6b IH |
1251 | } |
1252 | else { | |
1253 | if (c1 == c2) | |
4cadc6a9 | 1254 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1255 | else |
4cadc6a9 | 1256 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1257 | } |
1258 | break; | |
bbce6d69 | 1259 | case BOUNDL: |
3280af22 | 1260 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1261 | /* FALL THROUGH */ |
a0d0e21e | 1262 | case BOUND: |
ffc61ed2 | 1263 | if (do_utf8) { |
12d33761 | 1264 | if (s == PL_bostr) |
ffc61ed2 JH |
1265 | tmp = '\n'; |
1266 | else { | |
6136c704 | 1267 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1268 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1269 | } |
1270 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1271 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1272 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1273 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1274 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1275 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1276 | isALNUM_LC_utf8((U8*)s))) |
1277 | { | |
1278 | tmp = !tmp; | |
4cadc6a9 | 1279 | REXEC_FBC_TRYIT; |
a687059c | 1280 | } |
4cadc6a9 | 1281 | ); |
a0d0e21e | 1282 | } |
667bb95a | 1283 | else { |
12d33761 | 1284 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1285 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1286 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1287 | if (tmp == |
1288 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1289 | tmp = !tmp; | |
4cadc6a9 | 1290 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1291 | } |
4cadc6a9 | 1292 | ); |
a0ed51b3 | 1293 | } |
3b0527fe | 1294 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1295 | goto got_it; |
1296 | break; | |
bbce6d69 | 1297 | case NBOUNDL: |
3280af22 | 1298 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1299 | /* FALL THROUGH */ |
a0d0e21e | 1300 | case NBOUND: |
ffc61ed2 | 1301 | if (do_utf8) { |
12d33761 | 1302 | if (s == PL_bostr) |
ffc61ed2 JH |
1303 | tmp = '\n'; |
1304 | else { | |
6136c704 | 1305 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1306 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1307 | } |
1308 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1309 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1310 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1311 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1312 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1313 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1314 | isALNUM_LC_utf8((U8*)s))) |
1315 | tmp = !tmp; | |
4cadc6a9 YO |
1316 | else REXEC_FBC_TRYIT; |
1317 | ); | |
a0d0e21e | 1318 | } |
667bb95a | 1319 | else { |
12d33761 | 1320 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1321 | tmp = ((OP(c) == NBOUND ? |
1322 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1323 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1324 | if (tmp == |
1325 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1326 | tmp = !tmp; | |
4cadc6a9 YO |
1327 | else REXEC_FBC_TRYIT; |
1328 | ); | |
a0ed51b3 | 1329 | } |
3b0527fe | 1330 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1331 | goto got_it; |
1332 | break; | |
a0d0e21e | 1333 | case ALNUM: |
4cadc6a9 YO |
1334 | REXEC_FBC_CSCAN_PRELOAD( |
1335 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1336 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1337 | isALNUM(*s) | |
1338 | ); | |
bbce6d69 | 1339 | case ALNUML: |
4cadc6a9 YO |
1340 | REXEC_FBC_CSCAN_TAINT( |
1341 | isALNUM_LC_utf8((U8*)s), | |
1342 | isALNUM_LC(*s) | |
1343 | ); | |
a0d0e21e | 1344 | case NALNUM: |
4cadc6a9 YO |
1345 | REXEC_FBC_CSCAN_PRELOAD( |
1346 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1347 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1348 | !isALNUM(*s) | |
1349 | ); | |
bbce6d69 | 1350 | case NALNUML: |
4cadc6a9 YO |
1351 | REXEC_FBC_CSCAN_TAINT( |
1352 | !isALNUM_LC_utf8((U8*)s), | |
1353 | !isALNUM_LC(*s) | |
1354 | ); | |
a0d0e21e | 1355 | case SPACE: |
4cadc6a9 YO |
1356 | REXEC_FBC_CSCAN_PRELOAD( |
1357 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1358 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1359 | isSPACE(*s) | |
1360 | ); | |
bbce6d69 | 1361 | case SPACEL: |
4cadc6a9 YO |
1362 | REXEC_FBC_CSCAN_TAINT( |
1363 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1364 | isSPACE_LC(*s) | |
1365 | ); | |
a0d0e21e | 1366 | case NSPACE: |
4cadc6a9 YO |
1367 | REXEC_FBC_CSCAN_PRELOAD( |
1368 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1369 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1370 | !isSPACE(*s) | |
1371 | ); | |
bbce6d69 | 1372 | case NSPACEL: |
4cadc6a9 YO |
1373 | REXEC_FBC_CSCAN_TAINT( |
1374 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1375 | !isSPACE_LC(*s) | |
1376 | ); | |
a0d0e21e | 1377 | case DIGIT: |
4cadc6a9 YO |
1378 | REXEC_FBC_CSCAN_PRELOAD( |
1379 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1380 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1381 | isDIGIT(*s) | |
1382 | ); | |
b8c5462f | 1383 | case DIGITL: |
4cadc6a9 YO |
1384 | REXEC_FBC_CSCAN_TAINT( |
1385 | isDIGIT_LC_utf8((U8*)s), | |
1386 | isDIGIT_LC(*s) | |
1387 | ); | |
a0d0e21e | 1388 | case NDIGIT: |
4cadc6a9 YO |
1389 | REXEC_FBC_CSCAN_PRELOAD( |
1390 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1391 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1392 | !isDIGIT(*s) | |
1393 | ); | |
b8c5462f | 1394 | case NDIGITL: |
4cadc6a9 YO |
1395 | REXEC_FBC_CSCAN_TAINT( |
1396 | !isDIGIT_LC_utf8((U8*)s), | |
1397 | !isDIGIT_LC(*s) | |
1398 | ); | |
1de06328 YO |
1399 | case AHOCORASICKC: |
1400 | case AHOCORASICK: | |
07be1b83 YO |
1401 | { |
1402 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1403 | trie_type = do_utf8 ? | |
1404 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1405 | : trie_plain; | |
1406 | /* what trie are we using right now */ | |
1407 | reg_ac_data *aho | |
1408 | = (reg_ac_data*)prog->data->data[ ARG( c ) ]; | |
1409 | reg_trie_data *trie=aho->trie; | |
1410 | ||
1411 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1412 | #ifdef DEBUGGING |
07be1b83 | 1413 | const char *real_start = s; |
6148ee25 | 1414 | #endif |
07be1b83 | 1415 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1416 | SV *sv_points; |
1417 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1418 | when reading a given char. For ASCII this |
be8e71aa YO |
1419 | is unnecessary overhead as the relationship |
1420 | is always 1:1, but for unicode, especially | |
1421 | case folded unicode this is not true. */ | |
f9e705e8 | 1422 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1423 | U8 *bitmap=NULL; |
1424 | ||
07be1b83 YO |
1425 | |
1426 | GET_RE_DEBUG_FLAGS_DECL; | |
1427 | ||
be8e71aa YO |
1428 | /* We can't just allocate points here. We need to wrap it in |
1429 | * an SV so it gets freed properly if there is a croak while | |
1430 | * running the match */ | |
1431 | ENTER; | |
1432 | SAVETMPS; | |
1433 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1434 | SvCUR_set(sv_points, | |
1435 | maxlen * sizeof(U8 *)); | |
1436 | SvPOK_on(sv_points); | |
1437 | sv_2mortal(sv_points); | |
1438 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1439 | if ( trie_type != trie_utf8_fold |
1440 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1441 | { | |
786e8c11 YO |
1442 | if (trie->bitmap) |
1443 | bitmap=(U8*)trie->bitmap; | |
1444 | else | |
1445 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1446 | } |
786e8c11 YO |
1447 | /* this is the Aho-Corasick algorithm modified a touch |
1448 | to include special handling for long "unknown char" | |
1449 | sequences. The basic idea being that we use AC as long | |
1450 | as we are dealing with a possible matching char, when | |
1451 | we encounter an unknown char (and we have not encountered | |
1452 | an accepting state) we scan forward until we find a legal | |
1453 | starting char. | |
1454 | AC matching is basically that of trie matching, except | |
1455 | that when we encounter a failing transition, we fall back | |
1456 | to the current states "fail state", and try the current char | |
1457 | again, a process we repeat until we reach the root state, | |
1458 | state 1, or a legal transition. If we fail on the root state | |
1459 | then we can either terminate if we have reached an accepting | |
1460 | state previously, or restart the entire process from the beginning | |
1461 | if we have not. | |
1462 | ||
1463 | */ | |
07be1b83 YO |
1464 | while (s <= last_start) { |
1465 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1466 | U8 *uc = (U8*)s; | |
1467 | U16 charid = 0; | |
1468 | U32 base = 1; | |
1469 | U32 state = 1; | |
1470 | UV uvc = 0; | |
1471 | STRLEN len = 0; | |
1472 | STRLEN foldlen = 0; | |
1473 | U8 *uscan = (U8*)NULL; | |
1474 | U8 *leftmost = NULL; | |
786e8c11 YO |
1475 | #ifdef DEBUGGING |
1476 | U32 accepted_word= 0; | |
1477 | #endif | |
07be1b83 YO |
1478 | U32 pointpos = 0; |
1479 | ||
1480 | while ( state && uc <= (U8*)strend ) { | |
1481 | int failed=0; | |
786e8c11 YO |
1482 | U32 word = aho->states[ state ].wordnum; |
1483 | ||
1de06328 YO |
1484 | if( state==1 ) { |
1485 | if ( bitmap ) { | |
1486 | DEBUG_TRIE_EXECUTE_r( | |
1487 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1488 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1489 | (char *)uc, do_utf8 ); | |
1490 | PerlIO_printf( Perl_debug_log, | |
1491 | " Scanning for legal start char...\n"); | |
1492 | } | |
1493 | ); | |
1494 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1495 | uc++; | |
786e8c11 | 1496 | } |
1de06328 | 1497 | s= (char *)uc; |
786e8c11 | 1498 | } |
786e8c11 YO |
1499 | if (uc >(U8*)last_start) break; |
1500 | } | |
1501 | ||
1502 | if ( word ) { | |
1503 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1504 | if (!leftmost || lpos < leftmost) { | |
1505 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1506 | leftmost= lpos; |
786e8c11 | 1507 | } |
07be1b83 | 1508 | if (base==0) break; |
786e8c11 | 1509 | |
07be1b83 YO |
1510 | } |
1511 | points[pointpos++ % maxlen]= uc; | |
4cadc6a9 YO |
1512 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
1513 | uvc, charid, foldlen, foldbuf, uniflags); | |
786e8c11 YO |
1514 | DEBUG_TRIE_EXECUTE_r({ |
1515 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1516 | s, do_utf8 ); | |
07be1b83 | 1517 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1518 | " Charid:%3u CP:%4"UVxf" ", |
1519 | charid, uvc); | |
1520 | }); | |
07be1b83 YO |
1521 | |
1522 | do { | |
6148ee25 | 1523 | #ifdef DEBUGGING |
786e8c11 | 1524 | word = aho->states[ state ].wordnum; |
6148ee25 | 1525 | #endif |
07be1b83 YO |
1526 | base = aho->states[ state ].trans.base; |
1527 | ||
786e8c11 YO |
1528 | DEBUG_TRIE_EXECUTE_r({ |
1529 | if (failed) | |
1530 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1531 | s, do_utf8 ); | |
07be1b83 | 1532 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1533 | "%sState: %4"UVxf", word=%"UVxf, |
1534 | failed ? " Fail transition to " : "", | |
1535 | (UV)state, (UV)word); | |
1536 | }); | |
07be1b83 YO |
1537 | if ( base ) { |
1538 | U32 tmp; | |
1539 | if (charid && | |
1540 | (base + charid > trie->uniquecharcount ) | |
1541 | && (base + charid - 1 - trie->uniquecharcount | |
1542 | < trie->lasttrans) | |
1543 | && trie->trans[base + charid - 1 - | |
1544 | trie->uniquecharcount].check == state | |
1545 | && (tmp=trie->trans[base + charid - 1 - | |
1546 | trie->uniquecharcount ].next)) | |
1547 | { | |
786e8c11 YO |
1548 | DEBUG_TRIE_EXECUTE_r( |
1549 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1550 | state = tmp; |
1551 | break; | |
1552 | } | |
1553 | else { | |
786e8c11 YO |
1554 | DEBUG_TRIE_EXECUTE_r( |
1555 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1556 | failed = 1; | |
1557 | state = aho->fail[state]; | |
07be1b83 YO |
1558 | } |
1559 | } | |
1560 | else { | |
1561 | /* we must be accepting here */ | |
786e8c11 YO |
1562 | DEBUG_TRIE_EXECUTE_r( |
1563 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1564 | failed = 1; | |
07be1b83 YO |
1565 | break; |
1566 | } | |
1567 | } while(state); | |
786e8c11 | 1568 | uc += len; |
07be1b83 YO |
1569 | if (failed) { |
1570 | if (leftmost) | |
1571 | break; | |
786e8c11 | 1572 | if (!state) state = 1; |
07be1b83 YO |
1573 | } |
1574 | } | |
1575 | if ( aho->states[ state ].wordnum ) { | |
1576 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1577 | if (!leftmost || lpos < leftmost) { |
1578 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1579 | leftmost = lpos; |
786e8c11 | 1580 | } |
07be1b83 | 1581 | } |
07be1b83 YO |
1582 | if (leftmost) { |
1583 | s = (char*)leftmost; | |
786e8c11 YO |
1584 | DEBUG_TRIE_EXECUTE_r({ |
1585 | PerlIO_printf( | |
1586 | Perl_debug_log,"Matches word #%"UVxf" at position %d. Trying full pattern...\n", | |
1587 | (UV)accepted_word, s - real_start | |
1588 | ); | |
1589 | }); | |
be8e71aa YO |
1590 | if (!reginfo || regtry(reginfo, s)) { |
1591 | FREETMPS; | |
1592 | LEAVE; | |
07be1b83 | 1593 | goto got_it; |
be8e71aa | 1594 | } |
07be1b83 | 1595 | s = HOPc(s,1); |
786e8c11 YO |
1596 | DEBUG_TRIE_EXECUTE_r({ |
1597 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1598 | }); | |
07be1b83 | 1599 | } else { |
786e8c11 YO |
1600 | DEBUG_TRIE_EXECUTE_r( |
1601 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1602 | break; |
1603 | } | |
1604 | } | |
be8e71aa YO |
1605 | FREETMPS; |
1606 | LEAVE; | |
07be1b83 YO |
1607 | } |
1608 | break; | |
b3c9acc1 | 1609 | default: |
3c3eec57 GS |
1610 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1611 | break; | |
d6a28714 | 1612 | } |
6eb5f6b9 JH |
1613 | return 0; |
1614 | got_it: | |
1615 | return s; | |
1616 | } | |
1617 | ||
1618 | /* | |
1619 | - regexec_flags - match a regexp against a string | |
1620 | */ | |
1621 | I32 | |
1622 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1623 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1624 | /* strend: pointer to null at end of string */ | |
1625 | /* strbeg: real beginning of string */ | |
1626 | /* minend: end of match must be >=minend after stringarg. */ | |
1627 | /* data: May be used for some additional optimizations. */ | |
1628 | /* nosave: For optimizations. */ | |
1629 | { | |
97aff369 | 1630 | dVAR; |
6eb5f6b9 JH |
1631 | register char *s; |
1632 | register regnode *c; | |
1633 | register char *startpos = stringarg; | |
6eb5f6b9 JH |
1634 | I32 minlen; /* must match at least this many chars */ |
1635 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1636 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1637 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1638 | char *scream_olds = NULL; |
3dab1dad | 1639 | SV* const oreplsv = GvSV(PL_replgv); |
1df70142 | 1640 | const bool do_utf8 = DO_UTF8(sv); |
2757e526 | 1641 | I32 multiline; |
0df25f3d | 1642 | |
3b0527fe | 1643 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1644 | |
1645 | GET_RE_DEBUG_FLAGS_DECL; | |
1646 | ||
9d4ba2ae | 1647 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1648 | |
1649 | /* Be paranoid... */ | |
1650 | if (prog == NULL || startpos == NULL) { | |
1651 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1652 | return 0; | |
1653 | } | |
1654 | ||
2757e526 | 1655 | multiline = prog->reganch & PMf_MULTILINE; |
3b0527fe | 1656 | reginfo.prog = prog; |
2757e526 | 1657 | |
bac06658 | 1658 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1659 | DEBUG_EXECUTE_r( |
1660 | debug_start_match(prog, do_utf8, startpos, strend, | |
1661 | "Matching"); | |
1662 | ); | |
bac06658 | 1663 | |
6eb5f6b9 | 1664 | minlen = prog->minlen; |
1de06328 YO |
1665 | |
1666 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1667 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1668 | "String too short [regexec_flags]...\n")); |
1669 | goto phooey; | |
1aa99e6b | 1670 | } |
6eb5f6b9 | 1671 | |
1de06328 | 1672 | |
6eb5f6b9 JH |
1673 | /* Check validity of program. */ |
1674 | if (UCHARAT(prog->program) != REG_MAGIC) { | |
1675 | Perl_croak(aTHX_ "corrupted regexp program"); | |
1676 | } | |
1677 | ||
1678 | PL_reg_flags = 0; | |
1679 | PL_reg_eval_set = 0; | |
1680 | PL_reg_maxiter = 0; | |
1681 | ||
1682 | if (prog->reganch & ROPT_UTF8) | |
1683 | PL_reg_flags |= RF_utf8; | |
1684 | ||
1685 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1686 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1687 | PL_bostr = strbeg; |
3b0527fe | 1688 | reginfo.sv = sv; |
6eb5f6b9 JH |
1689 | |
1690 | /* Mark end of line for $ (and such) */ | |
1691 | PL_regeol = strend; | |
1692 | ||
1693 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1694 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1695 | |
6eb5f6b9 JH |
1696 | /* If there is a "must appear" string, look for it. */ |
1697 | s = startpos; | |
1698 | ||
3b0527fe | 1699 | if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1700 | MAGIC *mg; |
1701 | ||
1702 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
3b0527fe | 1703 | reginfo.ganch = startpos; |
6eb5f6b9 JH |
1704 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1705 | && SvMAGIC(sv) | |
14befaf4 DM |
1706 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1707 | && mg->mg_len >= 0) { | |
3b0527fe | 1708 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
6eb5f6b9 | 1709 | if (prog->reganch & ROPT_ANCH_GPOS) { |
3b0527fe | 1710 | if (s > reginfo.ganch) |
6eb5f6b9 | 1711 | goto phooey; |
3b0527fe | 1712 | s = reginfo.ganch; |
6eb5f6b9 JH |
1713 | } |
1714 | } | |
1715 | else /* pos() not defined */ | |
3b0527fe | 1716 | reginfo.ganch = strbeg; |
6eb5f6b9 JH |
1717 | } |
1718 | ||
a0714e2c | 1719 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1720 | re_scream_pos_data d; |
1721 | ||
1722 | d.scream_olds = &scream_olds; | |
1723 | d.scream_pos = &scream_pos; | |
1724 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1725 | if (!s) { |
a3621e74 | 1726 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1727 | goto phooey; /* not present */ |
3fa9c3d7 | 1728 | } |
6eb5f6b9 JH |
1729 | } |
1730 | ||
1de06328 | 1731 | |
6eb5f6b9 JH |
1732 | |
1733 | /* Simplest case: anchored match need be tried only once. */ | |
1734 | /* [unless only anchor is BOL and multiline is set] */ | |
1735 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { | |
3b0527fe | 1736 | if (s == startpos && regtry(®info, startpos)) |
6eb5f6b9 | 1737 | goto got_it; |
7fba1cd6 | 1738 | else if (multiline || (prog->reganch & ROPT_IMPLICIT) |
6eb5f6b9 JH |
1739 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
1740 | { | |
1741 | char *end; | |
1742 | ||
1743 | if (minlen) | |
1744 | dontbother = minlen - 1; | |
1aa99e6b | 1745 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1746 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1747 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1748 | if (s == startpos) |
1749 | goto after_try; | |
1750 | while (1) { | |
3b0527fe | 1751 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1752 | goto got_it; |
1753 | after_try: | |
1754 | if (s >= end) | |
1755 | goto phooey; | |
1756 | if (prog->reganch & RE_USE_INTUIT) { | |
1757 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); | |
1758 | if (!s) | |
1759 | goto phooey; | |
1760 | } | |
1761 | else | |
1762 | s++; | |
1763 | } | |
1764 | } else { | |
