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 | |
c476f425 | 153 | |
acfe0abc | 154 | static void restore_pos(pTHX_ void *arg); |
51371543 | 155 | |
76e3520e | 156 | STATIC CHECKPOINT |
cea2e8a9 | 157 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 158 | { |
97aff369 | 159 | dVAR; |
a3b680e6 | 160 | const int retval = PL_savestack_ix; |
b1ce53c5 | 161 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 162 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 163 | int p; |
40a82448 | 164 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 165 | |
e49a9654 IH |
166 | if (paren_elems_to_push < 0) |
167 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
168 | ||
6bda09f9 | 169 | #define REGCP_OTHER_ELEMS 8 |
4b3c1a47 | 170 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
3280af22 | 171 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 172 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
173 | SSPUSHINT(PL_regendp[p]); |
174 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 175 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 176 | SSPUSHINT(p); |
40a82448 DM |
177 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
178 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", | |
179 | (UV)p, (IV)PL_regstartp[p], | |
180 | (IV)(PL_reg_start_tmp[p] - PL_bostr), | |
181 | (IV)PL_regendp[p] | |
182 | )); | |
a0d0e21e | 183 | } |
b1ce53c5 | 184 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
6bda09f9 YO |
185 | SSPUSHPTR(PL_regstartp); |
186 | SSPUSHPTR(PL_regendp); | |
3280af22 NIS |
187 | SSPUSHINT(PL_regsize); |
188 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 189 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 190 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
191 | #define REGCP_FRAME_ELEMS 2 |
192 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
193 | * are needed for the regexp context stack bookkeeping. */ | |
194 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 195 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 196 | |
a0d0e21e LW |
197 | return retval; |
198 | } | |
199 | ||
c277df42 | 200 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
201 | #define REGCP_SET(cp) \ |
202 | DEBUG_STATE_r( \ | |
ab3bbdeb | 203 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 204 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
205 | (IV)PL_savestack_ix)); \ |
206 | cp = PL_savestack_ix | |
c3464db5 | 207 | |
ab3bbdeb | 208 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 209 | DEBUG_STATE_r( \ |
ab3bbdeb | 210 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
211 | PerlIO_printf(Perl_debug_log, \ |
212 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
213 | (IV)(cp), (IV)PL_savestack_ix)); \ |
214 | regcpblow(cp) | |
c277df42 | 215 | |
76e3520e | 216 | STATIC char * |
097eb12c | 217 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 218 | { |
97aff369 | 219 | dVAR; |
b1ce53c5 | 220 | I32 i; |
a0d0e21e | 221 | char *input; |
b1ce53c5 | 222 | |
a3621e74 YO |
223 | GET_RE_DEBUG_FLAGS_DECL; |
224 | ||
b1ce53c5 | 225 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 226 | i = SSPOPINT; |
b1ce53c5 JH |
227 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
228 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 229 | input = (char *) SSPOPPTR; |
a01268b5 | 230 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
231 | *PL_reglastparen = SSPOPINT; |
232 | PL_regsize = SSPOPINT; | |
6bda09f9 YO |
233 | PL_regendp=(I32 *) SSPOPPTR; |
234 | PL_regstartp=(I32 *) SSPOPPTR; | |
b1ce53c5 | 235 | |
6bda09f9 | 236 | |
b1ce53c5 | 237 | /* Now restore the parentheses context. */ |
41123dfd JH |
238 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
239 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 240 | I32 tmps; |
097eb12c | 241 | U32 paren = (U32)SSPOPINT; |
3280af22 | 242 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
243 | PL_regstartp[paren] = SSPOPINT; |
244 | tmps = SSPOPINT; | |
3280af22 NIS |
245 | if (paren <= *PL_reglastparen) |
246 | PL_regendp[paren] = tmps; | |
a3621e74 | 247 | DEBUG_EXECUTE_r( |
c3464db5 | 248 | PerlIO_printf(Perl_debug_log, |
b900a521 | 249 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 250 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 251 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 252 | (IV)PL_regendp[paren], |
3280af22 | 253 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 254 | ); |
a0d0e21e | 255 | } |
a3621e74 | 256 | DEBUG_EXECUTE_r( |
bb7a0f54 | 257 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 258 | PerlIO_printf(Perl_debug_log, |
faccc32b | 259 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 260 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
261 | } |
262 | ); | |
daf18116 | 263 | #if 1 |
dafc8851 JH |
264 | /* It would seem that the similar code in regtry() |
265 | * already takes care of this, and in fact it is in | |
266 | * a better location to since this code can #if 0-ed out | |
267 | * but the code in regtry() is needed or otherwise tests | |
268 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
269 | * (as of patchlevel 7877) will fail. Then again, |
270 | * this code seems to be necessary or otherwise | |
271 | * building DynaLoader will fail: | |
272 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
273 | * --jhi */ | |
bb7a0f54 | 274 | for (i = *PL_reglastparen + 1; (U32)i <= rex->nparens; i++) { |
097eb12c AL |
275 | if (i > PL_regsize) |
276 | PL_regstartp[i] = -1; | |
277 | PL_regendp[i] = -1; | |
a0d0e21e | 278 | } |
dafc8851 | 279 | #endif |
a0d0e21e LW |
280 | return input; |
281 | } | |
282 | ||
02db2b7b | 283 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 284 | |
a687059c | 285 | /* |
e50aee73 | 286 | * pregexec and friends |
a687059c LW |
287 | */ |
288 | ||
76234dfb | 289 | #ifndef PERL_IN_XSUB_RE |
a687059c | 290 | /* |
c277df42 | 291 | - pregexec - match a regexp against a string |
a687059c | 292 | */ |
c277df42 | 293 | I32 |
864dbfa3 | 294 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 295 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
296 | /* strend: pointer to null at end of string */ |
297 | /* strbeg: real beginning of string */ | |
298 | /* minend: end of match must be >=minend after stringarg. */ | |
299 | /* nosave: For optimizations. */ | |
300 | { | |
301 | return | |
9041c2e3 | 302 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
303 | nosave ? 0 : REXEC_COPY_STR); |
304 | } | |
76234dfb | 305 | #endif |
22e551b9 | 306 | |
9041c2e3 | 307 | /* |
cad2e5aa JH |
308 | * Need to implement the following flags for reg_anch: |
309 | * | |
310 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
311 | * USE_INTUIT_ML | |
312 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
313 | * INTUIT_AUTORITATIVE_ML | |
314 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
315 | * INTUIT_ONCE_ML | |
316 | * | |
317 | * Another flag for this function: SECOND_TIME (so that float substrs | |
318 | * with giant delta may be not rechecked). | |
319 | */ | |
320 | ||
321 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
322 | ||
3f7c398e | 323 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
324 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
325 | ||
326 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
327 | ||
6eb5f6b9 JH |
328 | /* XXXX Some places assume that there is a fixed substring. |
329 | An update may be needed if optimizer marks as "INTUITable" | |
330 | RExen without fixed substrings. Similarly, it is assumed that | |
331 | lengths of all the strings are no more than minlen, thus they | |
332 | cannot come from lookahead. | |
333 | (Or minlen should take into account lookahead.) */ | |
334 | ||
2c2d71f5 JH |
335 | /* A failure to find a constant substring means that there is no need to make |
336 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
337 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
338 | regtry() should be needed. |
339 | ||
340 | REx compiler's optimizer found 4 possible hints: | |
341 | a) Anchored substring; | |
342 | b) Fixed substring; | |
343 | c) Whether we are anchored (beginning-of-line or \G); | |
344 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 345 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
346 | string which does not contradict any of them. |
347 | */ | |
2c2d71f5 | 348 | |
6eb5f6b9 JH |
349 | /* Most of decisions we do here should have been done at compile time. |
350 | The nodes of the REx which we used for the search should have been | |
351 | deleted from the finite automaton. */ | |
352 | ||
cad2e5aa JH |
353 | char * |
354 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
355 | char *strend, U32 flags, re_scream_pos_data *data) | |
356 | { | |
97aff369 | 357 | dVAR; |
b7953727 | 358 | register I32 start_shift = 0; |
cad2e5aa | 359 | /* Should be nonnegative! */ |
b7953727 | 360 | register I32 end_shift = 0; |
2c2d71f5 JH |
361 | register char *s; |
362 | register SV *check; | |
a1933d95 | 363 | char *strbeg; |
cad2e5aa | 364 | char *t; |
1de06328 | 365 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 366 | I32 ml_anch; |
bd61b366 SS |
367 | register char *other_last = NULL; /* other substr checked before this */ |
368 | char *check_at = NULL; /* check substr found at this pos */ | |
1df70142 | 369 | const I32 multiline = prog->reganch & PMf_MULTILINE; |
30944b6d | 370 | #ifdef DEBUGGING |
890ce7af | 371 | const char * const i_strpos = strpos; |
30944b6d | 372 | #endif |
a3621e74 YO |
373 | |
374 | GET_RE_DEBUG_FLAGS_DECL; | |
375 | ||
a30b2f1f | 376 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 377 | |
b8d68ded | 378 | if (prog->reganch & ROPT_UTF8) { |
b8d68ded JH |
379 | PL_reg_flags |= RF_utf8; |
380 | } | |
ab3bbdeb YO |
381 | DEBUG_EXECUTE_r( |
382 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
383 | sv ? "Guessing start of match in sv for" |
384 | : "Guessing start of match in string for"); | |
2a782b5b | 385 | ); |
cad2e5aa | 386 | |
c344f387 JH |
387 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
388 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 389 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 390 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 391 | goto fail; |
2c2d71f5 | 392 | } |
1de06328 | 393 | |
a1933d95 | 394 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 395 | PL_regeol = strend; |
33b8afdf JH |
396 | if (do_utf8) { |
397 | if (!prog->check_utf8 && prog->check_substr) | |
398 | to_utf8_substr(prog); | |
399 | check = prog->check_utf8; | |
400 | } else { | |
401 | if (!prog->check_substr && prog->check_utf8) | |
402 | to_byte_substr(prog); | |
403 | check = prog->check_substr; | |
404 | } | |
1de06328 | 405 | if (check == &PL_sv_undef) { |
a3621e74 | 406 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 407 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
408 | goto fail; |
409 | } | |
2c2d71f5 | 410 | if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ |
cad2e5aa JH |
411 | ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) |
412 | || ( (prog->reganch & ROPT_ANCH_BOL) | |
7fba1cd6 | 413 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 414 | |
7e25d62c JH |
415 | if (!ml_anch) { |
416 | if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ | |
417 | | ROPT_IMPLICIT)) /* not a real BOL */ | |
3f7c398e | 418 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
419 | && sv && !SvROK(sv) |
420 | && (strpos != strbeg)) { | |
a3621e74 | 421 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
422 | goto fail; |
423 | } | |
424 | if (prog->check_offset_min == prog->check_offset_max && | |
cce850e4 | 425 | !(prog->reganch & ROPT_CANY_SEEN)) { |
2c2d71f5 | 426 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
427 | I32 slen; |
428 | ||
1aa99e6b | 429 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 430 | |
653099ff GS |
431 | if (SvTAIL(check)) { |
432 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 433 | |
9041c2e3 | 434 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 435 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 436 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 437 | goto fail_finish; |
cad2e5aa JH |
438 | } |
439 | /* Now should match s[0..slen-2] */ | |
440 | slen--; | |
3f7c398e | 441 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 442 | || (slen > 1 |
3f7c398e | 443 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 444 | report_neq: |
a3621e74 | 445 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
446 | goto fail_finish; |
447 | } | |
cad2e5aa | 448 | } |
3f7c398e | 449 | else if (*SvPVX_const(check) != *s |
653099ff | 450 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 451 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 452 | goto report_neq; |
c315bfe8 | 453 | check_at = s; |
2c2d71f5 | 454 | goto success_at_start; |
7e25d62c | 455 | } |
cad2e5aa | 456 | } |
2c2d71f5 | 457 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 458 | s = strpos; |
2c2d71f5 | 459 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
460 | end_shift = prog->check_end_shift; |
461 | ||
2c2d71f5 | 462 | if (!ml_anch) { |
a3b680e6 | 463 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 464 | - (SvTAIL(check) != 0); |
a3b680e6 | 465 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
466 | |
467 | if (end_shift < eshift) | |
468 | end_shift = eshift; | |
469 | } | |
cad2e5aa | 470 | } |
2c2d71f5 | 471 | else { /* Can match at random position */ |
cad2e5aa JH |
472 | ml_anch = 0; |
473 | s = strpos; | |
1de06328 YO |
474 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
475 | end_shift = prog->check_end_shift; | |
476 | ||
477 | /* end shift should be non negative here */ | |
cad2e5aa JH |
478 | } |
479 | ||
2c2d71f5 | 480 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 481 | if (end_shift < 0) |
1de06328 | 482 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 483 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
484 | #endif |
485 | ||
2c2d71f5 JH |
486 | restart: |
487 | /* Find a possible match in the region s..strend by looking for | |
488 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
489 | |
490 | { | |
491 | I32 srch_start_shift = start_shift; | |
492 | I32 srch_end_shift = end_shift; | |
493 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
494 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
495 | srch_start_shift = strbeg - s; | |
496 | } | |
6bda09f9 | 497 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
498 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
499 | (IV)prog->check_offset_min, | |
500 | (IV)srch_start_shift, | |
501 | (IV)srch_end_shift, | |
502 | (IV)prog->check_end_shift); | |
503 | }); | |
504 | ||
cad2e5aa | 505 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 506 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 507 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 508 | |
2c2d71f5 JH |
509 | if (PL_screamfirst[BmRARE(check)] >= 0 |
510 | || ( BmRARE(check) == '\n' | |
511 | && (BmPREVIOUS(check) == SvCUR(check) - 1) | |
512 | && SvTAIL(check) )) | |
9041c2e3 | 513 | s = screaminstr(sv, check, |
1de06328 | 514 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 515 | else |
2c2d71f5 | 516 | goto fail_finish; |
4addbd3b HS |
517 | /* we may be pointing at the wrong string */ |
518 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 519 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
520 | if (data) |
521 | *data->scream_olds = s; | |
522 | } | |
1de06328 YO |
523 | else { |
524 | U8* start_point; | |
525 | U8* end_point; | |
526 | if (prog->reganch & ROPT_CANY_SEEN) { | |
527 | start_point= (U8*)(s + srch_start_shift); | |
528 | end_point= (U8*)(strend - srch_end_shift); | |
529 | } else { | |
530 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
531 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
532 | } | |
6bda09f9 | 533 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 534 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 535 | (int)(end_point - start_point), |
fc8cd66c | 536 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
537 | start_point); |
538 | }); | |
539 | ||
540 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 541 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
542 | } |
543 | } | |
cad2e5aa JH |
544 | /* Update the count-of-usability, remove useless subpatterns, |
545 | unshift s. */ | |
2c2d71f5 | 546 | |
ab3bbdeb YO |
547 | DEBUG_EXECUTE_r({ |
548 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
549 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
550 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 551 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
552 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
553 | ? "anchored" : "floating"), | |
554 | quoted, | |
555 | RE_SV_TAIL(check), | |
556 | (s ? " at offset " : "...\n") ); | |
557 | }); | |
2c2d71f5 JH |
558 | |
559 | if (!s) | |
560 | goto fail_finish; | |
2c2d71f5 | 561 | /* Finish the diagnostic message */ |
a3621e74 | 562 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 563 | |
1de06328 YO |
564 | /* XXX dmq: first branch is for positive lookbehind... |
565 | Our check string is offset from the beginning of the pattern. | |
566 | So we need to do any stclass tests offset forward from that | |
567 | point. I think. :-( | |
568 | */ | |
569 | ||
570 | ||
571 | ||
572 | check_at=s; | |
573 | ||
574 | ||
2c2d71f5 JH |
575 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
576 | Start with the other substr. | |
577 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 578 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
579 | *always* match. Probably should be marked during compile... |
580 | Probably it is right to do no SCREAM here... | |
581 | */ | |
582 | ||
1de06328 YO |
583 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
584 | : (prog->float_substr && prog->anchored_substr)) | |
585 | { | |
30944b6d | 586 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
587 | /* XXXX May be hopelessly wrong for UTF... */ |
588 | if (!other_last) | |
6eb5f6b9 | 589 | other_last = strpos; |
33b8afdf | 590 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
591 | do_other_anchored: |
592 | { | |
890ce7af AL |
593 | char * const last = HOP3c(s, -start_shift, strbeg); |
594 | char *last1, *last2; | |
be8e71aa | 595 | char * const saved_s = s; |
33b8afdf | 596 | SV* must; |
2c2d71f5 | 597 | |
2c2d71f5 JH |
598 | t = s - prog->check_offset_max; |
599 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 600 | && (!do_utf8 |
0ce71af7 | 601 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 602 | && t > strpos))) |
6f207bd3 | 603 | NOOP; |
2c2d71f5 JH |
604 | else |
605 | t = strpos; | |
1aa99e6b | 606 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
607 | if (t < other_last) /* These positions already checked */ |
608 | t = other_last; | |
1aa99e6b | 609 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
610 | if (last < last1) |
611 | last1 = last; | |
1de06328 YO |
612 | /* XXXX It is not documented what units *_offsets are in. |
613 | We assume bytes, but this is clearly wrong. | |
614 | Meaning this code needs to be carefully reviewed for errors. | |
615 | dmq. | |
616 | */ | |
617 | ||
2c2d71f5 | 618 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
619 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
620 | if (must == &PL_sv_undef) { | |
621 | s = (char*)NULL; | |
1de06328 | 622 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
623 | } |
624 | else | |
625 | s = fbm_instr( | |
626 | (unsigned char*)t, | |
627 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
628 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
629 | must, | |
7fba1cd6 | 630 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 631 | ); |
ab3bbdeb YO |
632 | DEBUG_EXECUTE_r({ |
633 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
634 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
635 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 636 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
637 | quoted, RE_SV_TAIL(must)); |
638 | }); | |
639 | ||
640 | ||
2c2d71f5 JH |
641 | if (!s) { |
642 | if (last1 >= last2) { | |
a3621e74 | 643 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
644 | ", giving up...\n")); |
645 | goto fail_finish; | |
646 | } | |
a3621e74 | 647 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 648 | ", trying floating at offset %ld...\n", |
be8e71aa | 649 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
650 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
651 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
652 | goto restart; |
653 | } | |
654 | else { | |
a3621e74 | 655 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 656 | (long)(s - i_strpos))); |
1aa99e6b IH |
657 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
658 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 659 | s = saved_s; |
2c2d71f5 JH |
660 | if (t == strpos) |
661 | goto try_at_start; | |
2c2d71f5 JH |
662 | goto try_at_offset; |
663 | } | |
30944b6d | 664 | } |
2c2d71f5 JH |
665 | } |
666 | else { /* Take into account the floating substring. */ | |
33b8afdf | 667 | char *last, *last1; |
be8e71aa | 668 | char * const saved_s = s; |
33b8afdf JH |
669 | SV* must; |
670 | ||
671 | t = HOP3c(s, -start_shift, strbeg); | |
672 | last1 = last = | |
673 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
674 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
675 | last = HOP3c(t, prog->float_max_offset, strend); | |
676 | s = HOP3c(t, prog->float_min_offset, strend); | |
677 | if (s < other_last) | |
678 | s = other_last; | |
2c2d71f5 | 679 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
680 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
681 | /* fbm_instr() takes into account exact value of end-of-str | |
682 | if the check is SvTAIL(ed). Since false positives are OK, | |
683 | and end-of-str is not later than strend we are OK. */ | |
684 | if (must == &PL_sv_undef) { | |
685 | s = (char*)NULL; | |
1de06328 | 686 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
687 | } |
688 | else | |
2c2d71f5 | 689 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
690 | (unsigned char*)last + SvCUR(must) |
691 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 692 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
693 | DEBUG_EXECUTE_r({ |
694 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
695 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
696 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 697 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
698 | quoted, RE_SV_TAIL(must)); |
699 | }); | |
33b8afdf JH |
700 | if (!s) { |
701 | if (last1 == last) { | |
a3621e74 | 702 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
703 | ", giving up...\n")); |
704 | goto fail_finish; | |
2c2d71f5 | 705 | } |
a3621e74 | 706 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 707 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 708 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
709 | other_last = last; |
710 | s = HOP3c(t, 1, strend); | |
711 | goto restart; | |
712 | } | |
713 | else { | |
a3621e74 | 714 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
715 | (long)(s - i_strpos))); |
716 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 717 | s = saved_s; |
33b8afdf JH |
718 | if (t == strpos) |
719 | goto try_at_start; | |
720 | goto try_at_offset; | |
721 | } | |
2c2d71f5 | 722 | } |
cad2e5aa | 723 | } |
2c2d71f5 | 724 | |
1de06328 | 725 | |
9ef43ace | 726 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 727 | |
6bda09f9 | 728 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
729 | PerlIO_printf(Perl_debug_log, |
730 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
731 | (IV)prog->check_offset_min, | |
732 | (IV)prog->check_offset_max, | |
733 | (IV)(s-strpos), | |
734 | (IV)(t-strpos), | |
735 | (IV)(t-s), | |
736 | (IV)(strend-strpos) | |
737 | ) | |
738 | ); | |
739 | ||
2c2d71f5 | 740 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 741 | && (!do_utf8 |
9ef43ace | 742 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
743 | && t > strpos))) |
744 | { | |
2c2d71f5 JH |
745 | /* Fixed substring is found far enough so that the match |
746 | cannot start at strpos. */ | |
747 | try_at_offset: | |
cad2e5aa | 748 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
749 | /* Eventually fbm_*() should handle this, but often |
750 | anchored_offset is not 0, so this check will not be wasted. */ | |
751 | /* XXXX In the code below we prefer to look for "^" even in | |
752 | presence of anchored substrings. And we search even | |
753 | beyond the found float position. These pessimizations | |
