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); |
7f69552c | 171 | |
3280af22 | 172 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 173 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
174 | SSPUSHINT(PL_regendp[p]); |
175 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 176 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 177 | SSPUSHINT(p); |
40a82448 DM |
178 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
179 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", | |
180 | (UV)p, (IV)PL_regstartp[p], | |
181 | (IV)(PL_reg_start_tmp[p] - PL_bostr), | |
182 | (IV)PL_regendp[p] | |
183 | )); | |
a0d0e21e | 184 | } |
b1ce53c5 | 185 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
6bda09f9 YO |
186 | SSPUSHPTR(PL_regstartp); |
187 | SSPUSHPTR(PL_regendp); | |
3280af22 NIS |
188 | SSPUSHINT(PL_regsize); |
189 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 190 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 191 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
192 | #define REGCP_FRAME_ELEMS 2 |
193 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
194 | * are needed for the regexp context stack bookkeeping. */ | |
195 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 196 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 197 | |
a0d0e21e LW |
198 | return retval; |
199 | } | |
200 | ||
c277df42 | 201 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
202 | #define REGCP_SET(cp) \ |
203 | DEBUG_STATE_r( \ | |
ab3bbdeb | 204 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 205 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
206 | (IV)PL_savestack_ix)); \ |
207 | cp = PL_savestack_ix | |
c3464db5 | 208 | |
ab3bbdeb | 209 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 210 | DEBUG_STATE_r( \ |
ab3bbdeb | 211 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
212 | PerlIO_printf(Perl_debug_log, \ |
213 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
214 | (IV)(cp), (IV)PL_savestack_ix)); \ |
215 | regcpblow(cp) | |
c277df42 | 216 | |
76e3520e | 217 | STATIC char * |
097eb12c | 218 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 219 | { |
97aff369 | 220 | dVAR; |
b1ce53c5 | 221 | I32 i; |
a0d0e21e | 222 | char *input; |
b1ce53c5 | 223 | |
a3621e74 YO |
224 | GET_RE_DEBUG_FLAGS_DECL; |
225 | ||
b1ce53c5 | 226 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 227 | i = SSPOPINT; |
b1ce53c5 JH |
228 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
229 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 230 | input = (char *) SSPOPPTR; |
a01268b5 | 231 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
232 | *PL_reglastparen = SSPOPINT; |
233 | PL_regsize = SSPOPINT; | |
6bda09f9 YO |
234 | PL_regendp=(I32 *) SSPOPPTR; |
235 | PL_regstartp=(I32 *) SSPOPPTR; | |
b1ce53c5 | 236 | |
6bda09f9 | 237 | |
b1ce53c5 | 238 | /* Now restore the parentheses context. */ |
41123dfd JH |
239 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
240 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 241 | I32 tmps; |
097eb12c | 242 | U32 paren = (U32)SSPOPINT; |
3280af22 | 243 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
244 | PL_regstartp[paren] = SSPOPINT; |
245 | tmps = SSPOPINT; | |
3280af22 NIS |
246 | if (paren <= *PL_reglastparen) |
247 | PL_regendp[paren] = tmps; | |
a3621e74 | 248 | DEBUG_EXECUTE_r( |
c3464db5 | 249 | PerlIO_printf(Perl_debug_log, |
b900a521 | 250 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 251 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 252 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 253 | (IV)PL_regendp[paren], |
3280af22 | 254 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 255 | ); |
a0d0e21e | 256 | } |
a3621e74 | 257 | DEBUG_EXECUTE_r( |
bb7a0f54 | 258 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 259 | PerlIO_printf(Perl_debug_log, |
faccc32b | 260 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 261 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
262 | } |
263 | ); | |
daf18116 | 264 | #if 1 |
dafc8851 JH |
265 | /* It would seem that the similar code in regtry() |
266 | * already takes care of this, and in fact it is in | |
267 | * a better location to since this code can #if 0-ed out | |
268 | * but the code in regtry() is needed or otherwise tests | |
269 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
270 | * (as of patchlevel 7877) will fail. Then again, |
271 | * this code seems to be necessary or otherwise | |
272 | * building DynaLoader will fail: | |
273 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
274 | * --jhi */ | |
bb7a0f54 | 275 | for (i = *PL_reglastparen + 1; (U32)i <= rex->nparens; i++) { |
097eb12c AL |
276 | if (i > PL_regsize) |
277 | PL_regstartp[i] = -1; | |
278 | PL_regendp[i] = -1; | |
a0d0e21e | 279 | } |
dafc8851 | 280 | #endif |
a0d0e21e LW |
281 | return input; |
282 | } | |
283 | ||
02db2b7b | 284 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 285 | |
a687059c | 286 | /* |
e50aee73 | 287 | * pregexec and friends |
a687059c LW |
288 | */ |
289 | ||
76234dfb | 290 | #ifndef PERL_IN_XSUB_RE |
a687059c | 291 | /* |
c277df42 | 292 | - pregexec - match a regexp against a string |
a687059c | 293 | */ |
c277df42 | 294 | I32 |
864dbfa3 | 295 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 296 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
297 | /* strend: pointer to null at end of string */ |
298 | /* strbeg: real beginning of string */ | |
299 | /* minend: end of match must be >=minend after stringarg. */ | |
300 | /* nosave: For optimizations. */ | |
301 | { | |
302 | return | |
9041c2e3 | 303 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
304 | nosave ? 0 : REXEC_COPY_STR); |
305 | } | |
76234dfb | 306 | #endif |
22e551b9 | 307 | |
9041c2e3 | 308 | /* |
cad2e5aa JH |
309 | * Need to implement the following flags for reg_anch: |
310 | * | |
311 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
312 | * USE_INTUIT_ML | |
313 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
314 | * INTUIT_AUTORITATIVE_ML | |
315 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
316 | * INTUIT_ONCE_ML | |
317 | * | |
318 | * Another flag for this function: SECOND_TIME (so that float substrs | |
319 | * with giant delta may be not rechecked). | |
320 | */ | |
321 | ||
322 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
323 | ||
3f7c398e | 324 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
325 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
326 | ||
327 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
328 | ||
6eb5f6b9 JH |
329 | /* XXXX Some places assume that there is a fixed substring. |
330 | An update may be needed if optimizer marks as "INTUITable" | |
331 | RExen without fixed substrings. Similarly, it is assumed that | |
332 | lengths of all the strings are no more than minlen, thus they | |
333 | cannot come from lookahead. | |
40d049e4 YO |
334 | (Or minlen should take into account lookahead.) |
335 | NOTE: Some of this comment is not correct. minlen does now take account | |
336 | of lookahead/behind. Further research is required. -- demerphq | |
337 | ||
338 | */ | |
6eb5f6b9 | 339 | |
2c2d71f5 JH |
340 | /* A failure to find a constant substring means that there is no need to make |
341 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
342 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
343 | regtry() should be needed. |
344 | ||
345 | REx compiler's optimizer found 4 possible hints: | |
346 | a) Anchored substring; | |
347 | b) Fixed substring; | |
348 | c) Whether we are anchored (beginning-of-line or \G); | |
349 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 350 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
351 | string which does not contradict any of them. |
352 | */ | |
2c2d71f5 | 353 | |
6eb5f6b9 JH |
354 | /* Most of decisions we do here should have been done at compile time. |
355 | The nodes of the REx which we used for the search should have been | |
356 | deleted from the finite automaton. */ | |
357 | ||
cad2e5aa JH |
358 | char * |
359 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
360 | char *strend, U32 flags, re_scream_pos_data *data) | |
361 | { | |
97aff369 | 362 | dVAR; |
b7953727 | 363 | register I32 start_shift = 0; |
cad2e5aa | 364 | /* Should be nonnegative! */ |
b7953727 | 365 | register I32 end_shift = 0; |
2c2d71f5 JH |
366 | register char *s; |
367 | register SV *check; | |
a1933d95 | 368 | char *strbeg; |
cad2e5aa | 369 | char *t; |
1de06328 | 370 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 371 | I32 ml_anch; |
bd61b366 SS |
372 | register char *other_last = NULL; /* other substr checked before this */ |
373 | char *check_at = NULL; /* check substr found at this pos */ | |
1df70142 | 374 | const I32 multiline = prog->reganch & PMf_MULTILINE; |
30944b6d | 375 | #ifdef DEBUGGING |
890ce7af | 376 | const char * const i_strpos = strpos; |
30944b6d | 377 | #endif |
a3621e74 YO |
378 | |
379 | GET_RE_DEBUG_FLAGS_DECL; | |
380 | ||
a30b2f1f | 381 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 382 | |
b8d68ded | 383 | if (prog->reganch & ROPT_UTF8) { |
b8d68ded JH |
384 | PL_reg_flags |= RF_utf8; |
385 | } | |
ab3bbdeb YO |
386 | DEBUG_EXECUTE_r( |
387 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
388 | sv ? "Guessing start of match in sv for" |
389 | : "Guessing start of match in string for"); | |
2a782b5b | 390 | ); |
cad2e5aa | 391 | |
c344f387 JH |
392 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
393 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 394 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 395 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 396 | goto fail; |
2c2d71f5 | 397 | } |
1de06328 | 398 | |
a1933d95 | 399 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 400 | PL_regeol = strend; |
33b8afdf JH |
401 | if (do_utf8) { |
402 | if (!prog->check_utf8 && prog->check_substr) | |
403 | to_utf8_substr(prog); | |
404 | check = prog->check_utf8; | |
405 | } else { | |
406 | if (!prog->check_substr && prog->check_utf8) | |
407 | to_byte_substr(prog); | |
408 | check = prog->check_substr; | |
409 | } | |
1de06328 | 410 | if (check == &PL_sv_undef) { |
a3621e74 | 411 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 412 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
413 | goto fail; |
414 | } | |
2c2d71f5 | 415 | if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ |
cad2e5aa JH |
416 | ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) |