1765 | if (s > startpos) | |
1766 | s--; | |
1767 | while (s < end) { | |
1768 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
3b0527fe | 1769 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1770 | goto got_it; |
1771 | } | |
1772 | } | |
1773 | } | |
1774 | } | |
1775 | goto phooey; | |
1776 | } else if (prog->reganch & ROPT_ANCH_GPOS) { | |
3b0527fe | 1777 | if (regtry(®info, reginfo.ganch)) |
6eb5f6b9 JH |
1778 | goto got_it; |
1779 | goto phooey; | |
1780 | } | |
1781 | ||
1782 | /* Messy cases: unanchored match. */ | |
33b8afdf | 1783 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) { |
6eb5f6b9 JH |
1784 | /* we have /x+whatever/ */ |
1785 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1786 | char ch; |
bf93d4cc GS |
1787 | #ifdef DEBUGGING |
1788 | int did_match = 0; | |
1789 | #endif | |
33b8afdf JH |
1790 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1791 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1792 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1793 | |
1aa99e6b | 1794 | if (do_utf8) { |
4cadc6a9 | 1795 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1796 | if (*s == ch) { |
a3621e74 | 1797 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1798 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1799 | s += UTF8SKIP(s); |
1800 | while (s < strend && *s == ch) | |
1801 | s += UTF8SKIP(s); | |
1802 | } | |
4cadc6a9 | 1803 | ); |
6eb5f6b9 JH |
1804 | } |
1805 | else { | |
4cadc6a9 | 1806 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1807 | if (*s == ch) { |
a3621e74 | 1808 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1809 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1810 | s++; |
1811 | while (s < strend && *s == ch) | |
1812 | s++; | |
1813 | } | |
4cadc6a9 | 1814 | ); |
6eb5f6b9 | 1815 | } |
a3621e74 | 1816 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1817 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1818 | "Did not find anchored character...\n") |
1819 | ); | |
6eb5f6b9 | 1820 | } |
a0714e2c SS |
1821 | else if (prog->anchored_substr != NULL |
1822 | || prog->anchored_utf8 != NULL | |
1823 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1824 | && prog->float_max_offset < strend - s)) { |
1825 | SV *must; | |
1826 | I32 back_max; | |
1827 | I32 back_min; | |
1828 | char *last; | |
6eb5f6b9 | 1829 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1830 | #ifdef DEBUGGING |
1831 | int did_match = 0; | |
1832 | #endif | |
33b8afdf JH |
1833 | if (prog->anchored_substr || prog->anchored_utf8) { |
1834 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1835 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1836 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1837 | back_max = back_min = prog->anchored_offset; | |
1838 | } else { | |
1839 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1840 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1841 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1842 | back_max = prog->float_max_offset; | |
1843 | back_min = prog->float_min_offset; | |
1844 | } | |
1de06328 YO |
1845 | |
1846 | ||
33b8afdf JH |
1847 | if (must == &PL_sv_undef) |
1848 | /* could not downgrade utf8 check substring, so must fail */ | |
1849 | goto phooey; | |
1850 | ||
1de06328 YO |
1851 | if (back_min<0) { |
1852 | last = strend; | |
1853 | } else { | |
1854 | last = HOP3c(strend, /* Cannot start after this */ | |
1855 | -(I32)(CHR_SVLEN(must) | |
1856 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1857 | } | |
6eb5f6b9 JH |
1858 | if (s > PL_bostr) |
1859 | last1 = HOPc(s, -1); | |
1860 | else | |
1861 | last1 = s - 1; /* bogus */ | |
1862 | ||
a0288114 | 1863 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1864 | check_substr==must. */ |
1865 | scream_pos = -1; | |
1866 | dontbother = end_shift; | |
1867 | strend = HOPc(strend, -dontbother); | |
1868 | while ( (s <= last) && | |
9041c2e3 | 1869 | ((flags & REXEC_SCREAM) |
1de06328 | 1870 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1871 | end_shift, &scream_pos, 0)) |
1de06328 | 1872 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1873 | (unsigned char*)strend, must, |
7fba1cd6 | 1874 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1875 | /* we may be pointing at the wrong string */ |
1876 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1877 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1878 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1879 | if (HOPc(s, -back_max) > last1) { |
1880 | last1 = HOPc(s, -back_min); | |
1881 | s = HOPc(s, -back_max); | |
1882 | } | |
1883 | else { | |
52657f30 | 1884 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1885 | |
1886 | last1 = HOPc(s, -back_min); | |
52657f30 | 1887 | s = t; |
6eb5f6b9 | 1888 | } |
1aa99e6b | 1889 | if (do_utf8) { |
6eb5f6b9 | 1890 | while (s <= last1) { |
3b0527fe | 1891 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1892 | goto got_it; |
1893 | s += UTF8SKIP(s); | |
1894 | } | |
1895 | } | |
1896 | else { | |
1897 | while (s <= last1) { | |
3b0527fe | 1898 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1899 | goto got_it; |
1900 | s++; | |
1901 | } | |
1902 | } | |
1903 | } | |
ab3bbdeb YO |
1904 | DEBUG_EXECUTE_r(if (!did_match) { |
1905 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
1906 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
1907 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 1908 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 1909 | ? "anchored" : "floating"), |
ab3bbdeb YO |
1910 | quoted, RE_SV_TAIL(must)); |
1911 | }); | |
6eb5f6b9 JH |
1912 | goto phooey; |
1913 | } | |
786e8c11 | 1914 | else if ( (c = prog->regstclass) ) { |
f14c76ed | 1915 | if (minlen) { |
be8e71aa | 1916 | const OPCODE op = OP(prog->regstclass); |
66e933ab | 1917 | /* don't bother with what can't match */ |
786e8c11 | 1918 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
1919 | strend = HOPc(strend, -(minlen - 1)); |
1920 | } | |
a3621e74 | 1921 | DEBUG_EXECUTE_r({ |
be8e71aa | 1922 | SV * const prop = sv_newmortal(); |
32fc9b6a | 1923 | regprop(prog, prop, c); |
0df25f3d | 1924 | { |
ab3bbdeb YO |
1925 | RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1), |
1926 | s,strend-s,60); | |
0df25f3d | 1927 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 1928 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 1929 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 1930 | quoted, (int)(strend - s)); |
0df25f3d | 1931 | } |
ffc61ed2 | 1932 | }); |
3b0527fe | 1933 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1934 | goto got_it; |
07be1b83 | 1935 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
1936 | } |
1937 | else { | |
1938 | dontbother = 0; | |
a0714e2c | 1939 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 1940 | /* Trim the end. */ |
d6a28714 | 1941 | char *last; |
33b8afdf JH |
1942 | SV* float_real; |
1943 | ||
1944 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1945 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1946 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
1947 | |
1948 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 1949 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
1950 | end_shift, &scream_pos, 1); /* last one */ |
1951 | if (!last) | |
ffc61ed2 | 1952 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
1953 | /* we may be pointing at the wrong string */ |
1954 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 1955 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 1956 | } |
d6a28714 JH |
1957 | else { |
1958 | STRLEN len; | |
cfd0369c | 1959 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 1960 | |
33b8afdf | 1961 | if (SvTAIL(float_real)) { |
d6a28714 JH |
1962 | if (memEQ(strend - len + 1, little, len - 1)) |
1963 | last = strend - len + 1; | |
7fba1cd6 | 1964 | else if (!multiline) |
9041c2e3 | 1965 | last = memEQ(strend - len, little, len) |
bd61b366 | 1966 | ? strend - len : NULL; |
b8c5462f | 1967 | else |
d6a28714 JH |
1968 | goto find_last; |
1969 | } else { | |
1970 | find_last: | |
9041c2e3 | 1971 | if (len) |
d6a28714 | 1972 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 1973 | else |
a0288114 | 1974 | last = strend; /* matching "$" */ |
b8c5462f | 1975 | } |
b8c5462f | 1976 | } |
bf93d4cc | 1977 | if (last == NULL) { |
a3621e74 | 1978 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
bf93d4cc | 1979 | "%sCan't trim the tail, match fails (should not happen)%s\n", |
e4584336 | 1980 | PL_colors[4], PL_colors[5])); |
bf93d4cc GS |
1981 | goto phooey; /* Should not happen! */ |
1982 | } | |
d6a28714 JH |
1983 | dontbother = strend - last + prog->float_min_offset; |
1984 | } | |
1985 | if (minlen && (dontbother < minlen)) | |
1986 | dontbother = minlen - 1; | |
1987 | strend -= dontbother; /* this one's always in bytes! */ | |
1988 | /* We don't know much -- general case. */ | |
1aa99e6b | 1989 | if (do_utf8) { |
d6a28714 | 1990 | for (;;) { |
3b0527fe | 1991 | if (regtry(®info, s)) |
d6a28714 JH |
1992 | goto got_it; |
1993 | if (s >= strend) | |
1994 | break; | |
b8c5462f | 1995 | s += UTF8SKIP(s); |
d6a28714 JH |
1996 | }; |
1997 | } | |
1998 | else { | |
1999 | do { | |
3b0527fe | 2000 | if (regtry(®info, s)) |
d6a28714 JH |
2001 | goto got_it; |
2002 | } while (s++ < strend); | |
2003 | } | |
2004 | } | |
2005 | ||
2006 | /* Failure. */ | |
2007 | goto phooey; | |
2008 | ||
2009 | got_it: | |
2010 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2011 | ||
2012 | if (PL_reg_eval_set) { | |
2013 | /* Preserve the current value of $^R */ | |
2014 | if (oreplsv != GvSV(PL_replgv)) | |
2015 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2016 | restored, the value remains | |
2017 | the same. */ | |
4f639d21 | 2018 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2019 | } |
2020 | ||
2021 | /* make sure $`, $&, $', and $digit will work later */ | |
2022 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2023 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2024 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2025 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2026 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2027 | if ((SvIsCOW(sv) |
2028 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2029 | if (DEBUG_C_TEST) { | |
2030 | PerlIO_printf(Perl_debug_log, | |
2031 | "Copy on write: regexp capture, type %d\n", | |
2032 | (int) SvTYPE(sv)); | |
2033 | } | |
2034 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2035 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2036 | assert (SvPOKp(prog->saved_copy)); |
2037 | } else | |
2038 | #endif | |
2039 | { | |
2040 | RX_MATCH_COPIED_on(prog); | |
2041 | s = savepvn(strbeg, i); | |
2042 | prog->subbeg = s; | |
2043 | } | |
d6a28714 | 2044 | prog->sublen = i; |
d6a28714 JH |
2045 | } |
2046 | else { | |
2047 | prog->subbeg = strbeg; | |
2048 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2049 | } | |
2050 | } | |
9041c2e3 | 2051 | |
d6a28714 JH |
2052 | return 1; |
2053 | ||
2054 | phooey: | |
a3621e74 | 2055 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2056 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2057 | if (PL_reg_eval_set) |
4f639d21 | 2058 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2059 | return 0; |
2060 | } | |
2061 | ||
2062 | /* | |
2063 | - regtry - try match at specific point | |
2064 | */ | |
2065 | STATIC I32 /* 0 failure, 1 success */ | |
3b0527fe | 2066 | S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos) |
d6a28714 | 2067 | { |
97aff369 | 2068 | dVAR; |
d6a28714 JH |
2069 | register I32 *sp; |
2070 | register I32 *ep; | |
2071 | CHECKPOINT lastcp; | |
3b0527fe | 2072 | regexp *prog = reginfo->prog; |
a3621e74 | 2073 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2074 | |
2075 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { | |
2076 | MAGIC *mg; | |
2077 | ||
2078 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2079 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2080 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2081 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2082 | )); |
e8347627 | 2083 | SAVEI32(cxstack[cxstack_ix].blk_oldsp); |
d6a28714 JH |
2084 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2085 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2086 | SAVETMPS; | |
2087 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2088 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2089 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2090 | ||
3b0527fe | 2091 | if (reginfo->sv) { |
d6a28714 | 2092 | /* Make $_ available to executed code. */ |
3b0527fe | 2093 | if (reginfo->sv != DEFSV) { |
59f00321 | 2094 | SAVE_DEFSV; |
3b0527fe | 2095 | DEFSV = reginfo->sv; |
b8c5462f | 2096 | } |
d6a28714 | 2097 | |
3b0527fe DM |
2098 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2099 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2100 | /* prepare for quick setting of pos */ |
d300d9fa NC |
2101 | #ifdef PERL_OLD_COPY_ON_WRITE |
2102 | if (SvIsCOW(sv)) | |
2103 | sv_force_normal_flags(sv, 0); | |
2104 | #endif | |
3dab1dad | 2105 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2106 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2107 | mg->mg_len = -1; |
b8c5462f | 2108 | } |
d6a28714 JH |
2109 | PL_reg_magic = mg; |
2110 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2111 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2112 | } |
09687e5a | 2113 | if (!PL_reg_curpm) { |
a02a5408 | 2114 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2115 | #ifdef USE_ITHREADS |
2116 | { | |
be8e71aa | 2117 | SV* const repointer = newSViv(0); |
577e12cc | 2118 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2119 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2120 | av_push(PL_regex_padav,repointer); |
2121 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2122 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2123 | } | |
2124 | #endif | |
2125 | } | |
aaa362c4 | 2126 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2127 | PL_reg_oldcurpm = PL_curpm; |
2128 | PL_curpm = PL_reg_curpm; | |
2129 | if (RX_MATCH_COPIED(prog)) { | |
2130 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2131 | since it may be needed if this match fails. Thus | |
2132 | $` inside (?{}) could fail... */ | |
2133 | PL_reg_oldsaved = prog->subbeg; | |
2134 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2135 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2136 | PL_nrs = prog->saved_copy; |
2137 | #endif | |
d6a28714 JH |
2138 | RX_MATCH_COPIED_off(prog); |
2139 | } | |
2140 | else | |
bd61b366 | 2141 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2142 | prog->subbeg = PL_bostr; |
2143 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2144 | } | |
973dddac | 2145 | prog->startp[0] = startpos - PL_bostr; |
d6a28714 JH |
2146 | PL_reginput = startpos; |
2147 | PL_regstartp = prog->startp; | |
2148 | PL_regendp = prog->endp; | |
2149 | PL_reglastparen = &prog->lastparen; | |
a01268b5 | 2150 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2151 | prog->lastparen = 0; |
03994de8 | 2152 | prog->lastcloseparen = 0; |
d6a28714 | 2153 | PL_regsize = 0; |
a3621e74 | 2154 | DEBUG_EXECUTE_r(PL_reg_starttry = startpos); |
d6a28714 JH |
2155 | if (PL_reg_start_tmpl <= prog->nparens) { |
2156 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2157 | if(PL_reg_start_tmp) | |
2158 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2159 | else | |
a02a5408 | 2160 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2161 | } |
2162 | ||
2163 | /* XXXX What this code is doing here?!!! There should be no need | |
2164 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2165 | this! --ilya*/ |
dafc8851 JH |
2166 | |
2167 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2168 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2169 | * PL_reglastparen), is not needed at all by the test suite |
2170 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2171 | * enough, for building DynaLoader, or otherwise this | |
2172 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2173 | * will happen. Meanwhile, this code *is* needed for the | |
2174 | * above-mentioned test suite tests to succeed. The common theme | |
2175 | * on those tests seems to be returning null fields from matches. | |
2176 | * --jhi */ | |
dafc8851 | 2177 | #if 1 |
d6a28714 JH |
2178 | sp = prog->startp; |
2179 | ep = prog->endp; | |
2180 | if (prog->nparens) { | |
097eb12c | 2181 | register I32 i; |
eb160463 | 2182 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2183 | *++sp = -1; |
2184 | *++ep = -1; | |
2185 | } | |
2186 | } | |
dafc8851 | 2187 | #endif |
02db2b7b | 2188 | REGCP_SET(lastcp); |
3b0527fe | 2189 | if (regmatch(reginfo, prog->program + 1)) { |
d6a28714 JH |
2190 | prog->endp[0] = PL_reginput - PL_bostr; |
2191 | return 1; | |
2192 | } | |
02db2b7b | 2193 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2194 | return 0; |
2195 | } | |
2196 | ||
02db2b7b | 2197 | |
8ba1375e MJD |
2198 | #define sayYES goto yes |
2199 | #define sayNO goto no | |
e0f9d4a8 | 2200 | #define sayNO_ANYOF goto no_anyof |
8ba1375e | 2201 | #define sayYES_FINAL goto yes_final |
8ba1375e MJD |
2202 | #define sayNO_FINAL goto no_final |
2203 | #define sayNO_SILENT goto do_no | |
2204 | #define saySAME(x) if (x) goto yes; else goto no | |
2205 | ||
3ab3c9b4 | 2206 | #define CACHEsayNO STMT_START { \ |
3298f257 DM |
2207 | if (st->u.whilem.cache_offset | st->u.whilem.cache_bit) \ |
2208 | PL_reg_poscache[st->u.whilem.cache_offset] |= \ | |
2209 | (1<<st->u.whilem.cache_bit); \ | |
3ab3c9b4 HS |
2210 | sayNO; \ |
2211 | } STMT_END | |
2212 | ||
3298f257 | 2213 | |
a3621e74 | 2214 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2215 | 'failed...' are printed. It should be set such that messages |
2216 | are inline with the regop output that created them. | |
a3621e74 | 2217 | */ |
265c4333 | 2218 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2219 | |
2220 | ||
2221 | /* Make sure there is a test for this +1 options in re_tests */ | |
2222 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2223 | ||
40a82448 DM |
2224 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2225 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2226 | |
86545054 DM |
2227 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2228 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2229 | ||
5d9a96ca DM |
2230 | /* grab a new slab and return the first slot in it */ |
2231 | ||
2232 | STATIC regmatch_state * | |
2233 | S_push_slab(pTHX) | |
2234 | { | |
54df2634 NC |
2235 | #if PERL_VERSION < 9 |
2236 | dMY_CXT; | |
2237 | #endif | |
5d9a96ca DM |
2238 | regmatch_slab *s = PL_regmatch_slab->next; |
2239 | if (!s) { | |
2240 | Newx(s, 1, regmatch_slab); | |
2241 | s->prev = PL_regmatch_slab; | |
2242 | s->next = NULL; | |
2243 | PL_regmatch_slab->next = s; | |
2244 | } | |
2245 | PL_regmatch_slab = s; | |
86545054 | 2246 | return SLAB_FIRST(s); |
5d9a96ca | 2247 | } |
5b47454d | 2248 | |
95b24440 DM |
2249 | /* simulate a recursive call to regmatch */ |
2250 | ||
2251 | #define REGMATCH(ns, where) \ | |
5d9a96ca DM |
2252 | st->scan = scan; \ |
2253 | scan = (ns); \ | |
2254 | st->resume_state = resume_##where; \ | |
95b24440 DM |