754 | are historical artefacts only. */ | |
755 | find_anchor: | |
2c2d71f5 | 756 | while (t < strend - prog->minlen) { |
cad2e5aa | 757 | if (*t == '\n') { |
4ee3650e | 758 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 759 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
760 | /* Since we moved from the found position, |
761 | we definitely contradict the found anchored | |
30944b6d IZ |
762 | substr. Due to the above check we do not |
763 | contradict "check" substr. | |
764 | Thus we can arrive here only if check substr | |
765 | is float. Redo checking for "other"=="fixed". | |
766 | */ | |
9041c2e3 | 767 | strpos = t + 1; |
a3621e74 | 768 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 769 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
770 | goto do_other_anchored; |
771 | } | |
4ee3650e GS |
772 | /* We don't contradict the found floating substring. */ |
773 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 774 | s = t + 1; |
a3621e74 | 775 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 776 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
777 | goto set_useful; |
778 | } | |
4ee3650e GS |
779 | /* Position contradicts check-string */ |
780 | /* XXXX probably better to look for check-string | |
781 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 782 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 783 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 784 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
785 | goto restart; |
786 | } | |
787 | t++; | |
788 | } | |
a3621e74 | 789 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 790 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 791 | goto fail_finish; |
cad2e5aa | 792 | } |
f5952150 | 793 | else { |
a3621e74 | 794 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 795 | PL_colors[0], PL_colors[1])); |
f5952150 | 796 | } |
cad2e5aa JH |
797 | s = t; |
798 | set_useful: | |
33b8afdf | 799 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
800 | } |
801 | else { | |
f5952150 | 802 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 803 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
804 | unless it was an MBOL and we are not after MBOL, |
805 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
806 | try_at_start: |
807 | /* Even in this situation we may use MBOL flag if strpos is offset | |
808 | wrt the start of the string. */ | |
05b4157f | 809 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 810 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d IZ |
811 | /* May be due to an implicit anchor of m{.*foo} */ |
812 | && !(prog->reganch & ROPT_IMPLICIT)) | |
813 | { | |
cad2e5aa JH |
814 | t = strpos; |
815 | goto find_anchor; | |
816 | } | |
a3621e74 | 817 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 818 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
e4584336 | 819 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 820 | ); |
2c2d71f5 | 821 | success_at_start: |
30944b6d | 822 | if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
823 | && (do_utf8 ? ( |
824 | prog->check_utf8 /* Could be deleted already */ | |
825 | && --BmUSEFUL(prog->check_utf8) < 0 | |
826 | && (prog->check_utf8 == prog->float_utf8) | |
827 | ) : ( | |
828 | prog->check_substr /* Could be deleted already */ | |
829 | && --BmUSEFUL(prog->check_substr) < 0 | |
830 | && (prog->check_substr == prog->float_substr) | |
831 | ))) | |
66e933ab | 832 | { |
cad2e5aa | 833 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 834 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
835 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
836 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
837 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
838 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
839 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
840 | check = NULL; /* abort */ | |
cad2e5aa | 841 | s = strpos; |
3cf5c195 IZ |
842 | /* XXXX This is a remnant of the old implementation. It |
843 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 844 | other heuristics. */ |
cad2e5aa JH |
845 | prog->reganch &= ~RE_USE_INTUIT; |
846 | } | |
847 | else | |
848 | s = strpos; | |
849 | } | |
850 | ||
6eb5f6b9 JH |
851 | /* Last resort... */ |
852 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
853 | /* trie stclasses are too expensive to use here, we are better off to |
854 | leave it to regmatch itself */ | |
786e8c11 | 855 | if (prog->regstclass && PL_regkind[OP(prog->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
856 | /* minlen == 0 is possible if regstclass is \b or \B, |
857 | and the fixed substr is ''$. | |
858 | Since minlen is already taken into account, s+1 is before strend; | |
859 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
860 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
861 | regstclass does not come from lookahead... */ | |
862 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
863 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
890ce7af | 864 | const U8* const str = (U8*)STRING(prog->regstclass); |
3dab1dad | 865 | const int cl_l = (PL_regkind[OP(prog->regstclass)] == EXACT |
1aa99e6b | 866 | ? CHR_DIST(str+STR_LEN(prog->regstclass), str) |
66e933ab | 867 | : 1); |
1de06328 YO |
868 | char * endpos; |
869 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
870 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
871 | else if (prog->float_substr || prog->float_utf8) | |
872 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
873 | else | |
874 | endpos= strend; | |
875 | ||
56570a2c | 876 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %d s: %d endpos: %d\n", |
b57a0404 | 877 | (IV)start_shift, check_at - strbeg, s - strbeg, endpos - strbeg)); |
1de06328 | 878 | |
6eb5f6b9 | 879 | t = s; |
3b0527fe | 880 | s = find_byclass(prog, prog->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
881 | if (!s) { |
882 | #ifdef DEBUGGING | |
cbbf8932 | 883 | const char *what = NULL; |
6eb5f6b9 JH |
884 | #endif |
885 | if (endpos == strend) { | |
a3621e74 | 886 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
887 | "Could not match STCLASS...\n") ); |
888 | goto fail; | |
889 | } | |
a3621e74 | 890 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 891 | "This position contradicts STCLASS...\n") ); |
653099ff GS |
892 | if ((prog->reganch & ROPT_ANCH) && !ml_anch) |
893 | goto fail; | |
6eb5f6b9 | 894 | /* Contradict one of substrings */ |
33b8afdf JH |
895 | if (prog->anchored_substr || prog->anchored_utf8) { |
896 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 897 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 898 | hop_and_restart: |
1aa99e6b | 899 | s = HOP3c(t, 1, strend); |
66e933ab GS |
900 | if (s + start_shift + end_shift > strend) { |
901 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 902 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
903 | "Could not match STCLASS...\n") ); |
904 | goto fail; | |
905 | } | |
5e39e1e5 HS |
906 | if (!check) |
907 | goto giveup; | |
a3621e74 | 908 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 909 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
910 | what, (long)(s + start_shift - i_strpos)) ); |
911 | goto restart; | |
912 | } | |
66e933ab | 913 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
914 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
915 | goto retry_floating_check; | |
916 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 917 | s = check_at; |
5e39e1e5 HS |
918 | if (!check) |
919 | goto giveup; | |
a3621e74 | 920 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 921 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
922 | (long)(other_last - i_strpos)) ); |
923 | goto do_other_anchored; | |
924 | } | |
60e71179 GS |
925 | /* Another way we could have checked stclass at the |
926 | current position only: */ | |
927 | if (ml_anch) { | |
928 | s = t = t + 1; | |
5e39e1e5 HS |
929 | if (!check) |
930 | goto giveup; | |
a3621e74 | 931 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 932 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 933 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 934 | goto try_at_offset; |
66e933ab | 935 | } |
33b8afdf | 936 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 937 | goto fail; |
6eb5f6b9 JH |
938 | /* Check is floating subtring. */ |
939 | retry_floating_check: | |
940 | t = check_at - start_shift; | |
a3621e74 | 941 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
942 | goto hop_and_restart; |
943 | } | |
b7953727 | 944 | if (t != s) { |
a3621e74 | 945 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 946 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
947 | (long)(t - i_strpos), (long)(s - i_strpos)) |
948 | ); | |
949 | } | |
950 | else { | |
a3621e74 | 951 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
952 | "Does not contradict STCLASS...\n"); |
953 | ); | |
954 | } | |
6eb5f6b9 | 955 | } |
5e39e1e5 | 956 | giveup: |
a3621e74 | 957 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
958 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
959 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 960 | return s; |
2c2d71f5 JH |
961 | |
962 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
963 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
964 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 965 | fail: |
a3621e74 | 966 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 967 | PL_colors[4], PL_colors[5])); |
bd61b366 | 968 | return NULL; |
cad2e5aa | 969 | } |
9661b544 | 970 | |
786e8c11 | 971 | |
3b0527fe | 972 | |
4cadc6a9 YO |
973 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, uvc, charid, \ |
974 | foldlen, foldbuf, uniflags) STMT_START { \ | |
975 | switch (trie_type) { \ | |
976 | case trie_utf8_fold: \ | |
977 | if ( foldlen>0 ) { \ | |
978 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
979 | foldlen -= len; \ | |
980 | uscan += len; \ | |
981 | len=0; \ | |
982 | } else { \ | |
983 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
984 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
985 | foldlen -= UNISKIP( uvc ); \ | |
986 | uscan = foldbuf + UNISKIP( uvc ); \ | |
987 | } \ | |
988 | break; \ | |
989 | case trie_utf8: \ | |
990 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
991 | break; \ | |
992 | case trie_plain: \ | |
993 | uvc = (UV)*uc; \ | |
994 | len = 1; \ | |
995 | } \ | |
996 | \ | |
997 | if (uvc < 256) { \ | |
998 | charid = trie->charmap[ uvc ]; \ | |
999 | } \ | |
1000 | else { \ | |
1001 | charid = 0; \ | |
1002 | if (trie->widecharmap) { \ | |
1003 | SV** const svpp = hv_fetch(trie->widecharmap, \ | |
1004 | (char*)&uvc, sizeof(UV), 0); \ | |
1005 | if (svpp) \ | |
1006 | charid = (U16)SvIV(*svpp); \ | |
1007 | } \ | |
1008 | } \ | |
1009 | } STMT_END | |
1010 | ||
1011 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1012 | if ( (CoNd) \ | |
1013 | && (ln == len || \ | |
1014 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1015 | m, NULL, ln, (bool)UTF)) \ | |
1016 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1017 | goto got_it; \ | |
1018 | else { \ | |
1019 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1020 | uvchr_to_utf8(tmpbuf, c); \ | |
1021 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1022 | if ( f != c \ | |
1023 | && (f == c1 || f == c2) \ | |
1024 | && (ln == foldlen || \ | |
1025 | !ibcmp_utf8((char *) foldbuf, \ | |
1026 | NULL, foldlen, do_utf8, \ | |
1027 | m, \ | |
1028 | NULL, ln, (bool)UTF)) \ | |
1029 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1030 | goto got_it; \ | |
1031 | } \ | |
1032 | s += len | |
1033 | ||
1034 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1035 | STMT_START { \ | |
1036 | while (s <= e) { \ | |
1037 | if ( (CoNd) \ | |
1038 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1039 | ? ibcmp(s, m, ln) \ | |
1040 | : ibcmp_locale(s, m, ln))) \ | |
1041 | && (!reginfo || regtry(reginfo, s)) ) \ | |
1042 | goto got_it; \ | |
1043 | s++; \ | |
1044 | } \ | |
1045 | } STMT_END | |
1046 | ||
1047 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1048 | STMT_START { \ | |
1049 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1050 | CoDe \ | |
1051 | s += uskip; \ | |
1052 | } \ | |
1053 | } STMT_END | |
1054 | ||
1055 | #define REXEC_FBC_SCAN(CoDe) \ | |
1056 | STMT_START { \ | |
1057 | while (s < strend) { \ | |
1058 | CoDe \ | |
1059 | s++; \ | |
1060 | } \ | |
1061 | } STMT_END | |
1062 | ||
1063 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1064 | REXEC_FBC_UTF8_SCAN( \ | |
1065 | if (CoNd) { \ | |
1066 | if (tmp && (!reginfo || regtry(reginfo, s))) \ | |
1067 | goto got_it; \ | |
1068 | else \ | |
1069 | tmp = doevery; \ | |
1070 | } \ | |
1071 | else \ | |
1072 | tmp = 1; \ | |
1073 | ) | |
1074 | ||
1075 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1076 | REXEC_FBC_SCAN( \ | |
1077 | if (CoNd) { \ | |
1078 | if (tmp && (!reginfo || regtry(reginfo, s))) \ | |
1079 | goto got_it; \ | |
1080 | else \ | |
1081 | tmp = doevery; \ | |
1082 | } \ | |
1083 | else \ | |
1084 | tmp = 1; \ | |
1085 | ) | |
1086 | ||
1087 | #define REXEC_FBC_TRYIT \ | |
1088 | if ((!reginfo || regtry(reginfo, s))) \ | |
1089 | goto got_it | |
1090 | ||
1091 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ | |
1092 | if (do_utf8) { \ | |
1093 | UtFpReLoAd; \ | |
1094 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1095 | } \ | |
1096 | else { \ | |
1097 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1098 | } \ | |
1099 | break | |
1100 | ||
1101 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1102 | PL_reg_flags |= RF_tainted; \ | |
1103 | if (do_utf8) { \ | |
1104 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1105 | } \ | |
1106 | else { \ | |
1107 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1108 | } \ | |
1109 | break | |
1110 | ||
786e8c11 YO |
1111 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1112 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1113 | ||
1114 | /* We know what class REx starts with. Try to find this position... */ | |
1115 | /* if reginfo is NULL, its a dryrun */ | |
1116 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1117 | in regmatch. /grrr */ | |
1118 | ||
3c3eec57 | 1119 | STATIC char * |
07be1b83 YO |
1120 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
1121 | const char *strend, const regmatch_info *reginfo) | |
a687059c | 1122 | { |
27da23d5 | 1123 | dVAR; |
1df70142 | 1124 | const I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
6eb5f6b9 | 1125 | char *m; |
d8093b23 | 1126 | STRLEN ln; |
5dab1207 | 1127 | STRLEN lnc; |
078c425b | 1128 | register STRLEN uskip; |
d8093b23 G |
1129 | unsigned int c1; |
1130 | unsigned int c2; | |
6eb5f6b9 JH |
1131 | char *e; |
1132 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1133 | register const bool do_utf8 = PL_reg_match_utf8; |
cad2e5aa | 1134 | |
6eb5f6b9 JH |
1135 | /* We know what class it must start with. */ |
1136 | switch (OP(c)) { | |
6eb5f6b9 | 1137 | case ANYOF: |
388cc4de | 1138 | if (do_utf8) { |
4cadc6a9 | 1139 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1140 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1141 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1142 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1143 | } |
1144 | else { | |
1145 | while (s < strend) { | |
1146 | STRLEN skip = 1; | |
1147 | ||
32fc9b6a | 1148 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1149 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1150 | /* The assignment of 2 is intentional: | |
1151 | * for the folded sharp s, the skip is 2. */ | |
1152 | (skip = SHARP_S_SKIP))) { | |
3b0527fe | 1153 | if (tmp && (!reginfo || regtry(reginfo, s))) |
388cc4de HS |
1154 | goto got_it; |
1155 | else | |
1156 | tmp = doevery; | |
1157 | } | |
1158 | else | |
1159 | tmp = 1; | |
1160 | s += skip; | |
1161 | } | |
a0d0e21e | 1162 | } |
6eb5f6b9 | 1163 | break; |
f33976b4 | 1164 | case CANY: |
4cadc6a9 | 1165 | REXEC_FBC_SCAN( |
3b0527fe | 1166 | if (tmp && (!reginfo || regtry(reginfo, s))) |
f33976b4 DB |
1167 | goto got_it; |
1168 | else | |
1169 | tmp = doevery; | |
4cadc6a9 | 1170 | ); |
f33976b4 | 1171 | break; |
6eb5f6b9 | 1172 | case EXACTF: |
5dab1207 NIS |
1173 | m = STRING(c); |
1174 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1175 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1176 | if (UTF) { |
a2a2844f | 1177 | STRLEN ulen1, ulen2; |
5dab1207 | 1178 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1179 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1180 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1181 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1182 | |
1183 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1184 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1185 | ||
89ebb4a3 | 1186 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1187 | 0, uniflags); |
89ebb4a3 | 1188 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1189 | 0, uniflags); |
5dab1207 NIS |
1190 | lnc = 0; |
1191 | while (sm < ((U8 *) m + ln)) { | |
1192 | lnc++; | |
1193 | sm += UTF8SKIP(sm); | |
1194 | } | |
1aa99e6b IH |
1195 | } |
1196 | else { | |
1197 | c1 = *(U8*)m; | |
1198 | c2 = PL_fold[c1]; | |
1199 | } | |
6eb5f6b9 JH |
1200 | goto do_exactf; |
1201 | case EXACTFL: | |
5dab1207 NIS |
1202 | m = STRING(c); |
1203 | ln = STR_LEN(c); | |
1204 | lnc = (I32) ln; | |
d8093b23 | 1205 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1206 | c2 = PL_fold_locale[c1]; |
1207 | do_exactf: | |
db12adc6 | 1208 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1209 | |
3b0527fe | 1210 | if (!reginfo && e < s) |
6eb5f6b9 | 1211 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1212 | |
60a8b682 JH |
1213 | /* The idea in the EXACTF* cases is to first find the |
1214 | * first character of the EXACTF* node and then, if | |
1215 | * necessary, case-insensitively compare the full | |
1216 | * text of the node. The c1 and c2 are the first | |
1217 | * characters (though in Unicode it gets a bit | |
1218 | * more complicated because there are more cases | |
7f16dd3d JH |
1219 | * than just upper and lower: one needs to use |
1220 | * the so-called folding case for case-insensitive | |
1221 | * matching (called "loose matching" in Unicode). | |
1222 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1223 | |
1aa99e6b | 1224 | if (do_utf8) { |
575cac57 | 1225 | UV c, f; |
89ebb4a3 | 1226 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1227 | STRLEN len, foldlen; |
4ad0818d | 1228 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1229 | if (c1 == c2) { |
5dab1207 NIS |
1230 | /* Upper and lower of 1st char are equal - |
1231 | * probably not a "letter". */ | |
1aa99e6b | 1232 | while (s <= e) { |
89ebb4a3 | 1233 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1234 | uniflags); |
4cadc6a9 | 1235 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1236 | } |
09091399 JH |
1237 | } |
1238 | else { | |
1aa99e6b | 1239 | while (s <= e) { |
89ebb4a3 | 1240 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1241 | uniflags); |
80aecb99 | 1242 | |
60a8b682 | 1243 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1244 | * Note that not all the possible combinations |
1245 | * are handled here: some of them are handled | |
1246 | * by the standard folding rules, and some of | |
1247 | * them (the character class or ANYOF cases) | |
1248 | * are handled during compiletime in | |
1249 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1250 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1251 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1252 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1253 | |
4cadc6a9 | 1254 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1255 | } |
09091399 | 1256 | } |
1aa99e6b IH |
1257 | } |
1258 | else { | |
1259 | if (c1 == c2) | |
4cadc6a9 | 1260 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1261 | else |
4cadc6a9 | 1262 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1263 | } |
1264 | break; | |
bbce6d69 | 1265 | case BOUNDL: |
3280af22 | 1266 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1267 | /* FALL THROUGH */ |
a0d0e21e | 1268 | case BOUND: |
ffc61ed2 | 1269 | if (do_utf8) { |
12d33761 | 1270 | if (s == PL_bostr) |
ffc61ed2 JH |
1271 | tmp = '\n'; |
1272 | else { | |
6136c704 | 1273 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1274 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1275 | } |
1276 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1277 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1278 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1279 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1280 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1281 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1282 | isALNUM_LC_utf8((U8*)s))) |
1283 | { | |
1284 | tmp = !tmp; | |
4cadc6a9 | 1285 | REXEC_FBC_TRYIT; |
a687059c | 1286 | } |
4cadc6a9 | 1287 | ); |
a0d0e21e | 1288 | } |
667bb95a | 1289 | else { |
12d33761 | 1290 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1291 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1292 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1293 | if (tmp == |
1294 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1295 | tmp = !tmp; | |
4cadc6a9 | 1296 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1297 | } |
4cadc6a9 | 1298 | ); |
a0ed51b3 | 1299 | } |
3b0527fe | 1300 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1301 | goto got_it; |
1302 | break; | |
bbce6d69 | 1303 | case NBOUNDL: |
3280af22 | 1304 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1305 | /* FALL THROUGH */ |
a0d0e21e | 1306 | case NBOUND: |
ffc61ed2 | 1307 | if (do_utf8) { |
12d33761 | 1308 | if (s == PL_bostr) |
ffc61ed2 JH |
1309 | tmp = '\n'; |
1310 | else { | |
6136c704 | 1311 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1312 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1313 | } |
1314 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1315 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1316 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1317 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1318 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1319 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1320 | isALNUM_LC_utf8((U8*)s))) |
1321 | tmp = !tmp; | |
4cadc6a9 YO |
1322 | else REXEC_FBC_TRYIT; |
1323 | ); | |
a0d0e21e | 1324 | } |
667bb95a | 1325 | else { |
12d33761 | 1326 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1327 | tmp = ((OP(c) == NBOUND ? |
1328 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1329 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1330 | if (tmp == |
1331 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1332 | tmp = !tmp; | |
4cadc6a9 YO |
1333 | else REXEC_FBC_TRYIT; |
1334 | ); | |
a0ed51b3 | 1335 | } |
3b0527fe | 1336 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1337 | goto got_it; |
1338 | break; | |
a0d0e21e | 1339 | case ALNUM: |
4cadc6a9 YO |
1340 | REXEC_FBC_CSCAN_PRELOAD( |
1341 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1342 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1343 | isALNUM(*s) | |
1344 | ); | |
bbce6d69 | 1345 | case ALNUML: |
4cadc6a9 YO |
1346 | REXEC_FBC_CSCAN_TAINT( |
1347 | isALNUM_LC_utf8((U8*)s), | |
1348 | isALNUM_LC(*s) | |
1349 | ); | |
a0d0e21e | 1350 | case NALNUM: |
4cadc6a9 YO |
1351 | REXEC_FBC_CSCAN_PRELOAD( |
1352 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1353 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1354 | !isALNUM(*s) | |
1355 | ); | |
bbce6d69 | 1356 | case NALNUML: |
4cadc6a9 YO |
1357 | REXEC_FBC_CSCAN_TAINT( |
1358 | !isALNUM_LC_utf8((U8*)s), | |
1359 | !isALNUM_LC(*s) | |
1360 | ); | |
a0d0e21e | 1361 | case SPACE: |
4cadc6a9 YO |
1362 | REXEC_FBC_CSCAN_PRELOAD( |
1363 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1364 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1365 | isSPACE(*s) | |
1366 | ); | |
bbce6d69 | 1367 | case SPACEL: |
4cadc6a9 YO |
1368 | REXEC_FBC_CSCAN_TAINT( |
1369 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1370 | isSPACE_LC(*s) | |
1371 | ); | |
a0d0e21e | 1372 | case NSPACE: |
4cadc6a9 YO |
1373 | REXEC_FBC_CSCAN_PRELOAD( |
1374 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1375 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1376 | !isSPACE(*s) | |
1377 | ); | |
bbce6d69 | 1378 | case NSPACEL: |
4cadc6a9 YO |
1379 | REXEC_FBC_CSCAN_TAINT( |
1380 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1381 | !isSPACE_LC(*s) | |
1382 | ); | |
a0d0e21e | 1383 | case DIGIT: |
4cadc6a9 YO |
1384 | REXEC_FBC_CSCAN_PRELOAD( |
1385 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1386 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1387 | isDIGIT(*s) | |