417 | || ( (prog->reganch & ROPT_ANCH_BOL) | |
7fba1cd6 | 418 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 419 | |
7e25d62c JH |
420 | if (!ml_anch) { |
421 | if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ | |
422 | | ROPT_IMPLICIT)) /* not a real BOL */ | |
3f7c398e | 423 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
424 | && sv && !SvROK(sv) |
425 | && (strpos != strbeg)) { | |
a3621e74 | 426 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
427 | goto fail; |
428 | } | |
429 | if (prog->check_offset_min == prog->check_offset_max && | |
cce850e4 | 430 | !(prog->reganch & ROPT_CANY_SEEN)) { |
2c2d71f5 | 431 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
432 | I32 slen; |
433 | ||
1aa99e6b | 434 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 435 | |
653099ff GS |
436 | if (SvTAIL(check)) { |
437 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 438 | |
9041c2e3 | 439 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 440 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 441 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 442 | goto fail_finish; |
cad2e5aa JH |
443 | } |
444 | /* Now should match s[0..slen-2] */ | |
445 | slen--; | |
3f7c398e | 446 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 447 | || (slen > 1 |
3f7c398e | 448 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 449 | report_neq: |
a3621e74 | 450 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
451 | goto fail_finish; |
452 | } | |
cad2e5aa | 453 | } |
3f7c398e | 454 | else if (*SvPVX_const(check) != *s |
653099ff | 455 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 456 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 457 | goto report_neq; |
c315bfe8 | 458 | check_at = s; |
2c2d71f5 | 459 | goto success_at_start; |
7e25d62c | 460 | } |
cad2e5aa | 461 | } |
2c2d71f5 | 462 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 463 | s = strpos; |
2c2d71f5 | 464 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
465 | end_shift = prog->check_end_shift; |
466 | ||
2c2d71f5 | 467 | if (!ml_anch) { |
a3b680e6 | 468 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 469 | - (SvTAIL(check) != 0); |
a3b680e6 | 470 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
471 | |
472 | if (end_shift < eshift) | |
473 | end_shift = eshift; | |
474 | } | |
cad2e5aa | 475 | } |
2c2d71f5 | 476 | else { /* Can match at random position */ |
cad2e5aa JH |
477 | ml_anch = 0; |
478 | s = strpos; | |
1de06328 YO |
479 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
480 | end_shift = prog->check_end_shift; | |
481 | ||
482 | /* end shift should be non negative here */ | |
cad2e5aa JH |
483 | } |
484 | ||
2c2d71f5 | 485 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 486 | if (end_shift < 0) |
1de06328 | 487 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 488 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
489 | #endif |
490 | ||
2c2d71f5 JH |
491 | restart: |
492 | /* Find a possible match in the region s..strend by looking for | |
493 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
494 | |
495 | { | |
496 | I32 srch_start_shift = start_shift; | |
497 | I32 srch_end_shift = end_shift; | |
498 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
499 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
500 | srch_start_shift = strbeg - s; | |
501 | } | |
6bda09f9 | 502 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
503 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
504 | (IV)prog->check_offset_min, | |
505 | (IV)srch_start_shift, | |
506 | (IV)srch_end_shift, | |
507 | (IV)prog->check_end_shift); | |
508 | }); | |
509 | ||
cad2e5aa | 510 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 511 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 512 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 513 | |
2c2d71f5 JH |
514 | if (PL_screamfirst[BmRARE(check)] >= 0 |
515 | || ( BmRARE(check) == '\n' | |
516 | && (BmPREVIOUS(check) == SvCUR(check) - 1) | |
517 | && SvTAIL(check) )) | |
9041c2e3 | 518 | s = screaminstr(sv, check, |
1de06328 | 519 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 520 | else |
2c2d71f5 | 521 | goto fail_finish; |
4addbd3b HS |
522 | /* we may be pointing at the wrong string */ |
523 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 524 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
525 | if (data) |
526 | *data->scream_olds = s; | |
527 | } | |
1de06328 YO |
528 | else { |
529 | U8* start_point; | |
530 | U8* end_point; | |
531 | if (prog->reganch & ROPT_CANY_SEEN) { | |
532 | start_point= (U8*)(s + srch_start_shift); | |
533 | end_point= (U8*)(strend - srch_end_shift); | |
534 | } else { | |
535 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
536 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
537 | } | |
6bda09f9 | 538 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 539 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 540 | (int)(end_point - start_point), |
fc8cd66c | 541 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
542 | start_point); |
543 | }); | |
544 | ||
545 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 546 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
547 | } |
548 | } | |
cad2e5aa JH |
549 | /* Update the count-of-usability, remove useless subpatterns, |
550 | unshift s. */ | |
2c2d71f5 | 551 | |
ab3bbdeb YO |
552 | DEBUG_EXECUTE_r({ |
553 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
554 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
555 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 556 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
557 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
558 | ? "anchored" : "floating"), | |
559 | quoted, | |
560 | RE_SV_TAIL(check), | |
561 | (s ? " at offset " : "...\n") ); | |
562 | }); | |
2c2d71f5 JH |
563 | |
564 | if (!s) | |
565 | goto fail_finish; | |
2c2d71f5 | 566 | /* Finish the diagnostic message */ |
a3621e74 | 567 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 568 | |
1de06328 YO |
569 | /* XXX dmq: first branch is for positive lookbehind... |
570 | Our check string is offset from the beginning of the pattern. | |
571 | So we need to do any stclass tests offset forward from that | |
572 | point. I think. :-( | |
573 | */ | |
574 | ||
575 | ||
576 | ||
577 | check_at=s; | |
578 | ||
579 | ||
2c2d71f5 JH |
580 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
581 | Start with the other substr. | |
582 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 583 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
584 | *always* match. Probably should be marked during compile... |
585 | Probably it is right to do no SCREAM here... | |
586 | */ | |
587 | ||
1de06328 YO |
588 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
589 | : (prog->float_substr && prog->anchored_substr)) | |
590 | { | |
30944b6d | 591 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
592 | /* XXXX May be hopelessly wrong for UTF... */ |
593 | if (!other_last) | |
6eb5f6b9 | 594 | other_last = strpos; |
33b8afdf | 595 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
596 | do_other_anchored: |
597 | { | |
890ce7af AL |
598 | char * const last = HOP3c(s, -start_shift, strbeg); |
599 | char *last1, *last2; | |
be8e71aa | 600 | char * const saved_s = s; |
33b8afdf | 601 | SV* must; |
2c2d71f5 | 602 | |
2c2d71f5 JH |
603 | t = s - prog->check_offset_max; |
604 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 605 | && (!do_utf8 |
0ce71af7 | 606 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 607 | && t > strpos))) |
6f207bd3 | 608 | NOOP; |
2c2d71f5 JH |
609 | else |
610 | t = strpos; | |
1aa99e6b | 611 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
612 | if (t < other_last) /* These positions already checked */ |
613 | t = other_last; | |
1aa99e6b | 614 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
615 | if (last < last1) |
616 | last1 = last; | |
1de06328 YO |
617 | /* XXXX It is not documented what units *_offsets are in. |
618 | We assume bytes, but this is clearly wrong. | |
619 | Meaning this code needs to be carefully reviewed for errors. | |
620 | dmq. | |
621 | */ | |
622 | ||
2c2d71f5 | 623 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
624 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
625 | if (must == &PL_sv_undef) { | |
626 | s = (char*)NULL; | |
1de06328 | 627 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
628 | } |
629 | else | |
630 | s = fbm_instr( | |
631 | (unsigned char*)t, | |
632 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
633 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
634 | must, | |
7fba1cd6 | 635 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 636 | ); |
ab3bbdeb YO |
637 | DEBUG_EXECUTE_r({ |
638 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
639 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
640 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 641 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
642 | quoted, RE_SV_TAIL(must)); |
643 | }); | |
644 | ||
645 | ||
2c2d71f5 JH |
646 | if (!s) { |
647 | if (last1 >= last2) { | |
a3621e74 | 648 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
649 | ", giving up...\n")); |
650 | goto fail_finish; | |
651 | } | |
a3621e74 | 652 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 653 | ", trying floating at offset %ld...\n", |
be8e71aa | 654 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
655 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
656 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
657 | goto restart; |
658 | } | |
659 | else { | |
a3621e74 | 660 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 661 | (long)(s - i_strpos))); |
1aa99e6b IH |
662 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
663 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 664 | s = saved_s; |
2c2d71f5 JH |
665 | if (t == strpos) |
666 | goto try_at_start; | |
2c2d71f5 JH |
667 | goto try_at_offset; |
668 | } | |
30944b6d | 669 | } |
2c2d71f5 JH |
670 | } |
671 | else { /* Take into account the floating substring. */ | |
33b8afdf | 672 | char *last, *last1; |
be8e71aa | 673 | char * const saved_s = s; |
33b8afdf JH |
674 | SV* must; |
675 | ||
676 | t = HOP3c(s, -start_shift, strbeg); | |
677 | last1 = last = | |
678 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
679 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
680 | last = HOP3c(t, prog->float_max_offset, strend); | |
681 | s = HOP3c(t, prog->float_min_offset, strend); | |