2255 | goto start_recurse; \ |
2256 | resume_point_##where: | |
2257 | ||
40a82448 DM |
2258 | /* push a new state then goto it */ |
2259 | ||
2260 | #define PUSH_STATE_GOTO(state, node) \ | |
2261 | scan = node; \ | |
2262 | st->resume_state = state; \ | |
2263 | goto push_state; | |
2264 | ||
2265 | /* push a new state with success backtracking, then goto it */ | |
2266 | ||
2267 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2268 | scan = node; \ | |
2269 | st->resume_state = state; \ | |
2270 | goto push_yes_state; | |
2271 | ||
aa283a38 | 2272 | |
aa283a38 | 2273 | |
d6a28714 JH |
2274 | /* |
2275 | - regmatch - main matching routine | |
2276 | * | |
2277 | * Conceptually the strategy is simple: check to see whether the current | |
2278 | * node matches, call self recursively to see whether the rest matches, | |
2279 | * and then act accordingly. In practice we make some effort to avoid | |
2280 | * recursion, in particular by going through "ordinary" nodes (that don't | |
2281 | * need to know whether the rest of the match failed) by a loop instead of | |
2282 | * by recursion. | |
2283 | */ | |
2284 | /* [lwall] I've hoisted the register declarations to the outer block in order to | |
2285 | * maybe save a little bit of pushing and popping on the stack. It also takes | |
2286 | * advantage of machines that use a register save mask on subroutine entry. | |
95b24440 DM |
2287 | * |
2288 | * This function used to be heavily recursive, but since this had the | |
2289 | * effect of blowing the CPU stack on complex regexes, it has been | |
2290 | * restructured to be iterative, and to save state onto the heap rather | |
2291 | * than the stack. Essentially whereever regmatch() used to be called, it | |
2292 | * pushes the current state, notes where to return, then jumps back into | |
2293 | * the main loop. | |
2294 | * | |
2295 | * Originally the structure of this function used to look something like | |
2296 | ||
2297 | S_regmatch() { | |
2298 | int a = 1, b = 2; | |
2299 | ... | |
2300 | while (scan != NULL) { | |
5d9a96ca | 2301 | a++; // do stuff with a and b |
95b24440 DM |
2302 | ... |
2303 | switch (OP(scan)) { | |
2304 | case FOO: { | |
2305 | int local = 3; | |
2306 | ... | |
2307 | if (regmatch(...)) // recurse | |
2308 | goto yes; | |
2309 | } | |
2310 | ... | |
2311 | } | |
2312 | } | |
2313 | yes: | |
2314 | return 1; | |
2315 | } | |
2316 | ||
2317 | * Now it looks something like this: | |
2318 | ||
5d9a96ca | 2319 | typedef struct { |
95b24440 DM |
2320 | int a, b, local; |
2321 | int resume_state; | |
5d9a96ca | 2322 | } regmatch_state; |
95b24440 DM |
2323 | |
2324 | S_regmatch() { | |
5d9a96ca DM |
2325 | regmatch_state *st = new(); |
2326 | int depth=0; | |
2327 | st->a++; // do stuff with a and b | |
95b24440 DM |
2328 | ... |
2329 | while (scan != NULL) { | |
2330 | ... | |
2331 | switch (OP(scan)) { | |
2332 | case FOO: { | |
5d9a96ca | 2333 | st->local = 3; |
95b24440 | 2334 | ... |
5d9a96ca DM |
2335 | st->scan = scan; |
2336 | scan = ...; | |
2337 | st->resume_state = resume_FOO; | |
2338 | goto start_recurse; // recurse | |
95b24440 | 2339 | |
5d9a96ca DM |
2340 | resume_point_FOO: |
2341 | if (result) | |
95b24440 DM |
2342 | goto yes; |
2343 | } | |
2344 | ... | |
2345 | } | |
5d9a96ca DM |
2346 | start_recurse: |
2347 | st = new(); push a new state | |
2348 | st->a = 1; st->b = 2; | |
2349 | depth++; | |
95b24440 | 2350 | } |
5d9a96ca | 2351 | yes: |
95b24440 | 2352 | result = 1; |
5d9a96ca DM |
2353 | if (depth--) { |
2354 | st = pop(); | |
95b24440 DM |
2355 | switch (resume_state) { |
2356 | case resume_FOO: | |
2357 | goto resume_point_FOO; | |
2358 | ... | |
2359 | } | |
2360 | } | |
2361 | return result | |
2362 | } | |
2363 | ||
2364 | * WARNING: this means that any line in this function that contains a | |
2365 | * REGMATCH() or TRYPAREN() is actually simulating a recursive call to | |
2366 | * regmatch() using gotos instead. Thus the values of any local variables | |
2367 | * not saved in the regmatch_state structure will have been lost when | |
2368 | * execution resumes on the next line . | |
5d9a96ca DM |
2369 | * |
2370 | * States (ie the st pointer) are allocated in slabs of about 4K in size. | |
2371 | * PL_regmatch_state always points to the currently active state, and | |
2372 | * PL_regmatch_slab points to the slab currently containing PL_regmatch_state. | |
2373 | * The first time regmatch is called, the first slab is allocated, and is | |
2374 | * never freed until interpreter desctruction. When the slab is full, | |
2375 | * a new one is allocated chained to the end. At exit from regmatch, slabs | |
2376 | * allocated since entry are freed. | |
d6a28714 | 2377 | */ |
95b24440 | 2378 | |
40a82448 | 2379 | |
5bc10b2c | 2380 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2381 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2382 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2383 | PerlIO_printf(Perl_debug_log, \ | |
2384 | " %*s"pp" %s\n", \ | |
2385 | depth*2, "", \ | |
03363afd | 2386 | reg_name[st->resume_state] ); \ |
265c4333 | 2387 | }); |
5bc10b2c | 2388 | |
40a82448 | 2389 | |
3dab1dad | 2390 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2391 | |
3df15adc | 2392 | #ifdef DEBUGGING |
5bc10b2c | 2393 | |
ab3bbdeb YO |
2394 | STATIC void |
2395 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2396 | const char *start, const char *end, const char *blurb) | |
2397 | { | |
2398 | const bool utf8_pat= prog->reganch & ROPT_UTF8 ? 1 : 0; | |
2399 | if (!PL_colorset) | |
2400 | reginitcolors(); | |
2401 | { | |
2402 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2403 | prog->precomp, prog->prelen, 60); | |
2404 | ||
2405 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2406 | start, end - start, 60); | |
2407 | ||
2408 | PerlIO_printf(Perl_debug_log, | |
2409 | "%s%s REx%s %s against %s\n", | |
2410 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2411 | ||
2412 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2413 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2414 | utf8_pat ? "pattern" : "", | |
2415 | utf8_pat && do_utf8 ? " and " : "", | |
2416 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2417 | ); |
2418 | } | |
2419 | } | |
3df15adc YO |
2420 | |
2421 | STATIC void | |
786e8c11 YO |
2422 | S_dump_exec_pos(pTHX_ const char *locinput, |
2423 | const regnode *scan, | |
2424 | const char *loc_regeol, | |
2425 | const char *loc_bostr, | |
2426 | const char *loc_reg_starttry, | |
2427 | const bool do_utf8) | |
07be1b83 | 2428 | { |
786e8c11 | 2429 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2430 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2431 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2432 | /* The part of the string before starttry has one color |
2433 | (pref0_len chars), between starttry and current | |
2434 | position another one (pref_len - pref0_len chars), | |
2435 | after the current position the third one. | |
2436 | We assume that pref0_len <= pref_len, otherwise we | |
2437 | decrease pref0_len. */ | |
786e8c11 YO |
2438 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2439 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2440 | int pref0_len; |
2441 | ||
2442 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2443 | pref_len++; | |
786e8c11 YO |
2444 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2445 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2446 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2447 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2448 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2449 | l--; | |
2450 | if (pref0_len < 0) | |
2451 | pref0_len = 0; | |
2452 | if (pref0_len > pref_len) | |
2453 | pref0_len = pref_len; | |
2454 | { | |
3df15adc | 2455 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2456 | |
ab3bbdeb | 2457 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2458 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2459 | |
ab3bbdeb | 2460 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2461 | (locinput - pref_len + pref0_len), |
1de06328 | 2462 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2463 | |
ab3bbdeb | 2464 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2465 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2466 | |
1de06328 | 2467 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2468 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2469 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2470 | (IV)(locinput - loc_bostr), |
07be1b83 | 2471 | len0, s0, |
07be1b83 | 2472 | len1, s1, |
07be1b83 | 2473 | (docolor ? "" : "> <"), |
07be1b83 | 2474 | len2, s2, |
1de06328 | 2475 | tlen > 19 ? 0 : 19 - tlen, |
07be1b83 YO |
2476 | ""); |
2477 | } | |
2478 | } | |
3df15adc | 2479 | |
07be1b83 YO |
2480 | #endif |
2481 | ||
d6a28714 | 2482 | STATIC I32 /* 0 failure, 1 success */ |
3b0527fe | 2483 | S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2484 | { |
54df2634 NC |
2485 | #if PERL_VERSION < 9 |
2486 | dMY_CXT; | |
2487 | #endif | |
27da23d5 | 2488 | dVAR; |
95b24440 | 2489 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2490 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2491 | |
3b0527fe DM |
2492 | regexp *rex = reginfo->prog; |
2493 | ||
5d9a96ca DM |
2494 | regmatch_slab *orig_slab; |
2495 | regmatch_state *orig_state; | |
a3621e74 | 2496 | |
5d9a96ca DM |
2497 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2498 | register regmatch_state *st; | |
95b24440 | 2499 | |
5d9a96ca DM |
2500 | /* cache heavy used fields of st in registers */ |
2501 | register regnode *scan; | |
2502 | register regnode *next; | |
2503 | register I32 n = 0; /* initialize to shut up compiler warning */ | |
2504 | register char *locinput = PL_reginput; | |
95b24440 | 2505 | |
5d9a96ca DM |
2506 | /* these variables are NOT saved during a recusive RFEGMATCH: */ |
2507 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ | |
b69b0499 | 2508 | bool result = 0; /* return value of S_regmatch */ |
5d9a96ca | 2509 | int depth = 0; /* depth of recursion */ |
77cb431f DM |
2510 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2511 | subpattern */ | |
faec1544 | 2512 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2513 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2514 | U32 state_num; |
95b24440 | 2515 | |
10edeb5d JH |
2516 | I32 parenfloor = 0; |
2517 | ||
95b24440 | 2518 | #ifdef DEBUGGING |
e68ec53f | 2519 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2520 | #endif |
2521 | ||
5d9a96ca DM |
2522 | /* on first ever call to regmatch, allocate first slab */ |
2523 | if (!PL_regmatch_slab) { | |
2524 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2525 | PL_regmatch_slab->prev = NULL; | |
2526 | PL_regmatch_slab->next = NULL; | |
86545054 | 2527 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2528 | } |
2529 | ||
2530 | /* remember current high-water mark for exit */ | |
2531 | /* XXX this should be done with SAVE* instead */ | |
2532 | orig_slab = PL_regmatch_slab; | |
2533 | orig_state = PL_regmatch_state; | |
2534 | ||
2535 | /* grab next free state slot */ | |
2536 | st = ++PL_regmatch_state; | |
86545054 | 2537 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2538 | st = PL_regmatch_state = S_push_slab(aTHX); |
2539 | ||
2540 | st->minmod = 0; | |
2541 | st->sw = 0; | |
2542 | st->logical = 0; | |
b8591aee | 2543 | cur_curlyx = NULL; |
786e8c11 | 2544 | |
d6a28714 JH |
2545 | /* Note that nextchr is a byte even in UTF */ |
2546 | nextchr = UCHARAT(locinput); | |
2547 | scan = prog; | |
2548 | while (scan != NULL) { | |
8ba1375e | 2549 | |
a3621e74 | 2550 | DEBUG_EXECUTE_r( { |
6136c704 | 2551 | SV * const prop = sv_newmortal(); |
1de06328 | 2552 | regnode *rnext=regnext(scan); |
786e8c11 | 2553 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2554 | regprop(rex, prop, scan); |
07be1b83 YO |
2555 | |
2556 | PerlIO_printf(Perl_debug_log, | |
2557 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
5bc10b2c | 2558 | (IV)(scan - rex->program), depth*2, "", |
07be1b83 | 2559 | SvPVX_const(prop), |
1de06328 YO |
2560 | (PL_regkind[OP(scan)] == END || !rnext) ? |
2561 | 0 : (IV)(rnext - rex->program)); | |
2a782b5b | 2562 | }); |
d6a28714 JH |
2563 | |
2564 | next = scan + NEXT_OFF(scan); | |
2565 | if (next == scan) | |
2566 | next = NULL; | |
40a82448 | 2567 | state_num = OP(scan); |
d6a28714 | 2568 | |
40a82448 DM |
2569 | reenter_switch: |
2570 | switch (state_num) { | |
d6a28714 | 2571 | case BOL: |
7fba1cd6 | 2572 | if (locinput == PL_bostr) |
d6a28714 | 2573 | { |
3b0527fe | 2574 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2575 | break; |
2576 | } | |
d6a28714 JH |
2577 | sayNO; |
2578 | case MBOL: | |
12d33761 HS |
2579 | if (locinput == PL_bostr || |
2580 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2581 | { |
b8c5462f JH |
2582 | break; |
2583 | } | |
d6a28714 JH |
2584 | sayNO; |
2585 | case SBOL: | |
c2a73568 | 2586 | if (locinput == PL_bostr) |
b8c5462f | 2587 | break; |
d6a28714 JH |
2588 | sayNO; |
2589 | case GPOS: | |
3b0527fe | 2590 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2591 | break; |
2592 | sayNO; | |
2593 | case EOL: | |
d6a28714 JH |
2594 | goto seol; |
2595 | case MEOL: | |
d6a28714 | 2596 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2597 | sayNO; |
b8c5462f | 2598 | break; |
d6a28714 JH |
2599 | case SEOL: |
2600 | seol: | |
2601 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2602 | sayNO; |
d6a28714 | 2603 | if (PL_regeol - locinput > 1) |
b8c5462f | 2604 | sayNO; |
b8c5462f | 2605 | break; |
d6a28714 JH |
2606 | case EOS: |
2607 | if (PL_regeol != locinput) | |
b8c5462f | 2608 | sayNO; |
d6a28714 | 2609 | break; |
ffc61ed2 | 2610 | case SANY: |
d6a28714 | 2611 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2612 | sayNO; |
f33976b4 DB |
2613 | if (do_utf8) { |
2614 | locinput += PL_utf8skip[nextchr]; | |
2615 | if (locinput > PL_regeol) | |
2616 | sayNO; | |
2617 | nextchr = UCHARAT(locinput); | |
2618 | } | |
2619 | else | |
2620 | nextchr = UCHARAT(++locinput); | |
2621 | break; | |
2622 | case CANY: | |
2623 | if (!nextchr && locinput >= PL_regeol) | |
2624 | sayNO; | |
b8c5462f | 2625 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2626 | break; |
ffc61ed2 | 2627 | case REG_ANY: |
1aa99e6b IH |
2628 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2629 | sayNO; | |
2630 | if (do_utf8) { | |
b8c5462f | 2631 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2632 | if (locinput > PL_regeol) |
2633 | sayNO; | |
a0ed51b3 | 2634 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2635 | } |
1aa99e6b IH |
2636 | else |
2637 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2638 | break; |
166ba7cd DM |
2639 | |
2640 | #undef ST | |
2641 | #define ST st->u.trie | |
786e8c11 YO |
2642 | case TRIEC: |
2643 | /* In this case the charclass data is available inline so | |
2644 | we can fail fast without a lot of extra overhead. | |
2645 | */ | |
2646 | if (scan->flags == EXACT || !do_utf8) { | |
2647 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2648 | DEBUG_EXECUTE_r( | |
2649 | PerlIO_printf(Perl_debug_log, | |
2650 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2651 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2652 | ); |
2653 | sayNO_SILENT; | |
2654 | /* NOTREACHED */ | |
2655 | } | |
2656 | } | |
2657 | /* FALL THROUGH */ | |
5b47454d | 2658 | case TRIE: |
3dab1dad | 2659 | { |
07be1b83 | 2660 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2661 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2662 | trie_type = do_utf8 ? | |
2663 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2664 | : trie_plain; | |
2665 | ||
2666 | /* what trie are we using right now */ | |
be8e71aa | 2667 | reg_trie_data * const trie |
3dab1dad YO |
2668 | = (reg_trie_data*)rex->data->data[ ARG( scan ) ]; |
2669 | U32 state = trie->startstate; | |
166ba7cd | 2670 | |
3dab1dad YO |
2671 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2672 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2673 | ) { | |
2674 | if (trie->states[ state ].wordnum) { | |
2675 | DEBUG_EXECUTE_r( | |
2676 | PerlIO_printf(Perl_debug_log, | |
2677 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2678 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2679 | ); |
2680 | break; | |
2681 | } else { | |
2682 | DEBUG_EXECUTE_r( | |
2683 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2684 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2685 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2686 | ); |
2687 | sayNO_SILENT; | |
2688 | } | |
2689 | } | |
166ba7cd | 2690 | |
786e8c11 YO |
2691 | { |
2692 | U8 *uc = ( U8* )locinput; | |
2693 | ||
2694 | STRLEN len = 0; | |
2695 | STRLEN foldlen = 0; | |
2696 | U8 *uscan = (U8*)NULL; | |
2697 | STRLEN bufflen=0; | |
2698 | SV *sv_accept_buff = NULL; | |
2699 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2700 | ||
2701 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2702 | ST.B = next; | |
2703 | ST.jump = trie->jump; | |
2704 | ||
2705 | #ifdef DEBUGGING | |
2706 | ST.me = scan; | |
2707 | #endif | |
2708 | ||
2709 | ||
2710 | ||
07be1b83 YO |
2711 | /* |
2712 | traverse the TRIE keeping track of all accepting states | |
2713 | we transition through until we get to a failing node. | |
2714 | */ | |
2715 | ||
a3621e74 | 2716 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 YO |
2717 | U32 base = trie->states[ state ].trans.base; |
2718 | UV uvc; | |
2719 | U16 charid; | |
2720 | /* We use charid to hold the wordnum as we don't use it | |
2721 | for charid until after we have done the wordnum logic. | |
2722 | We define an alias just so that the wordnum logic reads | |
2723 | more naturally. */ | |
2724 | ||
2725 | #define got_wordnum charid | |
2726 | got_wordnum = trie->states[ state ].wordnum; | |
2727 | ||
2728 | if ( got_wordnum ) { | |
2729 | if ( ! ST.accepted ) { | |
5b47454d DM |
2730 | ENTER; |
2731 | SAVETMPS; | |
2732 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2733 | sv_accept_buff=newSV(bufflen * | |
2734 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2735 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2736 | SvPOK_on(sv_accept_buff); |
2737 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2738 | SAVETMPS; |
2739 | ST.accept_buff = | |
5b47454d DM |
2740 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2741 | } | |
786e8c11 | 2742 | do { |
166ba7cd | 2743 | if (ST.accepted >= bufflen) { |
5b47454d | 2744 | bufflen *= 2; |
166ba7cd | 2745 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2746 | SvGROW(sv_accept_buff, |
2747 | bufflen * sizeof(reg_trie_accepted)); | |
2748 | } | |
2749 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2750 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2751 | |
786e8c11 YO |
2752 | |
2753 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2754 | ST.accept_buff[ST.accepted].endpos = uc; | |
2755 | ++ST.accepted; | |
2756 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2757 | } | |
2758 | #undef got_wordnum | |
a3621e74 | 2759 | |
07be1b83 | 2760 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2761 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2762 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2763 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2764 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2765 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2766 | }); |