1388 | ); | |
b8c5462f | 1389 | case DIGITL: |
4cadc6a9 YO |
1390 | REXEC_FBC_CSCAN_TAINT( |
1391 | isDIGIT_LC_utf8((U8*)s), | |
1392 | isDIGIT_LC(*s) | |
1393 | ); | |
a0d0e21e | 1394 | case NDIGIT: |
4cadc6a9 YO |
1395 | REXEC_FBC_CSCAN_PRELOAD( |
1396 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1397 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1398 | !isDIGIT(*s) | |
1399 | ); | |
b8c5462f | 1400 | case NDIGITL: |
4cadc6a9 YO |
1401 | REXEC_FBC_CSCAN_TAINT( |
1402 | !isDIGIT_LC_utf8((U8*)s), | |
1403 | !isDIGIT_LC(*s) | |
1404 | ); | |
1de06328 YO |
1405 | case AHOCORASICKC: |
1406 | case AHOCORASICK: | |
07be1b83 YO |
1407 | { |
1408 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1409 | trie_type = do_utf8 ? | |
1410 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1411 | : trie_plain; | |
1412 | /* what trie are we using right now */ | |
1413 | reg_ac_data *aho | |
1414 | = (reg_ac_data*)prog->data->data[ ARG( c ) ]; | |
1415 | reg_trie_data *trie=aho->trie; | |
1416 | ||
1417 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1418 | #ifdef DEBUGGING |
07be1b83 | 1419 | const char *real_start = s; |
6148ee25 | 1420 | #endif |
07be1b83 | 1421 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1422 | SV *sv_points; |
1423 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1424 | when reading a given char. For ASCII this |
be8e71aa YO |
1425 | is unnecessary overhead as the relationship |
1426 | is always 1:1, but for unicode, especially | |
1427 | case folded unicode this is not true. */ | |
f9e705e8 | 1428 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1429 | U8 *bitmap=NULL; |
1430 | ||
07be1b83 YO |
1431 | |
1432 | GET_RE_DEBUG_FLAGS_DECL; | |
1433 | ||
be8e71aa YO |
1434 | /* We can't just allocate points here. We need to wrap it in |
1435 | * an SV so it gets freed properly if there is a croak while | |
1436 | * running the match */ | |
1437 | ENTER; | |
1438 | SAVETMPS; | |
1439 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1440 | SvCUR_set(sv_points, | |
1441 | maxlen * sizeof(U8 *)); | |
1442 | SvPOK_on(sv_points); | |
1443 | sv_2mortal(sv_points); | |
1444 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1445 | if ( trie_type != trie_utf8_fold |
1446 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1447 | { | |
786e8c11 YO |
1448 | if (trie->bitmap) |
1449 | bitmap=(U8*)trie->bitmap; | |
1450 | else | |
1451 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1452 | } |
786e8c11 YO |
1453 | /* this is the Aho-Corasick algorithm modified a touch |
1454 | to include special handling for long "unknown char" | |
1455 | sequences. The basic idea being that we use AC as long | |
1456 | as we are dealing with a possible matching char, when | |
1457 | we encounter an unknown char (and we have not encountered | |
1458 | an accepting state) we scan forward until we find a legal | |
1459 | starting char. | |
1460 | AC matching is basically that of trie matching, except | |
1461 | that when we encounter a failing transition, we fall back | |
1462 | to the current states "fail state", and try the current char | |
1463 | again, a process we repeat until we reach the root state, | |
1464 | state 1, or a legal transition. If we fail on the root state | |
1465 | then we can either terminate if we have reached an accepting | |
1466 | state previously, or restart the entire process from the beginning | |
1467 | if we have not. | |
1468 | ||
1469 | */ | |
07be1b83 YO |
1470 | while (s <= last_start) { |
1471 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1472 | U8 *uc = (U8*)s; | |
1473 | U16 charid = 0; | |
1474 | U32 base = 1; | |
1475 | U32 state = 1; | |
1476 | UV uvc = 0; | |
1477 | STRLEN len = 0; | |
1478 | STRLEN foldlen = 0; | |
1479 | U8 *uscan = (U8*)NULL; | |
1480 | U8 *leftmost = NULL; | |
786e8c11 YO |
1481 | #ifdef DEBUGGING |
1482 | U32 accepted_word= 0; | |
1483 | #endif | |
07be1b83 YO |
1484 | U32 pointpos = 0; |
1485 | ||
1486 | while ( state && uc <= (U8*)strend ) { | |
1487 | int failed=0; | |
786e8c11 YO |
1488 | U32 word = aho->states[ state ].wordnum; |
1489 | ||
1de06328 YO |
1490 | if( state==1 ) { |
1491 | if ( bitmap ) { | |
1492 | DEBUG_TRIE_EXECUTE_r( | |
1493 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1494 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1495 | (char *)uc, do_utf8 ); | |
1496 | PerlIO_printf( Perl_debug_log, | |
1497 | " Scanning for legal start char...\n"); | |
1498 | } | |
1499 | ); | |
1500 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1501 | uc++; | |
786e8c11 | 1502 | } |
1de06328 | 1503 | s= (char *)uc; |
786e8c11 | 1504 | } |
786e8c11 YO |
1505 | if (uc >(U8*)last_start) break; |
1506 | } | |
1507 | ||
1508 | if ( word ) { | |
1509 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1510 | if (!leftmost || lpos < leftmost) { | |
1511 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1512 | leftmost= lpos; |
786e8c11 | 1513 | } |
07be1b83 | 1514 | if (base==0) break; |
786e8c11 | 1515 | |
07be1b83 YO |
1516 | } |
1517 | points[pointpos++ % maxlen]= uc; | |
4cadc6a9 YO |
1518 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
1519 | uvc, charid, foldlen, foldbuf, uniflags); | |
786e8c11 YO |
1520 | DEBUG_TRIE_EXECUTE_r({ |
1521 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1522 | s, do_utf8 ); | |
07be1b83 | 1523 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1524 | " Charid:%3u CP:%4"UVxf" ", |
1525 | charid, uvc); | |
1526 | }); | |
07be1b83 YO |
1527 | |
1528 | do { | |
6148ee25 | 1529 | #ifdef DEBUGGING |
786e8c11 | 1530 | word = aho->states[ state ].wordnum; |
6148ee25 | 1531 | #endif |
07be1b83 YO |
1532 | base = aho->states[ state ].trans.base; |
1533 | ||
786e8c11 YO |
1534 | DEBUG_TRIE_EXECUTE_r({ |
1535 | if (failed) | |
1536 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1537 | s, do_utf8 ); | |
07be1b83 | 1538 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1539 | "%sState: %4"UVxf", word=%"UVxf, |
1540 | failed ? " Fail transition to " : "", | |
1541 | (UV)state, (UV)word); | |
1542 | }); | |
07be1b83 YO |
1543 | if ( base ) { |
1544 | U32 tmp; | |
1545 | if (charid && | |
1546 | (base + charid > trie->uniquecharcount ) | |
1547 | && (base + charid - 1 - trie->uniquecharcount | |
1548 | < trie->lasttrans) | |
1549 | && trie->trans[base + charid - 1 - | |
1550 | trie->uniquecharcount].check == state | |
1551 | && (tmp=trie->trans[base + charid - 1 - | |
1552 | trie->uniquecharcount ].next)) | |
1553 | { | |
786e8c11 YO |
1554 | DEBUG_TRIE_EXECUTE_r( |
1555 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1556 | state = tmp; |
1557 | break; | |
1558 | } | |
1559 | else { | |
786e8c11 YO |
1560 | DEBUG_TRIE_EXECUTE_r( |
1561 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1562 | failed = 1; | |
1563 | state = aho->fail[state]; | |
07be1b83 YO |
1564 | } |
1565 | } | |
1566 | else { | |
1567 | /* we must be accepting here */ | |
786e8c11 YO |
1568 | DEBUG_TRIE_EXECUTE_r( |
1569 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1570 | failed = 1; | |
07be1b83 YO |
1571 | break; |
1572 | } | |
1573 | } while(state); | |
786e8c11 | 1574 | uc += len; |
07be1b83 YO |
1575 | if (failed) { |
1576 | if (leftmost) | |
1577 | break; | |
786e8c11 | 1578 | if (!state) state = 1; |
07be1b83 YO |
1579 | } |
1580 | } | |
1581 | if ( aho->states[ state ].wordnum ) { | |
1582 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1583 | if (!leftmost || lpos < leftmost) { |
1584 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1585 | leftmost = lpos; |
786e8c11 | 1586 | } |
07be1b83 | 1587 | } |
07be1b83 YO |
1588 | if (leftmost) { |
1589 | s = (char*)leftmost; | |
786e8c11 YO |
1590 | DEBUG_TRIE_EXECUTE_r({ |
1591 | PerlIO_printf( | |
1592 | Perl_debug_log,"Matches word #%"UVxf" at position %d. Trying full pattern...\n", | |
1593 | (UV)accepted_word, s - real_start | |
1594 | ); | |
1595 | }); | |
be8e71aa YO |
1596 | if (!reginfo || regtry(reginfo, s)) { |
1597 | FREETMPS; | |
1598 | LEAVE; | |
07be1b83 | 1599 | goto got_it; |
be8e71aa | 1600 | } |
07be1b83 | 1601 | s = HOPc(s,1); |
786e8c11 YO |
1602 | DEBUG_TRIE_EXECUTE_r({ |
1603 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1604 | }); | |
07be1b83 | 1605 | } else { |
786e8c11 YO |
1606 | DEBUG_TRIE_EXECUTE_r( |
1607 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1608 | break; |
1609 | } | |
1610 | } | |
be8e71aa YO |
1611 | FREETMPS; |
1612 | LEAVE; | |
07be1b83 YO |
1613 | } |
1614 | break; | |
b3c9acc1 | 1615 | default: |
3c3eec57 GS |
1616 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1617 | break; | |
d6a28714 | 1618 | } |
6eb5f6b9 JH |
1619 | return 0; |
1620 | got_it: | |
1621 | return s; | |
1622 | } | |
1623 | ||
1624 | /* | |
1625 | - regexec_flags - match a regexp against a string | |
1626 | */ | |
1627 | I32 | |
1628 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1629 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1630 | /* strend: pointer to null at end of string */ | |
1631 | /* strbeg: real beginning of string */ | |
1632 | /* minend: end of match must be >=minend after stringarg. */ | |
1633 | /* data: May be used for some additional optimizations. */ | |
1634 | /* nosave: For optimizations. */ | |
1635 | { | |
97aff369 | 1636 | dVAR; |
6eb5f6b9 JH |
1637 | register char *s; |
1638 | register regnode *c; | |
1639 | register char *startpos = stringarg; | |
6eb5f6b9 JH |
1640 | I32 minlen; /* must match at least this many chars */ |
1641 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1642 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1643 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1644 | char *scream_olds = NULL; |
3dab1dad | 1645 | SV* const oreplsv = GvSV(PL_replgv); |
f9f4320a | 1646 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1647 | I32 multiline; |
0df25f3d | 1648 | |
3b0527fe | 1649 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1650 | |
1651 | GET_RE_DEBUG_FLAGS_DECL; | |
1652 | ||
9d4ba2ae | 1653 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1654 | |
1655 | /* Be paranoid... */ | |
1656 | if (prog == NULL || startpos == NULL) { | |
1657 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1658 | return 0; | |
1659 | } | |
1660 | ||
2757e526 | 1661 | multiline = prog->reganch & PMf_MULTILINE; |
3b0527fe | 1662 | reginfo.prog = prog; |
2757e526 | 1663 | |
bac06658 | 1664 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1665 | DEBUG_EXECUTE_r( |
1666 | debug_start_match(prog, do_utf8, startpos, strend, | |
1667 | "Matching"); | |
1668 | ); | |
bac06658 | 1669 | |
6eb5f6b9 | 1670 | minlen = prog->minlen; |
1de06328 YO |
1671 | |
1672 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1673 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1674 | "String too short [regexec_flags]...\n")); |
1675 | goto phooey; | |
1aa99e6b | 1676 | } |
6eb5f6b9 | 1677 | |
1de06328 | 1678 | |
6eb5f6b9 JH |
1679 | /* Check validity of program. */ |
1680 | if (UCHARAT(prog->program) != REG_MAGIC) { | |
1681 | Perl_croak(aTHX_ "corrupted regexp program"); | |
1682 | } | |
1683 | ||
1684 | PL_reg_flags = 0; | |
1685 | PL_reg_eval_set = 0; | |
1686 | PL_reg_maxiter = 0; | |
1687 | ||
1688 | if (prog->reganch & ROPT_UTF8) | |
1689 | PL_reg_flags |= RF_utf8; | |
1690 | ||
1691 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1692 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1693 | PL_bostr = strbeg; |
3b0527fe | 1694 | reginfo.sv = sv; |
6eb5f6b9 JH |
1695 | |
1696 | /* Mark end of line for $ (and such) */ | |
1697 | PL_regeol = strend; | |
1698 | ||
1699 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1700 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1701 | |
6eb5f6b9 JH |
1702 | /* If there is a "must appear" string, look for it. */ |
1703 | s = startpos; | |
1704 | ||
3b0527fe | 1705 | if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1706 | MAGIC *mg; |
1707 | ||
1708 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
3b0527fe | 1709 | reginfo.ganch = startpos; |
6eb5f6b9 JH |
1710 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1711 | && SvMAGIC(sv) | |
14befaf4 DM |
1712 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1713 | && mg->mg_len >= 0) { | |
3b0527fe | 1714 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
6eb5f6b9 | 1715 | if (prog->reganch & ROPT_ANCH_GPOS) { |
3b0527fe | 1716 | if (s > reginfo.ganch) |
6eb5f6b9 | 1717 | goto phooey; |
3b0527fe | 1718 | s = reginfo.ganch; |
6eb5f6b9 JH |
1719 | } |
1720 | } | |
1721 | else /* pos() not defined */ | |
3b0527fe | 1722 | reginfo.ganch = strbeg; |
6eb5f6b9 JH |
1723 | } |
1724 | ||
a0714e2c | 1725 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1726 | re_scream_pos_data d; |
1727 | ||
1728 | d.scream_olds = &scream_olds; | |
1729 | d.scream_pos = &scream_pos; | |
1730 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1731 | if (!s) { |
a3621e74 | 1732 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1733 | goto phooey; /* not present */ |
3fa9c3d7 | 1734 | } |
6eb5f6b9 JH |
1735 | } |
1736 | ||
1de06328 | 1737 | |
6eb5f6b9 JH |
1738 | |
1739 | /* Simplest case: anchored match need be tried only once. */ | |
1740 | /* [unless only anchor is BOL and multiline is set] */ | |
1741 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { | |
3b0527fe | 1742 | if (s == startpos && regtry(®info, startpos)) |
6eb5f6b9 | 1743 | goto got_it; |
7fba1cd6 | 1744 | else if (multiline || (prog->reganch & ROPT_IMPLICIT) |
6eb5f6b9 JH |
1745 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
1746 | { | |
1747 | char *end; | |
1748 | ||
1749 | if (minlen) | |
1750 | dontbother = minlen - 1; | |
1aa99e6b | 1751 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1752 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1753 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1754 | if (s == startpos) |
1755 | goto after_try; | |
1756 | while (1) { | |
3b0527fe | 1757 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1758 | goto got_it; |
1759 | after_try: | |
1760 | if (s >= end) | |
1761 | goto phooey; | |
1762 | if (prog->reganch & RE_USE_INTUIT) { | |
1763 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); | |
1764 | if (!s) | |
1765 | goto phooey; | |
1766 | } | |
1767 | else | |
1768 | s++; | |
1769 | } | |
1770 | } else { | |
1771 | if (s > startpos) | |
1772 | s--; | |
1773 | while (s < end) { | |
1774 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
3b0527fe | 1775 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1776 | goto got_it; |
1777 | } | |
1778 | } | |
1779 | } | |
1780 | } | |
1781 | goto phooey; | |
f9f4320a YO |
1782 | } else if (ROPT_GPOS_CHECK == (prog->reganch & ROPT_GPOS_CHECK)) |
1783 | { | |
1784 | /* the warning about reginfo.ganch being used without intialization | |
1785 | is bogus -- we set it above, when prog->reganch & ROPT_GPOS_SEEN | |
1786 | and we only enter this block when the same bit is set. */ | |
3b0527fe | 1787 | if (regtry(®info, reginfo.ganch)) |
6eb5f6b9 JH |
1788 | goto got_it; |
1789 | goto phooey; | |
1790 | } | |
1791 | ||
1792 | /* Messy cases: unanchored match. */ | |
33b8afdf | 1793 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) { |
6eb5f6b9 JH |
1794 | /* we have /x+whatever/ */ |
1795 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1796 | char ch; |
bf93d4cc GS |
1797 | #ifdef DEBUGGING |
1798 | int did_match = 0; | |
1799 | #endif | |
33b8afdf JH |
1800 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1801 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1802 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1803 | |
1aa99e6b | 1804 | if (do_utf8) { |
4cadc6a9 | 1805 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1806 | if (*s == ch) { |
a3621e74 | 1807 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1808 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1809 | s += UTF8SKIP(s); |
1810 | while (s < strend && *s == ch) | |
1811 | s += UTF8SKIP(s); | |
1812 | } | |
4cadc6a9 | 1813 | ); |
6eb5f6b9 JH |
1814 | } |
1815 | else { | |
4cadc6a9 | 1816 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1817 | if (*s == ch) { |
a3621e74 | 1818 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1819 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1820 | s++; |
1821 | while (s < strend && *s == ch) | |
1822 | s++; | |
1823 | } | |
4cadc6a9 | 1824 | ); |
6eb5f6b9 | 1825 | } |
a3621e74 | 1826 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1827 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1828 | "Did not find anchored character...\n") |
1829 | ); | |
6eb5f6b9 | 1830 | } |
a0714e2c SS |
1831 | else if (prog->anchored_substr != NULL |
1832 | || prog->anchored_utf8 != NULL | |
1833 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1834 | && prog->float_max_offset < strend - s)) { |
1835 | SV *must; | |
1836 | I32 back_max; | |
1837 | I32 back_min; | |
1838 | char *last; | |
6eb5f6b9 | 1839 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1840 | #ifdef DEBUGGING |
1841 | int did_match = 0; | |
1842 | #endif | |
33b8afdf JH |
1843 | if (prog->anchored_substr || prog->anchored_utf8) { |
1844 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1845 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1846 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1847 | back_max = back_min = prog->anchored_offset; | |
1848 | } else { | |
1849 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1850 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1851 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1852 | back_max = prog->float_max_offset; | |
1853 | back_min = prog->float_min_offset; | |
1854 | } | |
1de06328 YO |
1855 | |
1856 | ||
33b8afdf JH |
1857 | if (must == &PL_sv_undef) |
1858 | /* could not downgrade utf8 check substring, so must fail */ | |
1859 | goto phooey; | |
1860 | ||
1de06328 YO |
1861 | if (back_min<0) { |
1862 | last = strend; | |
1863 | } else { | |
1864 | last = HOP3c(strend, /* Cannot start after this */ | |
1865 | -(I32)(CHR_SVLEN(must) | |
1866 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1867 | } | |
6eb5f6b9 JH |
1868 | if (s > PL_bostr) |
1869 | last1 = HOPc(s, -1); | |
1870 | else | |
1871 | last1 = s - 1; /* bogus */ | |
1872 | ||
a0288114 | 1873 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1874 | check_substr==must. */ |
1875 | scream_pos = -1; | |
1876 | dontbother = end_shift; | |
1877 | strend = HOPc(strend, -dontbother); | |
1878 | while ( (s <= last) && | |
9041c2e3 | 1879 | ((flags & REXEC_SCREAM) |
1de06328 | 1880 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1881 | end_shift, &scream_pos, 0)) |
1de06328 | 1882 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1883 | (unsigned char*)strend, must, |
7fba1cd6 | 1884 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1885 | /* we may be pointing at the wrong string */ |
1886 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1887 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1888 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1889 | if (HOPc(s, -back_max) > last1) { |
1890 | last1 = HOPc(s, -back_min); | |
1891 | s = HOPc(s, -back_max); | |
1892 | } | |
1893 | else { | |
52657f30 | 1894 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1895 | |
1896 | last1 = HOPc(s, -back_min); | |
52657f30 | 1897 | s = t; |
6eb5f6b9 | 1898 | } |
1aa99e6b | 1899 | if (do_utf8) { |
6eb5f6b9 | 1900 | while (s <= last1) { |
3b0527fe | 1901 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1902 | goto got_it; |
1903 | s += UTF8SKIP(s); | |
1904 | } | |
1905 | } | |
1906 | else { | |
1907 | while (s <= last1) { | |
3b0527fe | 1908 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1909 | goto got_it; |
1910 | s++; | |
1911 | } | |
1912 | } | |
1913 | } | |
ab3bbdeb YO |
1914 | DEBUG_EXECUTE_r(if (!did_match) { |
1915 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
1916 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
1917 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 1918 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 1919 | ? "anchored" : "floating"), |
ab3bbdeb YO |
1920 | quoted, RE_SV_TAIL(must)); |
1921 | }); | |
6eb5f6b9 JH |
1922 | goto phooey; |
1923 | } | |
786e8c11 | 1924 | else if ( (c = prog->regstclass) ) { |
f14c76ed | 1925 | if (minlen) { |
be8e71aa | 1926 | const OPCODE op = OP(prog->regstclass); |
66e933ab | 1927 | /* don't bother with what can't match */ |
786e8c11 | 1928 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
1929 | strend = HOPc(strend, -(minlen - 1)); |
1930 | } | |
a3621e74 | 1931 | DEBUG_EXECUTE_r({ |
be8e71aa | 1932 | SV * const prop = sv_newmortal(); |
32fc9b6a | 1933 | regprop(prog, prop, c); |
0df25f3d | 1934 | { |
ab3bbdeb YO |
1935 | RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1), |
1936 | s,strend-s,60); | |
0df25f3d | 1937 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 1938 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 1939 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 1940 | quoted, (int)(strend - s)); |
0df25f3d | 1941 | } |
ffc61ed2 | 1942 | }); |
3b0527fe | 1943 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1944 | goto got_it; |
07be1b83 | 1945 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
1946 | } |
1947 | else { | |
1948 | dontbother = 0; | |
a0714e2c | 1949 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 1950 | /* Trim the end. */ |
d6a28714 | 1951 | char *last; |
33b8afdf JH |
1952 | SV* float_real; |
1953 | ||
1954 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1955 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1956 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
1957 | |
1958 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 1959 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
1960 | end_shift, &scream_pos, 1); /* last one */ |
1961 | if (!last) | |
ffc61ed2 | 1962 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
1963 | /* we may be pointing at the wrong string */ |
1964 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 1965 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 1966 | } |
d6a28714 JH |
1967 | else { |
1968 | STRLEN len; | |
cfd0369c | 1969 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 1970 | |
33b8afdf | 1971 | if (SvTAIL(float_real)) { |
d6a28714 JH |
1972 | if (memEQ(strend - len + 1, little, len - 1)) |
1973 | last = strend - len + 1; | |
7fba1cd6 | 1974 | else if (!multiline) |
9041c2e3 | 1975 | last = memEQ(strend - len, little, len) |
bd61b366 | 1976 | ? strend - len : NULL; |
b8c5462f | 1977 | else |
d6a28714 JH |
1978 | goto find_last; |
1979 | } else { | |
1980 | find_last: | |
9041c2e3 | 1981 | if (len) |
d6a28714 | 1982 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 1983 | else |
a0288114 | 1984 | last = strend; /* matching "$" */ |
b8c5462f | 1985 | } |
b8c5462f | 1986 | } |
bf93d4cc | 1987 | if (last == NULL) { |
6bda09f9 YO |
1988 | DEBUG_EXECUTE_r( |
1989 | PerlIO_printf(Perl_debug_log, | |
1990 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
1991 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
1992 | goto phooey; /* Should not happen! */ |
1993 | } | |
d6a28714 JH |
1994 | dontbother = strend - last + prog->float_min_offset; |
1995 | } | |
1996 | if (minlen && (dontbother < minlen)) | |
1997 | dontbother = minlen - 1; | |
1998 | strend -= dontbother; /* this one's always in bytes! */ | |
1999 | /* We don't know much -- general case. */ | |
1aa99e6b | 2000 | if (do_utf8) { |
d6a28714 | 2001 | for (;;) { |
3b0527fe | 2002 | if (regtry(®info, s)) |
d6a28714 JH |
2003 | goto got_it; |
2004 | if (s >= strend) | |
2005 | break; | |
b8c5462f | 2006 | s += UTF8SKIP(s); |
d6a28714 JH |
2007 | }; |
2008 | } | |
2009 | else { | |
2010 | do { | |
3b0527fe | 2011 | if (regtry(®info, s)) |
d6a28714 JH |
2012 | goto got_it; |
2013 | } while (s++ < strend); | |
2014 | } | |
2015 | } | |
2016 | ||
2017 | /* Failure. */ | |
2018 | goto phooey; | |
2019 | ||
2020 | got_it: | |
2021 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2022 | ||
2023 | if (PL_reg_eval_set) { | |
2024 | /* Preserve the current value of $^R */ | |
2025 | if (oreplsv != GvSV(PL_replgv)) | |
2026 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2027 | restored, the value remains | |
2028 | the same. */ | |
4f639d21 | 2029 | restore_pos(aTHX_ prog); |
d6a28714 | 2030 | } |
81714fb9 YO |
2031 | if (prog->paren_names) |
2032 | (void)hv_iterinit(prog->paren_names); | |
d6a28714 JH |
2033 | |
2034 | /* make sure $`, $&, $', and $digit will work later */ | |
2035 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2036 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2037 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2038 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2039 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2040 | if ((SvIsCOW(sv) |