682 | if (s < other_last) | |
683 | s = other_last; | |
2c2d71f5 | 684 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
685 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
686 | /* fbm_instr() takes into account exact value of end-of-str | |
687 | if the check is SvTAIL(ed). Since false positives are OK, | |
688 | and end-of-str is not later than strend we are OK. */ | |
689 | if (must == &PL_sv_undef) { | |
690 | s = (char*)NULL; | |
1de06328 | 691 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
692 | } |
693 | else | |
2c2d71f5 | 694 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
695 | (unsigned char*)last + SvCUR(must) |
696 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 697 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
698 | DEBUG_EXECUTE_r({ |
699 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
700 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
701 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 702 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
703 | quoted, RE_SV_TAIL(must)); |
704 | }); | |
33b8afdf JH |
705 | if (!s) { |
706 | if (last1 == last) { | |
a3621e74 | 707 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
708 | ", giving up...\n")); |
709 | goto fail_finish; | |
2c2d71f5 | 710 | } |
a3621e74 | 711 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 712 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 713 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
714 | other_last = last; |
715 | s = HOP3c(t, 1, strend); | |
716 | goto restart; | |
717 | } | |
718 | else { | |
a3621e74 | 719 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
720 | (long)(s - i_strpos))); |
721 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 722 | s = saved_s; |
33b8afdf JH |
723 | if (t == strpos) |
724 | goto try_at_start; | |
725 | goto try_at_offset; | |
726 | } | |
2c2d71f5 | 727 | } |
cad2e5aa | 728 | } |
2c2d71f5 | 729 | |
1de06328 | 730 | |
9ef43ace | 731 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 732 | |
6bda09f9 | 733 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
734 | PerlIO_printf(Perl_debug_log, |
735 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
736 | (IV)prog->check_offset_min, | |
737 | (IV)prog->check_offset_max, | |
738 | (IV)(s-strpos), | |
739 | (IV)(t-strpos), | |
740 | (IV)(t-s), | |
741 | (IV)(strend-strpos) | |
742 | ) | |
743 | ); | |
744 | ||
2c2d71f5 | 745 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 746 | && (!do_utf8 |
9ef43ace | 747 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
748 | && t > strpos))) |
749 | { | |
2c2d71f5 JH |
750 | /* Fixed substring is found far enough so that the match |
751 | cannot start at strpos. */ | |
752 | try_at_offset: | |
cad2e5aa | 753 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
754 | /* Eventually fbm_*() should handle this, but often |
755 | anchored_offset is not 0, so this check will not be wasted. */ | |
756 | /* XXXX In the code below we prefer to look for "^" even in | |
757 | presence of anchored substrings. And we search even | |
758 | beyond the found float position. These pessimizations | |
759 | are historical artefacts only. */ | |
760 | find_anchor: | |
2c2d71f5 | 761 | while (t < strend - prog->minlen) { |
cad2e5aa | 762 | if (*t == '\n') { |
4ee3650e | 763 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 764 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
765 | /* Since we moved from the found position, |
766 | we definitely contradict the found anchored | |
30944b6d IZ |
767 | substr. Due to the above check we do not |
768 | contradict "check" substr. | |
769 | Thus we can arrive here only if check substr | |
770 | is float. Redo checking for "other"=="fixed". | |
771 | */ | |
9041c2e3 | 772 | strpos = t + 1; |
a3621e74 | 773 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 774 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
775 | goto do_other_anchored; |
776 | } | |
4ee3650e GS |
777 | /* We don't contradict the found floating substring. */ |
778 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 779 | s = t + 1; |
a3621e74 | 780 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 781 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
782 | goto set_useful; |
783 | } | |
4ee3650e GS |
784 | /* Position contradicts check-string */ |
785 | /* XXXX probably better to look for check-string | |
786 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 787 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 788 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 789 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
790 | goto restart; |
791 | } | |
792 | t++; | |
793 | } | |
a3621e74 | 794 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 795 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 796 | goto fail_finish; |
cad2e5aa | 797 | } |
f5952150 | 798 | else { |
a3621e74 | 799 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 800 | PL_colors[0], PL_colors[1])); |
f5952150 | 801 | } |
cad2e5aa JH |
802 | s = t; |
803 | set_useful: | |
33b8afdf | 804 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
805 | } |
806 | else { | |
f5952150 | 807 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 808 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
809 | unless it was an MBOL and we are not after MBOL, |
810 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
811 | try_at_start: |
812 | /* Even in this situation we may use MBOL flag if strpos is offset | |
813 | wrt the start of the string. */ | |
05b4157f | 814 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 815 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d IZ |
816 | /* May be due to an implicit anchor of m{.*foo} */ |
817 | && !(prog->reganch & ROPT_IMPLICIT)) | |
818 | { | |
cad2e5aa JH |
819 | t = strpos; |
820 | goto find_anchor; | |
821 | } | |
a3621e74 | 822 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 823 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
e4584336 | 824 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 825 | ); |
2c2d71f5 | 826 | success_at_start: |
30944b6d | 827 | if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
828 | && (do_utf8 ? ( |
829 | prog->check_utf8 /* Could be deleted already */ | |
830 | && --BmUSEFUL(prog->check_utf8) < 0 | |
831 | && (prog->check_utf8 == prog->float_utf8) | |
832 | ) : ( | |
833 | prog->check_substr /* Could be deleted already */ | |
834 | && --BmUSEFUL(prog->check_substr) < 0 | |
835 | && (prog->check_substr == prog->float_substr) | |
836 | ))) | |
66e933ab | 837 | { |
cad2e5aa | 838 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 839 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
840 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
841 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
842 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
843 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
844 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
845 | check = NULL; /* abort */ | |
cad2e5aa | 846 | s = strpos; |
3cf5c195 IZ |
847 | /* XXXX This is a remnant of the old implementation. It |
848 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 849 | other heuristics. */ |
cad2e5aa JH |
850 | prog->reganch &= ~RE_USE_INTUIT; |
851 | } | |
852 | else | |
853 | s = strpos; | |
854 | } | |
855 | ||
6eb5f6b9 JH |
856 | /* Last resort... */ |
857 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
858 | /* trie stclasses are too expensive to use here, we are better off to |
859 | leave it to regmatch itself */ | |
786e8c11 | 860 | if (prog->regstclass && PL_regkind[OP(prog->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
861 | /* minlen == 0 is possible if regstclass is \b or \B, |
862 | and the fixed substr is ''$. | |
863 | Since minlen is already taken into account, s+1 is before strend; | |
864 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
865 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
866 | regstclass does not come from lookahead... */ | |
867 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
868 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
890ce7af | 869 | const U8* const str = (U8*)STRING(prog->regstclass); |
3dab1dad | 870 | const int cl_l = (PL_regkind[OP(prog->regstclass)] == EXACT |
1aa99e6b | 871 | ? CHR_DIST(str+STR_LEN(prog->regstclass), str) |
66e933ab | 872 | : 1); |
1de06328 YO |
873 | char * endpos; |
874 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
875 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
876 | else if (prog->float_substr || prog->float_utf8) | |
877 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
878 | else | |
879 | endpos= strend; | |
880 | ||
56570a2c | 881 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %d s: %d endpos: %d\n", |
b57a0404 | 882 | (IV)start_shift, check_at - strbeg, s - strbeg, endpos - strbeg)); |
1de06328 | 883 | |
6eb5f6b9 | 884 | t = s; |
3b0527fe | 885 | s = find_byclass(prog, prog->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
886 | if (!s) { |
887 | #ifdef DEBUGGING | |
cbbf8932 | 888 | const char *what = NULL; |
6eb5f6b9 JH |
889 | #endif |
890 | if (endpos == strend) { | |
a3621e74 | 891 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
892 | "Could not match STCLASS...\n") ); |
893 | goto fail; | |
894 | } | |
a3621e74 | 895 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 896 | "This position contradicts STCLASS...\n") ); |
653099ff GS |
897 | if ((prog->reganch & ROPT_ANCH) && !ml_anch) |
898 | goto fail; | |
6eb5f6b9 | 899 | /* Contradict one of substrings */ |
33b8afdf JH |
900 | if (prog->anchored_substr || prog->anchored_utf8) { |
901 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 902 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 903 | hop_and_restart: |
1aa99e6b | 904 | s = HOP3c(t, 1, strend); |
66e933ab GS |
905 | if (s + start_shift + end_shift > strend) { |
906 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 907 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
908 | "Could not match STCLASS...\n") ); |
909 | goto fail; | |
910 | } | |
5e39e1e5 HS |
911 | if (!check) |
912 | goto giveup; | |
a3621e74 | 913 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 914 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
915 | what, (long)(s + start_shift - i_strpos)) ); |
916 | goto restart; | |
917 | } | |
66e933ab | 918 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
919 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
920 | goto retry_floating_check; | |