a3621e74 YO |
2767 | |
2768 | if ( base ) { | |
4cadc6a9 YO |
2769 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
2770 | uvc, charid, foldlen, foldbuf, uniflags); | |
a3621e74 | 2771 | |
5b47454d DM |
2772 | if (charid && |
2773 | (base + charid > trie->uniquecharcount ) | |
2774 | && (base + charid - 1 - trie->uniquecharcount | |
2775 | < trie->lasttrans) | |
2776 | && trie->trans[base + charid - 1 - | |
2777 | trie->uniquecharcount].check == state) | |
2778 | { | |
2779 | state = trie->trans[base + charid - 1 - | |
2780 | trie->uniquecharcount ].next; | |
2781 | } | |
2782 | else { | |
2783 | state = 0; | |
2784 | } | |
2785 | uc += len; | |
2786 | ||
2787 | } | |
2788 | else { | |
a3621e74 YO |
2789 | state = 0; |
2790 | } | |
2791 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 2792 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2793 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 2794 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
2795 | ); |
2796 | } | |
166ba7cd | 2797 | if (!ST.accepted ) |
a3621e74 | 2798 | sayNO; |
a3621e74 | 2799 | |
166ba7cd DM |
2800 | DEBUG_EXECUTE_r( |
2801 | PerlIO_printf( Perl_debug_log, | |
2802 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 2803 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
2804 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
2805 | ); | |
786e8c11 | 2806 | }} |
166ba7cd DM |
2807 | |
2808 | /* FALL THROUGH */ | |
2809 | ||
2810 | case TRIE_next_fail: /* we failed - try next alterative */ | |
2811 | ||
2812 | if ( ST.accepted == 1 ) { | |
2813 | /* only one choice left - just continue */ | |
2814 | DEBUG_EXECUTE_r({ | |
2815 | reg_trie_data * const trie | |
2816 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2817 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2818 | ? av_fetch( trie->words, ST.accept_buff[ 0 ].wordnum-1, 0 ) | |
2819 | : NULL; | |
2820 | PerlIO_printf( Perl_debug_log, | |
2821 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 2822 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd DM |
2823 | ST.accept_buff[ 0 ].wordnum, |
2824 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", | |
2825 | PL_colors[5] ); | |
2826 | }); | |
2827 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
2828 | /* in this case we free tmps/leave before we call regmatch | |
2829 | as we wont be using accept_buff again. */ | |
2830 | FREETMPS; | |
2831 | LEAVE; | |
2832 | locinput = PL_reginput; | |
2833 | nextchr = UCHARAT(locinput); | |
786e8c11 YO |
2834 | |
2835 | if ( !ST.jump ) | |
2836 | scan = ST.B; | |
2837 | else | |
2838 | scan = ST.B - ST.jump[ST.accept_buff[0].wordnum]; | |
2839 | ||
166ba7cd DM |
2840 | continue; /* execute rest of RE */ |
2841 | } | |
2842 | ||
2843 | if (!ST.accepted-- ) { | |
2844 | FREETMPS; | |
2845 | LEAVE; | |
2846 | sayNO; | |
2847 | } | |
2848 | ||
a3621e74 | 2849 | /* |
166ba7cd DM |
2850 | There are at least two accepting states left. Presumably |
2851 | the number of accepting states is going to be low, | |
2852 | typically two. So we simply scan through to find the one | |
2853 | with lowest wordnum. Once we find it, we swap the last | |
2854 | state into its place and decrement the size. We then try to | |
2855 | match the rest of the pattern at the point where the word | |
2856 | ends. If we succeed, control just continues along the | |
2857 | regex; if we fail we return here to try the next accepting | |
2858 | state | |
2859 | */ | |
a3621e74 | 2860 | |
166ba7cd DM |
2861 | { |
2862 | U32 best = 0; | |
2863 | U32 cur; | |
2864 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
2865 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 2866 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 2867 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 2868 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
2869 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
2870 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
2871 | ); | |
2872 | ||
2873 | if (ST.accept_buff[cur].wordnum < | |
2874 | ST.accept_buff[best].wordnum) | |
2875 | best = cur; | |
a3621e74 | 2876 | } |
166ba7cd DM |
2877 | |
2878 | DEBUG_EXECUTE_r({ | |
2879 | reg_trie_data * const trie | |
2880 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2881 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2882 | ? av_fetch( trie->words, ST.accept_buff[ best ].wordnum - 1, 0 ) | |
2883 | : NULL; | |
265c4333 YO |
2884 | regnode *nextop=!ST.jump ? |
2885 | ST.B : | |
2886 | ST.B - ST.jump[ST.accept_buff[best].wordnum]; | |
2887 | PerlIO_printf( Perl_debug_log, | |
2888 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 2889 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 2890 | ST.accept_buff[best].wordnum, |
265c4333 YO |
2891 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", |
2892 | REG_NODE_NUM(nextop), | |
166ba7cd DM |
2893 | PL_colors[5] ); |
2894 | }); | |
2895 | ||
2896 | if ( best<ST.accepted ) { | |
2897 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
2898 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
2899 | ST.accept_buff[ ST.accepted ] = tmp; | |
2900 | best = ST.accepted; | |
a3621e74 | 2901 | } |
166ba7cd | 2902 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
786e8c11 YO |
2903 | if ( !ST.jump ) { |
2904 | PUSH_STATE_GOTO(TRIE_next, ST.B); | |
2905 | /* NOTREACHED */ | |
2906 | } else { | |
2907 | PUSH_STATE_GOTO(TRIE_next, ST.B - ST.jump[ST.accept_buff[best].wordnum]); | |
2908 | /* NOTREACHED */ | |
2909 | } | |
2910 | /* NOTREACHED */ | |
166ba7cd | 2911 | } |
166ba7cd DM |
2912 | /* NOTREACHED */ |
2913 | ||
2914 | #undef ST | |
2915 | ||
95b24440 DM |
2916 | case EXACT: { |
2917 | char *s = STRING(scan); | |
5d9a96ca | 2918 | st->ln = STR_LEN(scan); |
eb160463 | 2919 | if (do_utf8 != UTF) { |
bc517b45 | 2920 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 2921 | char *l = locinput; |
be8e71aa | 2922 | const char * const e = s + st->ln; |
a72c7584 | 2923 | |
5ff6fc6d JH |
2924 | if (do_utf8) { |
2925 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 2926 | while (s < e) { |
a3b680e6 | 2927 | STRLEN ulen; |
1aa99e6b | 2928 | if (l >= PL_regeol) |
5ff6fc6d JH |
2929 | sayNO; |
2930 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 2931 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 2932 | uniflags)) |
5ff6fc6d | 2933 | sayNO; |
bc517b45 | 2934 | l += ulen; |
5ff6fc6d | 2935 | s ++; |
1aa99e6b | 2936 | } |
5ff6fc6d JH |
2937 | } |
2938 | else { | |
2939 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 2940 | while (s < e) { |
a3b680e6 | 2941 | STRLEN ulen; |
1aa99e6b IH |
2942 | if (l >= PL_regeol) |
2943 | sayNO; | |
5ff6fc6d | 2944 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 2945 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 2946 | uniflags)) |
1aa99e6b | 2947 | sayNO; |
bc517b45 | 2948 | s += ulen; |
a72c7584 | 2949 | l ++; |
1aa99e6b | 2950 | } |
5ff6fc6d | 2951 | } |
1aa99e6b IH |
2952 | locinput = l; |
2953 | nextchr = UCHARAT(locinput); | |
2954 | break; | |
2955 | } | |
bc517b45 | 2956 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
2957 | /* Inline the first character, for speed. */ |
2958 | if (UCHARAT(s) != nextchr) | |
2959 | sayNO; | |
5d9a96ca | 2960 | if (PL_regeol - locinput < st->ln) |
d6a28714 | 2961 | sayNO; |
5d9a96ca | 2962 | if (st->ln > 1 && memNE(s, locinput, st->ln)) |
d6a28714 | 2963 | sayNO; |
5d9a96ca | 2964 | locinput += st->ln; |
d6a28714 JH |
2965 | nextchr = UCHARAT(locinput); |
2966 | break; | |
95b24440 | 2967 | } |
d6a28714 | 2968 | case EXACTFL: |
b8c5462f JH |
2969 | PL_reg_flags |= RF_tainted; |
2970 | /* FALL THROUGH */ | |
95b24440 | 2971 | case EXACTF: { |
be8e71aa | 2972 | char * const s = STRING(scan); |
5d9a96ca | 2973 | st->ln = STR_LEN(scan); |
d6a28714 | 2974 | |
d07ddd77 JH |
2975 | if (do_utf8 || UTF) { |
2976 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 2977 | const char * const l = locinput; |
d07ddd77 | 2978 | char *e = PL_regeol; |
bc517b45 | 2979 | |
5d9a96ca | 2980 | if (ibcmp_utf8(s, 0, st->ln, (bool)UTF, |
1feea2c7 | 2981 | l, &e, 0, do_utf8)) { |
5486206c JH |
2982 | /* One more case for the sharp s: |
2983 | * pack("U0U*", 0xDF) =~ /ss/i, | |
2984 | * the 0xC3 0x9F are the UTF-8 | |
2985 | * byte sequence for the U+00DF. */ | |
2986 | if (!(do_utf8 && | |
2987 | toLOWER(s[0]) == 's' && | |
5d9a96ca | 2988 | st->ln >= 2 && |
5486206c JH |
2989 | toLOWER(s[1]) == 's' && |
2990 | (U8)l[0] == 0xC3 && | |
2991 | e - l >= 2 && | |
2992 | (U8)l[1] == 0x9F)) | |
2993 | sayNO; | |
2994 | } | |
d07ddd77 JH |
2995 | locinput = e; |
2996 | nextchr = UCHARAT(locinput); | |
2997 | break; | |
a0ed51b3 | 2998 | } |
d6a28714 | 2999 | |
bc517b45 JH |
3000 | /* Neither the target and the pattern are utf8. */ |
3001 | ||
d6a28714 JH |
3002 | /* Inline the first character, for speed. */ |
3003 | if (UCHARAT(s) != nextchr && | |
3004 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3005 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3006 | sayNO; |
5d9a96ca | 3007 | if (PL_regeol - locinput < st->ln) |
b8c5462f | 3008 | sayNO; |
5d9a96ca DM |
3009 | if (st->ln > 1 && (OP(scan) == EXACTF |
3010 | ? ibcmp(s, locinput, st->ln) | |
3011 | : ibcmp_locale(s, locinput, st->ln))) | |
4633a7c4 | 3012 | sayNO; |
5d9a96ca | 3013 | locinput += st->ln; |
d6a28714 | 3014 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3015 | break; |
95b24440 | 3016 | } |
d6a28714 | 3017 | case ANYOF: |
ffc61ed2 | 3018 | if (do_utf8) { |
9e55ce06 JH |
3019 | STRLEN inclasslen = PL_regeol - locinput; |
3020 | ||
32fc9b6a | 3021 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
e0f9d4a8 | 3022 | sayNO_ANYOF; |
ffc61ed2 JH |
3023 | if (locinput >= PL_regeol) |
3024 | sayNO; | |
0f0076b4 | 3025 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3026 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3027 | break; |
ffc61ed2 JH |
3028 | } |
3029 | else { | |
3030 | if (nextchr < 0) | |
3031 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3032 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
e0f9d4a8 | 3033 | sayNO_ANYOF; |
ffc61ed2 JH |
3034 | if (!nextchr && locinput >= PL_regeol) |
3035 | sayNO; | |
3036 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3037 | break; |
3038 | } | |
3039 | no_anyof: | |
3040 | /* If we might have the case of the German sharp s | |
3041 | * in a casefolding Unicode character class. */ | |
3042 | ||
ebc501f0 JH |
3043 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3044 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3045 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3046 | } |
e0f9d4a8 JH |
3047 | else |
3048 | sayNO; | |
b8c5462f | 3049 | break; |
d6a28714 | 3050 | case ALNUML: |
b8c5462f JH |
3051 | PL_reg_flags |= RF_tainted; |
3052 | /* FALL THROUGH */ | |
d6a28714 | 3053 | case ALNUM: |
b8c5462f | 3054 | if (!nextchr) |
4633a7c4 | 3055 | sayNO; |
ffc61ed2 | 3056 | if (do_utf8) { |
1a4fad37 | 3057 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3058 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3059 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3060 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3061 | { |
3062 | sayNO; | |
a0ed51b3 | 3063 | } |
b8c5462f | 3064 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3065 | nextchr = UCHARAT(locinput); |
3066 | break; | |
3067 | } | |
ffc61ed2 | 3068 | if (!(OP(scan) == ALNUM |
d6a28714 | 3069 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3070 | sayNO; |
b8c5462f | 3071 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3072 | break; |
d6a28714 | 3073 | case NALNUML: |
b8c5462f JH |
3074 | PL_reg_flags |= RF_tainted; |
3075 | /* FALL THROUGH */ | |
d6a28714 JH |
3076 | case NALNUM: |
3077 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3078 | sayNO; |
ffc61ed2 | 3079 | if (do_utf8) { |
1a4fad37 | 3080 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3081 | if (OP(scan) == NALNUM |
bb7a0f54 | 3082 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3083 | : isALNUM_LC_utf8((U8*)locinput)) |
3084 | { | |
b8c5462f | 3085 | sayNO; |
d6a28714 | 3086 | } |
b8c5462f JH |
3087 | locinput += PL_utf8skip[nextchr]; |
3088 | nextchr = UCHARAT(locinput); | |
3089 | break; | |
3090 | } | |
ffc61ed2 | 3091 | if (OP(scan) == NALNUM |
d6a28714 | 3092 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3093 | sayNO; |
76e3520e | 3094 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3095 | break; |
d6a28714 JH |
3096 | case BOUNDL: |
3097 | case NBOUNDL: | |
3280af22 | 3098 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3099 | /* FALL THROUGH */ |
d6a28714 JH |
3100 | case BOUND: |
3101 | case NBOUND: | |
3102 | /* was last char in word? */ | |
ffc61ed2 | 3103 | if (do_utf8) { |
12d33761 | 3104 | if (locinput == PL_bostr) |
5d9a96ca | 3105 | st->ln = '\n'; |
ffc61ed2 | 3106 | else { |
a3b680e6 | 3107 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3108 | |
4ad0818d | 3109 | st->ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3110 | } |
3111 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
5d9a96ca | 3112 | st->ln = isALNUM_uni(st->ln); |
1a4fad37 | 3113 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3114 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3115 | } |
3116 | else { | |
5d9a96ca | 3117 | st->ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(st->ln)); |
ffc61ed2 JH |
3118 | n = isALNUM_LC_utf8((U8*)locinput); |
3119 | } | |
a0ed51b3 | 3120 | } |
d6a28714 | 3121 | else { |
5d9a96ca | 3122 | st->ln = (locinput != PL_bostr) ? |
12d33761 | 3123 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3124 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
5d9a96ca | 3125 | st->ln = isALNUM(st->ln); |
ffc61ed2 JH |
3126 | n = isALNUM(nextchr); |
3127 | } | |
3128 | else { | |
5d9a96ca | 3129 | st->ln = isALNUM_LC(st->ln); |
ffc61ed2 JH |
3130 | n = isALNUM_LC(nextchr); |
3131 | } | |
d6a28714 | 3132 | } |
5d9a96ca | 3133 | if (((!st->ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3134 | OP(scan) == BOUNDL)) |
3135 | sayNO; | |
a0ed51b3 | 3136 | break; |
d6a28714 | 3137 | case SPACEL: |
3280af22 | 3138 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3139 | /* FALL THROUGH */ |
d6a28714 | 3140 | case SPACE: |
9442cb0e | 3141 | if (!nextchr) |
4633a7c4 | 3142 | sayNO; |
1aa99e6b | 3143 | if (do_utf8) { |
fd400ab9 | 3144 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3145 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3146 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3147 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3148 | : isSPACE_LC_utf8((U8*)locinput))) |
3149 | { | |
3150 | sayNO; | |
3151 | } | |
3152 | locinput += PL_utf8skip[nextchr]; | |
3153 | nextchr = UCHARAT(locinput); | |
3154 | break; | |
d6a28714 | 3155 | } |
ffc61ed2 JH |
3156 | if (!(OP(scan) == SPACE |
3157 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3158 | sayNO; | |
3159 | nextchr = UCHARAT(++locinput); | |
3160 | } | |
3161 | else { | |
3162 | if (!(OP(scan) == SPACE | |
3163 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3164 | sayNO; | |
3165 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3166 | } |
a0ed51b3 | 3167 | break; |
d6a28714 | 3168 | case NSPACEL: |
3280af22 | 3169 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3170 | /* FALL THROUGH */ |
d6a28714 | 3171 | case NSPACE: |
9442cb0e | 3172 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3173 | sayNO; |
1aa99e6b | 3174 | if (do_utf8) { |
1a4fad37 | 3175 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3176 | if (OP(scan) == NSPACE |
bb7a0f54 | 3177 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3178 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3179 | { |
3180 | sayNO; | |
3181 | } | |
3182 | locinput += PL_utf8skip[nextchr]; | |
3183 | nextchr = UCHARAT(locinput); | |
3184 | break; | |
a0ed51b3 | 3185 | } |
ffc61ed2 | 3186 | if (OP(scan) == NSPACE |
d6a28714 | 3187 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3188 | sayNO; |
76e3520e | 3189 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3190 | break; |
d6a28714 | 3191 | case DIGITL: |
a0ed51b3 LW |
3192 | PL_reg_flags |= RF_tainted; |
3193 | /* FALL THROUGH */ | |
d6a28714 | 3194 | case DIGIT: |
9442cb0e | 3195 | if (!nextchr) |
a0ed51b3 | 3196 | sayNO; |
1aa99e6b | 3197 | if (do_utf8) { |
1a4fad37 | 3198 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3199 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3200 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3201 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3202 | { |
a0ed51b3 | 3203 | sayNO; |
dfe13c55 | 3204 | } |
6f06b55f | 3205 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3206 | nextchr = UCHARAT(locinput); |
3207 | break; | |
3208 | } | |
ffc61ed2 | 3209 | if (!(OP(scan) == DIGIT |
9442cb0e | 3210 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3211 | sayNO; |
3212 | nextchr = UCHARAT(++locinput); | |
3213 | break; | |
d6a28714 | 3214 | case NDIGITL: |
b8c5462f JH |
3215 | PL_reg_flags |= RF_tainted; |
3216 | /* FALL THROUGH */ | |
d6a28714 | 3217 | case NDIGIT: |
9442cb0e | 3218 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3219 | sayNO; |
1aa99e6b | 3220 | if (do_utf8) { |
1a4fad37 | 3221 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3222 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3223 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3224 | : isDIGIT_LC_utf8((U8*)locinput)) |
3225 | { | |
a0ed51b3 | 3226 | sayNO; |
9442cb0e | 3227 | } |
6f06b55f | 3228 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3229 | nextchr = UCHARAT(locinput); |
3230 | break; | |
3231 | } | |
ffc61ed2 | 3232 | if (OP(scan) == NDIGIT |
9442cb0e | 3233 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3234 | sayNO; |
3235 | nextchr = UCHARAT(++locinput); | |
3236 | break; | |
3237 | case CLUMP: | |
b7c83a7e | 3238 | if (locinput >= PL_regeol) |
a0ed51b3 | 3239 | sayNO; |
b7c83a7e | 3240 | if (do_utf8) { |
1a4fad37 | 3241 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3242 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3243 | sayNO; | |
3244 | locinput += PL_utf8skip[nextchr]; | |
3245 | while (locinput < PL_regeol && | |
3246 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3247 | locinput += UTF8SKIP(locinput); | |
3248 | if (locinput > PL_regeol) | |
3249 | sayNO; | |
eb08e2da JH |
3250 | } |
3251 | else | |
3252 | locinput++; | |
a0ed51b3 LW |
3253 | nextchr = UCHARAT(locinput); |
3254 | break; | |
c8756f30 | 3255 | case REFFL: |
3280af22 | 3256 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3257 | /* FALL THROUGH */ |
c277df42 | 3258 | case REF: |
95b24440 DM |
3259 | case REFF: { |
3260 | char *s; | |
c277df42 | 3261 | n = ARG(scan); /* which paren pair */ |
5d9a96ca | 3262 | st->ln = PL_regstartp[n]; |
2c2d71f5 | 3263 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
5d9a96ca | 3264 | if ((I32)*PL_reglastparen < n || st->ln == -1) |
af3f8c16 | 3265 | sayNO; /* Do not match unless seen CLOSEn. */ |
5d9a96ca | 3266 | if (st->ln == PL_regendp[n]) |
a0d0e21e | 3267 | break; |
a0ed51b3 | 3268 | |
5d9a96ca | 3269 | s = PL_bostr + st->ln; |