2041 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2042 | if (DEBUG_C_TEST) { | |
2043 | PerlIO_printf(Perl_debug_log, | |
2044 | "Copy on write: regexp capture, type %d\n", | |
2045 | (int) SvTYPE(sv)); | |
2046 | } | |
2047 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2048 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2049 | assert (SvPOKp(prog->saved_copy)); |
2050 | } else | |
2051 | #endif | |
2052 | { | |
2053 | RX_MATCH_COPIED_on(prog); | |
2054 | s = savepvn(strbeg, i); | |
2055 | prog->subbeg = s; | |
2056 | } | |
d6a28714 | 2057 | prog->sublen = i; |
d6a28714 JH |
2058 | } |
2059 | else { | |
2060 | prog->subbeg = strbeg; | |
2061 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2062 | } | |
2063 | } | |
9041c2e3 | 2064 | |
d6a28714 JH |
2065 | return 1; |
2066 | ||
2067 | phooey: | |
a3621e74 | 2068 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2069 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2070 | if (PL_reg_eval_set) |
4f639d21 | 2071 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2072 | return 0; |
2073 | } | |
2074 | ||
6bda09f9 | 2075 | |
d6a28714 JH |
2076 | /* |
2077 | - regtry - try match at specific point | |
2078 | */ | |
2079 | STATIC I32 /* 0 failure, 1 success */ | |
3b0527fe | 2080 | S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos) |
d6a28714 | 2081 | { |
97aff369 | 2082 | dVAR; |
d6a28714 JH |
2083 | register I32 *sp; |
2084 | register I32 *ep; | |
2085 | CHECKPOINT lastcp; | |
3b0527fe | 2086 | regexp *prog = reginfo->prog; |
a3621e74 | 2087 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2088 | |
2089 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { | |
2090 | MAGIC *mg; | |
2091 | ||
2092 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2093 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2094 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2095 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2096 | )); |
e8347627 | 2097 | SAVEI32(cxstack[cxstack_ix].blk_oldsp); |
d6a28714 JH |
2098 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2099 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2100 | SAVETMPS; | |
2101 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2102 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2103 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2104 | ||
3b0527fe | 2105 | if (reginfo->sv) { |
d6a28714 | 2106 | /* Make $_ available to executed code. */ |
3b0527fe | 2107 | if (reginfo->sv != DEFSV) { |
59f00321 | 2108 | SAVE_DEFSV; |
3b0527fe | 2109 | DEFSV = reginfo->sv; |
b8c5462f | 2110 | } |
d6a28714 | 2111 | |
3b0527fe DM |
2112 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2113 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2114 | /* prepare for quick setting of pos */ |
d300d9fa NC |
2115 | #ifdef PERL_OLD_COPY_ON_WRITE |
2116 | if (SvIsCOW(sv)) | |
2117 | sv_force_normal_flags(sv, 0); | |
2118 | #endif | |
3dab1dad | 2119 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2120 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2121 | mg->mg_len = -1; |
b8c5462f | 2122 | } |
d6a28714 JH |
2123 | PL_reg_magic = mg; |
2124 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2125 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2126 | } |
09687e5a | 2127 | if (!PL_reg_curpm) { |
a02a5408 | 2128 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2129 | #ifdef USE_ITHREADS |
2130 | { | |
be8e71aa | 2131 | SV* const repointer = newSViv(0); |
577e12cc | 2132 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2133 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2134 | av_push(PL_regex_padav,repointer); |
2135 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2136 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2137 | } | |
2138 | #endif | |
2139 | } | |
aaa362c4 | 2140 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2141 | PL_reg_oldcurpm = PL_curpm; |
2142 | PL_curpm = PL_reg_curpm; | |
2143 | if (RX_MATCH_COPIED(prog)) { | |
2144 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2145 | since it may be needed if this match fails. Thus | |
2146 | $` inside (?{}) could fail... */ | |
2147 | PL_reg_oldsaved = prog->subbeg; | |
2148 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2149 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2150 | PL_nrs = prog->saved_copy; |
2151 | #endif | |
d6a28714 JH |
2152 | RX_MATCH_COPIED_off(prog); |
2153 | } | |
2154 | else | |
bd61b366 | 2155 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2156 | prog->subbeg = PL_bostr; |
2157 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2158 | } | |
6bda09f9 | 2159 | DEBUG_EXECUTE_r(PL_reg_starttry = startpos); |
973dddac | 2160 | prog->startp[0] = startpos - PL_bostr; |
d6a28714 | 2161 | PL_reginput = startpos; |
d6a28714 | 2162 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2163 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2164 | prog->lastparen = 0; |
03994de8 | 2165 | prog->lastcloseparen = 0; |
d6a28714 | 2166 | PL_regsize = 0; |
6bda09f9 YO |
2167 | PL_regstartp = prog->startp; |
2168 | PL_regendp = prog->endp; | |
d6a28714 JH |
2169 | if (PL_reg_start_tmpl <= prog->nparens) { |
2170 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2171 | if(PL_reg_start_tmp) | |
2172 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2173 | else | |
a02a5408 | 2174 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2175 | } |
2176 | ||
2177 | /* XXXX What this code is doing here?!!! There should be no need | |
2178 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2179 | this! --ilya*/ |
dafc8851 JH |
2180 | |
2181 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2182 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2183 | * PL_reglastparen), is not needed at all by the test suite |
2184 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2185 | * enough, for building DynaLoader, or otherwise this | |
2186 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2187 | * will happen. Meanwhile, this code *is* needed for the | |
2188 | * above-mentioned test suite tests to succeed. The common theme | |
2189 | * on those tests seems to be returning null fields from matches. | |
2190 | * --jhi */ | |
dafc8851 | 2191 | #if 1 |
d6a28714 JH |
2192 | sp = prog->startp; |
2193 | ep = prog->endp; | |
2194 | if (prog->nparens) { | |
097eb12c | 2195 | register I32 i; |
eb160463 | 2196 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2197 | *++sp = -1; |
2198 | *++ep = -1; | |
2199 | } | |
2200 | } | |
dafc8851 | 2201 | #endif |
02db2b7b | 2202 | REGCP_SET(lastcp); |
3b0527fe | 2203 | if (regmatch(reginfo, prog->program + 1)) { |
d6a28714 JH |
2204 | prog->endp[0] = PL_reginput - PL_bostr; |
2205 | return 1; | |
2206 | } | |
02db2b7b | 2207 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2208 | return 0; |
2209 | } | |
2210 | ||
02db2b7b | 2211 | |
8ba1375e MJD |
2212 | #define sayYES goto yes |
2213 | #define sayNO goto no | |
262b90c4 | 2214 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2215 | |
f9f4320a YO |
2216 | /* we dont use STMT_START/END here because it leads to |
2217 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2218 | #define CACHEsayNO \ | |
c476f425 DM |
2219 | if (ST.cache_mask) \ |
2220 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2221 | sayNO |
3298f257 | 2222 | |
a3621e74 | 2223 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2224 | 'failed...' are printed. It should be set such that messages |
2225 | are inline with the regop output that created them. | |
a3621e74 | 2226 | */ |
265c4333 | 2227 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2228 | |
2229 | ||
2230 | /* Make sure there is a test for this +1 options in re_tests */ | |
2231 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2232 | ||
40a82448 DM |
2233 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2234 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2235 | |
86545054 DM |
2236 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2237 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2238 | ||
5d9a96ca DM |
2239 | /* grab a new slab and return the first slot in it */ |
2240 | ||
2241 | STATIC regmatch_state * | |
2242 | S_push_slab(pTHX) | |
2243 | { | |
54df2634 NC |
2244 | #if PERL_VERSION < 9 |
2245 | dMY_CXT; | |
2246 | #endif | |
5d9a96ca DM |
2247 | regmatch_slab *s = PL_regmatch_slab->next; |
2248 | if (!s) { | |
2249 | Newx(s, 1, regmatch_slab); | |
2250 | s->prev = PL_regmatch_slab; | |
2251 | s->next = NULL; | |
2252 | PL_regmatch_slab->next = s; | |
2253 | } | |
2254 | PL_regmatch_slab = s; | |
86545054 | 2255 | return SLAB_FIRST(s); |
5d9a96ca | 2256 | } |
5b47454d | 2257 | |
95b24440 | 2258 | |
40a82448 DM |
2259 | /* push a new state then goto it */ |
2260 | ||
2261 | #define PUSH_STATE_GOTO(state, node) \ | |
2262 | scan = node; \ | |
2263 | st->resume_state = state; \ | |
2264 | goto push_state; | |
2265 | ||
2266 | /* push a new state with success backtracking, then goto it */ | |
2267 | ||
2268 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2269 | scan = node; \ | |
2270 | st->resume_state = state; \ | |
2271 | goto push_yes_state; | |
2272 | ||
aa283a38 | 2273 | |
aa283a38 | 2274 | |
d6a28714 | 2275 | /* |
95b24440 | 2276 | |
bf1f174e DM |
2277 | regmatch() - main matching routine |
2278 | ||
2279 | This is basically one big switch statement in a loop. We execute an op, | |
2280 | set 'next' to point the next op, and continue. If we come to a point which | |
2281 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2282 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2283 | state, and re-enter the loop at the state indicated. If there are no more | |
2284 | states to pop, we return failure. | |
2285 | ||
2286 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2287 | after successfully matching one A, we need to go back and try to | |
2288 | match another one; similarly for lookahead assertions: if the assertion | |
2289 | completes successfully, we backtrack to the state just before the assertion | |
2290 | and then carry on. In these cases, the pushed state is marked as | |
2291 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2292 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2293 | the nearest yes state, discarding any intermediate failure-only states. | |
2294 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2295 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2296 | it to free the inner regex. | |
2297 | ||
2298 | Note that failure backtracking rewinds the cursor position, while | |
2299 | success backtracking leaves it alone. | |
2300 | ||
2301 | A pattern is complete when the END op is executed, while a subpattern | |
2302 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2303 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2304 | behaviour. | |
2305 | ||
2306 | A common convention in this function is to use A and B to refer to the two | |
2307 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2308 | the subpattern to be matched possibly multiple times, while B is the entire | |
2309 | rest of the pattern. Variable and state names reflect this convention. | |
2310 | ||
2311 | The states in the main switch are the union of ops and failure/success of | |
2312 | substates associated with with that op. For example, IFMATCH is the op | |
2313 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2314 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2315 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2316 | just failed to match A. Resume states always come in pairs. The backtrack | |
2317 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2318 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2319 | on success or failure. | |
2320 | ||
2321 | The struct that holds a backtracking state is actually a big union, with | |
2322 | one variant for each major type of op. The variable st points to the | |
2323 | top-most backtrack struct. To make the code clearer, within each | |
2324 | block of code we #define ST to alias the relevant union. | |
2325 | ||
2326 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2327 | implementation: | |
2328 | ||
2329 | switch (state) { | |
2330 | .... | |
2331 | ||
2332 | #define ST st->u.ifmatch | |
2333 | ||
2334 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2335 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2336 | ... |
bf1f174e DM |
2337 | // push a yes backtrack state with a resume value of |
2338 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2339 | // first node of A: | |
2340 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2341 | // NOTREACHED | |
2342 | ||
2343 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2344 | next = B; | |
2345 | bar = ST.foo; // do something with the preserved value | |
2346 | break; | |
2347 | ||
2348 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2349 | ...; // do some housekeeping, then ... | |
2350 | sayNO; // propagate the failure | |
2351 | ||
2352 | #undef ST | |
95b24440 | 2353 | |
bf1f174e DM |
2354 | ... |
2355 | } | |
95b24440 | 2356 | |
bf1f174e DM |
2357 | For any old-timers reading this who are familiar with the old recursive |
2358 | approach, the code above is equivalent to: | |
95b24440 | 2359 | |
bf1f174e DM |
2360 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2361 | { | |
2362 | int foo = ... | |
95b24440 | 2363 | ... |
bf1f174e DM |
2364 | if (regmatch(A)) { |
2365 | next = B; | |
2366 | bar = foo; | |
2367 | break; | |
95b24440 | 2368 | } |
bf1f174e DM |
2369 | ...; // do some housekeeping, then ... |
2370 | sayNO; // propagate the failure | |
95b24440 | 2371 | } |
bf1f174e DM |
2372 | |
2373 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2374 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2375 | save, then do one of | |
2376 | ||
2377 | PUSH_STATE_GOTO(resume_state, node); | |
2378 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2379 | ||
2380 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2381 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2382 | On backtracking, the free slot is popped, and the saved state becomes the | |
2383 | new free state. An ST.foo field in this new top state can be temporarily | |
2384 | accessed to retrieve values, but once the main loop is re-entered, it | |
2385 | becomes available for reuse. | |
2386 | ||
2387 | Note that the depth of the backtrack stack constantly increases during the | |
2388 | left-to-right execution of the pattern, rather than going up and down with | |
2389 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2390 | end of the pattern, rather than at X in the following: | |
2391 | ||
2392 | /(((X)+)+)+....(Y)+....Z/ | |
2393 | ||
2394 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2395 | (?>A), which pop all the backtrack states associated with A before | |
2396 | continuing. | |
2397 | ||
2398 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2399 | PL_regmatch_state and st always point to the currently active state, | |
2400 | and PL_regmatch_slab points to the slab currently containing | |
2401 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2402 | allocated, and is never freed until interpreter destruction. When the slab | |
2403 | is full, a new one is allocated and chained to the end. At exit from | |
2404 | regmatch(), slabs allocated since entry are freed. | |
2405 | ||
2406 | */ | |
95b24440 | 2407 | |
40a82448 | 2408 | |
5bc10b2c | 2409 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2410 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2411 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2412 | PerlIO_printf(Perl_debug_log, \ | |
2413 | " %*s"pp" %s\n", \ | |
2414 | depth*2, "", \ | |
03363afd | 2415 | reg_name[st->resume_state] ); \ |
265c4333 | 2416 | }); |
5bc10b2c | 2417 | |
40a82448 | 2418 | |
3dab1dad | 2419 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2420 | |
3df15adc | 2421 | #ifdef DEBUGGING |
5bc10b2c | 2422 | |
ab3bbdeb YO |
2423 | STATIC void |
2424 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2425 | const char *start, const char *end, const char *blurb) | |
2426 | { | |
2427 | const bool utf8_pat= prog->reganch & ROPT_UTF8 ? 1 : 0; | |
2428 | if (!PL_colorset) | |
2429 | reginitcolors(); | |
2430 | { | |
2431 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2432 | prog->precomp, prog->prelen, 60); | |
2433 | ||
2434 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2435 | start, end - start, 60); | |
2436 | ||
2437 | PerlIO_printf(Perl_debug_log, | |
2438 | "%s%s REx%s %s against %s\n", | |
2439 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2440 | ||
2441 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2442 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2443 | utf8_pat ? "pattern" : "", | |
2444 | utf8_pat && do_utf8 ? " and " : "", | |
2445 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2446 | ); |
2447 | } | |
2448 | } | |
3df15adc YO |
2449 | |
2450 | STATIC void | |
786e8c11 YO |
2451 | S_dump_exec_pos(pTHX_ const char *locinput, |
2452 | const regnode *scan, | |
2453 | const char *loc_regeol, | |
2454 | const char *loc_bostr, | |
2455 | const char *loc_reg_starttry, | |
2456 | const bool do_utf8) | |
07be1b83 | 2457 | { |
786e8c11 | 2458 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2459 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2460 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2461 | /* The part of the string before starttry has one color |
2462 | (pref0_len chars), between starttry and current | |
2463 | position another one (pref_len - pref0_len chars), | |
2464 | after the current position the third one. | |
2465 | We assume that pref0_len <= pref_len, otherwise we | |
2466 | decrease pref0_len. */ | |
786e8c11 YO |
2467 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2468 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2469 | int pref0_len; |
2470 | ||
2471 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2472 | pref_len++; | |
786e8c11 YO |
2473 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2474 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2475 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2476 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2477 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2478 | l--; | |
2479 | if (pref0_len < 0) | |
2480 | pref0_len = 0; | |
2481 | if (pref0_len > pref_len) | |
2482 | pref0_len = pref_len; | |
2483 | { | |
3df15adc | 2484 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2485 | |
ab3bbdeb | 2486 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2487 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2488 | |
ab3bbdeb | 2489 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2490 | (locinput - pref_len + pref0_len), |
1de06328 | 2491 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2492 | |
ab3bbdeb | 2493 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2494 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2495 | |
1de06328 | 2496 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2497 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2498 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2499 | (IV)(locinput - loc_bostr), |
07be1b83 | 2500 | len0, s0, |
07be1b83 | 2501 | len1, s1, |
07be1b83 | 2502 | (docolor ? "" : "> <"), |
07be1b83 | 2503 | len2, s2, |
f9f4320a | 2504 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2505 | ""); |
2506 | } | |
2507 | } | |
3df15adc | 2508 | |
07be1b83 YO |
2509 | #endif |
2510 | ||
d6a28714 | 2511 | STATIC I32 /* 0 failure, 1 success */ |
3b0527fe | 2512 | S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2513 | { |
54df2634 NC |
2514 | #if PERL_VERSION < 9 |
2515 | dMY_CXT; | |
2516 | #endif | |
27da23d5 | 2517 | dVAR; |
95b24440 | 2518 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2519 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2520 | |
3b0527fe DM |
2521 | regexp *rex = reginfo->prog; |
2522 | ||
5d9a96ca DM |
2523 | regmatch_slab *orig_slab; |
2524 | regmatch_state *orig_state; | |
a3621e74 | 2525 | |
5d9a96ca DM |
2526 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2527 | register regmatch_state *st; | |
95b24440 | 2528 | |
5d9a96ca DM |
2529 | /* cache heavy used fields of st in registers */ |
2530 | register regnode *scan; | |
2531 | register regnode *next; | |
24d3c4a9 DM |
2532 | register I32 n = 0; /* general value; init to avoid compiler warning */ |
2533 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ | |
5d9a96ca | 2534 | register char *locinput = PL_reginput; |
5d9a96ca | 2535 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2536 | |
b69b0499 | 2537 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2538 | int depth = 0; /* depth of backtrack stack */ |
6bda09f9 | 2539 | int nochange_depth = 0; /* depth of RECURSE recursion with nochange*/ |
77cb431f DM |
2540 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2541 | subpattern */ | |
faec1544 | 2542 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2543 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2544 | U32 state_num; |
10edeb5d | 2545 | |
24d3c4a9 DM |
2546 | /* these three flags are set by various ops to signal information to |
2547 | * the very next op. They have a useful lifetime of exactly one loop | |
2548 | * iteration, and are not preserved or restored by state pushes/pops | |
2549 | */ | |
2550 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2551 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2552 | int logical = 0; /* the following EVAL is: | |
2553 | 0: (?{...}) | |
2554 | 1: (?(?{...})X|Y) | |
2555 | 2: (??{...}) | |
2556 | or the following IFMATCH/UNLESSM is: | |
2557 | false: plain (?=foo) | |
2558 | true: used as a condition: (?(?=foo)) | |
2559 | */ | |
2560 | ||
95b24440 | 2561 | #ifdef DEBUGGING |
e68ec53f | 2562 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2563 | #endif |
2564 | ||
5d9a96ca DM |
2565 | /* on first ever call to regmatch, allocate first slab */ |
2566 | if (!PL_regmatch_slab) { | |
2567 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2568 | PL_regmatch_slab->prev = NULL; | |
2569 | PL_regmatch_slab->next = NULL; | |
86545054 | 2570 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2571 | } |
2572 | ||
2573 | /* remember current high-water mark for exit */ | |
2574 | /* XXX this should be done with SAVE* instead */ | |
2575 | orig_slab = PL_regmatch_slab; | |
2576 | orig_state = PL_regmatch_state; | |
2577 | ||
2578 | /* grab next free state slot */ | |
2579 | st = ++PL_regmatch_state; | |
86545054 | 2580 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2581 | st = PL_regmatch_state = S_push_slab(aTHX); |
2582 | ||
d6a28714 JH |
2583 | /* Note that nextchr is a byte even in UTF */ |
2584 | nextchr = UCHARAT(locinput); | |
2585 | scan = prog; | |
2586 | while (scan != NULL) { | |
8ba1375e | 2587 | |
a3621e74 | 2588 | DEBUG_EXECUTE_r( { |
6136c704 | 2589 | SV * const prop = sv_newmortal(); |
1de06328 | 2590 | regnode *rnext=regnext(scan); |
786e8c11 | 2591 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2592 | regprop(rex, prop, scan); |
07be1b83 YO |
2593 | |
2594 | PerlIO_printf(Perl_debug_log, | |
2595 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
5bc10b2c | 2596 | (IV)(scan - rex->program), depth*2, "", |
07be1b83 | 2597 | SvPVX_const(prop), |
1de06328 YO |
2598 | (PL_regkind[OP(scan)] == END || !rnext) ? |
2599 | 0 : (IV)(rnext - rex->program)); | |
2a782b5b | 2600 | }); |
d6a28714 JH |
2601 | |
2602 | next = scan + NEXT_OFF(scan); | |
2603 | if (next == scan) | |
2604 | next = NULL; | |
40a82448 | 2605 | state_num = OP(scan); |
d6a28714 | 2606 | |
40a82448 DM |
2607 | reenter_switch: |
2608 | switch (state_num) { | |
d6a28714 | 2609 | case BOL: |
7fba1cd6 | 2610 | if (locinput == PL_bostr) |
d6a28714 | 2611 | { |
3b0527fe | 2612 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2613 | break; |
2614 | } | |
d6a28714 JH |
2615 | sayNO; |
2616 | case MBOL: | |
12d33761 HS |
2617 | if (locinput == PL_bostr || |
2618 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2619 | { |
b8c5462f JH |
2620 | break; |
2621 | } | |
d6a28714 JH |
2622 | sayNO; |
2623 | case SBOL: | |
c2a73568 | 2624 | if (locinput == PL_bostr) |
b8c5462f | 2625 | break; |
d6a28714 JH |
2626 | sayNO; |
2627 | case GPOS: | |
3b0527fe | 2628 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2629 | break; |
2630 | sayNO; | |
2631 | case EOL: | |
d6a28714 JH |
2632 | goto seol; |
2633 | case MEOL: | |
d6a28714 | 2634 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2635 | sayNO; |
b8c5462f | 2636 | break; |
d6a28714 JH |
2637 | case SEOL: |