921 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 922 | s = check_at; |
5e39e1e5 HS |
923 | if (!check) |
924 | goto giveup; | |
a3621e74 | 925 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 926 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
927 | (long)(other_last - i_strpos)) ); |
928 | goto do_other_anchored; | |
929 | } | |
60e71179 GS |
930 | /* Another way we could have checked stclass at the |
931 | current position only: */ | |
932 | if (ml_anch) { | |
933 | s = t = t + 1; | |
5e39e1e5 HS |
934 | if (!check) |
935 | goto giveup; | |
a3621e74 | 936 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 937 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 938 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 939 | goto try_at_offset; |
66e933ab | 940 | } |
33b8afdf | 941 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 942 | goto fail; |
6eb5f6b9 JH |
943 | /* Check is floating subtring. */ |
944 | retry_floating_check: | |
945 | t = check_at - start_shift; | |
a3621e74 | 946 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
947 | goto hop_and_restart; |
948 | } | |
b7953727 | 949 | if (t != s) { |
a3621e74 | 950 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 951 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
952 | (long)(t - i_strpos), (long)(s - i_strpos)) |
953 | ); | |
954 | } | |
955 | else { | |
a3621e74 | 956 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
957 | "Does not contradict STCLASS...\n"); |
958 | ); | |
959 | } | |
6eb5f6b9 | 960 | } |
5e39e1e5 | 961 | giveup: |
a3621e74 | 962 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
963 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
964 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 965 | return s; |
2c2d71f5 JH |
966 | |
967 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
968 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
969 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 970 | fail: |
a3621e74 | 971 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 972 | PL_colors[4], PL_colors[5])); |
bd61b366 | 973 | return NULL; |
cad2e5aa | 974 | } |
9661b544 | 975 | |
786e8c11 | 976 | |
3b0527fe | 977 | |
4cadc6a9 YO |
978 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, uvc, charid, \ |
979 | foldlen, foldbuf, uniflags) STMT_START { \ | |
980 | switch (trie_type) { \ | |
981 | case trie_utf8_fold: \ | |
982 | if ( foldlen>0 ) { \ | |
983 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
984 | foldlen -= len; \ | |
985 | uscan += len; \ | |
986 | len=0; \ | |
987 | } else { \ | |
988 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
989 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
990 | foldlen -= UNISKIP( uvc ); \ | |
991 | uscan = foldbuf + UNISKIP( uvc ); \ | |
992 | } \ | |
993 | break; \ | |
994 | case trie_utf8: \ | |
995 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
996 | break; \ | |
997 | case trie_plain: \ | |
998 | uvc = (UV)*uc; \ | |
999 | len = 1; \ | |
1000 | } \ | |
1001 | \ | |
1002 | if (uvc < 256) { \ | |
1003 | charid = trie->charmap[ uvc ]; \ | |
1004 | } \ | |
1005 | else { \ | |
1006 | charid = 0; \ | |
1007 | if (trie->widecharmap) { \ | |
1008 | SV** const svpp = hv_fetch(trie->widecharmap, \ | |
1009 | (char*)&uvc, sizeof(UV), 0); \ | |
1010 | if (svpp) \ | |
1011 | charid = (U16)SvIV(*svpp); \ | |
1012 | } \ | |
1013 | } \ | |
1014 | } STMT_END | |
1015 | ||
1016 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1017 | if ( (CoNd) \ | |
1018 | && (ln == len || \ | |
1019 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1020 | m, NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1021 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1022 | goto got_it; \ |
1023 | else { \ | |
1024 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1025 | uvchr_to_utf8(tmpbuf, c); \ | |
1026 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1027 | if ( f != c \ | |
1028 | && (f == c1 || f == c2) \ | |
1029 | && (ln == foldlen || \ | |
1030 | !ibcmp_utf8((char *) foldbuf, \ | |
1031 | NULL, foldlen, do_utf8, \ | |
1032 | m, \ | |
1033 | NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1034 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1035 | goto got_it; \ |
1036 | } \ | |
1037 | s += len | |
1038 | ||
1039 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1040 | STMT_START { \ | |
1041 | while (s <= e) { \ | |
1042 | if ( (CoNd) \ | |
1043 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1044 | ? ibcmp(s, m, ln) \ | |
1045 | : ibcmp_locale(s, m, ln))) \ | |
24b23f37 | 1046 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1047 | goto got_it; \ |
1048 | s++; \ | |
1049 | } \ | |
1050 | } STMT_END | |
1051 | ||
1052 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1053 | STMT_START { \ | |
1054 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1055 | CoDe \ | |
1056 | s += uskip; \ | |
1057 | } \ | |
1058 | } STMT_END | |
1059 | ||
1060 | #define REXEC_FBC_SCAN(CoDe) \ | |
1061 | STMT_START { \ | |
1062 | while (s < strend) { \ | |
1063 | CoDe \ | |
1064 | s++; \ | |
1065 | } \ | |
1066 | } STMT_END | |
1067 | ||
1068 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1069 | REXEC_FBC_UTF8_SCAN( \ | |
1070 | if (CoNd) { \ | |
24b23f37 | 1071 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1072 | goto got_it; \ |
1073 | else \ | |
1074 | tmp = doevery; \ | |
1075 | } \ | |
1076 | else \ | |
1077 | tmp = 1; \ | |
1078 | ) | |
1079 | ||
1080 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1081 | REXEC_FBC_SCAN( \ | |
1082 | if (CoNd) { \ | |
24b23f37 | 1083 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1084 | goto got_it; \ |
1085 | else \ | |
1086 | tmp = doevery; \ | |
1087 | } \ | |
1088 | else \ | |
1089 | tmp = 1; \ | |
1090 | ) | |
1091 | ||
1092 | #define REXEC_FBC_TRYIT \ | |
24b23f37 | 1093 | if ((!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1094 | goto got_it |
1095 | ||
1096 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ | |
1097 | if (do_utf8) { \ | |
1098 | UtFpReLoAd; \ | |
1099 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1100 | } \ | |
1101 | else { \ | |
1102 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1103 | } \ | |
1104 | break | |
1105 | ||
1106 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1107 | PL_reg_flags |= RF_tainted; \ | |
1108 | if (do_utf8) { \ | |
1109 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1110 | } \ | |
1111 | else { \ | |
1112 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1113 | } \ | |
1114 | break | |
1115 | ||
786e8c11 YO |
1116 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1117 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1118 | ||
1119 | /* We know what class REx starts with. Try to find this position... */ | |
1120 | /* if reginfo is NULL, its a dryrun */ | |
1121 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1122 | in regmatch. /grrr */ | |
1123 | ||
3c3eec57 | 1124 | STATIC char * |
07be1b83 | 1125 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
24b23f37 | 1126 | const char *strend, regmatch_info *reginfo) |
a687059c | 1127 | { |
27da23d5 | 1128 | dVAR; |
1df70142 | 1129 | const I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
6eb5f6b9 | 1130 | char *m; |
d8093b23 | 1131 | STRLEN ln; |
5dab1207 | 1132 | STRLEN lnc; |
078c425b | 1133 | register STRLEN uskip; |
d8093b23 G |
1134 | unsigned int c1; |
1135 | unsigned int c2; | |
6eb5f6b9 JH |
1136 | char *e; |
1137 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1138 | register const bool do_utf8 = PL_reg_match_utf8; |
cad2e5aa | 1139 | |
6eb5f6b9 JH |
1140 | /* We know what class it must start with. */ |
1141 | switch (OP(c)) { | |
6eb5f6b9 | 1142 | case ANYOF: |
388cc4de | 1143 | if (do_utf8) { |
4cadc6a9 | 1144 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1145 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1146 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1147 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1148 | } |
1149 | else { | |
1150 | while (s < strend) { | |
1151 | STRLEN skip = 1; | |
1152 | ||
32fc9b6a | 1153 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1154 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1155 | /* The assignment of 2 is intentional: | |
1156 | * for the folded sharp s, the skip is 2. */ | |
1157 | (skip = SHARP_S_SKIP))) { | |
24b23f37 | 1158 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
388cc4de HS |
1159 | goto got_it; |
1160 | else | |
1161 | tmp = doevery; | |
1162 | } | |
1163 | else | |
1164 | tmp = 1; | |
1165 | s += skip; | |
1166 | } | |
a0d0e21e | 1167 | } |
6eb5f6b9 | 1168 | break; |
f33976b4 | 1169 | case CANY: |
4cadc6a9 | 1170 | REXEC_FBC_SCAN( |
24b23f37 | 1171 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
f33976b4 DB |
1172 | goto got_it; |
1173 | else | |
1174 | tmp = doevery; | |
4cadc6a9 | 1175 | ); |
f33976b4 | 1176 | break; |
6eb5f6b9 | 1177 | case EXACTF: |
5dab1207 NIS |
1178 | m = STRING(c); |
1179 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1180 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1181 | if (UTF) { |
a2a2844f | 1182 | STRLEN ulen1, ulen2; |
5dab1207 | 1183 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1184 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1185 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1186 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1187 | |
1188 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1189 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1190 | ||
89ebb4a3 | 1191 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1192 | 0, uniflags); |
89ebb4a3 | 1193 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1194 | 0, uniflags); |
5dab1207 NIS |
1195 | lnc = 0; |
1196 | while (sm < ((U8 *) m + ln)) { | |
1197 | lnc++; | |
1198 | sm += UTF8SKIP(sm); | |
1199 | } | |
1aa99e6b IH |
1200 | } |
1201 | else { | |
1202 | c1 = *(U8*)m; | |
1203 | c2 = PL_fold[c1]; | |
1204 | } | |
6eb5f6b9 JH |
1205 | goto do_exactf; |
1206 | case EXACTFL: | |
5dab1207 NIS |
1207 | m = STRING(c); |
1208 | ln = STR_LEN(c); | |
1209 | lnc = (I32) ln; | |
d8093b23 | 1210 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1211 | c2 = PL_fold_locale[c1]; |
1212 | do_exactf: | |
db12adc6 | 1213 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1214 | |
3b0527fe | 1215 | if (!reginfo && e < s) |
6eb5f6b9 | 1216 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1217 | |
60a8b682 JH |
1218 | /* The idea in the EXACTF* cases is to first find the |