1aa99e6b | 3270 | if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3271 | char *l = locinput; |
a3b680e6 | 3272 | const char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 LW |
3273 | /* |
3274 | * Note that we can't do the "other character" lookup trick as | |
3275 | * in the 8-bit case (no pun intended) because in Unicode we | |
3276 | * have to map both upper and title case to lower case. | |
3277 | */ | |
3278 | if (OP(scan) == REFF) { | |
3279 | while (s < e) { | |
a3b680e6 AL |
3280 | STRLEN ulen1, ulen2; |
3281 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3282 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3283 | ||
a0ed51b3 LW |
3284 | if (l >= PL_regeol) |
3285 | sayNO; | |
a2a2844f JH |
3286 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3287 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3288 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3289 | sayNO; |
a2a2844f JH |
3290 | s += ulen1; |
3291 | l += ulen2; | |
a0ed51b3 LW |
3292 | } |
3293 | } | |
3294 | locinput = l; | |
3295 | nextchr = UCHARAT(locinput); | |
3296 | break; | |
3297 | } | |
3298 | ||
a0d0e21e | 3299 | /* Inline the first character, for speed. */ |
76e3520e | 3300 | if (UCHARAT(s) != nextchr && |
c8756f30 AK |
3301 | (OP(scan) == REF || |
3302 | (UCHARAT(s) != ((OP(scan) == REFF | |
22c35a8c | 3303 | ? PL_fold : PL_fold_locale)[nextchr])))) |
4633a7c4 | 3304 | sayNO; |
5d9a96ca DM |
3305 | st->ln = PL_regendp[n] - st->ln; |
3306 | if (locinput + st->ln > PL_regeol) | |
4633a7c4 | 3307 | sayNO; |
5d9a96ca DM |
3308 | if (st->ln > 1 && (OP(scan) == REF |
3309 | ? memNE(s, locinput, st->ln) | |
c8756f30 | 3310 | : (OP(scan) == REFF |
5d9a96ca DM |
3311 | ? ibcmp(s, locinput, st->ln) |
3312 | : ibcmp_locale(s, locinput, st->ln)))) | |
4633a7c4 | 3313 | sayNO; |
5d9a96ca | 3314 | locinput += st->ln; |
76e3520e | 3315 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3316 | break; |
95b24440 | 3317 | } |
a0d0e21e LW |
3318 | |
3319 | case NOTHING: | |
c277df42 | 3320 | case TAIL: |
a0d0e21e LW |
3321 | break; |
3322 | case BACK: | |
3323 | break; | |
40a82448 DM |
3324 | |
3325 | #undef ST | |
3326 | #define ST st->u.eval | |
3327 | ||
3328 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
c277df42 | 3329 | { |
c277df42 | 3330 | SV *ret; |
8e5e9ebe | 3331 | { |
4aabdb9b DM |
3332 | /* execute the code in the {...} */ |
3333 | dSP; | |
6136c704 | 3334 | SV ** const before = SP; |
4aabdb9b DM |
3335 | OP_4tree * const oop = PL_op; |
3336 | COP * const ocurcop = PL_curcop; | |
3337 | PAD *old_comppad; | |
4aabdb9b DM |
3338 | |
3339 | n = ARG(scan); | |
32fc9b6a | 3340 | PL_op = (OP_4tree*)rex->data->data[n]; |
4aabdb9b | 3341 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); |
32fc9b6a | 3342 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]); |
4aabdb9b DM |
3343 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
3344 | ||
8e5e9ebe RGS |
3345 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3346 | SPAGAIN; | |
3347 | if (SP == before) | |
075aa684 | 3348 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3349 | else { |
3350 | ret = POPs; | |
3351 | PUTBACK; | |
3352 | } | |
4aabdb9b DM |
3353 | |
3354 | PL_op = oop; | |
3355 | PAD_RESTORE_LOCAL(old_comppad); | |
3356 | PL_curcop = ocurcop; | |
3357 | if (!st->logical) { | |
3358 | /* /(?{...})/ */ | |
3359 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3360 | break; |
3361 | } | |
8e5e9ebe | 3362 | } |
4aabdb9b DM |
3363 | if (st->logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3364 | regexp *re; | |
4aabdb9b | 3365 | { |
4f639d21 DM |
3366 | /* extract RE object from returned value; compiling if |
3367 | * necessary */ | |
3368 | ||
6136c704 | 3369 | MAGIC *mg = NULL; |
be8e71aa | 3370 | const SV *sv; |
faf82a0b AE |
3371 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3372 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3373 | else if (SvSMAGICAL(ret)) { | |
3374 | if (SvGMAGICAL(ret)) | |
3375 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3376 | else | |
3377 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3378 | } |
faf82a0b | 3379 | |
0f5d15d6 IZ |
3380 | if (mg) { |
3381 | re = (regexp *)mg->mg_obj; | |
df0003d4 | 3382 | (void)ReREFCNT_inc(re); |
0f5d15d6 IZ |
3383 | } |
3384 | else { | |
3385 | STRLEN len; | |
6136c704 | 3386 | const char * const t = SvPV_const(ret, len); |
0f5d15d6 | 3387 | PMOP pm; |
a3b680e6 | 3388 | const I32 osize = PL_regsize; |
0f5d15d6 | 3389 | |
5fcd1c1b | 3390 | Zero(&pm, 1, PMOP); |
4aabdb9b | 3391 | if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; |
83003860 | 3392 | re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm); |
9041c2e3 | 3393 | if (!(SvFLAGS(ret) |
faf82a0b AE |
3394 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3395 | | SVs_GMG))) | |
14befaf4 DM |
3396 | sv_magic(ret,(SV*)ReREFCNT_inc(re), |
3397 | PERL_MAGIC_qr,0,0); | |
0f5d15d6 | 3398 | PL_regsize = osize; |
0f5d15d6 | 3399 | } |
4aabdb9b | 3400 | } |
aa283a38 DM |
3401 | |
3402 | /* run the pattern returned from (??{...}) */ | |
40a82448 DM |
3403 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3404 | REGCP_SET(ST.lastcp); | |
4aabdb9b DM |
3405 | *PL_reglastparen = 0; |
3406 | *PL_reglastcloseparen = 0; | |
4aabdb9b | 3407 | PL_reginput = locinput; |
4aabdb9b DM |
3408 | |
3409 | /* XXXX This is too dramatic a measure... */ | |
3410 | PL_reg_maxiter = 0; | |
3411 | ||
5d9a96ca | 3412 | st->logical = 0; |
faec1544 DM |
3413 | ST.toggle_reg_flags = PL_reg_flags; |
3414 | if (re->reganch & ROPT_UTF8) | |
3415 | PL_reg_flags |= RF_utf8; | |
3416 | else | |
3417 | PL_reg_flags &= ~RF_utf8; | |
3418 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3419 | ||
40a82448 | 3420 | ST.prev_rex = rex; |
faec1544 | 3421 | ST.prev_curlyx = cur_curlyx; |
aa283a38 | 3422 | rex = re; |
faec1544 | 3423 | cur_curlyx = NULL; |
40a82448 | 3424 | ST.B = next; |
faec1544 DM |
3425 | ST.prev_eval = cur_eval; |
3426 | cur_eval = st; | |
3427 | ||
1de06328 YO |
3428 | DEBUG_EXECUTE_r( |
3429 | debug_start_match(re, do_utf8, locinput, PL_regeol, | |
3430 | "Matching embedded"); | |
3431 | ); | |
faec1544 DM |
3432 | /* now continue from first node in postoned RE */ |
3433 | PUSH_YES_STATE_GOTO(EVAL_AB, re->program + 1); | |
4aabdb9b | 3434 | /* NOTREACHED */ |
a0ed51b3 | 3435 | } |
4aabdb9b DM |
3436 | /* /(?(?{...})X|Y)/ */ |
3437 | st->sw = SvTRUE(ret); | |
3438 | st->logical = 0; | |
c277df42 IZ |
3439 | break; |
3440 | } | |
40a82448 | 3441 | |
faec1544 DM |
3442 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3443 | /* note: this is called twice; first after popping B, then A */ | |
3444 | PL_reg_flags ^= ST.toggle_reg_flags; | |
40a82448 DM |
3445 | ReREFCNT_dec(rex); |
3446 | rex = ST.prev_rex; | |
faec1544 DM |
3447 | regcpblow(ST.cp); |
3448 | cur_eval = ST.prev_eval; | |
3449 | cur_curlyx = ST.prev_curlyx; | |
40a82448 DM |
3450 | /* XXXX This is too dramatic a measure... */ |
3451 | PL_reg_maxiter = 0; | |
faec1544 | 3452 | sayYES_FINAL; |
40a82448 | 3453 | |
40a82448 | 3454 | |
faec1544 DM |
3455 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3456 | /* note: this is called twice; first after popping B, then A */ | |
3457 | PL_reg_flags ^= ST.toggle_reg_flags; | |
3458 | ReREFCNT_dec(rex); | |
3459 | rex = ST.prev_rex; | |
40a82448 DM |
3460 | PL_reginput = locinput; |
3461 | REGCP_UNWIND(ST.lastcp); | |
3462 | regcppop(rex); | |
faec1544 DM |
3463 | cur_eval = ST.prev_eval; |
3464 | cur_curlyx = ST.prev_curlyx; | |
3465 | /* XXXX This is too dramatic a measure... */ | |
3466 | PL_reg_maxiter = 0; | |
40a82448 DM |
3467 | sayNO_SILENT; |
3468 | ||
3469 | #undef ST | |
3470 | ||
a0d0e21e | 3471 | case OPEN: |
c277df42 | 3472 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3473 | PL_reg_start_tmp[n] = locinput; |
3474 | if (n > PL_regsize) | |
3475 | PL_regsize = n; | |
a0d0e21e LW |
3476 | break; |
3477 | case CLOSE: | |
c277df42 | 3478 | n = ARG(scan); /* which paren pair */ |
cf93c79d IZ |
3479 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
3480 | PL_regendp[n] = locinput - PL_bostr; | |
eb160463 | 3481 | if (n > (I32)*PL_reglastparen) |
3280af22 | 3482 | *PL_reglastparen = n; |
a01268b5 | 3483 | *PL_reglastcloseparen = n; |
a0d0e21e | 3484 | break; |
c277df42 IZ |
3485 | case GROUPP: |
3486 | n = ARG(scan); /* which paren pair */ | |
5d9a96ca | 3487 | st->sw = ((I32)*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 IZ |
3488 | break; |
3489 | case IFTHEN: | |
2c2d71f5 | 3490 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
5d9a96ca | 3491 | if (st->sw) |
c277df42 IZ |
3492 | next = NEXTOPER(NEXTOPER(scan)); |
3493 | else { | |
3494 | next = scan + ARG(scan); | |
3495 | if (OP(next) == IFTHEN) /* Fake one. */ | |
3496 | next = NEXTOPER(NEXTOPER(next)); | |
3497 | } | |
3498 | break; | |
3499 | case LOGICAL: | |
5d9a96ca | 3500 | st->logical = scan->flags; |
c277df42 | 3501 | break; |
2ab05381 | 3502 | /******************************************************************* |
b8591aee | 3503 | cur_curlyx points to the regmatch_state associated with the most recent CURLYX. |
a0374537 | 3504 | This struct contains info about the innermost (...)* loop (an |
b8591aee | 3505 | "infoblock"), and a pointer to the next outer cur_curlyx. |
2ab05381 IZ |
3506 | |
3507 | Here is how Y(A)*Z is processed (if it is compiled into CURLYX/WHILEM): | |
3508 | ||
95b24440 | 3509 | 1) After matching Y, regnode for CURLYX is processed; |
2ab05381 | 3510 | |
b8591aee | 3511 | 2) This regnode populates cur_curlyx, and calls regmatch() recursively |
95b24440 | 3512 | with the starting point at WHILEM node; |
2ab05381 IZ |
3513 | |
3514 | 3) Each hit of WHILEM node tries to match A and Z (in the order | |
3515 | depending on the current iteration, min/max of {min,max} and | |
3516 | greediness). The information about where are nodes for "A" | |
b8591aee | 3517 | and "Z" is read from cur_curlyx, as is info on how many times "A" |
2ab05381 IZ |
3518 | was already matched, and greediness. |
3519 | ||
3520 | 4) After A matches, the same WHILEM node is hit again. | |
3521 | ||
b8591aee | 3522 | 5) Each time WHILEM is hit, cur_curlyx is the infoblock created by CURLYX |
2ab05381 | 3523 | of the same pair. Thus when WHILEM tries to match Z, it temporarily |
b8591aee | 3524 | resets cur_curlyx, since this Y(A)*Z can be a part of some other loop: |
2ab05381 IZ |
3525 | as in (Y(A)*Z)*. If Z matches, the automaton will hit the WHILEM node |
3526 | of the external loop. | |
3527 | ||
b8591aee | 3528 | Currently present infoblocks form a tree with a stem formed by cur_curlyx |
2ab05381 IZ |
3529 | and whatever it mentions via ->next, and additional attached trees |
3530 | corresponding to temporarily unset infoblocks as in "5" above. | |
3531 | ||
95b24440 | 3532 | In the following picture, infoblocks for outer loop of |
2ab05381 IZ |
3533 | (Y(A)*?Z)*?T are denoted O, for inner I. NULL starting block |
3534 | is denoted by x. The matched string is YAAZYAZT. Temporarily postponed | |
3535 | infoblocks are drawn below the "reset" infoblock. | |
3536 | ||
3537 | In fact in the picture below we do not show failed matches for Z and T | |
3538 | by WHILEM blocks. [We illustrate minimal matches, since for them it is | |
3539 | more obvious *why* one needs to *temporary* unset infoblocks.] | |
3540 | ||
3541 | Matched REx position InfoBlocks Comment | |
3542 | (Y(A)*?Z)*?T x | |
3543 | Y(A)*?Z)*?T x <- O | |
3544 | Y (A)*?Z)*?T x <- O | |
3545 | Y A)*?Z)*?T x <- O <- I | |
3546 | YA )*?Z)*?T x <- O <- I | |
3547 | YA A)*?Z)*?T x <- O <- I | |
3548 | YAA )*?Z)*?T x <- O <- I | |
3549 | YAA Z)*?T x <- O # Temporary unset I | |
3550 | I | |
3551 | ||
3552 | YAAZ Y(A)*?Z)*?T x <- O | |
3553 | I | |
3554 | ||
3555 | YAAZY (A)*?Z)*?T x <- O | |
3556 | I | |
3557 | ||
3558 | YAAZY A)*?Z)*?T x <- O <- I | |
3559 | I | |
3560 | ||
3561 | YAAZYA )*?Z)*?T x <- O <- I | |
3562 | I | |
3563 | ||
3564 | YAAZYA Z)*?T x <- O # Temporary unset I | |
3565 | I,I | |
3566 | ||
3567 | YAAZYAZ )*?T x <- O | |
3568 | I,I | |
3569 | ||
3570 | YAAZYAZ T x # Temporary unset O | |
3571 | O | |
3572 | I,I | |
3573 | ||
3574 | YAAZYAZT x | |
3575 | O | |
3576 | I,I | |
3577 | *******************************************************************/ | |
95b24440 | 3578 | |
a0d0e21e | 3579 | case CURLYX: { |
cb434fcc | 3580 | /* No need to save/restore up to this paren */ |
10edeb5d JH |
3581 | parenfloor = scan->flags; |
3582 | ||
c2b7afd3 NC |
3583 | /* Dave says: |
3584 | ||
3585 | CURLYX and WHILEM are always paired: they're the moral | |
3586 | equivalent of pp_enteriter anbd pp_iter. | |
3587 | ||
3588 | The only time next could be null is if the node tree is | |
3589 | corrupt. This was mentioned on p5p a few days ago. | |
3590 | ||
3591 | See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-04/msg00556.html | |
3592 | So we'll assert that this is true: | |
3593 | */ | |
3594 | assert(next); | |
30b2893d | 3595 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ |
c277df42 | 3596 | next += ARG(next); |
cb434fcc IZ |
3597 | /* XXXX Probably it is better to teach regpush to support |
3598 | parenfloor > PL_regsize... */ | |
eb160463 | 3599 | if (parenfloor > (I32)*PL_reglastparen) |
cb434fcc | 3600 | parenfloor = *PL_reglastparen; /* Pessimization... */ |
a0374537 | 3601 | |
d8319b27 | 3602 | st->u.curlyx.cp = PL_savestack_ix; |
b8591aee DM |
3603 | st->u.curlyx.outercc = cur_curlyx; |
3604 | cur_curlyx = st; | |
a0374537 DM |
3605 | /* these fields contain the state of the current curly. |
3606 | * they are accessed by subsequent WHILEMs; | |
3607 | * cur and lastloc are also updated by WHILEM */ | |
d8319b27 DM |
3608 | st->u.curlyx.parenfloor = parenfloor; |
3609 | st->u.curlyx.cur = -1; /* this will be updated by WHILEM */ | |
3610 | st->u.curlyx.min = ARG1(scan); | |
3611 | st->u.curlyx.max = ARG2(scan); | |
3612 | st->u.curlyx.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
3613 | st->u.curlyx.lastloc = 0; | |
a0374537 DM |
3614 | /* st->next and st->minmod are also read by WHILEM */ |
3615 | ||
3280af22 | 3616 | PL_reginput = locinput; |
95b24440 DM |
3617 | REGMATCH(PREVOPER(next), CURLYX); /* start on the WHILEM */ |
3618 | /*** all unsaved local vars undefined at this point */ | |
d8319b27 | 3619 | regcpblow(st->u.curlyx.cp); |
b8591aee | 3620 | cur_curlyx = st->u.curlyx.outercc; |
95b24440 | 3621 | saySAME(result); |
a0d0e21e | 3622 | } |
5f66b61c | 3623 | /* NOTREACHED */ |
a0d0e21e LW |
3624 | case WHILEM: { |
3625 | /* | |
3626 | * This is really hard to understand, because after we match | |
3627 | * what we're trying to match, we must make sure the rest of | |
2c2d71f5 | 3628 | * the REx is going to match for sure, and to do that we have |
a0d0e21e LW |
3629 | * to go back UP the parse tree by recursing ever deeper. And |
3630 | * if it fails, we have to reset our parent's current state | |
3631 | * that we can try again after backing off. | |
3632 | */ | |
3633 | ||
c2b7afd3 NC |
3634 | /* Dave says: |
3635 | ||
b8591aee DM |
3636 | cur_curlyx gets initialised by CURLYX ready for use by WHILEM. |
3637 | So again, unless somethings been corrupted, cur_curlyx cannot | |
c2b7afd3 NC |
3638 | be null at that point in WHILEM. |
3639 | ||
3640 | See http://www.xray.mpe.mpg.de/mailing-lists/perl5-porters/2006-04/msg00556.html | |
3641 | So we'll assert that this is true: | |
3642 | */ | |
b8591aee DM |
3643 | assert(cur_curlyx); |
3644 | st->u.whilem.lastloc = cur_curlyx->u.curlyx.lastloc; /* Detection of 0-len. */ | |
d8319b27 DM |
3645 | st->u.whilem.cache_offset = 0; |
3646 | st->u.whilem.cache_bit = 0; | |
c277df42 | 3647 | |
b8591aee | 3648 | n = cur_curlyx->u.curlyx.cur + 1; /* how many we know we matched */ |
3280af22 | 3649 | PL_reginput = locinput; |
a0d0e21e | 3650 | |
a3621e74 | 3651 | DEBUG_EXECUTE_r( |
9041c2e3 | 3652 | PerlIO_printf(Perl_debug_log, |
91f3b821 | 3653 | "%*s %ld out of %ld..%ld cc=%"UVxf"\n", |
5bc10b2c | 3654 | REPORT_CODE_OFF+depth*2, "", |
b8591aee DM |
3655 | (long)n, (long)cur_curlyx->u.curlyx.min, |
3656 | (long)cur_curlyx->u.curlyx.max, | |
3657 | PTR2UV(cur_curlyx)) | |
c277df42 | 3658 | ); |
4633a7c4 | 3659 | |
a0d0e21e LW |
3660 | /* If degenerate scan matches "", assume scan done. */ |
3661 | ||
b8591aee DM |
3662 | if (locinput == cur_curlyx->u.curlyx.lastloc && n >= |
3663 | cur_curlyx->u.curlyx.min) | |
3664 | { | |
3665 | st->u.whilem.savecc = cur_curlyx; | |
3666 | cur_curlyx = cur_curlyx->u.curlyx.outercc; | |
3667 | if (cur_curlyx) | |
3668 | st->ln = cur_curlyx->u.curlyx.cur; | |
a3621e74 | 3669 | DEBUG_EXECUTE_r( |
c3464db5 DD |
3670 | PerlIO_printf(Perl_debug_log, |
3671 | "%*s empty match detected, try continuation...\n", | |
5bc10b2c | 3672 | REPORT_CODE_OFF+depth*2, "") |
c277df42 | 3673 | ); |
d8319b27 | 3674 | REGMATCH(st->u.whilem.savecc->next, WHILEM1); |
95b24440 | 3675 | /*** all unsaved local vars undefined at this point */ |
b8591aee | 3676 | cur_curlyx = st->u.whilem.savecc; |
95b24440 | 3677 | if (result) |
4633a7c4 | 3678 | sayYES; |
b8591aee DM |
3679 | if (cur_curlyx->u.curlyx.outercc) |
3680 | cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln; | |
4633a7c4 | 3681 | sayNO; |
a0d0e21e LW |
3682 | } |
3683 | ||
3684 | /* First just match a string of min scans. */ | |
3685 | ||
b8591aee DM |
3686 | if (n < cur_curlyx->u.curlyx.min) { |
3687 | cur_curlyx->u.curlyx.cur = n; | |
3688 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3689 | REGMATCH(cur_curlyx->u.curlyx.scan, WHILEM2); | |
95b24440 DM |
3690 | /*** all unsaved local vars undefined at this point */ |
3691 | if (result) | |
4633a7c4 | 3692 | sayYES; |
b8591aee DM |
3693 | cur_curlyx->u.curlyx.cur = n - 1; |
3694 | cur_curlyx->u.curlyx.lastloc = st->u.whilem.lastloc; | |
4633a7c4 | 3695 | sayNO; |
a0d0e21e LW |
3696 | } |
3697 | ||
2c2d71f5 JH |
3698 | if (scan->flags) { |
3699 | /* Check whether we already were at this position. | |
3700 | Postpone detection until we know the match is not | |
3701 | *that* much linear. */ | |
3702 | if (!PL_reg_maxiter) { | |
3703 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); | |
66bf836d DM |
3704 | /* possible overflow for long strings and many CURLYX's */ |
3705 | if (PL_reg_maxiter < 0) | |
3706 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
3707 | PL_reg_leftiter = PL_reg_maxiter; |
3708 | } | |
3709 | if (PL_reg_leftiter-- == 0) { | |
3298f257 | 3710 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 3711 | if (PL_reg_poscache) { |
eb160463 | 3712 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
3713 | Renew(PL_reg_poscache, size, char); |
3714 | PL_reg_poscache_size = size; | |
3715 | } | |
3716 | Zero(PL_reg_poscache, size, char); | |
3717 | } | |
3718 | else { | |
3719 | PL_reg_poscache_size = size; | |
a02a5408 | 3720 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 3721 | } |
a3621e74 | 3722 | DEBUG_EXECUTE_r( |
2c2d71f5 JH |
3723 | PerlIO_printf(Perl_debug_log, |
3724 | "%sDetected a super-linear match, switching on caching%s...\n", | |
3725 | PL_colors[4], PL_colors[5]) | |
3726 | ); | |
3727 | } | |
3728 | if (PL_reg_leftiter < 0) { | |
d8319b27 | 3729 | st->u.whilem.cache_offset = locinput - PL_bostr; |
2c2d71f5 | 3730 | |
3298f257 | 3731 | st->u.whilem.cache_offset = (scan->flags & 0xf) - 1 |
d8319b27 DM |