2638 | seol: | |
2639 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2640 | sayNO; |
d6a28714 | 2641 | if (PL_regeol - locinput > 1) |
b8c5462f | 2642 | sayNO; |
b8c5462f | 2643 | break; |
d6a28714 JH |
2644 | case EOS: |
2645 | if (PL_regeol != locinput) | |
b8c5462f | 2646 | sayNO; |
d6a28714 | 2647 | break; |
ffc61ed2 | 2648 | case SANY: |
d6a28714 | 2649 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2650 | sayNO; |
f33976b4 DB |
2651 | if (do_utf8) { |
2652 | locinput += PL_utf8skip[nextchr]; | |
2653 | if (locinput > PL_regeol) | |
2654 | sayNO; | |
2655 | nextchr = UCHARAT(locinput); | |
2656 | } | |
2657 | else | |
2658 | nextchr = UCHARAT(++locinput); | |
2659 | break; | |
2660 | case CANY: | |
2661 | if (!nextchr && locinput >= PL_regeol) | |
2662 | sayNO; | |
b8c5462f | 2663 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2664 | break; |
ffc61ed2 | 2665 | case REG_ANY: |
1aa99e6b IH |
2666 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2667 | sayNO; | |
2668 | if (do_utf8) { | |
b8c5462f | 2669 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2670 | if (locinput > PL_regeol) |
2671 | sayNO; | |
a0ed51b3 | 2672 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2673 | } |
1aa99e6b IH |
2674 | else |
2675 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2676 | break; |
166ba7cd DM |
2677 | |
2678 | #undef ST | |
2679 | #define ST st->u.trie | |
786e8c11 YO |
2680 | case TRIEC: |
2681 | /* In this case the charclass data is available inline so | |
2682 | we can fail fast without a lot of extra overhead. | |
2683 | */ | |
2684 | if (scan->flags == EXACT || !do_utf8) { | |
2685 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2686 | DEBUG_EXECUTE_r( | |
2687 | PerlIO_printf(Perl_debug_log, | |
2688 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2689 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2690 | ); |
2691 | sayNO_SILENT; | |
2692 | /* NOTREACHED */ | |
2693 | } | |
2694 | } | |
2695 | /* FALL THROUGH */ | |
5b47454d | 2696 | case TRIE: |
3dab1dad | 2697 | { |
07be1b83 | 2698 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2699 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2700 | trie_type = do_utf8 ? | |
2701 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2702 | : trie_plain; | |
2703 | ||
2704 | /* what trie are we using right now */ | |
be8e71aa | 2705 | reg_trie_data * const trie |
3dab1dad YO |
2706 | = (reg_trie_data*)rex->data->data[ ARG( scan ) ]; |
2707 | U32 state = trie->startstate; | |
166ba7cd | 2708 | |
3dab1dad YO |
2709 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2710 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2711 | ) { | |
2712 | if (trie->states[ state ].wordnum) { | |
2713 | DEBUG_EXECUTE_r( | |
2714 | PerlIO_printf(Perl_debug_log, | |
2715 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2716 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2717 | ); |
2718 | break; | |
2719 | } else { | |
2720 | DEBUG_EXECUTE_r( | |
2721 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2722 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2723 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2724 | ); |
2725 | sayNO_SILENT; | |
2726 | } | |
2727 | } | |
166ba7cd | 2728 | |
786e8c11 YO |
2729 | { |
2730 | U8 *uc = ( U8* )locinput; | |
2731 | ||
2732 | STRLEN len = 0; | |
2733 | STRLEN foldlen = 0; | |
2734 | U8 *uscan = (U8*)NULL; | |
2735 | STRLEN bufflen=0; | |
2736 | SV *sv_accept_buff = NULL; | |
2737 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2738 | ||
2739 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2740 | ST.B = next; | |
2741 | ST.jump = trie->jump; | |
2742 | ||
2743 | #ifdef DEBUGGING | |
2744 | ST.me = scan; | |
2745 | #endif | |
2746 | ||
2747 | ||
2748 | ||
07be1b83 YO |
2749 | /* |
2750 | traverse the TRIE keeping track of all accepting states | |
2751 | we transition through until we get to a failing node. | |
2752 | */ | |
2753 | ||
a3621e74 | 2754 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 2755 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 2756 | UV uvc = 0; |
786e8c11 YO |
2757 | U16 charid; |
2758 | /* We use charid to hold the wordnum as we don't use it | |
2759 | for charid until after we have done the wordnum logic. | |
2760 | We define an alias just so that the wordnum logic reads | |
2761 | more naturally. */ | |
2762 | ||
2763 | #define got_wordnum charid | |
2764 | got_wordnum = trie->states[ state ].wordnum; | |
2765 | ||
2766 | if ( got_wordnum ) { | |
2767 | if ( ! ST.accepted ) { | |
5b47454d DM |
2768 | ENTER; |
2769 | SAVETMPS; | |
2770 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2771 | sv_accept_buff=newSV(bufflen * | |
2772 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2773 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2774 | SvPOK_on(sv_accept_buff); |
2775 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2776 | SAVETMPS; |
2777 | ST.accept_buff = | |
5b47454d DM |
2778 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2779 | } | |
786e8c11 | 2780 | do { |
166ba7cd | 2781 | if (ST.accepted >= bufflen) { |
5b47454d | 2782 | bufflen *= 2; |
166ba7cd | 2783 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2784 | SvGROW(sv_accept_buff, |
2785 | bufflen * sizeof(reg_trie_accepted)); | |
2786 | } | |
2787 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2788 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2789 | |
786e8c11 YO |
2790 | |
2791 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2792 | ST.accept_buff[ST.accepted].endpos = uc; | |
2793 | ++ST.accepted; | |
2794 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2795 | } | |
2796 | #undef got_wordnum | |
a3621e74 | 2797 | |
07be1b83 | 2798 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2799 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2800 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2801 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2802 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2803 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2804 | }); |
a3621e74 YO |
2805 | |
2806 | if ( base ) { | |
4cadc6a9 YO |
2807 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
2808 | uvc, charid, foldlen, foldbuf, uniflags); | |
a3621e74 | 2809 | |
5b47454d DM |
2810 | if (charid && |
2811 | (base + charid > trie->uniquecharcount ) | |
2812 | && (base + charid - 1 - trie->uniquecharcount | |
2813 | < trie->lasttrans) | |
2814 | && trie->trans[base + charid - 1 - | |
2815 | trie->uniquecharcount].check == state) | |
2816 | { | |
2817 | state = trie->trans[base + charid - 1 - | |
2818 | trie->uniquecharcount ].next; | |
2819 | } | |
2820 | else { | |
2821 | state = 0; | |
2822 | } | |
2823 | uc += len; | |
2824 | ||
2825 | } | |
2826 | else { | |
a3621e74 YO |
2827 | state = 0; |
2828 | } | |
2829 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 2830 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2831 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 2832 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
2833 | ); |
2834 | } | |
166ba7cd | 2835 | if (!ST.accepted ) |
a3621e74 | 2836 | sayNO; |
a3621e74 | 2837 | |
166ba7cd DM |
2838 | DEBUG_EXECUTE_r( |
2839 | PerlIO_printf( Perl_debug_log, | |
2840 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 2841 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
2842 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
2843 | ); | |
786e8c11 | 2844 | }} |
166ba7cd DM |
2845 | |
2846 | /* FALL THROUGH */ | |
2847 | ||
2848 | case TRIE_next_fail: /* we failed - try next alterative */ | |
2849 | ||
2850 | if ( ST.accepted == 1 ) { | |
2851 | /* only one choice left - just continue */ | |
2852 | DEBUG_EXECUTE_r({ | |
2853 | reg_trie_data * const trie | |
2854 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2855 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2856 | ? av_fetch( trie->words, ST.accept_buff[ 0 ].wordnum-1, 0 ) | |
2857 | : NULL; | |
2858 | PerlIO_printf( Perl_debug_log, | |
2859 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 2860 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd DM |
2861 | ST.accept_buff[ 0 ].wordnum, |
2862 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", | |
2863 | PL_colors[5] ); | |
2864 | }); | |
2865 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
2866 | /* in this case we free tmps/leave before we call regmatch | |
2867 | as we wont be using accept_buff again. */ | |
2868 | FREETMPS; | |
2869 | LEAVE; | |
2870 | locinput = PL_reginput; | |
2871 | nextchr = UCHARAT(locinput); | |
786e8c11 YO |
2872 | |
2873 | if ( !ST.jump ) | |
2874 | scan = ST.B; | |
2875 | else | |
2876 | scan = ST.B - ST.jump[ST.accept_buff[0].wordnum]; | |
2877 | ||
166ba7cd DM |
2878 | continue; /* execute rest of RE */ |
2879 | } | |
2880 | ||
2881 | if (!ST.accepted-- ) { | |
2882 | FREETMPS; | |
2883 | LEAVE; | |
2884 | sayNO; | |
2885 | } | |
2886 | ||
a3621e74 | 2887 | /* |
166ba7cd DM |
2888 | There are at least two accepting states left. Presumably |
2889 | the number of accepting states is going to be low, | |
2890 | typically two. So we simply scan through to find the one | |
2891 | with lowest wordnum. Once we find it, we swap the last | |
2892 | state into its place and decrement the size. We then try to | |
2893 | match the rest of the pattern at the point where the word | |
2894 | ends. If we succeed, control just continues along the | |
2895 | regex; if we fail we return here to try the next accepting | |
2896 | state | |
2897 | */ | |
a3621e74 | 2898 | |
166ba7cd DM |
2899 | { |
2900 | U32 best = 0; | |
2901 | U32 cur; | |
2902 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
2903 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 2904 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 2905 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 2906 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
2907 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
2908 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
2909 | ); | |
2910 | ||
2911 | if (ST.accept_buff[cur].wordnum < | |
2912 | ST.accept_buff[best].wordnum) | |
2913 | best = cur; | |
a3621e74 | 2914 | } |
166ba7cd DM |
2915 | |
2916 | DEBUG_EXECUTE_r({ | |
2917 | reg_trie_data * const trie | |
2918 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2919 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2920 | ? av_fetch( trie->words, ST.accept_buff[ best ].wordnum - 1, 0 ) | |
2921 | : NULL; | |
265c4333 YO |
2922 | regnode *nextop=!ST.jump ? |
2923 | ST.B : | |
2924 | ST.B - ST.jump[ST.accept_buff[best].wordnum]; | |
2925 | PerlIO_printf( Perl_debug_log, | |
2926 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 2927 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 2928 | ST.accept_buff[best].wordnum, |
265c4333 YO |
2929 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", |
2930 | REG_NODE_NUM(nextop), | |
166ba7cd DM |
2931 | PL_colors[5] ); |
2932 | }); | |
2933 | ||
2934 | if ( best<ST.accepted ) { | |
2935 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
2936 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
2937 | ST.accept_buff[ ST.accepted ] = tmp; | |
2938 | best = ST.accepted; | |
a3621e74 | 2939 | } |
166ba7cd | 2940 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
786e8c11 YO |
2941 | if ( !ST.jump ) { |
2942 | PUSH_STATE_GOTO(TRIE_next, ST.B); | |
2943 | /* NOTREACHED */ | |
2944 | } else { | |
2945 | PUSH_STATE_GOTO(TRIE_next, ST.B - ST.jump[ST.accept_buff[best].wordnum]); | |
2946 | /* NOTREACHED */ | |
2947 | } | |
2948 | /* NOTREACHED */ | |
166ba7cd | 2949 | } |
166ba7cd DM |
2950 | /* NOTREACHED */ |
2951 | ||
2952 | #undef ST | |
2953 | ||
95b24440 DM |
2954 | case EXACT: { |
2955 | char *s = STRING(scan); | |
24d3c4a9 | 2956 | ln = STR_LEN(scan); |
eb160463 | 2957 | if (do_utf8 != UTF) { |
bc517b45 | 2958 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 2959 | char *l = locinput; |
24d3c4a9 | 2960 | const char * const e = s + ln; |
a72c7584 | 2961 | |
5ff6fc6d JH |
2962 | if (do_utf8) { |
2963 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 2964 | while (s < e) { |
a3b680e6 | 2965 | STRLEN ulen; |
1aa99e6b | 2966 | if (l >= PL_regeol) |
5ff6fc6d JH |
2967 | sayNO; |
2968 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 2969 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 2970 | uniflags)) |
5ff6fc6d | 2971 | sayNO; |
bc517b45 | 2972 | l += ulen; |
5ff6fc6d | 2973 | s ++; |
1aa99e6b | 2974 | } |
5ff6fc6d JH |
2975 | } |
2976 | else { | |
2977 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 2978 | while (s < e) { |
a3b680e6 | 2979 | STRLEN ulen; |
1aa99e6b IH |
2980 | if (l >= PL_regeol) |
2981 | sayNO; | |
5ff6fc6d | 2982 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 2983 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 2984 | uniflags)) |
1aa99e6b | 2985 | sayNO; |
bc517b45 | 2986 | s += ulen; |
a72c7584 | 2987 | l ++; |
1aa99e6b | 2988 | } |
5ff6fc6d | 2989 | } |
1aa99e6b IH |
2990 | locinput = l; |
2991 | nextchr = UCHARAT(locinput); | |
2992 | break; | |
2993 | } | |
bc517b45 | 2994 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
2995 | /* Inline the first character, for speed. */ |
2996 | if (UCHARAT(s) != nextchr) | |
2997 | sayNO; | |
24d3c4a9 | 2998 | if (PL_regeol - locinput < ln) |
d6a28714 | 2999 | sayNO; |
24d3c4a9 | 3000 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3001 | sayNO; |
24d3c4a9 | 3002 | locinput += ln; |
d6a28714 JH |
3003 | nextchr = UCHARAT(locinput); |
3004 | break; | |
95b24440 | 3005 | } |
d6a28714 | 3006 | case EXACTFL: |
b8c5462f JH |
3007 | PL_reg_flags |= RF_tainted; |
3008 | /* FALL THROUGH */ | |
95b24440 | 3009 | case EXACTF: { |
be8e71aa | 3010 | char * const s = STRING(scan); |
24d3c4a9 | 3011 | ln = STR_LEN(scan); |
d6a28714 | 3012 | |
d07ddd77 JH |
3013 | if (do_utf8 || UTF) { |
3014 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3015 | const char * const l = locinput; |
d07ddd77 | 3016 | char *e = PL_regeol; |
bc517b45 | 3017 | |
24d3c4a9 | 3018 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3019 | l, &e, 0, do_utf8)) { |
5486206c JH |
3020 | /* One more case for the sharp s: |
3021 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3022 | * the 0xC3 0x9F are the UTF-8 | |
3023 | * byte sequence for the U+00DF. */ | |
3024 | if (!(do_utf8 && | |
3025 | toLOWER(s[0]) == 's' && | |
24d3c4a9 | 3026 | ln >= 2 && |
5486206c JH |
3027 | toLOWER(s[1]) == 's' && |
3028 | (U8)l[0] == 0xC3 && | |
3029 | e - l >= 2 && | |
3030 | (U8)l[1] == 0x9F)) | |
3031 | sayNO; | |
3032 | } | |
d07ddd77 JH |
3033 | locinput = e; |
3034 | nextchr = UCHARAT(locinput); | |
3035 | break; | |
a0ed51b3 | 3036 | } |
d6a28714 | 3037 | |
bc517b45 JH |
3038 | /* Neither the target and the pattern are utf8. */ |
3039 | ||
d6a28714 JH |
3040 | /* Inline the first character, for speed. */ |
3041 | if (UCHARAT(s) != nextchr && | |
3042 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3043 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3044 | sayNO; |
24d3c4a9 | 3045 | if (PL_regeol - locinput < ln) |
b8c5462f | 3046 | sayNO; |
24d3c4a9 DM |
3047 | if (ln > 1 && (OP(scan) == EXACTF |
3048 | ? ibcmp(s, locinput, ln) | |
3049 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3050 | sayNO; |
24d3c4a9 | 3051 | locinput += ln; |
d6a28714 | 3052 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3053 | break; |
95b24440 | 3054 | } |
d6a28714 | 3055 | case ANYOF: |
ffc61ed2 | 3056 | if (do_utf8) { |
9e55ce06 JH |
3057 | STRLEN inclasslen = PL_regeol - locinput; |
3058 | ||
32fc9b6a | 3059 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3060 | goto anyof_fail; |
ffc61ed2 JH |
3061 | if (locinput >= PL_regeol) |
3062 | sayNO; | |
0f0076b4 | 3063 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3064 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3065 | break; |
ffc61ed2 JH |
3066 | } |
3067 | else { | |
3068 | if (nextchr < 0) | |
3069 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3070 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3071 | goto anyof_fail; |
ffc61ed2 JH |
3072 | if (!nextchr && locinput >= PL_regeol) |
3073 | sayNO; | |
3074 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3075 | break; |
3076 | } | |
262b90c4 | 3077 | anyof_fail: |
e0f9d4a8 JH |
3078 | /* If we might have the case of the German sharp s |
3079 | * in a casefolding Unicode character class. */ | |
3080 | ||
ebc501f0 JH |
3081 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3082 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3083 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3084 | } |
e0f9d4a8 JH |
3085 | else |
3086 | sayNO; | |
b8c5462f | 3087 | break; |
d6a28714 | 3088 | case ALNUML: |
b8c5462f JH |
3089 | PL_reg_flags |= RF_tainted; |
3090 | /* FALL THROUGH */ | |
d6a28714 | 3091 | case ALNUM: |
b8c5462f | 3092 | if (!nextchr) |
4633a7c4 | 3093 | sayNO; |
ffc61ed2 | 3094 | if (do_utf8) { |
1a4fad37 | 3095 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3096 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3097 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3098 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3099 | { |
3100 | sayNO; | |
a0ed51b3 | 3101 | } |
b8c5462f | 3102 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3103 | nextchr = UCHARAT(locinput); |
3104 | break; | |
3105 | } | |
ffc61ed2 | 3106 | if (!(OP(scan) == ALNUM |
d6a28714 | 3107 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3108 | sayNO; |
b8c5462f | 3109 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3110 | break; |
d6a28714 | 3111 | case NALNUML: |
b8c5462f JH |
3112 | PL_reg_flags |= RF_tainted; |
3113 | /* FALL THROUGH */ | |
d6a28714 JH |
3114 | case NALNUM: |
3115 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3116 | sayNO; |
ffc61ed2 | 3117 | if (do_utf8) { |
1a4fad37 | 3118 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3119 | if (OP(scan) == NALNUM |
bb7a0f54 | 3120 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3121 | : isALNUM_LC_utf8((U8*)locinput)) |
3122 | { | |
b8c5462f | 3123 | sayNO; |
d6a28714 | 3124 | } |
b8c5462f JH |
3125 | locinput += PL_utf8skip[nextchr]; |
3126 | nextchr = UCHARAT(locinput); | |
3127 | break; | |
3128 | } | |
ffc61ed2 | 3129 | if (OP(scan) == NALNUM |
d6a28714 | 3130 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3131 | sayNO; |
76e3520e | 3132 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3133 | break; |
d6a28714 JH |
3134 | case BOUNDL: |
3135 | case NBOUNDL: | |
3280af22 | 3136 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3137 | /* FALL THROUGH */ |
d6a28714 JH |
3138 | case BOUND: |
3139 | case NBOUND: | |
3140 | /* was last char in word? */ | |
ffc61ed2 | 3141 | if (do_utf8) { |
12d33761 | 3142 | if (locinput == PL_bostr) |
24d3c4a9 | 3143 | ln = '\n'; |
ffc61ed2 | 3144 | else { |
a3b680e6 | 3145 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3146 | |
24d3c4a9 | 3147 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3148 | } |
3149 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3150 | ln = isALNUM_uni(ln); |
1a4fad37 | 3151 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3152 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3153 | } |
3154 | else { | |
24d3c4a9 | 3155 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3156 | n = isALNUM_LC_utf8((U8*)locinput); |
3157 | } | |
a0ed51b3 | 3158 | } |
d6a28714 | 3159 | else { |
24d3c4a9 | 3160 | ln = (locinput != PL_bostr) ? |
12d33761 | 3161 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3162 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3163 | ln = isALNUM(ln); |
ffc61ed2 JH |
3164 | n = isALNUM(nextchr); |
3165 | } | |
3166 | else { | |
24d3c4a9 | 3167 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3168 | n = isALNUM_LC(nextchr); |
3169 | } | |
d6a28714 | 3170 | } |
24d3c4a9 | 3171 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3172 | OP(scan) == BOUNDL)) |
3173 | sayNO; | |
a0ed51b3 | 3174 | break; |
d6a28714 | 3175 | case SPACEL: |
3280af22 | 3176 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3177 | /* FALL THROUGH */ |
d6a28714 | 3178 | case SPACE: |
9442cb0e | 3179 | if (!nextchr) |
4633a7c4 | 3180 | sayNO; |
1aa99e6b | 3181 | if (do_utf8) { |
fd400ab9 | 3182 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3183 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3184 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3185 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3186 | : isSPACE_LC_utf8((U8*)locinput))) |
3187 | { | |
3188 | sayNO; | |
3189 | } | |
3190 | locinput += PL_utf8skip[nextchr]; | |
3191 | nextchr = UCHARAT(locinput); | |
3192 | break; | |
d6a28714 | 3193 | } |
ffc61ed2 JH |
3194 | if (!(OP(scan) == SPACE |
3195 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3196 | sayNO; | |
3197 | nextchr = UCHARAT(++locinput); | |
3198 | } | |
3199 | else { | |
3200 | if (!(OP(scan) == SPACE | |
3201 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3202 | sayNO; | |
3203 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3204 | } |
a0ed51b3 | 3205 | break; |
d6a28714 | 3206 | case NSPACEL: |
3280af22 | 3207 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3208 | /* FALL THROUGH */ |
d6a28714 | 3209 | case NSPACE: |
9442cb0e | 3210 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3211 | sayNO; |
1aa99e6b | 3212 | if (do_utf8) { |
1a4fad37 | 3213 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3214 | if (OP(scan) == NSPACE |
bb7a0f54 | 3215 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3216 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3217 | { |
3218 | sayNO; | |
3219 | } | |
3220 | locinput += PL_utf8skip[nextchr]; | |
3221 | nextchr = UCHARAT(locinput); | |
3222 | break; | |
a0ed51b3 | 3223 | } |
ffc61ed2 | 3224 | if (OP(scan) == NSPACE |
d6a28714 | 3225 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3226 | sayNO; |
76e3520e | 3227 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3228 | break; |
d6a28714 | 3229 | case DIGITL: |
a0ed51b3 LW |
3230 | PL_reg_flags |= RF_tainted; |
3231 | /* FALL THROUGH */ | |
d6a28714 | 3232 | case DIGIT: |
9442cb0e | 3233 | if (!nextchr) |
a0ed51b3 | 3234 | sayNO; |
1aa99e6b | 3235 | if (do_utf8) { |
1a4fad37 | 3236 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3237 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3238 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3239 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3240 | { |
a0ed51b3 | 3241 | sayNO; |
dfe13c55 | 3242 | } |
6f06b55f | 3243 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3244 | nextchr = UCHARAT(locinput); |
3245 | break; | |
3246 | } | |
ffc61ed2 | 3247 | if (!(OP(scan) == DIGIT |
9442cb0e | 3248 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3249 | sayNO; |
3250 | nextchr = UCHARAT(++locinput); | |
3251 | break; | |
d6a28714 | 3252 | case NDIGITL: |
b8c5462f JH |
3253 | PL_reg_flags |= RF_tainted; |
3254 | /* FALL THROUGH */ | |
d6a28714 | 3255 | case NDIGIT: |
9442cb0e | 3256 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3257 | sayNO; |
1aa99e6b | 3258 | if (do_utf8) { |
1a4fad37 | 3259 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3260 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3261 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3262 | : isDIGIT_LC_utf8((U8*)locinput)) |
3263 | { | |
a0ed51b3 | 3264 | sayNO; |
9442cb0e | 3265 | } |
6f06b55f | 3266 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3267 | nextchr = UCHARAT(locinput); |
3268 | break; | |
3269 | } | |
ffc61ed2 | 3270 | if (OP(scan) == NDIGIT |
9442cb0e | 3271 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3272 | sayNO; |
3273 | nextchr = UCHARAT(++locinput); | |
3274 | break; | |
3275 | case CLUMP: | |
b7c83a7e | 3276 | if (locinput >= PL_regeol) |
a0ed51b3 | 3277 | sayNO; |
b7c83a7e | 3278 | if (do_utf8) { |
1a4fad37 | 3279 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3280 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3281 | sayNO; | |
3282 | locinput += PL_utf8skip[nextchr]; | |
3283 | while (locinput < PL_regeol && | |
3284 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3285 | locinput += UTF8SKIP(locinput); | |
3286 | if (locinput > PL_regeol) | |
3287 | sayNO; | |
eb08e2da JH |
3288 | } |
3289 | else | |
3290 | locinput++; | |
a0ed51b3 LW |
3291 | nextchr = UCHARAT(locinput); |
3292 | break; | |
81714fb9 YO |
3293 | |
3294 | case NREFFL: | |
3295 | { | |
3296 | char *s; | |
3297 | char type = OP(scan); | |
3298 | PL_reg_flags |= RF_tainted; | |
3299 | /* FALL THROUGH */ | |
3300 | case NREF: | |
3301 | case NREFF: | |
3302 | { | |
3303 | SV *sv_dat=(SV*)rex->data->data[ ARG( scan ) ]; | |
3304 | I32 *nums=(I32*)SvPVX(sv_dat); | |