1219 | * first character of the EXACTF* node and then, if | |
1220 | * necessary, case-insensitively compare the full | |
1221 | * text of the node. The c1 and c2 are the first | |
1222 | * characters (though in Unicode it gets a bit | |
1223 | * more complicated because there are more cases | |
7f16dd3d JH |
1224 | * than just upper and lower: one needs to use |
1225 | * the so-called folding case for case-insensitive | |
1226 | * matching (called "loose matching" in Unicode). | |
1227 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1228 | |
1aa99e6b | 1229 | if (do_utf8) { |
575cac57 | 1230 | UV c, f; |
89ebb4a3 | 1231 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1232 | STRLEN len, foldlen; |
4ad0818d | 1233 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1234 | if (c1 == c2) { |
5dab1207 NIS |
1235 | /* Upper and lower of 1st char are equal - |
1236 | * probably not a "letter". */ | |
1aa99e6b | 1237 | while (s <= e) { |
89ebb4a3 | 1238 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1239 | uniflags); |
4cadc6a9 | 1240 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1241 | } |
09091399 JH |
1242 | } |
1243 | else { | |
1aa99e6b | 1244 | while (s <= e) { |
89ebb4a3 | 1245 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1246 | uniflags); |
80aecb99 | 1247 | |
60a8b682 | 1248 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1249 | * Note that not all the possible combinations |
1250 | * are handled here: some of them are handled | |
1251 | * by the standard folding rules, and some of | |
1252 | * them (the character class or ANYOF cases) | |
1253 | * are handled during compiletime in | |
1254 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1255 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1256 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1257 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1258 | |
4cadc6a9 | 1259 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1260 | } |
09091399 | 1261 | } |
1aa99e6b IH |
1262 | } |
1263 | else { | |
1264 | if (c1 == c2) | |
4cadc6a9 | 1265 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1266 | else |
4cadc6a9 | 1267 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1268 | } |
1269 | break; | |
bbce6d69 | 1270 | case BOUNDL: |
3280af22 | 1271 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1272 | /* FALL THROUGH */ |
a0d0e21e | 1273 | case BOUND: |
ffc61ed2 | 1274 | if (do_utf8) { |
12d33761 | 1275 | if (s == PL_bostr) |
ffc61ed2 JH |
1276 | tmp = '\n'; |
1277 | else { | |
6136c704 | 1278 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1279 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1280 | } |
1281 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1282 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1283 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1284 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1285 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1286 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1287 | isALNUM_LC_utf8((U8*)s))) |
1288 | { | |
1289 | tmp = !tmp; | |
4cadc6a9 | 1290 | REXEC_FBC_TRYIT; |
a687059c | 1291 | } |
4cadc6a9 | 1292 | ); |
a0d0e21e | 1293 | } |
667bb95a | 1294 | else { |
12d33761 | 1295 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1296 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1297 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1298 | if (tmp == |
1299 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1300 | tmp = !tmp; | |
4cadc6a9 | 1301 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1302 | } |
4cadc6a9 | 1303 | ); |
a0ed51b3 | 1304 | } |
24b23f37 | 1305 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1306 | goto got_it; |
1307 | break; | |
bbce6d69 | 1308 | case NBOUNDL: |
3280af22 | 1309 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1310 | /* FALL THROUGH */ |
a0d0e21e | 1311 | case NBOUND: |
ffc61ed2 | 1312 | if (do_utf8) { |
12d33761 | 1313 | if (s == PL_bostr) |
ffc61ed2 JH |
1314 | tmp = '\n'; |
1315 | else { | |
6136c704 | 1316 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1317 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1318 | } |
1319 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1320 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1321 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1322 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1323 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1324 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1325 | isALNUM_LC_utf8((U8*)s))) |
1326 | tmp = !tmp; | |
4cadc6a9 YO |
1327 | else REXEC_FBC_TRYIT; |
1328 | ); | |
a0d0e21e | 1329 | } |
667bb95a | 1330 | else { |
12d33761 | 1331 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1332 | tmp = ((OP(c) == NBOUND ? |
1333 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1334 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1335 | if (tmp == |
1336 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1337 | tmp = !tmp; | |
4cadc6a9 YO |
1338 | else REXEC_FBC_TRYIT; |
1339 | ); | |
a0ed51b3 | 1340 | } |
24b23f37 | 1341 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1342 | goto got_it; |
1343 | break; | |
a0d0e21e | 1344 | case ALNUM: |
4cadc6a9 YO |
1345 | REXEC_FBC_CSCAN_PRELOAD( |
1346 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1347 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1348 | isALNUM(*s) | |
1349 | ); | |
bbce6d69 | 1350 | case ALNUML: |
4cadc6a9 YO |
1351 | REXEC_FBC_CSCAN_TAINT( |
1352 | isALNUM_LC_utf8((U8*)s), | |
1353 | isALNUM_LC(*s) | |
1354 | ); | |
a0d0e21e | 1355 | case NALNUM: |
4cadc6a9 YO |
1356 | REXEC_FBC_CSCAN_PRELOAD( |
1357 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1358 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1359 | !isALNUM(*s) | |
1360 | ); | |
bbce6d69 | 1361 | case NALNUML: |
4cadc6a9 YO |
1362 | REXEC_FBC_CSCAN_TAINT( |
1363 | !isALNUM_LC_utf8((U8*)s), | |
1364 | !isALNUM_LC(*s) | |
1365 | ); | |
a0d0e21e | 1366 | case SPACE: |
4cadc6a9 YO |
1367 | REXEC_FBC_CSCAN_PRELOAD( |
1368 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1369 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1370 | isSPACE(*s) | |
1371 | ); | |
bbce6d69 | 1372 | case SPACEL: |
4cadc6a9 YO |
1373 | REXEC_FBC_CSCAN_TAINT( |
1374 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1375 | isSPACE_LC(*s) | |
1376 | ); | |
a0d0e21e | 1377 | case NSPACE: |
4cadc6a9 YO |
1378 | REXEC_FBC_CSCAN_PRELOAD( |
1379 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1380 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1381 | !isSPACE(*s) | |
1382 | ); | |
bbce6d69 | 1383 | case NSPACEL: |
4cadc6a9 YO |
1384 | REXEC_FBC_CSCAN_TAINT( |
1385 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1386 | !isSPACE_LC(*s) | |
1387 | ); | |
a0d0e21e | 1388 | case DIGIT: |
4cadc6a9 YO |
1389 | REXEC_FBC_CSCAN_PRELOAD( |
1390 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1391 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1392 | isDIGIT(*s) | |
1393 | ); | |
b8c5462f | 1394 | case DIGITL: |
4cadc6a9 YO |
1395 | REXEC_FBC_CSCAN_TAINT( |
1396 | isDIGIT_LC_utf8((U8*)s), | |
1397 | isDIGIT_LC(*s) | |
1398 | ); | |
a0d0e21e | 1399 | case NDIGIT: |
4cadc6a9 YO |
1400 | REXEC_FBC_CSCAN_PRELOAD( |
1401 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1402 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1403 | !isDIGIT(*s) | |
1404 | ); | |
b8c5462f | 1405 | case NDIGITL: |
4cadc6a9 YO |
1406 | REXEC_FBC_CSCAN_TAINT( |
1407 | !isDIGIT_LC_utf8((U8*)s), | |
1408 | !isDIGIT_LC(*s) | |
1409 | ); | |
1de06328 YO |
1410 | case AHOCORASICKC: |
1411 | case AHOCORASICK: | |
07be1b83 YO |
1412 | { |
1413 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1414 | trie_type = do_utf8 ? | |
1415 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1416 | : trie_plain; | |
1417 | /* what trie are we using right now */ | |
1418 | reg_ac_data *aho | |
1419 | = (reg_ac_data*)prog->data->data[ ARG( c ) ]; | |
1420 | reg_trie_data *trie=aho->trie; | |
1421 | ||
1422 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1423 | #ifdef DEBUGGING |
07be1b83 | 1424 | const char *real_start = s; |
6148ee25 | 1425 | #endif |
07be1b83 | 1426 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1427 | SV *sv_points; |
1428 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1429 | when reading a given char. For ASCII this |
be8e71aa YO |
1430 | is unnecessary overhead as the relationship |
1431 | is always 1:1, but for unicode, especially | |
1432 | case folded unicode this is not true. */ | |
f9e705e8 | 1433 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1434 | U8 *bitmap=NULL; |
1435 | ||
07be1b83 YO |
1436 | |
1437 | GET_RE_DEBUG_FLAGS_DECL; | |
1438 | ||
be8e71aa YO |
1439 | /* We can't just allocate points here. We need to wrap it in |
1440 | * an SV so it gets freed properly if there is a croak while | |
1441 | * running the match */ | |
1442 | ENTER; | |
1443 | SAVETMPS; | |
1444 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1445 | SvCUR_set(sv_points, | |
1446 | maxlen * sizeof(U8 *)); | |
1447 | SvPOK_on(sv_points); | |
1448 | sv_2mortal(sv_points); | |
1449 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1450 | if ( trie_type != trie_utf8_fold |
1451 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1452 | { | |
786e8c11 YO |
1453 | if (trie->bitmap) |
1454 | bitmap=(U8*)trie->bitmap; | |
1455 | else | |
1456 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1457 | } |
786e8c11 YO |
1458 | /* this is the Aho-Corasick algorithm modified a touch |
1459 | to include special handling for long "unknown char" | |
1460 | sequences. The basic idea being that we use AC as long | |
1461 | as we are dealing with a possible matching char, when | |
1462 | we encounter an unknown char (and we have not encountered | |
1463 | an accepting state) we scan forward until we find a legal | |
1464 | starting char. | |
1465 | AC matching is basically that of trie matching, except | |
1466 | that when we encounter a failing transition, we fall back | |
1467 | to the current states "fail state", and try the current char | |
1468 | again, a process we repeat until we reach the root state, | |
1469 | state 1, or a legal transition. If we fail on the root state | |
1470 | then we can either terminate if we have reached an accepting | |
1471 | state previously, or restart the entire process from the beginning | |
1472 | if we have not. | |
1473 | ||
1474 | */ | |
07be1b83 YO |
1475 | while (s <= last_start) { |
1476 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1477 | U8 *uc = (U8*)s; | |