3732 | + st->u.whilem.cache_offset * (scan->flags>>4); |
3733 | st->u.whilem.cache_bit = st->u.whilem.cache_offset % 8; | |
3734 | st->u.whilem.cache_offset /= 8; | |
3735 | if (PL_reg_poscache[st->u.whilem.cache_offset] & (1<<st->u.whilem.cache_bit)) { | |
a3621e74 | 3736 | DEBUG_EXECUTE_r( |
2c2d71f5 JH |
3737 | PerlIO_printf(Perl_debug_log, |
3738 | "%*s already tried at this position...\n", | |
5bc10b2c | 3739 | REPORT_CODE_OFF+depth*2, "") |
2c2d71f5 | 3740 | ); |
3298f257 | 3741 | sayNO; /* cache records failure */ |
2c2d71f5 | 3742 | } |
2c2d71f5 JH |
3743 | } |
3744 | } | |
3745 | ||
a0d0e21e LW |
3746 | /* Prefer next over scan for minimal matching. */ |
3747 | ||
b8591aee DM |
3748 | if (cur_curlyx->minmod) { |
3749 | st->u.whilem.savecc = cur_curlyx; | |
3750 | cur_curlyx = cur_curlyx->u.curlyx.outercc; | |
3751 | if (cur_curlyx) | |
3752 | st->ln = cur_curlyx->u.curlyx.cur; | |
d8319b27 DM |
3753 | st->u.whilem.cp = regcppush(st->u.whilem.savecc->u.curlyx.parenfloor); |
3754 | REGCP_SET(st->u.whilem.lastcp); | |
3755 | REGMATCH(st->u.whilem.savecc->next, WHILEM3); | |
95b24440 | 3756 | /*** all unsaved local vars undefined at this point */ |
b8591aee | 3757 | cur_curlyx = st->u.whilem.savecc; |
95b24440 | 3758 | if (result) { |
d8319b27 | 3759 | regcpblow(st->u.whilem.cp); |
3298f257 | 3760 | sayYES; /* All done. */ |
5f05dabc | 3761 | } |
d8319b27 | 3762 | REGCP_UNWIND(st->u.whilem.lastcp); |
4f639d21 | 3763 | regcppop(rex); |
b8591aee DM |
3764 | if (cur_curlyx->u.curlyx.outercc) |
3765 | cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln; | |
a0d0e21e | 3766 | |
b8591aee | 3767 | if (n >= cur_curlyx->u.curlyx.max) { /* Maximum greed exceeded? */ |
9041c2e3 | 3768 | if (ckWARN(WARN_REGEXP) && n >= REG_INFTY |
3280af22 NIS |
3769 | && !(PL_reg_flags & RF_warned)) { |
3770 | PL_reg_flags |= RF_warned; | |
9014280d | 3771 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", |
2f3ca594 GS |
3772 | "Complex regular subexpression recursion", |
3773 | REG_INFTY - 1); | |
c277df42 | 3774 | } |
3ab3c9b4 | 3775 | CACHEsayNO; |
c277df42 | 3776 | } |
a687059c | 3777 | |
a3621e74 | 3778 | DEBUG_EXECUTE_r( |
c3464db5 DD |
3779 | PerlIO_printf(Perl_debug_log, |
3780 | "%*s trying longer...\n", | |
5bc10b2c | 3781 | REPORT_CODE_OFF+depth*2, "") |
c277df42 | 3782 | ); |
a0d0e21e | 3783 | /* Try scanning more and see if it helps. */ |
3280af22 | 3784 | PL_reginput = locinput; |
b8591aee DM |
3785 | cur_curlyx->u.curlyx.cur = n; |
3786 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3787 | st->u.whilem.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
d8319b27 | 3788 | REGCP_SET(st->u.whilem.lastcp); |
b8591aee | 3789 | REGMATCH(cur_curlyx->u.curlyx.scan, WHILEM4); |
95b24440 DM |
3790 | /*** all unsaved local vars undefined at this point */ |
3791 | if (result) { | |
d8319b27 | 3792 | regcpblow(st->u.whilem.cp); |
3298f257 | 3793 | sayYES; |
5f05dabc | 3794 | } |
d8319b27 | 3795 | REGCP_UNWIND(st->u.whilem.lastcp); |
4f639d21 | 3796 | regcppop(rex); |
b8591aee DM |
3797 | cur_curlyx->u.curlyx.cur = n - 1; |
3798 | cur_curlyx->u.curlyx.lastloc = st->u.whilem.lastloc; | |
3ab3c9b4 | 3799 | CACHEsayNO; |
a0d0e21e LW |
3800 | } |
3801 | ||
3802 | /* Prefer scan over next for maximal matching. */ | |
3803 | ||
b8591aee DM |
3804 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ |
3805 | st->u.whilem.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
3806 | cur_curlyx->u.curlyx.cur = n; | |
3807 | cur_curlyx->u.curlyx.lastloc = locinput; | |
d8319b27 | 3808 | REGCP_SET(st->u.whilem.lastcp); |
b8591aee | 3809 | REGMATCH(cur_curlyx->u.curlyx.scan, WHILEM5); |
95b24440 DM |
3810 | /*** all unsaved local vars undefined at this point */ |
3811 | if (result) { | |
d8319b27 | 3812 | regcpblow(st->u.whilem.cp); |
3298f257 | 3813 | sayYES; |
5f05dabc | 3814 | } |
d8319b27 | 3815 | REGCP_UNWIND(st->u.whilem.lastcp); |
4f639d21 | 3816 | regcppop(rex); /* Restore some previous $<digit>s? */ |
3280af22 | 3817 | PL_reginput = locinput; |
a3621e74 | 3818 | DEBUG_EXECUTE_r( |
c3464db5 DD |
3819 | PerlIO_printf(Perl_debug_log, |
3820 | "%*s failed, try continuation...\n", | |
5bc10b2c | 3821 | REPORT_CODE_OFF+depth*2, "") |
c277df42 IZ |
3822 | ); |
3823 | } | |
9041c2e3 | 3824 | if (ckWARN(WARN_REGEXP) && n >= REG_INFTY |
599cee73 | 3825 | && !(PL_reg_flags & RF_warned)) { |
3280af22 | 3826 | PL_reg_flags |= RF_warned; |
9014280d | 3827 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", |
cb5d145d GS |
3828 | "Complex regular subexpression recursion", |
3829 | REG_INFTY - 1); | |
a0d0e21e LW |
3830 | } |
3831 | ||
3832 | /* Failed deeper matches of scan, so see if this one works. */ | |
b8591aee DM |
3833 | st->u.whilem.savecc = cur_curlyx; |
3834 | cur_curlyx = cur_curlyx->u.curlyx.outercc; | |
3835 | if (cur_curlyx) | |
3836 | st->ln = cur_curlyx->u.curlyx.cur; | |
d8319b27 | 3837 | REGMATCH(st->u.whilem.savecc->next, WHILEM6); |
95b24440 | 3838 | /*** all unsaved local vars undefined at this point */ |
b8591aee | 3839 | cur_curlyx = st->u.whilem.savecc; |
95b24440 | 3840 | if (result) |
3298f257 | 3841 | sayYES; |
b8591aee DM |
3842 | if (cur_curlyx->u.curlyx.outercc) |
3843 | cur_curlyx->u.curlyx.outercc->u.curlyx.cur = st->ln; | |
3844 | cur_curlyx->u.curlyx.cur = n - 1; | |
3845 | cur_curlyx->u.curlyx.lastloc = st->u.whilem.lastloc; | |
3ab3c9b4 | 3846 | CACHEsayNO; |
a0d0e21e | 3847 | } |
5f66b61c | 3848 | /* NOTREACHED */ |
40a82448 DM |
3849 | |
3850 | #undef ST | |
3851 | #define ST st->u.branch | |
3852 | ||
3853 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
3854 | next = scan + ARG(scan); |
3855 | if (next == scan) | |
3856 | next = NULL; | |
40a82448 DM |
3857 | scan = NEXTOPER(scan); |
3858 | /* FALL THROUGH */ | |
c277df42 | 3859 | |
40a82448 DM |
3860 | case BRANCH: /* /(...|A|...)/ */ |
3861 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
3862 | if (!next || (OP(next) != BRANCH && OP(next) != BRANCHJ)) | |
3863 | /* last branch; skip state push and jump direct to node */ | |
3864 | continue; | |
3865 | ST.lastparen = *PL_reglastparen; | |
3866 | ST.next_branch = next; | |
3867 | REGCP_SET(ST.cp); | |
3868 | PL_reginput = locinput; | |
02db2b7b | 3869 | |
40a82448 DM |
3870 | /* Now go into the branch */ |
3871 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
3872 | /* NOTREACHED */ | |
3873 | ||
3874 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ | |
3875 | REGCP_UNWIND(ST.cp); | |
3876 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
3877 | PL_regendp[n] = -1; | |
3878 | *PL_reglastparen = n; | |
3879 | scan = ST.next_branch; | |
3880 | /* no more branches? */ | |
3881 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) | |
3882 | sayNO; | |
3883 | continue; /* execute next BRANCH[J] op */ | |
3884 | /* NOTREACHED */ | |
3885 | ||
a0d0e21e | 3886 | case MINMOD: |
5d9a96ca | 3887 | st->minmod = 1; |
a0d0e21e | 3888 | break; |
40a82448 DM |
3889 | |
3890 | #undef ST | |
3891 | #define ST st->u.curlym | |
3892 | ||
3893 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
3894 | ||
3895 | /* This is an optimisation of CURLYX that enables us to push | |
3896 | * only a single backtracking state, no matter now many matches | |
3897 | * there are in {m,n}. It relies on the pattern being constant | |
3898 | * length, with no parens to influence future backrefs | |
3899 | */ | |
3900 | ||
3901 | ST.me = scan; | |
dc45a647 | 3902 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
3903 | |
3904 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
3905 | if (ST.me->flags) { | |
3906 | I32 paren = ST.me->flags; | |
3907 | if (paren > PL_regsize) | |
3908 | PL_regsize = paren; | |
3909 | if (paren > (I32)*PL_reglastparen) | |
3910 | *PL_reglastparen = paren; | |
c277df42 | 3911 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 3912 | } |
40a82448 DM |
3913 | ST.A = scan; |
3914 | ST.B = next; | |
3915 | ST.alen = 0; | |
3916 | ST.count = 0; | |
3917 | ST.minmod = st->minmod; | |
3918 | st->minmod = 0; | |
3919 | ST.c1 = CHRTEST_UNINIT; | |
3920 | REGCP_SET(ST.cp); | |
6407bf3b | 3921 | |
40a82448 DM |
3922 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
3923 | goto curlym_do_B; | |
3924 | ||
3925 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 3926 | PL_reginput = locinput; |
40a82448 DM |
3927 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
3928 | /* NOTREACHED */ | |
5f80c4cf | 3929 | |
40a82448 DM |
3930 | case CURLYM_A: /* we've just matched an A */ |
3931 | locinput = st->locinput; | |
3932 | nextchr = UCHARAT(locinput); | |
3933 | ||
3934 | ST.count++; | |
3935 | /* after first match, determine A's length: u.curlym.alen */ | |
3936 | if (ST.count == 1) { | |
3937 | if (PL_reg_match_utf8) { | |
3938 | char *s = locinput; | |
3939 | while (s < PL_reginput) { | |
3940 | ST.alen++; | |
3941 | s += UTF8SKIP(s); | |
3942 | } | |
3943 | } | |
3944 | else { | |
3945 | ST.alen = PL_reginput - locinput; | |
3946 | } | |
3947 | if (ST.alen == 0) | |
3948 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
3949 | } | |
0cadcf80 DM |
3950 | DEBUG_EXECUTE_r( |
3951 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 3952 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 3953 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 3954 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
3955 | ); |
3956 | ||
40a82448 DM |
3957 | locinput = PL_reginput; |
3958 | if (ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) | |
3959 | goto curlym_do_A; /* try to match another A */ | |
3960 | goto curlym_do_B; /* try to match B */ | |
5f80c4cf | 3961 | |
40a82448 DM |
3962 | case CURLYM_A_fail: /* just failed to match an A */ |
3963 | REGCP_UNWIND(ST.cp); | |
3964 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ ) | |
3965 | sayNO; | |
0cadcf80 | 3966 | |
40a82448 DM |
3967 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
3968 | PL_reginput = locinput; | |
3969 | if (ST.c1 == CHRTEST_UNINIT) { | |
3970 | /* calculate c1 and c2 for possible match of 1st char | |
3971 | * following curly */ | |
3972 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
3973 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
3974 | regnode *text_node = ST.B; | |
3975 | if (! HAS_TEXT(text_node)) | |
3976 | FIND_NEXT_IMPT(text_node); | |
3977 | if (HAS_TEXT(text_node) | |
3978 | && PL_regkind[OP(text_node)] != REF) | |
3979 | { | |
3980 | ST.c1 = (U8)*STRING(text_node); | |
3981 | ST.c2 = | |
3982 | (OP(text_node) == EXACTF || OP(text_node) == REFF) | |
3983 | ? PL_fold[ST.c1] | |
3984 | : (OP(text_node) == EXACTFL || OP(text_node) == REFFL) | |
3985 | ? PL_fold_locale[ST.c1] | |
3986 | : ST.c1; | |
c277df42 | 3987 | } |
c277df42 | 3988 | } |
40a82448 DM |
3989 | } |
3990 | ||
3991 | DEBUG_EXECUTE_r( | |
3992 | PerlIO_printf(Perl_debug_log, | |
3993 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 3994 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
3995 | "", (IV)ST.count) |
3996 | ); | |
3997 | if (ST.c1 != CHRTEST_VOID | |
3998 | && UCHARAT(PL_reginput) != ST.c1 | |
3999 | && UCHARAT(PL_reginput) != ST.c2) | |
4000 | { | |
4001 | /* simulate B failing */ | |
4002 | state_num = CURLYM_B_fail; | |
4003 | goto reenter_switch; | |
4004 | } | |
4005 | ||
4006 | if (ST.me->flags) { | |
4007 | /* mark current A as captured */ | |
4008 | I32 paren = ST.me->flags; | |
4009 | if (ST.count) { | |
4010 | PL_regstartp[paren] | |
4011 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; | |
4012 | PL_regendp[paren] = PL_reginput - PL_bostr; | |
c277df42 | 4013 | } |
40a82448 DM |
4014 | else |
4015 | PL_regendp[paren] = -1; | |
c277df42 | 4016 | } |
40a82448 | 4017 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4018 | /* NOTREACHED */ |
40a82448 DM |
4019 | |
4020 | case CURLYM_B_fail: /* just failed to match a B */ | |
4021 | REGCP_UNWIND(ST.cp); | |
4022 | if (ST.minmod) { | |
4023 | if (ST.count == ARG2(ST.me) /* max */) | |
4024 | sayNO; | |
4025 | goto curlym_do_A; /* try to match a further A */ | |
4026 | } | |
4027 | /* backtrack one A */ | |
4028 | if (ST.count == ARG1(ST.me) /* min */) | |
4029 | sayNO; | |
4030 | ST.count--; | |
4031 | locinput = HOPc(locinput, -ST.alen); | |
4032 | goto curlym_do_B; /* try to match B */ | |
4033 | ||
c255a977 DM |
4034 | #undef ST |
4035 | #define ST st->u.curly | |
40a82448 | 4036 | |
c255a977 DM |
4037 | #define CURLY_SETPAREN(paren, success) \ |
4038 | if (paren) { \ | |
4039 | if (success) { \ | |
4040 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; \ | |
4041 | PL_regendp[paren] = locinput - PL_bostr; \ | |
4042 | } \ | |
4043 | else \ | |
4044 | PL_regendp[paren] = -1; \ | |
4045 | } | |
4046 | ||
4047 | case STAR: /* /A*B/ where A is width 1 */ | |
4048 | ST.paren = 0; | |
4049 | ST.min = 0; | |
4050 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4051 | scan = NEXTOPER(scan); |
4052 | goto repeat; | |
c255a977 DM |
4053 | case PLUS: /* /A+B/ where A is width 1 */ |
4054 | ST.paren = 0; | |
4055 | ST.min = 1; | |
4056 | ST.max = REG_INFTY; | |
c277df42 | 4057 | scan = NEXTOPER(scan); |
c255a977 DM |
4058 | goto repeat; |
4059 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4060 | ST.paren = scan->flags; /* Which paren to set */ | |
4061 | if (ST.paren > PL_regsize) | |
4062 | PL_regsize = ST.paren; | |
4063 | if (ST.paren > (I32)*PL_reglastparen) | |
4064 | *PL_reglastparen = ST.paren; | |
4065 | ST.min = ARG1(scan); /* min to match */ | |
4066 | ST.max = ARG2(scan); /* max to match */ | |
4067 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); | |
4068 | goto repeat; | |
4069 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4070 | ST.paren = 0; | |
4071 | ST.min = ARG1(scan); /* min to match */ | |
4072 | ST.max = ARG2(scan); /* max to match */ | |
4073 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4074 | repeat: |
a0d0e21e LW |
4075 | /* |
4076 | * Lookahead to avoid useless match attempts | |
4077 | * when we know what character comes next. | |
c255a977 | 4078 | * |
5f80c4cf JP |
4079 | * Used to only do .*x and .*?x, but now it allows |
4080 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4081 | * of the quantifier and the EXACT-like node. -- japhy | |
4082 | */ | |
4083 | ||
c255a977 DM |
4084 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4085 | sayNO; | |
cca55fe3 | 4086 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4087 | U8 *s; |
4088 | regnode *text_node = next; | |
4089 | ||
3dab1dad YO |
4090 | if (! HAS_TEXT(text_node)) |
4091 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4092 | |
9e137952 | 4093 | if (! HAS_TEXT(text_node)) |
c255a977 | 4094 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4095 | else { |
3dab1dad | 4096 | if (PL_regkind[OP(text_node)] == REF) { |
c255a977 | 4097 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4098 | goto assume_ok_easy; |
cca55fe3 | 4099 | } |
be8e71aa YO |
4100 | else |
4101 | s = (U8*)STRING(text_node); | |
5f80c4cf JP |
4102 | |
4103 | if (!UTF) { | |
c255a977 | 4104 | ST.c2 = ST.c1 = *s; |
f65d3ee7 | 4105 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) |
c255a977 | 4106 | ST.c2 = PL_fold[ST.c1]; |
f65d3ee7 | 4107 | else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) |
c255a977 | 4108 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4109 | } |
5f80c4cf | 4110 | else { /* UTF */ |
f65d3ee7 | 4111 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) { |
a2a2844f | 4112 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4113 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4114 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4115 | |
4116 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4117 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4118 | #ifdef EBCDIC |
4119 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4120 | ckWARN(WARN_UTF8) ? | |
4121 | 0 : UTF8_ALLOW_ANY); | |
4122 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4123 | ckWARN(WARN_UTF8) ? | |
4124 | 0 : UTF8_ALLOW_ANY); | |
4125 | #else | |
c255a977 | 4126 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4127 | uniflags); |
c255a977 | 4128 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4129 | uniflags); |
4130 | #endif | |
5f80c4cf JP |
4131 | } |
4132 | else { | |
c255a977 | 4133 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4134 | uniflags); |
5f80c4cf | 4135 | } |
1aa99e6b IH |
4136 | } |
4137 | } | |
bbce6d69 | 4138 | } |
a0d0e21e | 4139 | else |
c255a977 | 4140 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4141 | assume_ok_easy: |
c255a977 DM |
4142 | |
4143 | ST.A = scan; | |
4144 | ST.B = next; | |
3280af22 | 4145 | PL_reginput = locinput; |
5d9a96ca DM |
4146 | if (st->minmod) { |
4147 | st->minmod = 0; | |
c255a977 | 4148 | if (ST.min && regrepeat(rex, ST.A, ST.min) < ST.min) |
4633a7c4 | 4149 | sayNO; |
c255a977 | 4150 | ST.count = ST.min; |
a0ed51b3 | 4151 | locinput = PL_reginput; |
c255a977 DM |
4152 | REGCP_SET(ST.cp); |
4153 | if (ST.c1 == CHRTEST_VOID) | |
4154 | goto curly_try_B_min; | |
4155 | ||
4156 | ST.oldloc = locinput; | |
4157 | ||
4158 | /* set ST.maxpos to the furthest point along the | |
4159 | * string that could possibly match */ | |
4160 | if (ST.max == REG_INFTY) { | |
4161 | ST.maxpos = PL_regeol - 1; | |
4162 | if (do_utf8) | |
4163 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4164 | ST.maxpos--; | |
4165 | } | |
4166 | else if (do_utf8) { | |
4167 | int m = ST.max - ST.min; | |
4168 | for (ST.maxpos = locinput; | |
4169 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4170 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4171 | } | |
4172 | else { | |
4173 | ST.maxpos = locinput + ST.max - ST.min; | |
4174 | if (ST.maxpos >= PL_regeol) | |
4175 | ST.maxpos = PL_regeol - 1; | |
4176 | } | |
4177 | goto curly_try_B_min_known; | |
4178 | ||
4179 | } | |
4180 | else { | |
4181 | ST.count = regrepeat(rex, ST.A, ST.max); | |
4182 | locinput = PL_reginput; | |
4183 | if (ST.count < ST.min) | |
4184 | sayNO; | |
4185 | if ((ST.count > ST.min) | |
4186 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4187 | { | |
4188 | /* A{m,n} must come at the end of the string, there's | |
4189 | * no point in backing off ... */ | |
4190 | ST.min = ST.count; | |
4191 | /* ...except that $ and \Z can match before *and* after | |
4192 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4193 | We may back off by one in this case. */ | |
4194 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4195 | ST.min--; | |
4196 | } | |
4197 | REGCP_SET(ST.cp); | |
4198 | goto curly_try_B_max; | |
4199 | } | |
4200 | /* NOTREACHED */ | |
4201 | ||
4202 | ||
4203 | case CURLY_B_min_known_fail: | |
4204 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4205 | if (ST.paren && ST.count) | |
4206 | PL_regendp[ST.paren] = -1; | |
4207 | ||
4208 | PL_reginput = locinput; /* Could be reset... */ | |
4209 | REGCP_UNWIND(ST.cp); | |
4210 | /* Couldn't or didn't -- move forward. */ | |
4211 | ST.oldloc = locinput; | |
4212 | if (do_utf8) | |
4213 | locinput += UTF8SKIP(locinput); | |
4214 | else | |
4215 | locinput++; | |
4216 | ST.count++; | |
4217 | curly_try_B_min_known: | |
4218 | /* find the next place where 'B' could work, then call B */ | |
4219 | { | |
4220 | int n; | |
4221 | if (do_utf8) { | |
4222 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4223 | if (ST.c1 == ST.c2) { | |