3305 | for ( n=0; n<SvIVX(sv_dat); n++ ) { | |
3306 | if ((I32)*PL_reglastparen >= nums[n] && | |
3307 | PL_regstartp[nums[n]] != -1 && | |
3308 | PL_regendp[nums[n]] != -1) | |
3309 | { | |
3310 | n = nums[n]; | |
3311 | type = REF + ( type - NREF ); | |
3312 | goto do_ref; | |
3313 | } | |
3314 | } | |
3315 | sayNO; | |
3316 | /* unreached */ | |
3317 | } | |
c8756f30 | 3318 | case REFFL: |
3280af22 | 3319 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3320 | /* FALL THROUGH */ |
c277df42 | 3321 | case REF: |
81714fb9 | 3322 | case REFF: |
c277df42 | 3323 | n = ARG(scan); /* which paren pair */ |
81714fb9 YO |
3324 | type = OP(scan); |
3325 | do_ref: | |
24d3c4a9 | 3326 | ln = PL_regstartp[n]; |
2c2d71f5 | 3327 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3328 | if ((I32)*PL_reglastparen < n || ln == -1) |
af3f8c16 | 3329 | sayNO; /* Do not match unless seen CLOSEn. */ |
24d3c4a9 | 3330 | if (ln == PL_regendp[n]) |
a0d0e21e | 3331 | break; |
a0ed51b3 | 3332 | |
24d3c4a9 | 3333 | s = PL_bostr + ln; |
81714fb9 | 3334 | if (do_utf8 && type != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3335 | char *l = locinput; |
a3b680e6 | 3336 | const char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 LW |
3337 | /* |
3338 | * Note that we can't do the "other character" lookup trick as | |
3339 | * in the 8-bit case (no pun intended) because in Unicode we | |
3340 | * have to map both upper and title case to lower case. | |
3341 | */ | |
81714fb9 | 3342 | if (type == REFF) { |
a0ed51b3 | 3343 | while (s < e) { |
a3b680e6 AL |
3344 | STRLEN ulen1, ulen2; |
3345 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3346 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3347 | ||
a0ed51b3 LW |
3348 | if (l >= PL_regeol) |
3349 | sayNO; | |
a2a2844f JH |
3350 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3351 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3352 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3353 | sayNO; |
a2a2844f JH |
3354 | s += ulen1; |
3355 | l += ulen2; | |
a0ed51b3 LW |
3356 | } |
3357 | } | |
3358 | locinput = l; | |
3359 | nextchr = UCHARAT(locinput); | |
3360 | break; | |
3361 | } | |
3362 | ||
a0d0e21e | 3363 | /* Inline the first character, for speed. */ |
76e3520e | 3364 | if (UCHARAT(s) != nextchr && |
81714fb9 YO |
3365 | (type == REF || |
3366 | (UCHARAT(s) != (type == REFF | |
3367 | ? PL_fold : PL_fold_locale)[nextchr]))) | |
4633a7c4 | 3368 | sayNO; |
24d3c4a9 DM |
3369 | ln = PL_regendp[n] - ln; |
3370 | if (locinput + ln > PL_regeol) | |
4633a7c4 | 3371 | sayNO; |
81714fb9 | 3372 | if (ln > 1 && (type == REF |
24d3c4a9 | 3373 | ? memNE(s, locinput, ln) |
81714fb9 | 3374 | : (type == REFF |
24d3c4a9 DM |
3375 | ? ibcmp(s, locinput, ln) |
3376 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 3377 | sayNO; |
24d3c4a9 | 3378 | locinput += ln; |
76e3520e | 3379 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3380 | break; |
81714fb9 | 3381 | } |
a0d0e21e | 3382 | case NOTHING: |
c277df42 | 3383 | case TAIL: |
a0d0e21e LW |
3384 | break; |
3385 | case BACK: | |
3386 | break; | |
40a82448 DM |
3387 | |
3388 | #undef ST | |
3389 | #define ST st->u.eval | |
c277df42 | 3390 | { |
c277df42 | 3391 | SV *ret; |
6bda09f9 YO |
3392 | regexp *re; |
3393 | regnode *startpoint; | |
3394 | ||
3395 | case SRECURSE: | |
3396 | case RECURSE: /* /(...(?1))/ */ | |
3397 | if (cur_eval && cur_eval->locinput==locinput) { | |
3398 | if (cur_eval->u.eval.close_paren == ARG(scan)) | |
3399 | Perl_croak(aTHX_ "Infinite recursion in RECURSE in regexp"); | |
3400 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) | |
3401 | Perl_croak(aTHX_ "RECURSE without pos change exceeded limit in regexp"); | |
3402 | } else { | |
3403 | nochange_depth = 0; | |
3404 | } | |
3405 | re = rex; | |
3406 | (void)ReREFCNT_inc(rex); | |
3407 | if (OP(scan)==RECURSE) { | |
3408 | startpoint = scan + ARG2L(scan); | |
3409 | ST.close_paren = ARG(scan); | |
3410 | } else { | |
3411 | startpoint = re->program+1; | |
3412 | ST.close_paren = 0; | |
3413 | } | |
3414 | goto eval_recurse_doit; | |
3415 | /* NOTREACHED */ | |
3416 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
3417 | if (cur_eval && cur_eval->locinput==locinput) { | |
3418 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) | |
3419 | Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regexp"); | |
3420 | } else { | |
3421 | nochange_depth = 0; | |
3422 | } | |
8e5e9ebe | 3423 | { |
4aabdb9b DM |
3424 | /* execute the code in the {...} */ |
3425 | dSP; | |
6136c704 | 3426 | SV ** const before = SP; |
4aabdb9b DM |
3427 | OP_4tree * const oop = PL_op; |
3428 | COP * const ocurcop = PL_curcop; | |
3429 | PAD *old_comppad; | |
4aabdb9b DM |
3430 | |
3431 | n = ARG(scan); | |
32fc9b6a | 3432 | PL_op = (OP_4tree*)rex->data->data[n]; |
4aabdb9b | 3433 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); |
32fc9b6a | 3434 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]); |
4aabdb9b DM |
3435 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
3436 | ||
8e5e9ebe RGS |
3437 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3438 | SPAGAIN; | |
3439 | if (SP == before) | |
075aa684 | 3440 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3441 | else { |
3442 | ret = POPs; | |
3443 | PUTBACK; | |
3444 | } | |
4aabdb9b DM |
3445 | |
3446 | PL_op = oop; | |
3447 | PAD_RESTORE_LOCAL(old_comppad); | |
3448 | PL_curcop = ocurcop; | |
24d3c4a9 | 3449 | if (!logical) { |
4aabdb9b DM |
3450 | /* /(?{...})/ */ |
3451 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3452 | break; |
3453 | } | |
8e5e9ebe | 3454 | } |
24d3c4a9 DM |
3455 | if (logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3456 | logical = 0; | |
4aabdb9b | 3457 | { |
4f639d21 DM |
3458 | /* extract RE object from returned value; compiling if |
3459 | * necessary */ | |
3460 | ||
6136c704 | 3461 | MAGIC *mg = NULL; |
be8e71aa | 3462 | const SV *sv; |
faf82a0b AE |
3463 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3464 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3465 | else if (SvSMAGICAL(ret)) { | |
3466 | if (SvGMAGICAL(ret)) | |
3467 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3468 | else | |
3469 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3470 | } |
faf82a0b | 3471 | |
0f5d15d6 IZ |
3472 | if (mg) { |
3473 | re = (regexp *)mg->mg_obj; | |
df0003d4 | 3474 | (void)ReREFCNT_inc(re); |
0f5d15d6 IZ |
3475 | } |
3476 | else { | |
3477 | STRLEN len; | |
6136c704 | 3478 | const char * const t = SvPV_const(ret, len); |
0f5d15d6 | 3479 | PMOP pm; |
a3b680e6 | 3480 | const I32 osize = PL_regsize; |
0f5d15d6 | 3481 | |
5fcd1c1b | 3482 | Zero(&pm, 1, PMOP); |
4aabdb9b | 3483 | if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; |
f9f4320a | 3484 | re = CALLREGCOMP((char*)t, (char*)t + len, &pm); |
9041c2e3 | 3485 | if (!(SvFLAGS(ret) |
faf82a0b AE |
3486 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3487 | | SVs_GMG))) | |
14befaf4 DM |
3488 | sv_magic(ret,(SV*)ReREFCNT_inc(re), |
3489 | PERL_MAGIC_qr,0,0); | |
0f5d15d6 | 3490 | PL_regsize = osize; |
0f5d15d6 | 3491 | } |
4aabdb9b | 3492 | } |
6bda09f9 YO |
3493 | DEBUG_EXECUTE_r( |
3494 | debug_start_match(re, do_utf8, locinput, PL_regeol, | |
3495 | "Matching embedded"); | |
3496 | ); | |
3497 | startpoint = re->program + 1; | |
3498 | ST.close_paren = 0; /* only used for RECURSE */ | |
3499 | /* borrowed from regtry */ | |
3500 | if (PL_reg_start_tmpl <= re->nparens) { | |
3501 | PL_reg_start_tmpl = re->nparens*3/2 + 3; | |
3502 | if(PL_reg_start_tmp) | |
3503 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3504 | else | |
3505 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3506 | } | |
aa283a38 | 3507 | |
6bda09f9 | 3508 | eval_recurse_doit: /* Share code with RECURSE below this line */ |
aa283a38 | 3509 | /* run the pattern returned from (??{...}) */ |
40a82448 DM |
3510 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3511 | REGCP_SET(ST.lastcp); | |
6bda09f9 YO |
3512 | |
3513 | PL_regstartp = re->startp; /* essentially NOOP on RECURSE */ | |
3514 | PL_regendp = re->endp; /* essentially NOOP on RECURSE */ | |
3515 | ||
4aabdb9b DM |
3516 | *PL_reglastparen = 0; |
3517 | *PL_reglastcloseparen = 0; | |
4aabdb9b | 3518 | PL_reginput = locinput; |
4aabdb9b DM |
3519 | |
3520 | /* XXXX This is too dramatic a measure... */ | |
3521 | PL_reg_maxiter = 0; | |
3522 | ||
faec1544 DM |
3523 | ST.toggle_reg_flags = PL_reg_flags; |
3524 | if (re->reganch & ROPT_UTF8) | |
3525 | PL_reg_flags |= RF_utf8; | |
3526 | else | |
3527 | PL_reg_flags &= ~RF_utf8; | |
3528 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3529 | ||
40a82448 | 3530 | ST.prev_rex = rex; |
faec1544 | 3531 | ST.prev_curlyx = cur_curlyx; |
aa283a38 | 3532 | rex = re; |
faec1544 | 3533 | cur_curlyx = NULL; |
40a82448 | 3534 | ST.B = next; |
faec1544 DM |
3535 | ST.prev_eval = cur_eval; |
3536 | cur_eval = st; | |
faec1544 | 3537 | /* now continue from first node in postoned RE */ |
6bda09f9 | 3538 | PUSH_YES_STATE_GOTO(EVAL_AB, startpoint); |
4aabdb9b | 3539 | /* NOTREACHED */ |
a0ed51b3 | 3540 | } |
24d3c4a9 DM |
3541 | /* logical is 1, /(?(?{...})X|Y)/ */ |
3542 | sw = (bool)SvTRUE(ret); | |
3543 | logical = 0; | |
c277df42 IZ |
3544 | break; |
3545 | } | |
40a82448 | 3546 | |
faec1544 DM |
3547 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3548 | /* note: this is called twice; first after popping B, then A */ | |
3549 | PL_reg_flags ^= ST.toggle_reg_flags; | |
40a82448 DM |
3550 | ReREFCNT_dec(rex); |
3551 | rex = ST.prev_rex; | |
faec1544 DM |
3552 | regcpblow(ST.cp); |
3553 | cur_eval = ST.prev_eval; | |
3554 | cur_curlyx = ST.prev_curlyx; | |
40a82448 DM |
3555 | /* XXXX This is too dramatic a measure... */ |
3556 | PL_reg_maxiter = 0; | |
262b90c4 | 3557 | sayYES; |
40a82448 | 3558 | |
40a82448 | 3559 | |
faec1544 DM |
3560 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3561 | /* note: this is called twice; first after popping B, then A */ | |
3562 | PL_reg_flags ^= ST.toggle_reg_flags; | |
3563 | ReREFCNT_dec(rex); | |
3564 | rex = ST.prev_rex; | |
40a82448 DM |
3565 | PL_reginput = locinput; |
3566 | REGCP_UNWIND(ST.lastcp); | |
3567 | regcppop(rex); | |
faec1544 DM |
3568 | cur_eval = ST.prev_eval; |
3569 | cur_curlyx = ST.prev_curlyx; | |
3570 | /* XXXX This is too dramatic a measure... */ | |
3571 | PL_reg_maxiter = 0; | |
40a82448 | 3572 | sayNO_SILENT; |
40a82448 DM |
3573 | #undef ST |
3574 | ||
a0d0e21e | 3575 | case OPEN: |
c277df42 | 3576 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3577 | PL_reg_start_tmp[n] = locinput; |
3578 | if (n > PL_regsize) | |
3579 | PL_regsize = n; | |
a0d0e21e LW |
3580 | break; |
3581 | case CLOSE: | |
c277df42 | 3582 | n = ARG(scan); /* which paren pair */ |
cf93c79d IZ |
3583 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
3584 | PL_regendp[n] = locinput - PL_bostr; | |
eb160463 | 3585 | if (n > (I32)*PL_reglastparen) |
3280af22 | 3586 | *PL_reglastparen = n; |
a01268b5 | 3587 | *PL_reglastcloseparen = n; |
6bda09f9 YO |
3588 | if (cur_eval && cur_eval->u.eval.close_paren == (U32)n) { |
3589 | goto fake_end; | |
3590 | } | |
a0d0e21e | 3591 | break; |
c277df42 IZ |
3592 | case GROUPP: |
3593 | n = ARG(scan); /* which paren pair */ | |
24d3c4a9 | 3594 | sw = (bool)((I32)*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 IZ |
3595 | break; |
3596 | case IFTHEN: | |
2c2d71f5 | 3597 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3598 | if (sw) |
c277df42 IZ |
3599 | next = NEXTOPER(NEXTOPER(scan)); |
3600 | else { | |
3601 | next = scan + ARG(scan); | |
3602 | if (OP(next) == IFTHEN) /* Fake one. */ | |
3603 | next = NEXTOPER(NEXTOPER(next)); | |
3604 | } | |
3605 | break; | |
3606 | case LOGICAL: | |
24d3c4a9 | 3607 | logical = scan->flags; |
c277df42 | 3608 | break; |
c476f425 | 3609 | |
2ab05381 | 3610 | /******************************************************************* |
2ab05381 | 3611 | |
c476f425 DM |
3612 | The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/ |
3613 | pattern, where A and B are subpatterns. (For simple A, CURLYM or | |
3614 | STAR/PLUS/CURLY/CURLYN are used instead.) | |
2ab05381 | 3615 | |
c476f425 | 3616 | A*B is compiled as <CURLYX><A><WHILEM><B> |
2ab05381 | 3617 | |
c476f425 DM |
3618 | On entry to the subpattern, CURLYX is called. This pushes a CURLYX |
3619 | state, which contains the current count, initialised to -1. It also sets | |
3620 | cur_curlyx to point to this state, with any previous value saved in the | |
3621 | state block. | |
2ab05381 | 3622 | |
c476f425 DM |
3623 | CURLYX then jumps straight to the WHILEM op, rather than executing A, |
3624 | since the pattern may possibly match zero times (i.e. it's a while {} loop | |
3625 | rather than a do {} while loop). | |
2ab05381 | 3626 | |
c476f425 DM |
3627 | Each entry to WHILEM represents a successful match of A. The count in the |
3628 | CURLYX block is incremented, another WHILEM state is pushed, and execution | |
3629 | passes to A or B depending on greediness and the current count. | |
2ab05381 | 3630 | |
c476f425 DM |
3631 | For example, if matching against the string a1a2a3b (where the aN are |
3632 | substrings that match /A/), then the match progresses as follows: (the | |
3633 | pushed states are interspersed with the bits of strings matched so far): | |
2ab05381 | 3634 | |
c476f425 DM |
3635 | <CURLYX cnt=-1> |
3636 | <CURLYX cnt=0><WHILEM> | |
3637 | <CURLYX cnt=1><WHILEM> a1 <WHILEM> | |
3638 | <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM> | |
3639 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> | |
3640 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b | |
2ab05381 | 3641 | |
c476f425 DM |
3642 | (Contrast this with something like CURLYM, which maintains only a single |
3643 | backtrack state: | |
2ab05381 | 3644 | |
c476f425 DM |
3645 | <CURLYM cnt=0> a1 |
3646 | a1 <CURLYM cnt=1> a2 | |
3647 | a1 a2 <CURLYM cnt=2> a3 | |
3648 | a1 a2 a3 <CURLYM cnt=3> b | |
3649 | ) | |
2ab05381 | 3650 | |
c476f425 DM |
3651 | Each WHILEM state block marks a point to backtrack to upon partial failure |
3652 | of A or B, and also contains some minor state data related to that | |
3653 | iteration. The CURLYX block, pointed to by cur_curlyx, contains the | |
3654 | overall state, such as the count, and pointers to the A and B ops. | |
2ab05381 | 3655 | |
c476f425 DM |
3656 | This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx |
3657 | must always point to the *current* CURLYX block, the rules are: | |
2ab05381 | 3658 | |
c476f425 DM |
3659 | When executing CURLYX, save the old cur_curlyx in the CURLYX state block, |
3660 | and set cur_curlyx to point the new block. | |
2ab05381 | 3661 | |
c476f425 DM |
3662 | When popping the CURLYX block after a successful or unsuccessful match, |
3663 | restore the previous cur_curlyx. | |
2ab05381 | 3664 | |
c476f425 DM |
3665 | When WHILEM is about to execute B, save the current cur_curlyx, and set it |
3666 | to the outer one saved in the CURLYX block. | |
2ab05381 | 3667 | |
c476f425 DM |
3668 | When popping the WHILEM block after a successful or unsuccessful B match, |
3669 | restore the previous cur_curlyx. | |
2ab05381 | 3670 | |
c476f425 DM |
3671 | Here's an example for the pattern (AI* BI)*BO |
3672 | I and O refer to inner and outer, C and W refer to CURLYX and WHILEM: | |
2ab05381 | 3673 | |
c476f425 DM |
3674 | cur_ |
3675 | curlyx backtrack stack | |
3676 | ------ --------------- | |
3677 | NULL | |
3678 | CO <CO prev=NULL> <WO> | |
3679 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3680 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi | |
3681 | NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo | |
2ab05381 | 3682 | |
c476f425 DM |
3683 | At this point the pattern succeeds, and we work back down the stack to |
3684 | clean up, restoring as we go: | |
95b24440 | 3685 | |
c476f425 DM |
3686 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi |
3687 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3688 | CO <CO prev=NULL> <WO> | |
3689 | NULL | |
a0374537 | 3690 | |
c476f425 DM |
3691 | *******************************************************************/ |
3692 | ||
3693 | #define ST st->u.curlyx | |
3694 | ||
3695 | case CURLYX: /* start of /A*B/ (for complex A) */ | |
3696 | { | |
3697 | /* No need to save/restore up to this paren */ | |
3698 | I32 parenfloor = scan->flags; | |
3699 | ||
3700 | assert(next); /* keep Coverity happy */ | |
3701 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
3702 | next += ARG(next); | |
3703 | ||
3704 | /* XXXX Probably it is better to teach regpush to support | |
3705 | parenfloor > PL_regsize... */ | |
3706 | if (parenfloor > (I32)*PL_reglastparen) | |
3707 | parenfloor = *PL_reglastparen; /* Pessimization... */ | |
3708 | ||
3709 | ST.prev_curlyx= cur_curlyx; | |
3710 | cur_curlyx = st; | |
3711 | ST.cp = PL_savestack_ix; | |
3712 | ||
3713 | /* these fields contain the state of the current curly. | |
3714 | * they are accessed by subsequent WHILEMs */ | |
3715 | ST.parenfloor = parenfloor; | |
3716 | ST.min = ARG1(scan); | |
3717 | ST.max = ARG2(scan); | |
3718 | ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
3719 | ST.B = next; | |
24d3c4a9 DM |
3720 | ST.minmod = minmod; |
3721 | minmod = 0; | |
c476f425 DM |
3722 | ST.count = -1; /* this will be updated by WHILEM */ |
3723 | ST.lastloc = NULL; /* this will be updated by WHILEM */ | |
3724 | ||
3725 | PL_reginput = locinput; | |
3726 | PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next)); | |
5f66b61c | 3727 | /* NOTREACHED */ |
c476f425 | 3728 | } |
a0d0e21e | 3729 | |
c476f425 DM |
3730 | case CURLYX_end: /* just finished matching all of A*B */ |
3731 | regcpblow(ST.cp); | |
3732 | cur_curlyx = ST.prev_curlyx; | |
3733 | sayYES; | |
3734 | /* NOTREACHED */ | |
a0d0e21e | 3735 | |
c476f425 DM |
3736 | case CURLYX_end_fail: /* just failed to match all of A*B */ |
3737 | regcpblow(ST.cp); | |
3738 | cur_curlyx = ST.prev_curlyx; | |
3739 | sayNO; | |
3740 | /* NOTREACHED */ | |
4633a7c4 | 3741 | |
a0d0e21e | 3742 | |
c476f425 DM |
3743 | #undef ST |
3744 | #define ST st->u.whilem | |
3745 | ||
3746 | case WHILEM: /* just matched an A in /A*B/ (for complex A) */ | |
3747 | { | |
3748 | /* see the discussion above about CURLYX/WHILEM */ | |
3749 | ||
3750 | I32 n; | |
3751 | assert(cur_curlyx); /* keep Coverity happy */ | |
3752 | n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */ | |
3753 | ST.save_lastloc = cur_curlyx->u.curlyx.lastloc; | |
3754 | ST.cache_offset = 0; | |
3755 | ST.cache_mask = 0; | |
3756 | ||
3757 | PL_reginput = locinput; | |
3758 | ||
3759 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3760 | "%*s whilem: matched %ld out of %ld..%ld\n", | |
3761 | REPORT_CODE_OFF+depth*2, "", (long)n, | |
3762 | (long)cur_curlyx->u.curlyx.min, | |
3763 | (long)cur_curlyx->u.curlyx.max) | |
3764 | ); | |
a0d0e21e | 3765 | |
c476f425 | 3766 | /* First just match a string of min A's. */ |
a0d0e21e | 3767 | |
c476f425 DM |
3768 | if (n < cur_curlyx->u.curlyx.min) { |
3769 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3770 | PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A); | |
3771 | /* NOTREACHED */ | |
3772 | } | |
3773 | ||
3774 | /* If degenerate A matches "", assume A done. */ | |
3775 | ||
3776 | if (locinput == cur_curlyx->u.curlyx.lastloc) { | |
3777 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3778 | "%*s whilem: empty match detected, trying continuation...\n", | |
3779 | REPORT_CODE_OFF+depth*2, "") | |
3780 | ); | |
3781 | goto do_whilem_B_max; | |
3782 | } | |
3783 | ||
3784 | /* super-linear cache processing */ | |
3785 | ||
3786 | if (scan->flags) { | |
a0d0e21e | 3787 | |
2c2d71f5 | 3788 | if (!PL_reg_maxiter) { |
c476f425 DM |
3789 | /* start the countdown: Postpone detection until we |
3790 | * know the match is not *that* much linear. */ | |
2c2d71f5 | 3791 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
66bf836d DM |
3792 | /* possible overflow for long strings and many CURLYX's */ |
3793 | if (PL_reg_maxiter < 0) | |
3794 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
3795 | PL_reg_leftiter = PL_reg_maxiter; |
3796 | } | |
c476f425 | 3797 | |
2c2d71f5 | 3798 | if (PL_reg_leftiter-- == 0) { |
c476f425 | 3799 | /* initialise cache */ |
3298f257 | 3800 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 3801 | if (PL_reg_poscache) { |
eb160463 | 3802 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
3803 | Renew(PL_reg_poscache, size, char); |
3804 | PL_reg_poscache_size = size; | |
3805 | } | |
3806 | Zero(PL_reg_poscache, size, char); | |
3807 | } | |
3808 | else { | |
3809 | PL_reg_poscache_size = size; | |
a02a5408 | 3810 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 3811 | } |
c476f425 DM |
3812 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
3813 | "%swhilem: Detected a super-linear match, switching on caching%s...\n", | |
3814 | PL_colors[4], PL_colors[5]) | |
3815 | ); | |
2c2d71f5 | 3816 | } |
c476f425 | 3817 | |
2c2d71f5 | 3818 | if (PL_reg_leftiter < 0) { |
c476f425 DM |
3819 | /* have we already failed at this position? */ |
3820 | I32 offset, mask; | |
3821 | offset = (scan->flags & 0xf) - 1 | |
3822 | + (locinput - PL_bostr) * (scan->flags>>4); | |
3823 | mask = 1 << (offset % 8); | |
3824 | offset /= 8; | |
3825 | if (PL_reg_poscache[offset] & mask) { | |
3826 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3827 | "%*s whilem: (cache) already tried at this position...\n", | |
3828 | REPORT_CODE_OFF+depth*2, "") | |
2c2d71f5 | 3829 | ); |
3298f257 | 3830 | sayNO; /* cache records failure */ |
2c2d71f5 | 3831 | } |
c476f425 DM |
3832 | ST.cache_offset = offset; |
3833 | ST.cache_mask = mask; | |
2c2d71f5 | 3834 | } |
c476f425 | 3835 | } |
2c2d71f5 | 3836 | |
c476f425 | 3837 | /* Prefer B over A for minimal matching. */ |
a687059c | 3838 | |
c476f425 DM |
3839 | if (cur_curlyx->u.curlyx.minmod) { |
3840 | ST.save_curlyx = cur_curlyx; | |
3841 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
3842 | ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor); | |
3843 | REGCP_SET(ST.lastcp); | |
3844 | PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B); | |
3845 | /* NOTREACHED */ | |
3846 | } | |
a0d0e21e | 3847 | |
c476f425 DM |
3848 | /* Prefer A over B for maximal matching. */ |
3849 | ||
3850 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ | |
3851 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
3852 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3853 | REGCP_SET(ST.lastcp); | |
3854 | PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A); | |
3855 | /* NOTREACHED */ | |
3856 | } | |
3857 | goto do_whilem_B_max; | |
3858 | } | |
3859 | /* NOTREACHED */ | |
3860 | ||
3861 | case WHILEM_B_min: /* just matched B in a minimal match */ | |
3862 | case WHILEM_B_max: /* just matched B in a maximal match */ | |
3863 | cur_curlyx = ST.save_curlyx; | |
3864 | sayYES; | |
3865 | /* NOTREACHED */ | |
3866 | ||
3867 | case WHILEM_B_max_fail: /* just failed to match B in a maximal match */ | |
3868 | cur_curlyx = ST.save_curlyx; | |
3869 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
3870 | cur_curlyx->u.curlyx.count--; | |
3871 | CACHEsayNO; | |
3872 | /* NOTREACHED */ | |
3873 | ||
3874 | case WHILEM_A_min_fail: /* just failed to match A in a minimal match */ | |
3875 | REGCP_UNWIND(ST.lastcp); | |
3876 | regcppop(rex); | |
3877 | /* FALL THROUGH */ | |
3878 | case WHILEM_A_pre_fail: /* just failed to match even minimal A */ | |
3879 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
3880 | cur_curlyx->u.curlyx.count--; | |
3881 | CACHEsayNO; | |
3882 | /* NOTREACHED */ | |
3883 | ||
3884 | case WHILEM_A_max_fail: /* just failed to match A in a maximal match */ | |
3885 | REGCP_UNWIND(ST.lastcp); | |
3886 | regcppop(rex); /* Restore some previous $<digit>s? */ | |
3887 | PL_reginput = locinput; | |
3888 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
3889 | "%*s whilem: failed, trying continuation...\n", | |
3890 | REPORT_CODE_OFF+depth*2, "") | |
3891 | ); | |
3892 | do_whilem_B_max: | |
3893 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
3894 | && ckWARN(WARN_REGEXP) | |
3895 | && !(PL_reg_flags & RF_warned)) | |
3896 | { | |
3897 | PL_reg_flags |= RF_warned; | |
3898 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", | |
3899 | "Complex regular subexpression recursion", | |
3900 | REG_INFTY - 1); | |
3901 | } | |
3902 | ||
3903 | /* now try B */ | |
3904 | ST.save_curlyx = cur_curlyx; | |
3905 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
3906 | PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B); | |
3907 | /* NOTREACHED */ | |
3908 | ||
3909 | case WHILEM_B_min_fail: /* just failed to match B in a minimal match */ | |