1478 | U16 charid = 0; | |
1479 | U32 base = 1; | |
1480 | U32 state = 1; | |
1481 | UV uvc = 0; | |
1482 | STRLEN len = 0; | |
1483 | STRLEN foldlen = 0; | |
1484 | U8 *uscan = (U8*)NULL; | |
1485 | U8 *leftmost = NULL; | |
786e8c11 YO |
1486 | #ifdef DEBUGGING |
1487 | U32 accepted_word= 0; | |
1488 | #endif | |
07be1b83 YO |
1489 | U32 pointpos = 0; |
1490 | ||
1491 | while ( state && uc <= (U8*)strend ) { | |
1492 | int failed=0; | |
786e8c11 YO |
1493 | U32 word = aho->states[ state ].wordnum; |
1494 | ||
1de06328 YO |
1495 | if( state==1 ) { |
1496 | if ( bitmap ) { | |
1497 | DEBUG_TRIE_EXECUTE_r( | |
1498 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1499 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1500 | (char *)uc, do_utf8 ); | |
1501 | PerlIO_printf( Perl_debug_log, | |
1502 | " Scanning for legal start char...\n"); | |
1503 | } | |
1504 | ); | |
1505 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1506 | uc++; | |
786e8c11 | 1507 | } |
1de06328 | 1508 | s= (char *)uc; |
786e8c11 | 1509 | } |
786e8c11 YO |
1510 | if (uc >(U8*)last_start) break; |
1511 | } | |
1512 | ||
1513 | if ( word ) { | |
1514 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1515 | if (!leftmost || lpos < leftmost) { | |
1516 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1517 | leftmost= lpos; |
786e8c11 | 1518 | } |
07be1b83 | 1519 | if (base==0) break; |
786e8c11 | 1520 | |
07be1b83 YO |
1521 | } |
1522 | points[pointpos++ % maxlen]= uc; | |
4cadc6a9 YO |
1523 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
1524 | uvc, charid, foldlen, foldbuf, uniflags); | |
786e8c11 YO |
1525 | DEBUG_TRIE_EXECUTE_r({ |
1526 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1527 | s, do_utf8 ); | |
07be1b83 | 1528 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1529 | " Charid:%3u CP:%4"UVxf" ", |
1530 | charid, uvc); | |
1531 | }); | |
07be1b83 YO |
1532 | |
1533 | do { | |
6148ee25 | 1534 | #ifdef DEBUGGING |
786e8c11 | 1535 | word = aho->states[ state ].wordnum; |
6148ee25 | 1536 | #endif |
07be1b83 YO |
1537 | base = aho->states[ state ].trans.base; |
1538 | ||
786e8c11 YO |
1539 | DEBUG_TRIE_EXECUTE_r({ |
1540 | if (failed) | |
1541 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1542 | s, do_utf8 ); | |
07be1b83 | 1543 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1544 | "%sState: %4"UVxf", word=%"UVxf, |
1545 | failed ? " Fail transition to " : "", | |
1546 | (UV)state, (UV)word); | |
1547 | }); | |
07be1b83 YO |
1548 | if ( base ) { |
1549 | U32 tmp; | |
1550 | if (charid && | |
1551 | (base + charid > trie->uniquecharcount ) | |
1552 | && (base + charid - 1 - trie->uniquecharcount | |
1553 | < trie->lasttrans) | |
1554 | && trie->trans[base + charid - 1 - | |
1555 | trie->uniquecharcount].check == state | |
1556 | && (tmp=trie->trans[base + charid - 1 - | |
1557 | trie->uniquecharcount ].next)) | |
1558 | { | |
786e8c11 YO |
1559 | DEBUG_TRIE_EXECUTE_r( |
1560 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1561 | state = tmp; |
1562 | break; | |
1563 | } | |
1564 | else { | |
786e8c11 YO |
1565 | DEBUG_TRIE_EXECUTE_r( |
1566 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1567 | failed = 1; | |
1568 | state = aho->fail[state]; | |
07be1b83 YO |
1569 | } |
1570 | } | |
1571 | else { | |
1572 | /* we must be accepting here */ | |
786e8c11 YO |
1573 | DEBUG_TRIE_EXECUTE_r( |
1574 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1575 | failed = 1; | |
07be1b83 YO |
1576 | break; |
1577 | } | |
1578 | } while(state); | |
786e8c11 | 1579 | uc += len; |
07be1b83 YO |
1580 | if (failed) { |
1581 | if (leftmost) | |
1582 | break; | |
786e8c11 | 1583 | if (!state) state = 1; |
07be1b83 YO |
1584 | } |
1585 | } | |
1586 | if ( aho->states[ state ].wordnum ) { | |
1587 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1588 | if (!leftmost || lpos < leftmost) { |
1589 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1590 | leftmost = lpos; |
786e8c11 | 1591 | } |
07be1b83 | 1592 | } |
07be1b83 YO |
1593 | if (leftmost) { |
1594 | s = (char*)leftmost; | |
786e8c11 YO |
1595 | DEBUG_TRIE_EXECUTE_r({ |
1596 | PerlIO_printf( | |
1597 | Perl_debug_log,"Matches word #%"UVxf" at position %d. Trying full pattern...\n", | |
1598 | (UV)accepted_word, s - real_start | |
1599 | ); | |
1600 | }); | |
24b23f37 | 1601 | if (!reginfo || regtry(reginfo, &s)) { |
be8e71aa YO |
1602 | FREETMPS; |
1603 | LEAVE; | |
07be1b83 | 1604 | goto got_it; |
be8e71aa | 1605 | } |
07be1b83 | 1606 | s = HOPc(s,1); |
786e8c11 YO |
1607 | DEBUG_TRIE_EXECUTE_r({ |
1608 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1609 | }); | |
07be1b83 | 1610 | } else { |
786e8c11 YO |
1611 | DEBUG_TRIE_EXECUTE_r( |
1612 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1613 | break; |
1614 | } | |
1615 | } | |
be8e71aa YO |
1616 | FREETMPS; |
1617 | LEAVE; | |
07be1b83 YO |
1618 | } |
1619 | break; | |
b3c9acc1 | 1620 | default: |
3c3eec57 GS |
1621 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1622 | break; | |
d6a28714 | 1623 | } |
6eb5f6b9 JH |
1624 | return 0; |
1625 | got_it: | |
1626 | return s; | |
1627 | } | |
1628 | ||
1629 | /* | |
1630 | - regexec_flags - match a regexp against a string | |
1631 | */ | |
1632 | I32 | |
1633 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1634 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1635 | /* strend: pointer to null at end of string */ | |
1636 | /* strbeg: real beginning of string */ | |
1637 | /* minend: end of match must be >=minend after stringarg. */ | |
1638 | /* data: May be used for some additional optimizations. */ | |
1639 | /* nosave: For optimizations. */ | |
1640 | { | |
97aff369 | 1641 | dVAR; |
24b23f37 | 1642 | /*register*/ char *s; |
6eb5f6b9 | 1643 | register regnode *c; |
24b23f37 | 1644 | /*register*/ char *startpos = stringarg; |
6eb5f6b9 JH |
1645 | I32 minlen; /* must match at least this many chars */ |
1646 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1647 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1648 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1649 | char *scream_olds = NULL; |
3dab1dad | 1650 | SV* const oreplsv = GvSV(PL_replgv); |
f9f4320a | 1651 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1652 | I32 multiline; |
0df25f3d | 1653 | |
3b0527fe | 1654 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1655 | |
1656 | GET_RE_DEBUG_FLAGS_DECL; | |
1657 | ||
9d4ba2ae | 1658 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1659 | |
1660 | /* Be paranoid... */ | |
1661 | if (prog == NULL || startpos == NULL) { | |
1662 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1663 | return 0; | |
1664 | } | |
1665 | ||
2757e526 | 1666 | multiline = prog->reganch & PMf_MULTILINE; |
3b0527fe | 1667 | reginfo.prog = prog; |
2757e526 | 1668 | |
bac06658 | 1669 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1670 | DEBUG_EXECUTE_r( |
1671 | debug_start_match(prog, do_utf8, startpos, strend, | |
1672 | "Matching"); | |
1673 | ); | |
bac06658 | 1674 | |
6eb5f6b9 | 1675 | minlen = prog->minlen; |
1de06328 YO |
1676 | |
1677 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1678 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1679 | "String too short [regexec_flags]...\n")); |
1680 | goto phooey; | |
1aa99e6b | 1681 | } |
6eb5f6b9 | 1682 | |
1de06328 | 1683 | |
6eb5f6b9 JH |
1684 | /* Check validity of program. */ |
1685 | if (UCHARAT(prog->program) != REG_MAGIC) { | |
1686 | Perl_croak(aTHX_ "corrupted regexp program"); | |
1687 | } | |
1688 | ||
1689 | PL_reg_flags = 0; | |
1690 | PL_reg_eval_set = 0; | |
1691 | PL_reg_maxiter = 0; | |
1692 | ||
1693 | if (prog->reganch & ROPT_UTF8) | |
1694 | PL_reg_flags |= RF_utf8; | |
1695 | ||
1696 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1697 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1698 | PL_bostr = strbeg; |
3b0527fe | 1699 | reginfo.sv = sv; |
6eb5f6b9 JH |
1700 | |
1701 | /* Mark end of line for $ (and such) */ | |
1702 | PL_regeol = strend; | |
1703 | ||
1704 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1705 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1706 | |
6eb5f6b9 JH |
1707 | /* If there is a "must appear" string, look for it. */ |
1708 | s = startpos; | |
1709 | ||
3b0527fe | 1710 | if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1711 | MAGIC *mg; |
1712 | ||
1713 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
3b0527fe | 1714 | reginfo.ganch = startpos; |
6eb5f6b9 JH |
1715 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1716 | && SvMAGIC(sv) | |
14befaf4 DM |
1717 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1718 | && mg->mg_len >= 0) { | |
3b0527fe | 1719 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
6eb5f6b9 | 1720 | if (prog->reganch & ROPT_ANCH_GPOS) { |
3b0527fe | 1721 | if (s > reginfo.ganch) |
6eb5f6b9 | 1722 | goto phooey; |
3b0527fe | 1723 | s = reginfo.ganch; |
6eb5f6b9 JH |
1724 | } |
1725 | } | |
1726 | else /* pos() not defined */ | |
3b0527fe | 1727 | reginfo.ganch = strbeg; |
6eb5f6b9 JH |
1728 | } |
1729 | ||
a0714e2c | 1730 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1731 | re_scream_pos_data d; |
1732 | ||
1733 | d.scream_olds = &scream_olds; | |
1734 | d.scream_pos = &scream_pos; | |
1735 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1736 | if (!s) { |
a3621e74 | 1737 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1738 | goto phooey; /* not present */ |
3fa9c3d7 | 1739 | } |
6eb5f6b9 JH |
1740 | } |
1741 | ||
1de06328 | 1742 | |
6eb5f6b9 JH |
1743 | |
1744 | /* Simplest case: anchored match need be tried only once. */ | |
1745 | /* [unless only anchor is BOL and multiline is set] */ | |
1746 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { | |
24b23f37 | 1747 | if (s == startpos && regtry(®info, &startpos)) |
6eb5f6b9 | 1748 | goto got_it; |
7fba1cd6 | 1749 | else if (multiline || (prog->reganch & ROPT_IMPLICIT) |
6eb5f6b9 JH |
1750 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
1751 | { | |
1752 | char *end; | |
1753 | ||
1754 | if (minlen) | |
1755 | dontbother = minlen - 1; | |
1aa99e6b | 1756 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1757 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1758 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1759 | if (s == startpos) |
1760 | goto after_try; | |
1761 | while (1) { | |
24b23f37 | 1762 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1763 | goto got_it; |
1764 | after_try: | |
1765 | if (s >= end) | |
1766 | goto phooey; | |