4224 | STRLEN len; | |
4225 | /* set n to utf8_distance(oldloc, locinput) */ | |
4226 | while (locinput <= ST.maxpos && | |
4227 | utf8n_to_uvchr((U8*)locinput, | |
4228 | UTF8_MAXBYTES, &len, | |
4229 | uniflags) != (UV)ST.c1) { | |
4230 | locinput += len; | |
4231 | n++; | |
4232 | } | |
1aa99e6b IH |
4233 | } |
4234 | else { | |
c255a977 DM |
4235 | /* set n to utf8_distance(oldloc, locinput) */ |
4236 | while (locinput <= ST.maxpos) { | |
4237 | STRLEN len; | |
4238 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4239 | UTF8_MAXBYTES, &len, | |
4240 | uniflags); | |
4241 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4242 | break; | |
4243 | locinput += len; | |
4244 | n++; | |
1aa99e6b | 4245 | } |
0fe9bf95 IZ |
4246 | } |
4247 | } | |
c255a977 DM |
4248 | else { |
4249 | if (ST.c1 == ST.c2) { | |
4250 | while (locinput <= ST.maxpos && | |
4251 | UCHARAT(locinput) != ST.c1) | |
4252 | locinput++; | |
bbce6d69 | 4253 | } |
c255a977 DM |
4254 | else { |
4255 | while (locinput <= ST.maxpos | |
4256 | && UCHARAT(locinput) != ST.c1 | |
4257 | && UCHARAT(locinput) != ST.c2) | |
4258 | locinput++; | |
a0ed51b3 | 4259 | } |
c255a977 DM |
4260 | n = locinput - ST.oldloc; |
4261 | } | |
4262 | if (locinput > ST.maxpos) | |
4263 | sayNO; | |
4264 | /* PL_reginput == oldloc now */ | |
4265 | if (n) { | |
4266 | ST.count += n; | |
4267 | if (regrepeat(rex, ST.A, n) < n) | |
4633a7c4 | 4268 | sayNO; |
a0d0e21e | 4269 | } |
c255a977 DM |
4270 | PL_reginput = locinput; |
4271 | CURLY_SETPAREN(ST.paren, ST.count); | |
4272 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); | |
a0d0e21e | 4273 | } |
c255a977 DM |
4274 | /* NOTREACHED */ |
4275 | ||
4276 | ||
4277 | case CURLY_B_min_fail: | |
4278 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4279 | if (ST.paren && ST.count) | |
4280 | PL_regendp[ST.paren] = -1; | |
4281 | ||
4282 | REGCP_UNWIND(ST.cp); | |
4283 | /* failed -- move forward one */ | |
4284 | PL_reginput = locinput; | |
4285 | if (regrepeat(rex, ST.A, 1)) { | |
4286 | ST.count++; | |
a0ed51b3 | 4287 | locinput = PL_reginput; |
c255a977 DM |
4288 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4289 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4290 | { |
c255a977 DM |
4291 | curly_try_B_min: |
4292 | CURLY_SETPAREN(ST.paren, ST.count); | |
4293 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); | |
a0d0e21e LW |
4294 | } |
4295 | } | |
4633a7c4 | 4296 | sayNO; |
c255a977 DM |
4297 | /* NOTREACHED */ |
4298 | ||
4299 | ||
4300 | curly_try_B_max: | |
4301 | /* a successful greedy match: now try to match B */ | |
4302 | { | |
4303 | UV c = 0; | |
4304 | if (ST.c1 != CHRTEST_VOID) | |
4305 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4306 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4307 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4308 | /* If it could work, try it. */ |
4309 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4310 | CURLY_SETPAREN(ST.paren, ST.count); | |
4311 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4312 | /* NOTREACHED */ | |
4313 | } | |
4314 | } | |
4315 | /* FALL THROUGH */ | |
4316 | case CURLY_B_max_fail: | |
4317 | /* failed to find B in a greedy match */ | |
4318 | if (ST.paren && ST.count) | |
4319 | PL_regendp[ST.paren] = -1; | |
4320 | ||
4321 | REGCP_UNWIND(ST.cp); | |
4322 | /* back up. */ | |
4323 | if (--ST.count < ST.min) | |
4324 | sayNO; | |
4325 | PL_reginput = locinput = HOPc(locinput, -1); | |
4326 | goto curly_try_B_max; | |
4327 | ||
4328 | #undef ST | |
4329 | ||
4330 | ||
a0d0e21e | 4331 | case END: |
faec1544 DM |
4332 | if (cur_eval) { |
4333 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4334 | I32 tmpix; | |
4335 | ||
4336 | ||
4337 | st->u.eval.toggle_reg_flags | |
4338 | = cur_eval->u.eval.toggle_reg_flags; | |
4339 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4340 | ||
4341 | st->u.eval.prev_rex = rex; /* inner */ | |
4342 | rex = cur_eval->u.eval.prev_rex; /* outer */ | |
4343 | cur_curlyx = cur_eval->u.eval.prev_curlyx; | |
4344 | ReREFCNT_inc(rex); | |
4345 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ | |
4346 | REGCP_SET(st->u.eval.lastcp); | |
4347 | PL_reginput = locinput; | |
4348 | ||
4349 | /* Restore parens of the outer rex without popping the | |
4350 | * savestack */ | |
4351 | tmpix = PL_savestack_ix; | |
4352 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4353 | regcppop(rex); | |
4354 | PL_savestack_ix = tmpix; | |
4355 | ||
4356 | st->u.eval.prev_eval = cur_eval; | |
4357 | cur_eval = cur_eval->u.eval.prev_eval; | |
4358 | DEBUG_EXECUTE_r( | |
4359 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ...\n", | |
5bc10b2c | 4360 | REPORT_CODE_OFF+depth*2, "");); |
faec1544 DM |
4361 | PUSH_YES_STATE_GOTO(EVAL_AB, |
4362 | st->u.eval.prev_eval->u.eval.B); /* match B */ | |
4363 | } | |
4364 | ||
3b0527fe | 4365 | if (locinput < reginfo->till) { |
a3621e74 | 4366 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4367 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4368 | PL_colors[4], | |
4369 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4370 | (long)(reginfo->till - PL_reg_starttry), |
7821416a IZ |
4371 | PL_colors[5])); |
4372 | sayNO_FINAL; /* Cannot match: too short. */ | |
4373 | } | |
4374 | PL_reginput = locinput; /* put where regtry can find it */ | |
4375 | sayYES_FINAL; /* Success! */ | |
dad79028 DM |
4376 | |
4377 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4378 | DEBUG_EXECUTE_r( | |
4379 | PerlIO_printf(Perl_debug_log, | |
4380 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4381 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4382 | PL_reginput = locinput; /* put where regtry can find it */ |
dad79028 DM |
4383 | sayYES_FINAL; /* Success! */ |
4384 | ||
40a82448 DM |
4385 | #undef ST |
4386 | #define ST st->u.ifmatch | |
4387 | ||
4388 | case SUSPEND: /* (?>A) */ | |
4389 | ST.wanted = 1; | |
9fe1d20c | 4390 | PL_reginput = locinput; |
9041c2e3 | 4391 | goto do_ifmatch; |
dad79028 | 4392 | |
40a82448 DM |
4393 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4394 | ST.wanted = 0; | |
dad79028 DM |
4395 | goto ifmatch_trivial_fail_test; |
4396 | ||
40a82448 DM |
4397 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4398 | ST.wanted = 1; | |
dad79028 | 4399 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4400 | if (scan->flags) { |
52657f30 | 4401 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4402 | if (!s) { |
4403 | /* trivial fail */ | |
4404 | if (st->logical) { | |
4405 | st->logical = 0; | |
40a82448 | 4406 | st->sw = 1 - (bool)ST.wanted; |
dad79028 | 4407 | } |
40a82448 | 4408 | else if (ST.wanted) |
dad79028 DM |
4409 | sayNO; |
4410 | next = scan + ARG(scan); | |
4411 | if (next == scan) | |
4412 | next = NULL; | |
4413 | break; | |
4414 | } | |
efb30f32 | 4415 | PL_reginput = s; |
a0ed51b3 LW |
4416 | } |
4417 | else | |
4418 | PL_reginput = locinput; | |
4419 | ||
c277df42 | 4420 | do_ifmatch: |
40a82448 DM |
4421 | ST.me = scan; |
4422 | /* execute body of (?...A) */ | |
4423 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
4424 | /* NOTREACHED */ | |
4425 | ||
4426 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
4427 | ST.wanted = !ST.wanted; | |
4428 | /* FALL THROUGH */ | |
4429 | ||
4430 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
4431 | if (st->logical) { | |
4432 | st->logical = 0; | |
4433 | st->sw = (bool)ST.wanted; | |
4434 | } | |
4435 | else if (!ST.wanted) | |
4436 | sayNO; | |
4437 | ||
4438 | if (OP(ST.me) == SUSPEND) | |
4439 | locinput = PL_reginput; | |
4440 | else { | |
4441 | locinput = PL_reginput = st->locinput; | |
4442 | nextchr = UCHARAT(locinput); | |
4443 | } | |
4444 | scan = ST.me + ARG(ST.me); | |
4445 | if (scan == ST.me) | |
4446 | scan = NULL; | |
4447 | continue; /* execute B */ | |
4448 | ||
4449 | #undef ST | |
dad79028 | 4450 | |
c277df42 | 4451 | case LONGJMP: |
c277df42 IZ |
4452 | next = scan + ARG(scan); |
4453 | if (next == scan) | |
4454 | next = NULL; | |
a0d0e21e LW |
4455 | break; |
4456 | default: | |
b900a521 | 4457 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 4458 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 4459 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c | 4460 | } |
95b24440 | 4461 | |
a0d0e21e | 4462 | scan = next; |
95b24440 DM |
4463 | continue; |
4464 | /* NOTREACHED */ | |
4465 | ||
40a82448 DM |
4466 | push_yes_state: |
4467 | /* push a state that backtracks on success */ | |
4468 | st->u.yes.prev_yes_state = yes_state; | |
4469 | yes_state = st; | |
4470 | /* FALL THROUGH */ | |
4471 | push_state: | |
4472 | /* push a new regex state, then continue at scan */ | |
4473 | { | |
4474 | regmatch_state *newst; | |
4475 | ||
5bc10b2c | 4476 | DEBUG_STATE_pp("push"); |
40a82448 | 4477 | depth++; |
40a82448 DM |
4478 | st->locinput = locinput; |
4479 | newst = st+1; | |
4480 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
4481 | newst = S_push_slab(aTHX); | |
4482 | PL_regmatch_state = newst; | |
40a82448 DM |
4483 | /* XXX probably don't need to initialise these */ |
4484 | newst->minmod = 0; | |
4485 | newst->sw = 0; | |
4486 | newst->logical = 0; | |
4487 | ||
786e8c11 | 4488 | |
40a82448 DM |
4489 | locinput = PL_reginput; |
4490 | nextchr = UCHARAT(locinput); | |
4491 | st = newst; | |
4492 | continue; | |
4493 | /* NOTREACHED */ | |
4494 | } | |
4495 | ||
95b24440 DM |
4496 | /* simulate recursively calling regmatch(), but without actually |
4497 | * recursing - ie save the current state on the heap rather than on | |
4498 | * the stack, then re-enter the loop. This avoids complex regexes | |
4499 | * blowing the processor stack */ | |
4500 | ||
4501 | start_recurse: | |
4502 | { | |
5d9a96ca DM |
4503 | /* push new state */ |
4504 | regmatch_state *oldst = st; | |
4505 | ||
5bc10b2c | 4506 | DEBUG_STATE_pp("push"); |
5d9a96ca DM |
4507 | depth++; |
4508 | ||
4509 | /* grab the next free state slot */ | |
4510 | st++; | |
86545054 | 4511 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
4512 | st = S_push_slab(aTHX); |
4513 | PL_regmatch_state = st; | |
4514 | ||
4515 | oldst->next = next; | |
4516 | oldst->n = n; | |
4517 | oldst->locinput = locinput; | |
5d9a96ca | 4518 | |
95b24440 DM |
4519 | locinput = PL_reginput; |
4520 | nextchr = UCHARAT(locinput); | |
5d9a96ca DM |
4521 | st->minmod = 0; |
4522 | st->sw = 0; | |
4523 | st->logical = 0; | |
786e8c11 | 4524 | |
95b24440 | 4525 | } |
a0d0e21e | 4526 | } |
a687059c | 4527 | |
aa283a38 DM |
4528 | |
4529 | ||
a0d0e21e LW |
4530 | /* |
4531 | * We get here only if there's trouble -- normally "case END" is | |
4532 | * the terminating point. | |
4533 | */ | |
cea2e8a9 | 4534 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 4535 | /*NOTREACHED*/ |
4633a7c4 LW |
4536 | sayNO; |
4537 | ||
7821416a | 4538 | yes_final: |
77cb431f DM |
4539 | |
4540 | if (yes_state) { | |
4541 | /* we have successfully completed a subexpression, but we must now | |
4542 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 4543 | assert(st != yes_state); |
5bc10b2c DM |
4544 | #ifdef DEBUGGING |
4545 | while (st != yes_state) { | |
4546 | st--; | |
4547 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4548 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4549 | st = SLAB_LAST(PL_regmatch_slab); | |
4550 | } | |
4551 | DEBUG_STATE_pp("pop (yes)"); | |
4552 | depth--; | |
4553 | } | |
4554 | #else | |
77cb431f DM |
4555 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
4556 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
4557 | { | |
4558 | /* not in this slab, pop slab */ | |
4559 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
4560 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4561 | st = SLAB_LAST(PL_regmatch_slab); | |
4562 | } | |
4563 | depth -= (st - yes_state); | |
5bc10b2c | 4564 | #endif |
77cb431f DM |
4565 | st = yes_state; |
4566 | yes_state = st->u.yes.prev_yes_state; | |
4567 | PL_regmatch_state = st; | |
4568 | ||
4569 | switch (st->resume_state) { | |
faec1544 | 4570 | case EVAL_AB: |
40a82448 DM |
4571 | case IFMATCH_A: |
4572 | case CURLYM_A: | |
40a82448 DM |
4573 | state_num = st->resume_state; |
4574 | goto reenter_switch; | |
4575 | ||
4576 | case CURLYM_B: | |
166ba7cd DM |
4577 | case BRANCH_next: |
4578 | case TRIE_next: | |
c255a977 | 4579 | case CURLY_B_max: |
77cb431f | 4580 | default: |
40a82448 | 4581 | Perl_croak(aTHX_ "unexpected yes resume state"); |
77cb431f DM |
4582 | } |
4583 | } | |
4584 | ||
a3621e74 | 4585 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 4586 | PL_colors[4], PL_colors[5])); |
4633a7c4 | 4587 | yes: |
02db2b7b | 4588 | |
95b24440 | 4589 | result = 1; |
aa283a38 | 4590 | /* XXX this is duplicate(ish) code to that in the do_no section. |
40a82448 | 4591 | * will disappear when REGFMATCH goes */ |
aa283a38 DM |
4592 | if (depth) { |
4593 | /* restore previous state and re-enter */ | |
40a82448 DM |
4594 | st--; |
4595 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4596 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4597 | st = SLAB_LAST(PL_regmatch_slab); | |
4598 | } | |
4599 | PL_regmatch_state = st; | |
4600 | scan = st->scan; | |
4601 | next = st->next; | |
4602 | n = st->n; | |
4603 | locinput= st->locinput; | |
4604 | nextchr = UCHARAT(locinput); | |
aa283a38 | 4605 | |
5bc10b2c DM |
4606 | DEBUG_STATE_pp("pop"); |
4607 | depth--; | |
4608 | ||
aa283a38 | 4609 | switch (st->resume_state) { |
aa283a38 DM |
4610 | case resume_CURLYX: |
4611 | goto resume_point_CURLYX; | |
4612 | case resume_WHILEM1: | |
4613 | goto resume_point_WHILEM1; | |
4614 | case resume_WHILEM2: | |
4615 | goto resume_point_WHILEM2; | |
4616 | case resume_WHILEM3: | |
4617 | goto resume_point_WHILEM3; | |
4618 | case resume_WHILEM4: | |
4619 | goto resume_point_WHILEM4; | |
4620 | case resume_WHILEM5: | |
4621 | goto resume_point_WHILEM5; | |
4622 | case resume_WHILEM6: | |
4623 | goto resume_point_WHILEM6; | |
77cb431f | 4624 | |
166ba7cd | 4625 | case TRIE_next: |
40a82448 DM |
4626 | case CURLYM_A: |
4627 | case CURLYM_B: | |
faec1544 | 4628 | case EVAL_AB: |
40a82448 DM |
4629 | case IFMATCH_A: |
4630 | case BRANCH_next: | |
c255a977 DM |
4631 | case CURLY_B_max: |
4632 | case CURLY_B_min: | |
4633 | case CURLY_B_min_known: | |
40a82448 DM |
4634 | break; |
4635 | ||
aa283a38 DM |
4636 | default: |
4637 | Perl_croak(aTHX_ "regexp resume memory corruption"); | |
4638 | } | |
4639 | } | |
4640 | goto final_exit; | |
4633a7c4 LW |
4641 | |
4642 | no: | |
a3621e74 | 4643 | DEBUG_EXECUTE_r( |
7821416a | 4644 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 4645 | "%*s %sfailed...%s\n", |
5bc10b2c | 4646 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 4647 | PL_colors[4], PL_colors[5]) |
7821416a | 4648 | ); |
7821416a IZ |
4649 | no_final: |
4650 | do_no: | |
aa283a38 | 4651 | |
95b24440 | 4652 | result = 0; |
5d9a96ca | 4653 | |
aa283a38 DM |
4654 | if (depth) { |
4655 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
4656 | st--; |
4657 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4658 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4659 | st = SLAB_LAST(PL_regmatch_slab); | |
4660 | } | |
4661 | PL_regmatch_state = st; | |
4662 | scan = st->scan; | |
4663 | next = st->next; | |
4664 | n = st->n; | |
4665 | locinput= st->locinput; | |
4666 | nextchr = UCHARAT(locinput); | |
4667 | ||
5bc10b2c DM |
4668 | DEBUG_STATE_pp("pop"); |
4669 | depth--; | |
4670 | ||
5d9a96ca | 4671 | switch (st->resume_state) { |
95b24440 DM |
4672 | case resume_CURLYX: |
4673 | goto resume_point_CURLYX; | |
4674 | case resume_WHILEM1: | |
4675 | goto resume_point_WHILEM1; | |
4676 | case resume_WHILEM2: | |
4677 | goto resume_point_WHILEM2; | |
4678 | case resume_WHILEM3: | |
4679 | goto resume_point_WHILEM3; | |
4680 | case resume_WHILEM4: | |
4681 | goto resume_point_WHILEM4; | |
4682 | case resume_WHILEM5: | |
4683 | goto resume_point_WHILEM5; | |
4684 | case resume_WHILEM6: | |
4685 | goto resume_point_WHILEM6; | |
dad79028 | 4686 | |
166ba7cd | 4687 | case TRIE_next: |
faec1544 | 4688 | case EVAL_AB: |
40a82448 DM |
4689 | case BRANCH_next: |
4690 | case CURLYM_A: | |
4691 | case CURLYM_B: | |
4692 | case IFMATCH_A: | |
c255a977 DM |
4693 | case CURLY_B_max: |
4694 | case CURLY_B_min: | |
4695 | case CURLY_B_min_known: | |
40a82448 DM |
4696 | if (yes_state == st) |
4697 | yes_state = st->u.yes.prev_yes_state; | |
4698 | state_num = st->resume_state + 1; /* failure = success + 1 */ | |
4699 | goto reenter_switch; | |
dad79028 | 4700 | |
95b24440 DM |
4701 | default: |
4702 | Perl_croak(aTHX_ "regexp resume memory corruption"); | |
4703 | } | |
95b24440 | 4704 | } |
aa283a38 DM |
4705 | |
4706 | final_exit: | |
4707 | ||
5d9a96ca DM |
4708 | /* restore original high-water mark */ |
4709 | PL_regmatch_slab = orig_slab; | |
4710 | PL_regmatch_state = orig_state; | |
4711 | ||
4712 | /* free all slabs above current one */ | |
4713 | if (orig_slab->next) { | |
c4fd8992 | 4714 | regmatch_slab *sl = orig_slab->next; |
5d9a96ca DM |
4715 | orig_slab->next = NULL; |
4716 | while (sl) { | |
c4fd8992 | 4717 | regmatch_slab * const osl = sl; |
5d9a96ca | 4718 | sl = sl->next; |
ad65c075 | 4719 | Safefree(osl); |
5d9a96ca DM |
4720 | } |
4721 | } | |
4722 | ||
95b24440 DM |
4723 | return result; |
4724 | ||
a687059c LW |
4725 | } |
4726 | ||
4727 | /* | |
4728 | - regrepeat - repeatedly match something simple, report how many | |
4729 | */ | |
4730 | /* | |
4731 | * [This routine now assumes that it will only match on things of length 1. | |
4732 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 4733 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 4734 | */ |
76e3520e | 4735 | STATIC I32 |
32fc9b6a | 4736 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max) |
a687059c | 4737 | { |
27da23d5 | 4738 | dVAR; |
a0d0e21e | 4739 | register char *scan; |
a0d0e21e | 4740 | register I32 c; |
3280af22 | 4741 | register char *loceol = PL_regeol; |
a0ed51b3 | 4742 | register I32 hardcount = 0; |
53c4c00c | 4743 | register bool do_utf8 = PL_reg_match_utf8; |
a0d0e21e | 4744 | |
3280af22 | 4745 | scan = PL_reginput; |
faf11cac HS |
4746 | if (max == REG_INFTY) |
4747 | max = I32_MAX; | |
4748 | else if (max < loceol - scan) | |
7f596f4c | 4749 | loceol = scan + max; |
a0d0e21e | 4750 | switch (OP(p)) { |
22c35a8c | 4751 | case REG_ANY: |
1aa99e6b | 4752 | if (do_utf8) { |
ffc61ed2 | 4753 | loceol = PL_regeol; |
1aa99e6b | 4754 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
4755 | scan += UTF8SKIP(scan); |
4756 | hardcount++; | |
4757 | } | |
4758 | } else { | |
4759 | while (scan < loceol && *scan != '\n') | |
4760 | scan++; | |
a0ed51b3 LW |
4761 | } |
4762 | break; | |
ffc61ed2 | 4763 | case SANY: |
def8e4ea JH |
4764 | if (do_utf8) { |
4765 | loceol = PL_regeol; | |
a0804c9e | 4766 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
4767 | scan += UTF8SKIP(scan); |
4768 | hardcount++; | |
4769 | } | |
4770 | } | |
4771 | else | |
4772 | scan = loceol; | |
a0ed51b3 | 4773 | break; |
f33976b4 DB |
4774 | case CANY: |
4775 | scan = loceol; | |
4776 | break; | |
090f7165 JH |
4777 | case EXACT: /* length of string is 1 */ |
4778 | c = (U8)*STRING(p); | |
4779 | while (scan < loceol && UCHARAT(scan) == c) | |
4780 | scan++; | |
bbce6d69 | 4781 | break; |