3910 | cur_curlyx = ST.save_curlyx; | |
3911 | REGCP_UNWIND(ST.lastcp); | |
3912 | regcppop(rex); | |
3913 | ||
3914 | if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) { | |
3915 | /* Maximum greed exceeded */ | |
3916 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
3917 | && ckWARN(WARN_REGEXP) | |
3918 | && !(PL_reg_flags & RF_warned)) | |
3919 | { | |
3280af22 | 3920 | PL_reg_flags |= RF_warned; |
c476f425 DM |
3921 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), |
3922 | "%s limit (%d) exceeded", | |
3923 | "Complex regular subexpression recursion", | |
3924 | REG_INFTY - 1); | |
a0d0e21e | 3925 | } |
c476f425 | 3926 | cur_curlyx->u.curlyx.count--; |
3ab3c9b4 | 3927 | CACHEsayNO; |
a0d0e21e | 3928 | } |
c476f425 DM |
3929 | |
3930 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
3931 | "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "") | |
3932 | ); | |
3933 | /* Try grabbing another A and see if it helps. */ | |
3934 | PL_reginput = locinput; | |
3935 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3936 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
3937 | REGCP_SET(ST.lastcp); | |
3938 | PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A); | |
5f66b61c | 3939 | /* NOTREACHED */ |
40a82448 DM |
3940 | |
3941 | #undef ST | |
3942 | #define ST st->u.branch | |
3943 | ||
3944 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
3945 | next = scan + ARG(scan); |
3946 | if (next == scan) | |
3947 | next = NULL; | |
40a82448 DM |
3948 | scan = NEXTOPER(scan); |
3949 | /* FALL THROUGH */ | |
c277df42 | 3950 | |
40a82448 DM |
3951 | case BRANCH: /* /(...|A|...)/ */ |
3952 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
3953 | if (!next || (OP(next) != BRANCH && OP(next) != BRANCHJ)) | |
3954 | /* last branch; skip state push and jump direct to node */ | |
3955 | continue; | |
3956 | ST.lastparen = *PL_reglastparen; | |
3957 | ST.next_branch = next; | |
3958 | REGCP_SET(ST.cp); | |
3959 | PL_reginput = locinput; | |
02db2b7b | 3960 | |
40a82448 DM |
3961 | /* Now go into the branch */ |
3962 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
3963 | /* NOTREACHED */ | |
3964 | ||
3965 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ | |
3966 | REGCP_UNWIND(ST.cp); | |
3967 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
3968 | PL_regendp[n] = -1; | |
3969 | *PL_reglastparen = n; | |
3970 | scan = ST.next_branch; | |
3971 | /* no more branches? */ | |
3972 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) | |
3973 | sayNO; | |
3974 | continue; /* execute next BRANCH[J] op */ | |
3975 | /* NOTREACHED */ | |
3976 | ||
a0d0e21e | 3977 | case MINMOD: |
24d3c4a9 | 3978 | minmod = 1; |
a0d0e21e | 3979 | break; |
40a82448 DM |
3980 | |
3981 | #undef ST | |
3982 | #define ST st->u.curlym | |
3983 | ||
3984 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
3985 | ||
3986 | /* This is an optimisation of CURLYX that enables us to push | |
3987 | * only a single backtracking state, no matter now many matches | |
3988 | * there are in {m,n}. It relies on the pattern being constant | |
3989 | * length, with no parens to influence future backrefs | |
3990 | */ | |
3991 | ||
3992 | ST.me = scan; | |
dc45a647 | 3993 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
3994 | |
3995 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
3996 | if (ST.me->flags) { | |
3997 | I32 paren = ST.me->flags; | |
3998 | if (paren > PL_regsize) | |
3999 | PL_regsize = paren; | |
4000 | if (paren > (I32)*PL_reglastparen) | |
4001 | *PL_reglastparen = paren; | |
c277df42 | 4002 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 4003 | } |
40a82448 DM |
4004 | ST.A = scan; |
4005 | ST.B = next; | |
4006 | ST.alen = 0; | |
4007 | ST.count = 0; | |
24d3c4a9 DM |
4008 | ST.minmod = minmod; |
4009 | minmod = 0; | |
40a82448 DM |
4010 | ST.c1 = CHRTEST_UNINIT; |
4011 | REGCP_SET(ST.cp); | |
6407bf3b | 4012 | |
40a82448 DM |
4013 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
4014 | goto curlym_do_B; | |
4015 | ||
4016 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 4017 | PL_reginput = locinput; |
40a82448 DM |
4018 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
4019 | /* NOTREACHED */ | |
5f80c4cf | 4020 | |
40a82448 DM |
4021 | case CURLYM_A: /* we've just matched an A */ |
4022 | locinput = st->locinput; | |
4023 | nextchr = UCHARAT(locinput); | |
4024 | ||
4025 | ST.count++; | |
4026 | /* after first match, determine A's length: u.curlym.alen */ | |
4027 | if (ST.count == 1) { | |
4028 | if (PL_reg_match_utf8) { | |
4029 | char *s = locinput; | |
4030 | while (s < PL_reginput) { | |
4031 | ST.alen++; | |
4032 | s += UTF8SKIP(s); | |
4033 | } | |
4034 | } | |
4035 | else { | |
4036 | ST.alen = PL_reginput - locinput; | |
4037 | } | |
4038 | if (ST.alen == 0) | |
4039 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
4040 | } | |
0cadcf80 DM |
4041 | DEBUG_EXECUTE_r( |
4042 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 4043 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 4044 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 4045 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
4046 | ); |
4047 | ||
40a82448 DM |
4048 | locinput = PL_reginput; |
4049 | if (ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) | |
4050 | goto curlym_do_A; /* try to match another A */ | |
4051 | goto curlym_do_B; /* try to match B */ | |
5f80c4cf | 4052 | |
40a82448 DM |
4053 | case CURLYM_A_fail: /* just failed to match an A */ |
4054 | REGCP_UNWIND(ST.cp); | |
4055 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ ) | |
4056 | sayNO; | |
0cadcf80 | 4057 | |
40a82448 DM |
4058 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
4059 | PL_reginput = locinput; | |
4060 | if (ST.c1 == CHRTEST_UNINIT) { | |
4061 | /* calculate c1 and c2 for possible match of 1st char | |
4062 | * following curly */ | |
4063 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
4064 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
4065 | regnode *text_node = ST.B; | |
4066 | if (! HAS_TEXT(text_node)) | |
4067 | FIND_NEXT_IMPT(text_node); | |
4068 | if (HAS_TEXT(text_node) | |
4069 | && PL_regkind[OP(text_node)] != REF) | |
4070 | { | |
4071 | ST.c1 = (U8)*STRING(text_node); | |
4072 | ST.c2 = | |
4073 | (OP(text_node) == EXACTF || OP(text_node) == REFF) | |
4074 | ? PL_fold[ST.c1] | |
4075 | : (OP(text_node) == EXACTFL || OP(text_node) == REFFL) | |
4076 | ? PL_fold_locale[ST.c1] | |
4077 | : ST.c1; | |
c277df42 | 4078 | } |
c277df42 | 4079 | } |
40a82448 DM |
4080 | } |
4081 | ||
4082 | DEBUG_EXECUTE_r( | |
4083 | PerlIO_printf(Perl_debug_log, | |
4084 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 4085 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
4086 | "", (IV)ST.count) |
4087 | ); | |
4088 | if (ST.c1 != CHRTEST_VOID | |
4089 | && UCHARAT(PL_reginput) != ST.c1 | |
4090 | && UCHARAT(PL_reginput) != ST.c2) | |
4091 | { | |
4092 | /* simulate B failing */ | |
4093 | state_num = CURLYM_B_fail; | |
4094 | goto reenter_switch; | |
4095 | } | |
4096 | ||
4097 | if (ST.me->flags) { | |
4098 | /* mark current A as captured */ | |
4099 | I32 paren = ST.me->flags; | |
4100 | if (ST.count) { | |
4101 | PL_regstartp[paren] | |
4102 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; | |
4103 | PL_regendp[paren] = PL_reginput - PL_bostr; | |
c277df42 | 4104 | } |
40a82448 DM |
4105 | else |
4106 | PL_regendp[paren] = -1; | |
c277df42 | 4107 | } |
40a82448 | 4108 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4109 | /* NOTREACHED */ |
40a82448 DM |
4110 | |
4111 | case CURLYM_B_fail: /* just failed to match a B */ | |
4112 | REGCP_UNWIND(ST.cp); | |
4113 | if (ST.minmod) { | |
4114 | if (ST.count == ARG2(ST.me) /* max */) | |
4115 | sayNO; | |
4116 | goto curlym_do_A; /* try to match a further A */ | |
4117 | } | |
4118 | /* backtrack one A */ | |
4119 | if (ST.count == ARG1(ST.me) /* min */) | |
4120 | sayNO; | |
4121 | ST.count--; | |
4122 | locinput = HOPc(locinput, -ST.alen); | |
4123 | goto curlym_do_B; /* try to match B */ | |
4124 | ||
c255a977 DM |
4125 | #undef ST |
4126 | #define ST st->u.curly | |
40a82448 | 4127 | |
c255a977 DM |
4128 | #define CURLY_SETPAREN(paren, success) \ |
4129 | if (paren) { \ | |
4130 | if (success) { \ | |
4131 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; \ | |
4132 | PL_regendp[paren] = locinput - PL_bostr; \ | |
4133 | } \ | |
4134 | else \ | |
4135 | PL_regendp[paren] = -1; \ | |
4136 | } | |
4137 | ||
4138 | case STAR: /* /A*B/ where A is width 1 */ | |
4139 | ST.paren = 0; | |
4140 | ST.min = 0; | |
4141 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4142 | scan = NEXTOPER(scan); |
4143 | goto repeat; | |
c255a977 DM |
4144 | case PLUS: /* /A+B/ where A is width 1 */ |
4145 | ST.paren = 0; | |
4146 | ST.min = 1; | |
4147 | ST.max = REG_INFTY; | |
c277df42 | 4148 | scan = NEXTOPER(scan); |
c255a977 DM |
4149 | goto repeat; |
4150 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4151 | ST.paren = scan->flags; /* Which paren to set */ | |
4152 | if (ST.paren > PL_regsize) | |
4153 | PL_regsize = ST.paren; | |
4154 | if (ST.paren > (I32)*PL_reglastparen) | |
4155 | *PL_reglastparen = ST.paren; | |
4156 | ST.min = ARG1(scan); /* min to match */ | |
4157 | ST.max = ARG2(scan); /* max to match */ | |
4158 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); | |
4159 | goto repeat; | |
4160 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4161 | ST.paren = 0; | |
4162 | ST.min = ARG1(scan); /* min to match */ | |
4163 | ST.max = ARG2(scan); /* max to match */ | |
4164 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4165 | repeat: |
a0d0e21e LW |
4166 | /* |
4167 | * Lookahead to avoid useless match attempts | |
4168 | * when we know what character comes next. | |
c255a977 | 4169 | * |
5f80c4cf JP |
4170 | * Used to only do .*x and .*?x, but now it allows |
4171 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4172 | * of the quantifier and the EXACT-like node. -- japhy | |
4173 | */ | |
4174 | ||
c255a977 DM |
4175 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4176 | sayNO; | |
cca55fe3 | 4177 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4178 | U8 *s; |
4179 | regnode *text_node = next; | |
4180 | ||
3dab1dad YO |
4181 | if (! HAS_TEXT(text_node)) |
4182 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4183 | |
9e137952 | 4184 | if (! HAS_TEXT(text_node)) |
c255a977 | 4185 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4186 | else { |
3dab1dad | 4187 | if (PL_regkind[OP(text_node)] == REF) { |
c255a977 | 4188 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4189 | goto assume_ok_easy; |
cca55fe3 | 4190 | } |
be8e71aa YO |
4191 | else |
4192 | s = (U8*)STRING(text_node); | |
5f80c4cf JP |
4193 | |
4194 | if (!UTF) { | |
c255a977 | 4195 | ST.c2 = ST.c1 = *s; |
f65d3ee7 | 4196 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) |
c255a977 | 4197 | ST.c2 = PL_fold[ST.c1]; |
f65d3ee7 | 4198 | else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) |
c255a977 | 4199 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4200 | } |
5f80c4cf | 4201 | else { /* UTF */ |
f65d3ee7 | 4202 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) { |
a2a2844f | 4203 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4204 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4205 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4206 | |
4207 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4208 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4209 | #ifdef EBCDIC |
4210 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4211 | ckWARN(WARN_UTF8) ? | |
4212 | 0 : UTF8_ALLOW_ANY); | |
4213 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4214 | ckWARN(WARN_UTF8) ? | |
4215 | 0 : UTF8_ALLOW_ANY); | |
4216 | #else | |
c255a977 | 4217 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4218 | uniflags); |
c255a977 | 4219 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4220 | uniflags); |
4221 | #endif | |
5f80c4cf JP |
4222 | } |
4223 | else { | |
c255a977 | 4224 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4225 | uniflags); |
5f80c4cf | 4226 | } |
1aa99e6b IH |
4227 | } |
4228 | } | |
bbce6d69 | 4229 | } |
a0d0e21e | 4230 | else |
c255a977 | 4231 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4232 | assume_ok_easy: |
c255a977 DM |
4233 | |
4234 | ST.A = scan; | |
4235 | ST.B = next; | |
3280af22 | 4236 | PL_reginput = locinput; |
24d3c4a9 DM |
4237 | if (minmod) { |
4238 | minmod = 0; | |
c255a977 | 4239 | if (ST.min && regrepeat(rex, ST.A, ST.min) < ST.min) |
4633a7c4 | 4240 | sayNO; |
c255a977 | 4241 | ST.count = ST.min; |
a0ed51b3 | 4242 | locinput = PL_reginput; |
c255a977 DM |
4243 | REGCP_SET(ST.cp); |
4244 | if (ST.c1 == CHRTEST_VOID) | |
4245 | goto curly_try_B_min; | |
4246 | ||
4247 | ST.oldloc = locinput; | |
4248 | ||
4249 | /* set ST.maxpos to the furthest point along the | |
4250 | * string that could possibly match */ | |
4251 | if (ST.max == REG_INFTY) { | |
4252 | ST.maxpos = PL_regeol - 1; | |
4253 | if (do_utf8) | |
4254 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4255 | ST.maxpos--; | |
4256 | } | |
4257 | else if (do_utf8) { | |
4258 | int m = ST.max - ST.min; | |
4259 | for (ST.maxpos = locinput; | |
4260 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4261 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4262 | } | |
4263 | else { | |
4264 | ST.maxpos = locinput + ST.max - ST.min; | |
4265 | if (ST.maxpos >= PL_regeol) | |
4266 | ST.maxpos = PL_regeol - 1; | |
4267 | } | |
4268 | goto curly_try_B_min_known; | |
4269 | ||
4270 | } | |
4271 | else { | |
4272 | ST.count = regrepeat(rex, ST.A, ST.max); | |
4273 | locinput = PL_reginput; | |
4274 | if (ST.count < ST.min) | |
4275 | sayNO; | |
4276 | if ((ST.count > ST.min) | |
4277 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4278 | { | |
4279 | /* A{m,n} must come at the end of the string, there's | |
4280 | * no point in backing off ... */ | |
4281 | ST.min = ST.count; | |
4282 | /* ...except that $ and \Z can match before *and* after | |
4283 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4284 | We may back off by one in this case. */ | |
4285 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4286 | ST.min--; | |
4287 | } | |
4288 | REGCP_SET(ST.cp); | |
4289 | goto curly_try_B_max; | |
4290 | } | |
4291 | /* NOTREACHED */ | |
4292 | ||
4293 | ||
4294 | case CURLY_B_min_known_fail: | |
4295 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4296 | if (ST.paren && ST.count) | |
4297 | PL_regendp[ST.paren] = -1; | |
4298 | ||
4299 | PL_reginput = locinput; /* Could be reset... */ | |
4300 | REGCP_UNWIND(ST.cp); | |
4301 | /* Couldn't or didn't -- move forward. */ | |
4302 | ST.oldloc = locinput; | |
4303 | if (do_utf8) | |
4304 | locinput += UTF8SKIP(locinput); | |
4305 | else | |
4306 | locinput++; | |
4307 | ST.count++; | |
4308 | curly_try_B_min_known: | |
4309 | /* find the next place where 'B' could work, then call B */ | |
4310 | { | |
4311 | int n; | |
4312 | if (do_utf8) { | |
4313 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4314 | if (ST.c1 == ST.c2) { | |
4315 | STRLEN len; | |
4316 | /* set n to utf8_distance(oldloc, locinput) */ | |
4317 | while (locinput <= ST.maxpos && | |
4318 | utf8n_to_uvchr((U8*)locinput, | |
4319 | UTF8_MAXBYTES, &len, | |
4320 | uniflags) != (UV)ST.c1) { | |
4321 | locinput += len; | |
4322 | n++; | |
4323 | } | |
1aa99e6b IH |
4324 | } |
4325 | else { | |
c255a977 DM |
4326 | /* set n to utf8_distance(oldloc, locinput) */ |
4327 | while (locinput <= ST.maxpos) { | |
4328 | STRLEN len; | |
4329 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4330 | UTF8_MAXBYTES, &len, | |
4331 | uniflags); | |
4332 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4333 | break; | |
4334 | locinput += len; | |
4335 | n++; | |
1aa99e6b | 4336 | } |
0fe9bf95 IZ |
4337 | } |
4338 | } | |
c255a977 DM |
4339 | else { |
4340 | if (ST.c1 == ST.c2) { | |
4341 | while (locinput <= ST.maxpos && | |
4342 | UCHARAT(locinput) != ST.c1) | |
4343 | locinput++; | |
bbce6d69 | 4344 | } |
c255a977 DM |
4345 | else { |
4346 | while (locinput <= ST.maxpos | |
4347 | && UCHARAT(locinput) != ST.c1 | |
4348 | && UCHARAT(locinput) != ST.c2) | |
4349 | locinput++; | |
a0ed51b3 | 4350 | } |
c255a977 DM |
4351 | n = locinput - ST.oldloc; |
4352 | } | |
4353 | if (locinput > ST.maxpos) | |
4354 | sayNO; | |
4355 | /* PL_reginput == oldloc now */ | |
4356 | if (n) { | |
4357 | ST.count += n; | |
4358 | if (regrepeat(rex, ST.A, n) < n) | |
4633a7c4 | 4359 | sayNO; |
a0d0e21e | 4360 | } |
c255a977 DM |
4361 | PL_reginput = locinput; |
4362 | CURLY_SETPAREN(ST.paren, ST.count); | |
4363 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); | |
a0d0e21e | 4364 | } |
c255a977 DM |
4365 | /* NOTREACHED */ |
4366 | ||
4367 | ||
4368 | case CURLY_B_min_fail: | |
4369 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4370 | if (ST.paren && ST.count) | |
4371 | PL_regendp[ST.paren] = -1; | |
4372 | ||
4373 | REGCP_UNWIND(ST.cp); | |
4374 | /* failed -- move forward one */ | |
4375 | PL_reginput = locinput; | |
4376 | if (regrepeat(rex, ST.A, 1)) { | |
4377 | ST.count++; | |
a0ed51b3 | 4378 | locinput = PL_reginput; |
c255a977 DM |
4379 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4380 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4381 | { |
c255a977 DM |
4382 | curly_try_B_min: |
4383 | CURLY_SETPAREN(ST.paren, ST.count); | |
4384 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); | |
a0d0e21e LW |
4385 | } |
4386 | } | |
4633a7c4 | 4387 | sayNO; |
c255a977 DM |
4388 | /* NOTREACHED */ |
4389 | ||
4390 | ||
4391 | curly_try_B_max: | |
4392 | /* a successful greedy match: now try to match B */ | |
4393 | { | |
4394 | UV c = 0; | |
4395 | if (ST.c1 != CHRTEST_VOID) | |
4396 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4397 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4398 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4399 | /* If it could work, try it. */ |
4400 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4401 | CURLY_SETPAREN(ST.paren, ST.count); | |
4402 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4403 | /* NOTREACHED */ | |
4404 | } | |
4405 | } | |
4406 | /* FALL THROUGH */ | |
4407 | case CURLY_B_max_fail: | |
4408 | /* failed to find B in a greedy match */ | |
4409 | if (ST.paren && ST.count) | |
4410 | PL_regendp[ST.paren] = -1; | |
4411 | ||
4412 | REGCP_UNWIND(ST.cp); | |
4413 | /* back up. */ | |
4414 | if (--ST.count < ST.min) | |
4415 | sayNO; | |
4416 | PL_reginput = locinput = HOPc(locinput, -1); | |
4417 | goto curly_try_B_max; | |
4418 | ||
4419 | #undef ST | |
4420 | ||
4421 | ||
a0d0e21e | 4422 | case END: |
6bda09f9 | 4423 | fake_end: |
faec1544 DM |
4424 | if (cur_eval) { |
4425 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4426 | I32 tmpix; | |
4427 | ||
4428 | ||
4429 | st->u.eval.toggle_reg_flags | |
4430 | = cur_eval->u.eval.toggle_reg_flags; | |
4431 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4432 | ||
4433 | st->u.eval.prev_rex = rex; /* inner */ | |
4434 | rex = cur_eval->u.eval.prev_rex; /* outer */ | |
4435 | cur_curlyx = cur_eval->u.eval.prev_curlyx; | |
4436 | ReREFCNT_inc(rex); | |
4437 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ | |
4438 | REGCP_SET(st->u.eval.lastcp); | |
4439 | PL_reginput = locinput; | |
4440 | ||
4441 | /* Restore parens of the outer rex without popping the | |
4442 | * savestack */ | |
4443 | tmpix = PL_savestack_ix; | |
4444 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4445 | regcppop(rex); | |
4446 | PL_savestack_ix = tmpix; | |
4447 | ||
4448 | st->u.eval.prev_eval = cur_eval; | |
4449 | cur_eval = cur_eval->u.eval.prev_eval; | |
4450 | DEBUG_EXECUTE_r( | |
6bda09f9 YO |
4451 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %x\n", |
4452 | REPORT_CODE_OFF+depth*2, "",(int)cur_eval);); | |
faec1544 DM |
4453 | PUSH_YES_STATE_GOTO(EVAL_AB, |
4454 | st->u.eval.prev_eval->u.eval.B); /* match B */ | |
4455 | } | |
4456 | ||
3b0527fe | 4457 | if (locinput < reginfo->till) { |
a3621e74 | 4458 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4459 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4460 | PL_colors[4], | |
4461 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4462 | (long)(reginfo->till - PL_reg_starttry), |
7821416a | 4463 | PL_colors[5])); |
262b90c4 | 4464 | sayNO_SILENT; /* Cannot match: too short. */ |
7821416a IZ |
4465 | } |
4466 | PL_reginput = locinput; /* put where regtry can find it */ | |
262b90c4 | 4467 | sayYES; /* Success! */ |
dad79028 DM |
4468 | |
4469 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4470 | DEBUG_EXECUTE_r( | |
4471 | PerlIO_printf(Perl_debug_log, | |
4472 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4473 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4474 | PL_reginput = locinput; /* put where regtry can find it */ |
262b90c4 | 4475 | sayYES; /* Success! */ |
dad79028 | 4476 | |
40a82448 DM |
4477 | #undef ST |
4478 | #define ST st->u.ifmatch | |
4479 | ||
4480 | case SUSPEND: /* (?>A) */ | |
4481 | ST.wanted = 1; | |
9fe1d20c | 4482 | PL_reginput = locinput; |
9041c2e3 | 4483 | goto do_ifmatch; |
dad79028 | 4484 | |
40a82448 DM |
4485 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4486 | ST.wanted = 0; | |
dad79028 DM |
4487 | goto ifmatch_trivial_fail_test; |
4488 | ||
40a82448 DM |
4489 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4490 | ST.wanted = 1; | |
dad79028 | 4491 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4492 | if (scan->flags) { |
52657f30 | 4493 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4494 | if (!s) { |
4495 | /* trivial fail */ | |
24d3c4a9 DM |
4496 | if (logical) { |
4497 | logical = 0; | |
4498 | sw = 1 - (bool)ST.wanted; | |
dad79028 | 4499 | } |
40a82448 | 4500 | else if (ST.wanted) |
dad79028 DM |
4501 | sayNO; |
4502 | next = scan + ARG(scan); | |
4503 | if (next == scan) | |
4504 | next = NULL; | |
4505 | break; | |
4506 | } | |
efb30f32 | 4507 | PL_reginput = s; |
a0ed51b3 LW |
4508 | } |
4509 | else | |
4510 | PL_reginput = locinput; | |
4511 | ||
c277df42 | 4512 | do_ifmatch: |
40a82448 | 4513 | ST.me = scan; |
24d3c4a9 | 4514 | ST.logical = logical; |
40a82448 DM |
4515 | /* execute body of (?...A) */ |
4516 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
4517 | /* NOTREACHED */ | |
4518 | ||
4519 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
4520 | ST.wanted = !ST.wanted; | |
4521 | /* FALL THROUGH */ | |
4522 | ||
4523 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
24d3c4a9 DM |
4524 | if (ST.logical) { |
4525 | sw = (bool)ST.wanted; | |
40a82448 DM |
4526 | } |
4527 | else if (!ST.wanted) | |
4528 | sayNO; | |
4529 | ||
4530 | if (OP(ST.me) == SUSPEND) | |
4531 | locinput = PL_reginput; | |
4532 | else { | |
4533 | locinput = PL_reginput = st->locinput; | |
4534 | nextchr = UCHARAT(locinput); | |
4535 | } | |
4536 | scan = ST.me + ARG(ST.me); | |
4537 | if (scan == ST.me) | |
4538 | scan = NULL; | |
4539 | continue; /* execute B */ | |
4540 | ||
4541 | #undef ST | |
dad79028 | 4542 | |
c277df42 | 4543 | case LONGJMP: |
c277df42 IZ |
4544 | next = scan + ARG(scan); |
4545 | if (next == scan) | |
4546 | next = NULL; | |
a0d0e21e LW |
4547 | break; |
4548 | default: | |
b900a521 | 4549 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 4550 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 4551 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c | 4552 | } |
95b24440 | 4553 | |
a0d0e21e | 4554 | scan = next; |
95b24440 DM |
4555 | continue; |
4556 | /* NOTREACHED */ | |
4557 | ||
40a82448 DM |
4558 | push_yes_state: |
4559 | /* push a state that backtracks on success */ | |
4560 | st->u.yes.prev_yes_state = yes_state; | |
4561 | yes_state = st; | |
4562 | /* FALL THROUGH */ | |
4563 | push_state: | |
4564 | /* push a new regex state, then continue at scan */ | |
4565 | { | |
4566 | regmatch_state *newst; | |
4567 | ||
5bc10b2c | 4568 | DEBUG_STATE_pp("push"); |
40a82448 | 4569 | depth++; |
40a82448 DM |
4570 | st->locinput = locinput; |
4571 | newst = st+1; | |
4572 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
4573 | newst = S_push_slab(aTHX); | |
4574 | PL_regmatch_state = newst; | |
786e8c11 | 4575 | |
40a82448 DM |
4576 | locinput = PL_reginput; |
4577 | nextchr = UCHARAT(locinput); | |
4578 | st = newst; | |
4579 | continue; | |
4580 | /* NOTREACHED */ | |
4581 | } | |
a0d0e21e | 4582 | } |
a687059c | 4583 | |
a0d0e21e LW |
4584 | /* |
4585 | * We get here only if there's trouble -- normally "case END" is | |
4586 | * the terminating point. | |
4587 | */ | |
cea2e8a9 | 4588 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 4589 | /*NOTREACHED*/ |
4633a7c4 LW |
4590 | sayNO; |
4591 | ||
262b90c4 | 4592 | yes: |