1767 | if (prog->reganch & RE_USE_INTUIT) { | |
1768 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); | |
1769 | if (!s) | |
1770 | goto phooey; | |
1771 | } | |
1772 | else | |
1773 | s++; | |
1774 | } | |
1775 | } else { | |
1776 | if (s > startpos) | |
1777 | s--; | |
1778 | while (s < end) { | |
1779 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
24b23f37 | 1780 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1781 | goto got_it; |
1782 | } | |
1783 | } | |
1784 | } | |
1785 | } | |
1786 | goto phooey; | |
f9f4320a YO |
1787 | } else if (ROPT_GPOS_CHECK == (prog->reganch & ROPT_GPOS_CHECK)) |
1788 | { | |
1789 | /* the warning about reginfo.ganch being used without intialization | |
1790 | is bogus -- we set it above, when prog->reganch & ROPT_GPOS_SEEN | |
1791 | and we only enter this block when the same bit is set. */ | |
24b23f37 | 1792 | if (regtry(®info, ®info.ganch)) |
6eb5f6b9 JH |
1793 | goto got_it; |
1794 | goto phooey; | |
1795 | } | |
1796 | ||
1797 | /* Messy cases: unanchored match. */ | |
33b8afdf | 1798 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) { |
6eb5f6b9 JH |
1799 | /* we have /x+whatever/ */ |
1800 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1801 | char ch; |
bf93d4cc GS |
1802 | #ifdef DEBUGGING |
1803 | int did_match = 0; | |
1804 | #endif | |
33b8afdf JH |
1805 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1806 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1807 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1808 | |
1aa99e6b | 1809 | if (do_utf8) { |
4cadc6a9 | 1810 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1811 | if (*s == ch) { |
a3621e74 | 1812 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1813 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1814 | s += UTF8SKIP(s); |
1815 | while (s < strend && *s == ch) | |
1816 | s += UTF8SKIP(s); | |
1817 | } | |
4cadc6a9 | 1818 | ); |
6eb5f6b9 JH |
1819 | } |
1820 | else { | |
4cadc6a9 | 1821 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1822 | if (*s == ch) { |
a3621e74 | 1823 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1824 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1825 | s++; |
1826 | while (s < strend && *s == ch) | |
1827 | s++; | |
1828 | } | |
4cadc6a9 | 1829 | ); |
6eb5f6b9 | 1830 | } |
a3621e74 | 1831 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1832 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1833 | "Did not find anchored character...\n") |
1834 | ); | |
6eb5f6b9 | 1835 | } |
a0714e2c SS |
1836 | else if (prog->anchored_substr != NULL |
1837 | || prog->anchored_utf8 != NULL | |
1838 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1839 | && prog->float_max_offset < strend - s)) { |
1840 | SV *must; | |
1841 | I32 back_max; | |
1842 | I32 back_min; | |
1843 | char *last; | |
6eb5f6b9 | 1844 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1845 | #ifdef DEBUGGING |
1846 | int did_match = 0; | |
1847 | #endif | |
33b8afdf JH |
1848 | if (prog->anchored_substr || prog->anchored_utf8) { |
1849 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1850 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1851 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1852 | back_max = back_min = prog->anchored_offset; | |
1853 | } else { | |
1854 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1855 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1856 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1857 | back_max = prog->float_max_offset; | |
1858 | back_min = prog->float_min_offset; | |
1859 | } | |
1de06328 YO |
1860 | |
1861 | ||
33b8afdf JH |
1862 | if (must == &PL_sv_undef) |
1863 | /* could not downgrade utf8 check substring, so must fail */ | |
1864 | goto phooey; | |
1865 | ||
1de06328 YO |
1866 | if (back_min<0) { |
1867 | last = strend; | |
1868 | } else { | |
1869 | last = HOP3c(strend, /* Cannot start after this */ | |
1870 | -(I32)(CHR_SVLEN(must) | |
1871 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1872 | } | |
6eb5f6b9 JH |
1873 | if (s > PL_bostr) |
1874 | last1 = HOPc(s, -1); | |
1875 | else | |
1876 | last1 = s - 1; /* bogus */ | |
1877 | ||
a0288114 | 1878 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1879 | check_substr==must. */ |
1880 | scream_pos = -1; | |
1881 | dontbother = end_shift; | |
1882 | strend = HOPc(strend, -dontbother); | |
1883 | while ( (s <= last) && | |
9041c2e3 | 1884 | ((flags & REXEC_SCREAM) |
1de06328 | 1885 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1886 | end_shift, &scream_pos, 0)) |
1de06328 | 1887 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1888 | (unsigned char*)strend, must, |
7fba1cd6 | 1889 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1890 | /* we may be pointing at the wrong string */ |
1891 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1892 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1893 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1894 | if (HOPc(s, -back_max) > last1) { |
1895 | last1 = HOPc(s, -back_min); | |
1896 | s = HOPc(s, -back_max); | |
1897 | } | |
1898 | else { | |
52657f30 | 1899 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1900 | |
1901 | last1 = HOPc(s, -back_min); | |
52657f30 | 1902 | s = t; |
6eb5f6b9 | 1903 | } |
1aa99e6b | 1904 | if (do_utf8) { |
6eb5f6b9 | 1905 | while (s <= last1) { |
24b23f37 | 1906 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1907 | goto got_it; |
1908 | s += UTF8SKIP(s); | |
1909 | } | |
1910 | } | |
1911 | else { | |
1912 | while (s <= last1) { | |
24b23f37 | 1913 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1914 | goto got_it; |
1915 | s++; | |
1916 | } | |
1917 | } | |
1918 | } | |
ab3bbdeb YO |
1919 | DEBUG_EXECUTE_r(if (!did_match) { |
1920 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
1921 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
1922 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 1923 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 1924 | ? "anchored" : "floating"), |
ab3bbdeb YO |
1925 | quoted, RE_SV_TAIL(must)); |
1926 | }); | |
6eb5f6b9 JH |
1927 | goto phooey; |
1928 | } | |
786e8c11 | 1929 | else if ( (c = prog->regstclass) ) { |
f14c76ed | 1930 | if (minlen) { |
be8e71aa | 1931 | const OPCODE op = OP(prog->regstclass); |
66e933ab | 1932 | /* don't bother with what can't match */ |
786e8c11 | 1933 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
1934 | strend = HOPc(strend, -(minlen - 1)); |
1935 | } | |
a3621e74 | 1936 | DEBUG_EXECUTE_r({ |
be8e71aa | 1937 | SV * const prop = sv_newmortal(); |
32fc9b6a | 1938 | regprop(prog, prop, c); |
0df25f3d | 1939 | { |
ab3bbdeb YO |
1940 | RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1), |
1941 | s,strend-s,60); | |
0df25f3d | 1942 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 1943 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 1944 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 1945 | quoted, (int)(strend - s)); |
0df25f3d | 1946 | } |
ffc61ed2 | 1947 | }); |
3b0527fe | 1948 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1949 | goto got_it; |
07be1b83 | 1950 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
1951 | } |
1952 | else { | |
1953 | dontbother = 0; | |
a0714e2c | 1954 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 1955 | /* Trim the end. */ |
d6a28714 | 1956 | char *last; |
33b8afdf JH |
1957 | SV* float_real; |
1958 | ||
1959 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1960 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1961 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
1962 | |
1963 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 1964 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
1965 | end_shift, &scream_pos, 1); /* last one */ |
1966 | if (!last) | |
ffc61ed2 | 1967 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
1968 | /* we may be pointing at the wrong string */ |
1969 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 1970 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 1971 | } |
d6a28714 JH |
1972 | else { |
1973 | STRLEN len; | |
cfd0369c | 1974 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 1975 | |
33b8afdf | 1976 | if (SvTAIL(float_real)) { |
d6a28714 JH |
1977 | if (memEQ(strend - len + 1, little, len - 1)) |
1978 | last = strend - len + 1; | |
7fba1cd6 | 1979 | else if (!multiline) |
9041c2e3 | 1980 | last = memEQ(strend - len, little, len) |
bd61b366 | 1981 | ? strend - len : NULL; |
b8c5462f | 1982 | else |
d6a28714 JH |
1983 | goto find_last; |
1984 | } else { | |
1985 | find_last: | |
9041c2e3 | 1986 | if (len) |
d6a28714 | 1987 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 1988 | else |
a0288114 | 1989 | last = strend; /* matching "$" */ |
b8c5462f | 1990 | } |
b8c5462f | 1991 | } |
bf93d4cc | 1992 | if (last == NULL) { |
6bda09f9 YO |
1993 | DEBUG_EXECUTE_r( |
1994 | PerlIO_printf(Perl_debug_log, | |
1995 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
1996 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
1997 | goto phooey; /* Should not happen! */ |
1998 | } | |
d6a28714 JH |
1999 | dontbother = strend - last + prog->float_min_offset; |
2000 | } | |
2001 | if (minlen && (dontbother < minlen)) | |
2002 | dontbother = minlen - 1; | |
2003 | strend -= dontbother; /* this one's always in bytes! */ | |
2004 | /* We don't know much -- general case. */ | |
1aa99e6b | 2005 | if (do_utf8) { |
d6a28714 | 2006 | for (;;) { |
24b23f37 | 2007 | if (regtry(®info, &s)) |
d6a28714 JH |
2008 | goto got_it; |
2009 | if (s >= strend) | |
2010 | break; | |
b8c5462f | 2011 | s += UTF8SKIP(s); |
d6a28714 JH |
2012 | }; |
2013 | } | |
2014 | else { | |
2015 | do { | |
24b23f37 | 2016 | if (regtry(®info, &s)) |
d6a28714 JH |
2017 | goto got_it; |
2018 | } while (s++ < strend); | |
2019 | } | |
2020 | } | |
2021 | ||
2022 | /* Failure. */ | |
2023 | goto phooey; | |
2024 | ||
2025 | got_it: | |
2026 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2027 | ||
2028 | if (PL_reg_eval_set) { | |
2029 | /* Preserve the current value of $^R */ | |
2030 | if (oreplsv != GvSV(PL_replgv)) | |
2031 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2032 | restored, the value remains | |
2033 | the same. */ | |
4f639d21 | 2034 | restore_pos(aTHX_ prog); |
d6a28714 | 2035 | } |
81714fb9 YO |
2036 | if (prog->paren_names) |
2037 | (void)hv_iterinit(prog->paren_names); | |
d6a28714 JH |
2038 | |
2039 | /* make sure $`, $&, $', and $digit will work later */ | |