4782 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 4783 | c = (U8)*STRING(p); |
bbce6d69 | 4784 | while (scan < loceol && |
22c35a8c | 4785 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 4786 | scan++; |
4787 | break; | |
4788 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 4789 | PL_reg_flags |= RF_tainted; |
cd439c50 | 4790 | c = (U8)*STRING(p); |
bbce6d69 | 4791 | while (scan < loceol && |
22c35a8c | 4792 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
4793 | scan++; |
4794 | break; | |
4795 | case ANYOF: | |
ffc61ed2 JH |
4796 | if (do_utf8) { |
4797 | loceol = PL_regeol; | |
cfc92286 | 4798 | while (hardcount < max && scan < loceol && |
32fc9b6a | 4799 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
4800 | scan += UTF8SKIP(scan); |
4801 | hardcount++; | |
4802 | } | |
4803 | } else { | |
32fc9b6a | 4804 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
4805 | scan++; |
4806 | } | |
a0d0e21e LW |
4807 | break; |
4808 | case ALNUM: | |
1aa99e6b | 4809 | if (do_utf8) { |
ffc61ed2 | 4810 | loceol = PL_regeol; |
1a4fad37 | 4811 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4812 | while (hardcount < max && scan < loceol && |
3568d838 | 4813 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4814 | scan += UTF8SKIP(scan); |
4815 | hardcount++; | |
4816 | } | |
4817 | } else { | |
4818 | while (scan < loceol && isALNUM(*scan)) | |
4819 | scan++; | |
a0ed51b3 LW |
4820 | } |
4821 | break; | |
bbce6d69 | 4822 | case ALNUML: |
3280af22 | 4823 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4824 | if (do_utf8) { |
ffc61ed2 | 4825 | loceol = PL_regeol; |
1aa99e6b IH |
4826 | while (hardcount < max && scan < loceol && |
4827 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4828 | scan += UTF8SKIP(scan); |
4829 | hardcount++; | |
4830 | } | |
4831 | } else { | |
4832 | while (scan < loceol && isALNUM_LC(*scan)) | |
4833 | scan++; | |
a0ed51b3 LW |
4834 | } |
4835 | break; | |
a0d0e21e | 4836 | case NALNUM: |
1aa99e6b | 4837 | if (do_utf8) { |
ffc61ed2 | 4838 | loceol = PL_regeol; |
1a4fad37 | 4839 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4840 | while (hardcount < max && scan < loceol && |
3568d838 | 4841 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4842 | scan += UTF8SKIP(scan); |
4843 | hardcount++; | |
4844 | } | |
4845 | } else { | |
4846 | while (scan < loceol && !isALNUM(*scan)) | |
4847 | scan++; | |
a0ed51b3 LW |
4848 | } |
4849 | break; | |
bbce6d69 | 4850 | case NALNUML: |
3280af22 | 4851 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4852 | if (do_utf8) { |
ffc61ed2 | 4853 | loceol = PL_regeol; |
1aa99e6b IH |
4854 | while (hardcount < max && scan < loceol && |
4855 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4856 | scan += UTF8SKIP(scan); |
4857 | hardcount++; | |
4858 | } | |
4859 | } else { | |
4860 | while (scan < loceol && !isALNUM_LC(*scan)) | |
4861 | scan++; | |
a0ed51b3 LW |
4862 | } |
4863 | break; | |
a0d0e21e | 4864 | case SPACE: |
1aa99e6b | 4865 | if (do_utf8) { |
ffc61ed2 | 4866 | loceol = PL_regeol; |
1a4fad37 | 4867 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4868 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4869 | (*scan == ' ' || |
4870 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4871 | scan += UTF8SKIP(scan); |
4872 | hardcount++; | |
4873 | } | |
4874 | } else { | |
4875 | while (scan < loceol && isSPACE(*scan)) | |
4876 | scan++; | |
a0ed51b3 LW |
4877 | } |
4878 | break; | |
bbce6d69 | 4879 | case SPACEL: |
3280af22 | 4880 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4881 | if (do_utf8) { |
ffc61ed2 | 4882 | loceol = PL_regeol; |
1aa99e6b | 4883 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4884 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4885 | scan += UTF8SKIP(scan); | |
4886 | hardcount++; | |
4887 | } | |
4888 | } else { | |
4889 | while (scan < loceol && isSPACE_LC(*scan)) | |
4890 | scan++; | |
a0ed51b3 LW |
4891 | } |
4892 | break; | |
a0d0e21e | 4893 | case NSPACE: |
1aa99e6b | 4894 | if (do_utf8) { |
ffc61ed2 | 4895 | loceol = PL_regeol; |
1a4fad37 | 4896 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4897 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4898 | !(*scan == ' ' || |
4899 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4900 | scan += UTF8SKIP(scan); |
4901 | hardcount++; | |
4902 | } | |
4903 | } else { | |
4904 | while (scan < loceol && !isSPACE(*scan)) | |
4905 | scan++; | |
4906 | break; | |
a0ed51b3 | 4907 | } |
bbce6d69 | 4908 | case NSPACEL: |
3280af22 | 4909 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4910 | if (do_utf8) { |
ffc61ed2 | 4911 | loceol = PL_regeol; |
1aa99e6b | 4912 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4913 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4914 | scan += UTF8SKIP(scan); | |
4915 | hardcount++; | |
4916 | } | |
4917 | } else { | |
4918 | while (scan < loceol && !isSPACE_LC(*scan)) | |
4919 | scan++; | |
a0ed51b3 LW |
4920 | } |
4921 | break; | |
a0d0e21e | 4922 | case DIGIT: |
1aa99e6b | 4923 | if (do_utf8) { |
ffc61ed2 | 4924 | loceol = PL_regeol; |
1a4fad37 | 4925 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 4926 | while (hardcount < max && scan < loceol && |
3568d838 | 4927 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4928 | scan += UTF8SKIP(scan); |
4929 | hardcount++; | |
4930 | } | |
4931 | } else { | |
4932 | while (scan < loceol && isDIGIT(*scan)) | |
4933 | scan++; | |
a0ed51b3 LW |
4934 | } |
4935 | break; | |
a0d0e21e | 4936 | case NDIGIT: |
1aa99e6b | 4937 | if (do_utf8) { |
ffc61ed2 | 4938 | loceol = PL_regeol; |
1a4fad37 | 4939 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 4940 | while (hardcount < max && scan < loceol && |
3568d838 | 4941 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4942 | scan += UTF8SKIP(scan); |
4943 | hardcount++; | |
4944 | } | |
4945 | } else { | |
4946 | while (scan < loceol && !isDIGIT(*scan)) | |
4947 | scan++; | |
a0ed51b3 LW |
4948 | } |
4949 | break; | |
a0d0e21e LW |
4950 | default: /* Called on something of 0 width. */ |
4951 | break; /* So match right here or not at all. */ | |
4952 | } | |
a687059c | 4953 | |
a0ed51b3 LW |
4954 | if (hardcount) |
4955 | c = hardcount; | |
4956 | else | |
4957 | c = scan - PL_reginput; | |
3280af22 | 4958 | PL_reginput = scan; |
a687059c | 4959 | |
a3621e74 | 4960 | DEBUG_r({ |
e68ec53f | 4961 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 4962 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
4963 | SV * const prop = sv_newmortal(); |
4964 | regprop(prog, prop, p); | |
4965 | PerlIO_printf(Perl_debug_log, | |
be8e71aa YO |
4966 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
4967 | REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max); | |
a3621e74 | 4968 | }); |
be8e71aa | 4969 | }); |
9041c2e3 | 4970 | |
a0d0e21e | 4971 | return(c); |
a687059c LW |
4972 | } |
4973 | ||
c277df42 | 4974 | |
be8e71aa | 4975 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 4976 | /* |
ffc61ed2 JH |
4977 | - regclass_swash - prepare the utf8 swash |
4978 | */ | |
4979 | ||
4980 | SV * | |
32fc9b6a | 4981 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 4982 | { |
97aff369 | 4983 | dVAR; |
9e55ce06 JH |
4984 | SV *sw = NULL; |
4985 | SV *si = NULL; | |
4986 | SV *alt = NULL; | |
3dab1dad | 4987 | const struct reg_data * const data = prog ? prog->data : NULL; |
ffc61ed2 | 4988 | |
4f639d21 | 4989 | if (data && data->count) { |
a3b680e6 | 4990 | const U32 n = ARG(node); |
ffc61ed2 | 4991 | |
4f639d21 DM |
4992 | if (data->what[n] == 's') { |
4993 | SV * const rv = (SV*)data->data[n]; | |
890ce7af | 4994 | AV * const av = (AV*)SvRV((SV*)rv); |
2d03de9c | 4995 | SV **const ary = AvARRAY(av); |
9e55ce06 | 4996 | SV **a, **b; |
9041c2e3 | 4997 | |
711a919c | 4998 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
4999 | * documentation of these array elements. */ |
5000 | ||
b11f357e | 5001 | si = *ary; |
8f7f7219 | 5002 | a = SvROK(ary[1]) ? &ary[1] : 0; |
b11f357e JH |
5003 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0; |
5004 | ||
ffc61ed2 JH |
5005 | if (a) |
5006 | sw = *a; | |
5007 | else if (si && doinit) { | |
5008 | sw = swash_init("utf8", "", si, 1, 0); | |
5009 | (void)av_store(av, 1, sw); | |
5010 | } | |
9e55ce06 JH |
5011 | if (b) |
5012 | alt = *b; | |
ffc61ed2 JH |
5013 | } |
5014 | } | |
5015 | ||
9e55ce06 JH |
5016 | if (listsvp) |
5017 | *listsvp = si; | |
5018 | if (altsvp) | |
5019 | *altsvp = alt; | |
ffc61ed2 JH |
5020 | |
5021 | return sw; | |
5022 | } | |
76234dfb | 5023 | #endif |
ffc61ed2 JH |
5024 | |
5025 | /* | |
ba7b4546 | 5026 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
5027 | |
5028 | The n is the ANYOF regnode, the p is the target string, lenp | |
5029 | is pointer to the maximum length of how far to go in the p | |
5030 | (if the lenp is zero, UTF8SKIP(p) is used), | |
5031 | do_utf8 tells whether the target string is in UTF-8. | |
5032 | ||
bbce6d69 | 5033 | */ |
5034 | ||
76e3520e | 5035 | STATIC bool |
32fc9b6a | 5036 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 5037 | { |
27da23d5 | 5038 | dVAR; |
a3b680e6 | 5039 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 5040 | bool match = FALSE; |
cc07378b | 5041 | UV c = *p; |
ae9ddab8 | 5042 | STRLEN len = 0; |
9e55ce06 | 5043 | STRLEN plen; |
1aa99e6b | 5044 | |
19f67299 TS |
5045 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5046 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5047 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5048 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5049 | if (len == (STRLEN)-1) |
5050 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5051 | } |
bbce6d69 | 5052 | |
0f0076b4 | 5053 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5054 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5055 | if (lenp) |
5056 | *lenp = 0; | |
ffc61ed2 | 5057 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5058 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5059 | match = TRUE; | |
bbce6d69 | 5060 | } |
3568d838 | 5061 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5062 | match = TRUE; |
ffc61ed2 | 5063 | if (!match) { |
9e55ce06 | 5064 | AV *av; |
32fc9b6a | 5065 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5066 | |
5067 | if (sw) { | |
3568d838 | 5068 | if (swash_fetch(sw, p, do_utf8)) |
ffc61ed2 JH |
5069 | match = TRUE; |
5070 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5071 | if (!match && lenp && av) { |
5072 | I32 i; | |
9e55ce06 | 5073 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5074 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5075 | STRLEN len; |
890ce7af | 5076 | const char * const s = SvPV_const(sv, len); |
9e55ce06 | 5077 | |
061b10df | 5078 | if (len <= plen && memEQ(s, (char*)p, len)) { |
9e55ce06 JH |
5079 | *lenp = len; |
5080 | match = TRUE; | |
5081 | break; | |
5082 | } | |
5083 | } | |
5084 | } | |
5085 | if (!match) { | |
89ebb4a3 | 5086 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 JH |
5087 | STRLEN tmplen; |
5088 | ||
9e55ce06 JH |
5089 | to_utf8_fold(p, tmpbuf, &tmplen); |
5090 | if (swash_fetch(sw, tmpbuf, do_utf8)) | |
5091 | match = TRUE; | |
5092 | } | |
ffc61ed2 JH |
5093 | } |
5094 | } | |
bbce6d69 | 5095 | } |
9e55ce06 | 5096 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5097 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5098 | } |
1aa99e6b | 5099 | if (!match && c < 256) { |
ffc61ed2 JH |
5100 | if (ANYOF_BITMAP_TEST(n, c)) |
5101 | match = TRUE; | |
5102 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5103 | U8 f; |
a0ed51b3 | 5104 | |
ffc61ed2 JH |
5105 | if (flags & ANYOF_LOCALE) { |
5106 | PL_reg_flags |= RF_tainted; | |
5107 | f = PL_fold_locale[c]; | |
5108 | } | |
5109 | else | |
5110 | f = PL_fold[c]; | |
5111 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5112 | match = TRUE; | |
5113 | } | |
5114 | ||
5115 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5116 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5117 | if ( |
5118 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5119 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5120 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5121 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5122 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5123 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5124 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5125 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5126 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5127 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5128 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5129 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5130 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5131 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5132 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5133 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5134 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5135 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5136 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5137 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5138 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5139 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5140 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5141 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5142 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5143 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5144 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5145 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5146 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5147 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5148 | ) /* How's that for a conditional? */ | |
5149 | { | |
5150 | match = TRUE; | |
5151 | } | |
a0ed51b3 | 5152 | } |
a0ed51b3 LW |
5153 | } |
5154 | ||
a0ed51b3 LW |
5155 | return (flags & ANYOF_INVERT) ? !match : match; |
5156 | } | |
161b471a | 5157 | |
dfe13c55 | 5158 | STATIC U8 * |
0ce71af7 | 5159 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5160 | { |
97aff369 | 5161 | dVAR; |
a0ed51b3 | 5162 | if (off >= 0) { |
1aa99e6b | 5163 | while (off-- && s < lim) { |
ffc61ed2 | 5164 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5165 | s += UTF8SKIP(s); |
ffc61ed2 | 5166 | } |
a0ed51b3 LW |
5167 | } |
5168 | else { | |
1de06328 YO |
5169 | while (off++ && s > lim) { |
5170 | s--; | |
5171 | if (UTF8_IS_CONTINUED(*s)) { | |
5172 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5173 | s--; | |
a0ed51b3 | 5174 | } |
1de06328 | 5175 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5176 | } |
5177 | } | |
5178 | return s; | |
5179 | } | |
161b471a | 5180 | |
dfe13c55 | 5181 | STATIC U8 * |
1de06328 YO |
5182 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5183 | { | |
5184 | dVAR; | |
5185 | if (off >= 0) { | |
5186 | while (off-- && s < rlim) { | |
5187 | /* XXX could check well-formedness here */ | |
5188 | s += UTF8SKIP(s); | |
5189 | } | |
5190 | } | |
5191 | else { | |
5192 | while (off++ && s > llim) { | |
5193 | s--; | |
5194 | if (UTF8_IS_CONTINUED(*s)) { | |
5195 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5196 | s--; | |
5197 | } | |
5198 | /* XXX could check well-formedness here */ | |
5199 | } | |
5200 | } | |
5201 | return s; | |
5202 | } | |
5203 | ||
5204 | ||
5205 | STATIC U8 * | |
0ce71af7 | 5206 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5207 | { |
97aff369 | 5208 | dVAR; |
a0ed51b3 | 5209 | if (off >= 0) { |
1aa99e6b | 5210 | while (off-- && s < lim) { |
ffc61ed2 | 5211 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5212 | s += UTF8SKIP(s); |
ffc61ed2 | 5213 | } |
a0ed51b3 | 5214 | if (off >= 0) |
3dab1dad | 5215 | return NULL; |
a0ed51b3 LW |
5216 | } |
5217 | else { | |
1de06328 YO |
5218 | while (off++ && s > lim) { |
5219 | s--; | |
5220 | if (UTF8_IS_CONTINUED(*s)) { | |
5221 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5222 | s--; | |
a0ed51b3 | 5223 | } |
1de06328 | 5224 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5225 | } |
5226 | if (off <= 0) | |
3dab1dad | 5227 | return NULL; |
a0ed51b3 LW |
5228 | } |
5229 | return s; | |
5230 | } | |
51371543 | 5231 | |
51371543 | 5232 | static void |
acfe0abc | 5233 | restore_pos(pTHX_ void *arg) |
51371543 | 5234 | { |
97aff369 | 5235 | dVAR; |
097eb12c | 5236 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5237 | if (PL_reg_eval_set) { |
5238 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5239 | rex->subbeg = PL_reg_oldsaved; |
5240 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5241 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5242 | rex->saved_copy = PL_nrs; |
ed252734 | 5243 | #endif |
4f639d21 | 5244 | RX_MATCH_COPIED_on(rex); |
51371543 GS |
5245 | } |
5246 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
5247 | PL_reg_eval_set = 0; | |
5248 | PL_curpm = PL_reg_oldcurpm; | |
5249 | } | |
5250 | } | |
33b8afdf JH |
5251 | |
5252 | STATIC void | |
5253 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
5254 | { | |
33b8afdf | 5255 | if (prog->float_substr && !prog->float_utf8) { |
097eb12c AL |
5256 | SV* const sv = newSVsv(prog->float_substr); |
5257 | prog->float_utf8 = sv; | |
33b8afdf JH |
5258 | sv_utf8_upgrade(sv); |
5259 | if (SvTAIL(prog->float_substr)) | |
5260 | SvTAIL_on(sv); | |
5261 | if (prog->float_substr == prog->check_substr) | |
5262 | prog->check_utf8 = sv; | |
5263 | } | |
5264 | if (prog->anchored_substr && !prog->anchored_utf8) { | |
097eb12c AL |
5265 | SV* const sv = newSVsv(prog->anchored_substr); |
5266 | prog->anchored_utf8 = sv; | |
33b8afdf JH |
5267 | sv_utf8_upgrade(sv); |
5268 | if (SvTAIL(prog->anchored_substr)) | |
5269 | SvTAIL_on(sv); | |
5270 | if (prog->anchored_substr == prog->check_substr) | |
5271 | prog->check_utf8 = sv; | |
5272 | } | |
5273 | } | |
5274 | ||
5275 | STATIC void | |
5276 | S_to_byte_substr(pTHX_ register regexp *prog) | |
5277 | { | |
97aff369 | 5278 | dVAR; |
33b8afdf | 5279 | if (prog->float_utf8 && !prog->float_substr) { |
097eb12c AL |
5280 | SV* sv = newSVsv(prog->float_utf8); |
5281 | prog->float_substr = sv; | |
33b8afdf JH |
5282 | if (sv_utf8_downgrade(sv, TRUE)) { |
5283 | if (SvTAIL(prog->float_utf8)) | |
5284 | SvTAIL_on(sv); | |
5285 | } else { | |
5286 | SvREFCNT_dec(sv); | |
5287 | prog->float_substr = sv = &PL_sv_undef; | |
5288 | } | |
5289 | if (prog->float_utf8 == prog->check_utf8) | |
5290 | prog->check_substr = sv; | |
5291 | } | |
5292 | if (prog->anchored_utf8 && !prog->anchored_substr) { | |
097eb12c AL |
5293 | SV* sv = newSVsv(prog->anchored_utf8); |
5294 | prog->anchored_substr = sv; | |
33b8afdf JH |
5295 | if (sv_utf8_downgrade(sv, TRUE)) { |
5296 | if (SvTAIL(prog->anchored_utf8)) | |
5297 | SvTAIL_on(sv); | |
5298 | } else { | |
5299 | SvREFCNT_dec(sv); | |
5300 | prog->anchored_substr = sv = &PL_sv_undef; | |
5301 | } | |
5302 | if (prog->anchored_utf8 == prog->check_utf8) | |
5303 | prog->check_substr = sv; | |
5304 | } | |
5305 | } | |
66610fdd RGS |
5306 | |
5307 | /* | |
5308 | * Local variables: | |
5309 | * c-indentation-style: bsd | |
5310 | * c-basic-offset: 4 | |
5311 | * indent-tabs-mode: t | |
5312 | * End: | |
5313 | * | |
37442d52 RGS |
5314 | * ex: set ts=8 sts=4 sw=4 noet: |
5315 | */ |