77cb431f DM |
4593 | if (yes_state) { |
4594 | /* we have successfully completed a subexpression, but we must now | |
4595 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 4596 | assert(st != yes_state); |
5bc10b2c DM |
4597 | #ifdef DEBUGGING |
4598 | while (st != yes_state) { | |
4599 | st--; | |
4600 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4601 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4602 | st = SLAB_LAST(PL_regmatch_slab); | |
4603 | } | |
4604 | DEBUG_STATE_pp("pop (yes)"); | |
4605 | depth--; | |
4606 | } | |
4607 | #else | |
77cb431f DM |
4608 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
4609 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
4610 | { | |
4611 | /* not in this slab, pop slab */ | |
4612 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
4613 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4614 | st = SLAB_LAST(PL_regmatch_slab); | |
4615 | } | |
4616 | depth -= (st - yes_state); | |
5bc10b2c | 4617 | #endif |
77cb431f DM |
4618 | st = yes_state; |
4619 | yes_state = st->u.yes.prev_yes_state; | |
4620 | PL_regmatch_state = st; | |
4621 | ||
24d3c4a9 DM |
4622 | state_num = st->resume_state; |
4623 | goto reenter_switch; | |
77cb431f DM |
4624 | } |
4625 | ||
a3621e74 | 4626 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 4627 | PL_colors[4], PL_colors[5])); |
02db2b7b | 4628 | |
95b24440 | 4629 | result = 1; |
aa283a38 | 4630 | goto final_exit; |
4633a7c4 LW |
4631 | |
4632 | no: | |
a3621e74 | 4633 | DEBUG_EXECUTE_r( |
7821416a | 4634 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 4635 | "%*s %sfailed...%s\n", |
5bc10b2c | 4636 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 4637 | PL_colors[4], PL_colors[5]) |
7821416a | 4638 | ); |
aa283a38 | 4639 | |
262b90c4 | 4640 | no_silent: |
aa283a38 DM |
4641 | if (depth) { |
4642 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
4643 | st--; |
4644 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4645 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4646 | st = SLAB_LAST(PL_regmatch_slab); | |
4647 | } | |
4648 | PL_regmatch_state = st; | |
40a82448 DM |
4649 | locinput= st->locinput; |
4650 | nextchr = UCHARAT(locinput); | |
4651 | ||
5bc10b2c DM |
4652 | DEBUG_STATE_pp("pop"); |
4653 | depth--; | |
262b90c4 DM |
4654 | if (yes_state == st) |
4655 | yes_state = st->u.yes.prev_yes_state; | |
5bc10b2c | 4656 | |
24d3c4a9 DM |
4657 | state_num = st->resume_state + 1; /* failure = success + 1 */ |
4658 | goto reenter_switch; | |
95b24440 | 4659 | } |
24d3c4a9 | 4660 | result = 0; |
aa283a38 | 4661 | |
262b90c4 | 4662 | final_exit: |
aa283a38 | 4663 | |
5d9a96ca DM |
4664 | /* restore original high-water mark */ |
4665 | PL_regmatch_slab = orig_slab; | |
4666 | PL_regmatch_state = orig_state; | |
4667 | ||
4668 | /* free all slabs above current one */ | |
4669 | if (orig_slab->next) { | |
c4fd8992 | 4670 | regmatch_slab *sl = orig_slab->next; |
5d9a96ca DM |
4671 | orig_slab->next = NULL; |
4672 | while (sl) { | |
c4fd8992 | 4673 | regmatch_slab * const osl = sl; |
5d9a96ca | 4674 | sl = sl->next; |
ad65c075 | 4675 | Safefree(osl); |
5d9a96ca DM |
4676 | } |
4677 | } | |
4678 | ||
95b24440 | 4679 | return result; |
a687059c LW |
4680 | } |
4681 | ||
4682 | /* | |
4683 | - regrepeat - repeatedly match something simple, report how many | |
4684 | */ | |
4685 | /* | |
4686 | * [This routine now assumes that it will only match on things of length 1. | |
4687 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 4688 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 4689 | */ |
76e3520e | 4690 | STATIC I32 |
32fc9b6a | 4691 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max) |
a687059c | 4692 | { |
27da23d5 | 4693 | dVAR; |
a0d0e21e | 4694 | register char *scan; |
a0d0e21e | 4695 | register I32 c; |
3280af22 | 4696 | register char *loceol = PL_regeol; |
a0ed51b3 | 4697 | register I32 hardcount = 0; |
53c4c00c | 4698 | register bool do_utf8 = PL_reg_match_utf8; |
a0d0e21e | 4699 | |
3280af22 | 4700 | scan = PL_reginput; |
faf11cac HS |
4701 | if (max == REG_INFTY) |
4702 | max = I32_MAX; | |
4703 | else if (max < loceol - scan) | |
7f596f4c | 4704 | loceol = scan + max; |
a0d0e21e | 4705 | switch (OP(p)) { |
22c35a8c | 4706 | case REG_ANY: |
1aa99e6b | 4707 | if (do_utf8) { |
ffc61ed2 | 4708 | loceol = PL_regeol; |
1aa99e6b | 4709 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
4710 | scan += UTF8SKIP(scan); |
4711 | hardcount++; | |
4712 | } | |
4713 | } else { | |
4714 | while (scan < loceol && *scan != '\n') | |
4715 | scan++; | |
a0ed51b3 LW |
4716 | } |
4717 | break; | |
ffc61ed2 | 4718 | case SANY: |
def8e4ea JH |
4719 | if (do_utf8) { |
4720 | loceol = PL_regeol; | |
a0804c9e | 4721 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
4722 | scan += UTF8SKIP(scan); |
4723 | hardcount++; | |
4724 | } | |
4725 | } | |
4726 | else | |
4727 | scan = loceol; | |
a0ed51b3 | 4728 | break; |
f33976b4 DB |
4729 | case CANY: |
4730 | scan = loceol; | |
4731 | break; | |
090f7165 JH |
4732 | case EXACT: /* length of string is 1 */ |
4733 | c = (U8)*STRING(p); | |
4734 | while (scan < loceol && UCHARAT(scan) == c) | |
4735 | scan++; | |
bbce6d69 | 4736 | break; |
4737 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 4738 | c = (U8)*STRING(p); |
bbce6d69 | 4739 | while (scan < loceol && |
22c35a8c | 4740 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 4741 | scan++; |
4742 | break; | |
4743 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 4744 | PL_reg_flags |= RF_tainted; |
cd439c50 | 4745 | c = (U8)*STRING(p); |
bbce6d69 | 4746 | while (scan < loceol && |
22c35a8c | 4747 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
4748 | scan++; |
4749 | break; | |
4750 | case ANYOF: | |
ffc61ed2 JH |
4751 | if (do_utf8) { |
4752 | loceol = PL_regeol; | |
cfc92286 | 4753 | while (hardcount < max && scan < loceol && |
32fc9b6a | 4754 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
4755 | scan += UTF8SKIP(scan); |
4756 | hardcount++; | |
4757 | } | |
4758 | } else { | |
32fc9b6a | 4759 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
4760 | scan++; |
4761 | } | |
a0d0e21e LW |
4762 | break; |
4763 | case ALNUM: | |
1aa99e6b | 4764 | if (do_utf8) { |
ffc61ed2 | 4765 | loceol = PL_regeol; |
1a4fad37 | 4766 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4767 | while (hardcount < max && scan < loceol && |
3568d838 | 4768 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4769 | scan += UTF8SKIP(scan); |
4770 | hardcount++; | |
4771 | } | |
4772 | } else { | |
4773 | while (scan < loceol && isALNUM(*scan)) | |
4774 | scan++; | |
a0ed51b3 LW |
4775 | } |
4776 | break; | |
bbce6d69 | 4777 | case ALNUML: |
3280af22 | 4778 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4779 | if (do_utf8) { |
ffc61ed2 | 4780 | loceol = PL_regeol; |
1aa99e6b IH |
4781 | while (hardcount < max && scan < loceol && |
4782 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4783 | scan += UTF8SKIP(scan); |
4784 | hardcount++; | |
4785 | } | |
4786 | } else { | |
4787 | while (scan < loceol && isALNUM_LC(*scan)) | |
4788 | scan++; | |
a0ed51b3 LW |
4789 | } |
4790 | break; | |
a0d0e21e | 4791 | case NALNUM: |
1aa99e6b | 4792 | if (do_utf8) { |
ffc61ed2 | 4793 | loceol = PL_regeol; |
1a4fad37 | 4794 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4795 | while (hardcount < max && scan < loceol && |
3568d838 | 4796 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4797 | scan += UTF8SKIP(scan); |
4798 | hardcount++; | |
4799 | } | |
4800 | } else { | |
4801 | while (scan < loceol && !isALNUM(*scan)) | |
4802 | scan++; | |
a0ed51b3 LW |
4803 | } |
4804 | break; | |
bbce6d69 | 4805 | case NALNUML: |
3280af22 | 4806 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4807 | if (do_utf8) { |
ffc61ed2 | 4808 | loceol = PL_regeol; |
1aa99e6b IH |
4809 | while (hardcount < max && scan < loceol && |
4810 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4811 | scan += UTF8SKIP(scan); |
4812 | hardcount++; | |
4813 | } | |
4814 | } else { | |
4815 | while (scan < loceol && !isALNUM_LC(*scan)) | |
4816 | scan++; | |
a0ed51b3 LW |
4817 | } |
4818 | break; | |
a0d0e21e | 4819 | case SPACE: |
1aa99e6b | 4820 | if (do_utf8) { |
ffc61ed2 | 4821 | loceol = PL_regeol; |
1a4fad37 | 4822 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4823 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4824 | (*scan == ' ' || |
4825 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4826 | scan += UTF8SKIP(scan); |
4827 | hardcount++; | |
4828 | } | |
4829 | } else { | |
4830 | while (scan < loceol && isSPACE(*scan)) | |
4831 | scan++; | |
a0ed51b3 LW |
4832 | } |
4833 | break; | |
bbce6d69 | 4834 | case SPACEL: |
3280af22 | 4835 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4836 | if (do_utf8) { |
ffc61ed2 | 4837 | loceol = PL_regeol; |
1aa99e6b | 4838 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4839 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4840 | scan += UTF8SKIP(scan); | |
4841 | hardcount++; | |
4842 | } | |
4843 | } else { | |
4844 | while (scan < loceol && isSPACE_LC(*scan)) | |
4845 | scan++; | |
a0ed51b3 LW |
4846 | } |
4847 | break; | |
a0d0e21e | 4848 | case NSPACE: |
1aa99e6b | 4849 | if (do_utf8) { |
ffc61ed2 | 4850 | loceol = PL_regeol; |
1a4fad37 | 4851 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4852 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4853 | !(*scan == ' ' || |
4854 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4855 | scan += UTF8SKIP(scan); |
4856 | hardcount++; | |
4857 | } | |
4858 | } else { | |
4859 | while (scan < loceol && !isSPACE(*scan)) | |
4860 | scan++; | |
4861 | break; | |
a0ed51b3 | 4862 | } |
bbce6d69 | 4863 | case NSPACEL: |
3280af22 | 4864 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4865 | if (do_utf8) { |
ffc61ed2 | 4866 | loceol = PL_regeol; |
1aa99e6b | 4867 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4868 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4869 | scan += UTF8SKIP(scan); | |
4870 | hardcount++; | |
4871 | } | |
4872 | } else { | |
4873 | while (scan < loceol && !isSPACE_LC(*scan)) | |
4874 | scan++; | |
a0ed51b3 LW |
4875 | } |
4876 | break; | |
a0d0e21e | 4877 | case DIGIT: |
1aa99e6b | 4878 | if (do_utf8) { |
ffc61ed2 | 4879 | loceol = PL_regeol; |
1a4fad37 | 4880 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 4881 | while (hardcount < max && scan < loceol && |
3568d838 | 4882 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4883 | scan += UTF8SKIP(scan); |
4884 | hardcount++; | |
4885 | } | |
4886 | } else { | |
4887 | while (scan < loceol && isDIGIT(*scan)) | |
4888 | scan++; | |
a0ed51b3 LW |
4889 | } |
4890 | break; | |
a0d0e21e | 4891 | case NDIGIT: |
1aa99e6b | 4892 | if (do_utf8) { |
ffc61ed2 | 4893 | loceol = PL_regeol; |
1a4fad37 | 4894 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 4895 | while (hardcount < max && scan < loceol && |
3568d838 | 4896 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4897 | scan += UTF8SKIP(scan); |
4898 | hardcount++; | |
4899 | } | |
4900 | } else { | |
4901 | while (scan < loceol && !isDIGIT(*scan)) | |
4902 | scan++; | |
a0ed51b3 LW |
4903 | } |
4904 | break; | |
a0d0e21e LW |
4905 | default: /* Called on something of 0 width. */ |
4906 | break; /* So match right here or not at all. */ | |
4907 | } | |
a687059c | 4908 | |
a0ed51b3 LW |
4909 | if (hardcount) |
4910 | c = hardcount; | |
4911 | else | |
4912 | c = scan - PL_reginput; | |
3280af22 | 4913 | PL_reginput = scan; |
a687059c | 4914 | |
a3621e74 | 4915 | DEBUG_r({ |
e68ec53f | 4916 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 4917 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
4918 | SV * const prop = sv_newmortal(); |
4919 | regprop(prog, prop, p); | |
4920 | PerlIO_printf(Perl_debug_log, | |
be8e71aa YO |
4921 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
4922 | REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max); | |
a3621e74 | 4923 | }); |
be8e71aa | 4924 | }); |
9041c2e3 | 4925 | |
a0d0e21e | 4926 | return(c); |
a687059c LW |
4927 | } |
4928 | ||
c277df42 | 4929 | |
be8e71aa | 4930 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 4931 | /* |
ffc61ed2 JH |
4932 | - regclass_swash - prepare the utf8 swash |
4933 | */ | |
4934 | ||
4935 | SV * | |
32fc9b6a | 4936 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 4937 | { |
97aff369 | 4938 | dVAR; |
9e55ce06 JH |
4939 | SV *sw = NULL; |
4940 | SV *si = NULL; | |
4941 | SV *alt = NULL; | |
3dab1dad | 4942 | const struct reg_data * const data = prog ? prog->data : NULL; |
ffc61ed2 | 4943 | |
4f639d21 | 4944 | if (data && data->count) { |
a3b680e6 | 4945 | const U32 n = ARG(node); |
ffc61ed2 | 4946 | |
4f639d21 DM |
4947 | if (data->what[n] == 's') { |
4948 | SV * const rv = (SV*)data->data[n]; | |
890ce7af | 4949 | AV * const av = (AV*)SvRV((SV*)rv); |
2d03de9c | 4950 | SV **const ary = AvARRAY(av); |
9e55ce06 | 4951 | SV **a, **b; |
9041c2e3 | 4952 | |
711a919c | 4953 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
4954 | * documentation of these array elements. */ |
4955 | ||
b11f357e | 4956 | si = *ary; |
8f7f7219 | 4957 | a = SvROK(ary[1]) ? &ary[1] : 0; |
b11f357e JH |
4958 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0; |
4959 | ||
ffc61ed2 JH |
4960 | if (a) |
4961 | sw = *a; | |
4962 | else if (si && doinit) { | |
4963 | sw = swash_init("utf8", "", si, 1, 0); | |
4964 | (void)av_store(av, 1, sw); | |
4965 | } | |
9e55ce06 JH |
4966 | if (b) |
4967 | alt = *b; | |
ffc61ed2 JH |
4968 | } |
4969 | } | |
4970 | ||
9e55ce06 JH |
4971 | if (listsvp) |
4972 | *listsvp = si; | |
4973 | if (altsvp) | |
4974 | *altsvp = alt; | |
ffc61ed2 JH |
4975 | |
4976 | return sw; | |
4977 | } | |
76234dfb | 4978 | #endif |
ffc61ed2 JH |
4979 | |
4980 | /* | |
ba7b4546 | 4981 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
4982 | |
4983 | The n is the ANYOF regnode, the p is the target string, lenp | |
4984 | is pointer to the maximum length of how far to go in the p | |
4985 | (if the lenp is zero, UTF8SKIP(p) is used), | |
4986 | do_utf8 tells whether the target string is in UTF-8. | |
4987 | ||
bbce6d69 | 4988 | */ |
4989 | ||
76e3520e | 4990 | STATIC bool |
32fc9b6a | 4991 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 4992 | { |
27da23d5 | 4993 | dVAR; |
a3b680e6 | 4994 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 4995 | bool match = FALSE; |
cc07378b | 4996 | UV c = *p; |
ae9ddab8 | 4997 | STRLEN len = 0; |
9e55ce06 | 4998 | STRLEN plen; |
1aa99e6b | 4999 | |
19f67299 TS |
5000 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5001 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5002 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5003 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5004 | if (len == (STRLEN)-1) |
5005 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5006 | } |
bbce6d69 | 5007 | |
0f0076b4 | 5008 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5009 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5010 | if (lenp) |
5011 | *lenp = 0; | |
ffc61ed2 | 5012 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5013 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5014 | match = TRUE; | |
bbce6d69 | 5015 | } |
3568d838 | 5016 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5017 | match = TRUE; |
ffc61ed2 | 5018 | if (!match) { |
9e55ce06 | 5019 | AV *av; |
32fc9b6a | 5020 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5021 | |
5022 | if (sw) { | |
3568d838 | 5023 | if (swash_fetch(sw, p, do_utf8)) |
ffc61ed2 JH |
5024 | match = TRUE; |
5025 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5026 | if (!match && lenp && av) { |
5027 | I32 i; | |
9e55ce06 | 5028 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5029 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5030 | STRLEN len; |
890ce7af | 5031 | const char * const s = SvPV_const(sv, len); |
9e55ce06 | 5032 | |
061b10df | 5033 | if (len <= plen && memEQ(s, (char*)p, len)) { |
9e55ce06 JH |
5034 | *lenp = len; |
5035 | match = TRUE; | |
5036 | break; | |
5037 | } | |
5038 | } | |
5039 | } | |
5040 | if (!match) { | |
89ebb4a3 | 5041 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 JH |
5042 | STRLEN tmplen; |
5043 | ||
9e55ce06 JH |
5044 | to_utf8_fold(p, tmpbuf, &tmplen); |
5045 | if (swash_fetch(sw, tmpbuf, do_utf8)) | |
5046 | match = TRUE; | |
5047 | } | |
ffc61ed2 JH |
5048 | } |
5049 | } | |
bbce6d69 | 5050 | } |
9e55ce06 | 5051 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5052 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5053 | } |
1aa99e6b | 5054 | if (!match && c < 256) { |
ffc61ed2 JH |
5055 | if (ANYOF_BITMAP_TEST(n, c)) |
5056 | match = TRUE; | |
5057 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5058 | U8 f; |
a0ed51b3 | 5059 | |
ffc61ed2 JH |
5060 | if (flags & ANYOF_LOCALE) { |
5061 | PL_reg_flags |= RF_tainted; | |
5062 | f = PL_fold_locale[c]; | |
5063 | } | |
5064 | else | |
5065 | f = PL_fold[c]; | |
5066 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5067 | match = TRUE; | |
5068 | } | |
5069 | ||
5070 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5071 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5072 | if ( |
5073 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5074 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5075 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5076 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5077 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5078 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5079 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5080 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5081 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5082 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5083 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5084 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5085 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5086 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5087 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5088 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5089 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5090 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5091 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5092 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5093 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5094 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5095 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5096 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5097 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5098 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5099 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5100 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5101 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5102 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5103 | ) /* How's that for a conditional? */ | |
5104 | { | |
5105 | match = TRUE; | |
5106 | } | |
a0ed51b3 | 5107 | } |
a0ed51b3 LW |
5108 | } |
5109 | ||
a0ed51b3 LW |
5110 | return (flags & ANYOF_INVERT) ? !match : match; |
5111 | } | |
161b471a | 5112 | |
dfe13c55 | 5113 | STATIC U8 * |
0ce71af7 | 5114 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5115 | { |
97aff369 | 5116 | dVAR; |
a0ed51b3 | 5117 | if (off >= 0) { |
1aa99e6b | 5118 | while (off-- && s < lim) { |
ffc61ed2 | 5119 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5120 | s += UTF8SKIP(s); |
ffc61ed2 | 5121 | } |
a0ed51b3 LW |
5122 | } |
5123 | else { | |
1de06328 YO |
5124 | while (off++ && s > lim) { |
5125 | s--; | |
5126 | if (UTF8_IS_CONTINUED(*s)) { | |
5127 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5128 | s--; | |
a0ed51b3 | 5129 | } |
1de06328 | 5130 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5131 | } |
5132 | } | |
5133 | return s; | |
5134 | } | |
161b471a | 5135 | |
f9f4320a YO |
5136 | #ifdef XXX_dmq |
5137 | /* there are a bunch of places where we use two reghop3's that should | |
5138 | be replaced with this routine. but since thats not done yet | |
5139 | we ifdef it out - dmq | |
5140 | */ | |
dfe13c55 | 5141 | STATIC U8 * |
1de06328 YO |
5142 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5143 | { | |
5144 | dVAR; | |
5145 | if (off >= 0) { | |
5146 | while (off-- && s < rlim) { | |
5147 | /* XXX could check well-formedness here */ | |
5148 | s += UTF8SKIP(s); | |
5149 | } | |
5150 | } | |
5151 | else { | |
5152 | while (off++ && s > llim) { | |
5153 | s--; | |
5154 | if (UTF8_IS_CONTINUED(*s)) { | |
5155 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5156 | s--; | |
5157 | } | |
5158 | /* XXX could check well-formedness here */ | |
5159 | } | |
5160 | } | |
5161 | return s; | |
5162 | } | |
f9f4320a | 5163 | #endif |
1de06328 YO |
5164 | |
5165 | STATIC U8 * | |
0ce71af7 | 5166 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5167 | { |
97aff369 | 5168 | dVAR; |
a0ed51b3 | 5169 | if (off >= 0) { |
1aa99e6b | 5170 | while (off-- && s < lim) { |
ffc61ed2 | 5171 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5172 | s += UTF8SKIP(s); |
ffc61ed2 | 5173 | } |
a0ed51b3 | 5174 | if (off >= 0) |
3dab1dad | 5175 | return NULL; |
a0ed51b3 LW |
5176 | } |
5177 | else { | |
1de06328 YO |
5178 | while (off++ && s > lim) { |
5179 | s--; | |
5180 | if (UTF8_IS_CONTINUED(*s)) { | |
5181 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5182 | s--; | |
a0ed51b3 | 5183 | } |
1de06328 | 5184 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5185 | } |
5186 | if (off <= 0) | |
3dab1dad | 5187 | return NULL; |
a0ed51b3 LW |
5188 | } |
5189 | return s; | |
5190 | } | |
51371543 | 5191 | |
51371543 | 5192 | static void |
acfe0abc | 5193 | restore_pos(pTHX_ void *arg) |
51371543 | 5194 | { |
97aff369 | 5195 | dVAR; |
097eb12c | 5196 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5197 | if (PL_reg_eval_set) { |
5198 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5199 | rex->subbeg = PL_reg_oldsaved; |
5200 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5201 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5202 | rex->saved_copy = PL_nrs; |
ed252734 | 5203 | #endif |
4f639d21 | 5204 | RX_MATCH_COPIED_on(rex); |
51371543 GS |
5205 | } |
5206 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
5207 | PL_reg_eval_set = 0; | |
5208 | PL_curpm = PL_reg_oldcurpm; | |
5209 | } | |
5210 | } | |
33b8afdf JH |
5211 | |
5212 | STATIC void | |
5213 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
5214 | { | |
33b8afdf | 5215 | if (prog->float_substr && !prog->float_utf8) { |
097eb12c AL |
5216 | SV* const sv = newSVsv(prog->float_substr); |
5217 | prog->float_utf8 = sv; | |
33b8afdf JH |
5218 | sv_utf8_upgrade(sv); |
5219 | if (SvTAIL(prog->float_substr)) | |
5220 | SvTAIL_on(sv); | |
5221 | if (prog->float_substr == prog->check_substr) | |
5222 | prog->check_utf8 = sv; | |
5223 | } | |
5224 | if (prog->anchored_substr && !prog->anchored_utf8) { | |
097eb12c AL |
5225 | SV* const sv = newSVsv(prog->anchored_substr); |
5226 | prog->anchored_utf8 = sv; | |
33b8afdf JH |
5227 | sv_utf8_upgrade(sv); |
5228 | if (SvTAIL(prog->anchored_substr)) | |
5229 | SvTAIL_on(sv); | |
5230 | if (prog->anchored_substr == prog->check_substr) | |
5231 | prog->check_utf8 = sv; | |
5232 | } | |
5233 | } | |
5234 | ||
5235 | STATIC void | |
5236 | S_to_byte_substr(pTHX_ register regexp *prog) | |
5237 | { | |
97aff369 | 5238 | dVAR; |
33b8afdf | 5239 | if (prog->float_utf8 && !prog->float_substr) { |
097eb12c AL |
5240 | SV* sv = newSVsv(prog->float_utf8); |
5241 | prog->float_substr = sv; | |
33b8afdf JH |
5242 | if (sv_utf8_downgrade(sv, TRUE)) { |
5243 | if (SvTAIL(prog->float_utf8)) | |
5244 | SvTAIL_on(sv); | |
5245 | } else { | |
5246 | SvREFCNT_dec(sv); | |
5247 | prog->float_substr = sv = &PL_sv_undef; | |
5248 | } | |
5249 | if (prog->float_utf8 == prog->check_utf8) | |
5250 | prog->check_substr = sv; | |
5251 | } | |
5252 | if (prog->anchored_utf8 && !prog->anchored_substr) { | |
097eb12c AL |
5253 | SV* sv = newSVsv(prog->anchored_utf8); |
5254 | prog->anchored_substr = sv; | |
33b8afdf JH |
5255 | if (sv_utf8_downgrade(sv, TRUE)) { |
5256 | if (SvTAIL(prog->anchored_utf8)) | |
5257 | SvTAIL_on(sv); | |
5258 | } else { | |
5259 | SvREFCNT_dec(sv); | |
5260 | prog->anchored_substr = sv = &PL_sv_undef; | |
5261 | } | |
5262 | if (prog->anchored_utf8 == prog->check_utf8) | |
5263 | prog->check_substr = sv; | |
5264 | } | |
5265 | } | |
66610fdd RGS |
5266 | |
5267 | /* | |
5268 | * Local variables: | |
5269 | * c-indentation-style: bsd | |
5270 | * c-basic-offset: 4 | |
5271 | * indent-tabs-mode: t | |
5272 | * End: | |
5273 | * | |
37442d52 RGS |
5274 | * ex: set ts=8 sts=4 sw=4 noet: |
5275 | */ |