2040 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2041 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2042 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2043 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2044 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2045 | if ((SvIsCOW(sv) |
2046 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2047 | if (DEBUG_C_TEST) { | |
2048 | PerlIO_printf(Perl_debug_log, | |
2049 | "Copy on write: regexp capture, type %d\n", | |
2050 | (int) SvTYPE(sv)); | |
2051 | } | |
2052 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2053 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2054 | assert (SvPOKp(prog->saved_copy)); |
2055 | } else | |
2056 | #endif | |
2057 | { | |
2058 | RX_MATCH_COPIED_on(prog); | |
2059 | s = savepvn(strbeg, i); | |
2060 | prog->subbeg = s; | |
2061 | } | |
d6a28714 | 2062 | prog->sublen = i; |
d6a28714 JH |
2063 | } |
2064 | else { | |
2065 | prog->subbeg = strbeg; | |
2066 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2067 | } | |
2068 | } | |
9041c2e3 | 2069 | |
d6a28714 JH |
2070 | return 1; |
2071 | ||
2072 | phooey: | |
a3621e74 | 2073 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2074 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2075 | if (PL_reg_eval_set) |
4f639d21 | 2076 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2077 | return 0; |
2078 | } | |
2079 | ||
6bda09f9 | 2080 | |
d6a28714 JH |
2081 | /* |
2082 | - regtry - try match at specific point | |
2083 | */ | |
2084 | STATIC I32 /* 0 failure, 1 success */ | |
24b23f37 | 2085 | S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) |
d6a28714 | 2086 | { |
97aff369 | 2087 | dVAR; |
d6a28714 JH |
2088 | register I32 *sp; |
2089 | register I32 *ep; | |
2090 | CHECKPOINT lastcp; | |
3b0527fe | 2091 | regexp *prog = reginfo->prog; |
a3621e74 | 2092 | GET_RE_DEBUG_FLAGS_DECL; |
24b23f37 | 2093 | reginfo->cutpoint=NULL; |
d6a28714 JH |
2094 | |
2095 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { | |
2096 | MAGIC *mg; | |
2097 | ||
2098 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2099 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2100 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2101 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2102 | )); |
ea8d6ae1 | 2103 | SAVESTACK_CXPOS(); |
d6a28714 JH |
2104 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2105 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2106 | SAVETMPS; | |
2107 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2108 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2109 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2110 | ||
3b0527fe | 2111 | if (reginfo->sv) { |
d6a28714 | 2112 | /* Make $_ available to executed code. */ |
3b0527fe | 2113 | if (reginfo->sv != DEFSV) { |
59f00321 | 2114 | SAVE_DEFSV; |
3b0527fe | 2115 | DEFSV = reginfo->sv; |
b8c5462f | 2116 | } |
d6a28714 | 2117 | |
3b0527fe DM |
2118 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2119 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2120 | /* prepare for quick setting of pos */ |
d300d9fa NC |
2121 | #ifdef PERL_OLD_COPY_ON_WRITE |
2122 | if (SvIsCOW(sv)) | |
2123 | sv_force_normal_flags(sv, 0); | |
2124 | #endif | |
3dab1dad | 2125 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2126 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2127 | mg->mg_len = -1; |
b8c5462f | 2128 | } |
d6a28714 JH |
2129 | PL_reg_magic = mg; |
2130 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2131 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2132 | } |
09687e5a | 2133 | if (!PL_reg_curpm) { |
a02a5408 | 2134 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2135 | #ifdef USE_ITHREADS |
2136 | { | |
be8e71aa | 2137 | SV* const repointer = newSViv(0); |
577e12cc | 2138 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2139 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2140 | av_push(PL_regex_padav,repointer); |
2141 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2142 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2143 | } | |
2144 | #endif | |
2145 | } | |
aaa362c4 | 2146 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2147 | PL_reg_oldcurpm = PL_curpm; |
2148 | PL_curpm = PL_reg_curpm; | |
2149 | if (RX_MATCH_COPIED(prog)) { | |
2150 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2151 | since it may be needed if this match fails. Thus | |
2152 | $` inside (?{}) could fail... */ | |
2153 | PL_reg_oldsaved = prog->subbeg; | |
2154 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2155 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2156 | PL_nrs = prog->saved_copy; |
2157 | #endif | |
d6a28714 JH |
2158 | RX_MATCH_COPIED_off(prog); |
2159 | } | |
2160 | else | |
bd61b366 | 2161 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2162 | prog->subbeg = PL_bostr; |
2163 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2164 | } | |
24b23f37 YO |
2165 | DEBUG_EXECUTE_r(PL_reg_starttry = *startpos); |
2166 | prog->startp[0] = *startpos - PL_bostr; | |
2167 | PL_reginput = *startpos; | |
d6a28714 | 2168 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2169 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2170 | prog->lastparen = 0; |
03994de8 | 2171 | prog->lastcloseparen = 0; |
d6a28714 | 2172 | PL_regsize = 0; |
6bda09f9 YO |
2173 | PL_regstartp = prog->startp; |
2174 | PL_regendp = prog->endp; | |
d6a28714 JH |
2175 | if (PL_reg_start_tmpl <= prog->nparens) { |
2176 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2177 | if(PL_reg_start_tmp) | |
2178 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2179 | else | |
a02a5408 | 2180 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2181 | } |
2182 | ||
2183 | /* XXXX What this code is doing here?!!! There should be no need | |
2184 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2185 | this! --ilya*/ |
dafc8851 JH |
2186 | |
2187 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2188 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2189 | * PL_reglastparen), is not needed at all by the test suite |
2190 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2191 | * enough, for building DynaLoader, or otherwise this | |
2192 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2193 | * will happen. Meanwhile, this code *is* needed for the | |
2194 | * above-mentioned test suite tests to succeed. The common theme | |
2195 | * on those tests seems to be returning null fields from matches. | |
2196 | * --jhi */ | |
dafc8851 | 2197 | #if 1 |
d6a28714 JH |
2198 | sp = prog->startp; |
2199 | ep = prog->endp; | |
2200 | if (prog->nparens) { | |
097eb12c | 2201 | register I32 i; |
eb160463 | 2202 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2203 | *++sp = -1; |
2204 | *++ep = -1; | |
2205 | } | |
2206 | } | |
dafc8851 | 2207 | #endif |
02db2b7b | 2208 | REGCP_SET(lastcp); |
3b0527fe | 2209 | if (regmatch(reginfo, prog->program + 1)) { |
d6a28714 JH |
2210 | prog->endp[0] = PL_reginput - PL_bostr; |
2211 | return 1; | |
2212 | } | |
24b23f37 YO |
2213 | if (reginfo->cutpoint) |
2214 | *startpos= reginfo->cutpoint; | |
02db2b7b | 2215 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2216 | return 0; |
2217 | } | |
2218 | ||
02db2b7b | 2219 | |
8ba1375e MJD |
2220 | #define sayYES goto yes |
2221 | #define sayNO goto no | |
262b90c4 | 2222 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2223 | |
f9f4320a YO |
2224 | /* we dont use STMT_START/END here because it leads to |
2225 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2226 | #define CACHEsayNO \ | |
c476f425 DM |
2227 | if (ST.cache_mask) \ |
2228 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2229 | sayNO |
3298f257 | 2230 | |
a3621e74 | 2231 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2232 | 'failed...' are printed. It should be set such that messages |
2233 | are inline with the regop output that created them. | |
a3621e74 | 2234 | */ |
265c4333 | 2235 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2236 | |
2237 | ||
2238 | /* Make sure there is a test for this +1 options in re_tests */ | |
2239 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2240 | ||
40a82448 DM |
2241 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2242 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2243 | |
86545054 DM |
2244 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2245 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2246 | ||
5d9a96ca DM |
2247 | /* grab a new slab and return the first slot in it */ |
2248 | ||
2249 | STATIC regmatch_state * | |
2250 | S_push_slab(pTHX) | |
2251 | { | |
54df2634 NC |
2252 | #if PERL_VERSION < 9 |
2253 | dMY_CXT; | |
2254 | #endif | |
5d9a96ca DM |
2255 | regmatch_slab *s = PL_regmatch_slab->next; |
2256 | if (!s) { | |
2257 | Newx(s, 1, regmatch_slab); | |
2258 | s->prev = PL_regmatch_slab; | |
2259 | s->next = NULL; | |
2260 | PL_regmatch_slab->next = s; | |
2261 | } | |
2262 | PL_regmatch_slab = s; | |
86545054 | 2263 | return SLAB_FIRST(s); |
5d9a96ca | 2264 | } |
5b47454d | 2265 | |
95b24440 | 2266 | |
40a82448 DM |
2267 | /* push a new state then goto it */ |
2268 | ||
2269 | #define PUSH_STATE_GOTO(state, node) \ | |
2270 | scan = node; \ | |
2271 | st->resume_state = state; \ | |
2272 | goto push_state; | |
2273 | ||
2274 | /* push a new state with success backtracking, then goto it */ | |
2275 | ||
2276 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2277 | scan = node; \ | |
2278 | st->resume_state = state; \ | |
2279 | goto push_yes_state; | |
2280 | ||
aa283a38 | 2281 | |
aa283a38 | 2282 | |
d6a28714 | 2283 | /* |
95b24440 | 2284 | |
bf1f174e DM |
2285 | regmatch() - main matching routine |
2286 | ||
2287 | This is basically one big switch statement in a loop. We execute an op, | |
2288 | set 'next' to point the next op, and continue. If we come to a point which | |
2289 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2290 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2291 | state, and re-enter the loop at the state indicated. If there are no more | |
2292 | states to pop, we return failure. | |
2293 | ||
2294 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2295 | after successfully matching one A, we need to go back and try to | |
2296 | match another one; similarly for lookahead assertions: if the assertion | |
2297 | completes successfully, we backtrack to the state just before the assertion | |
2298 | and then carry on. In these cases, the pushed state is marked as | |
2299 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2300 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2301 | the nearest yes state, discarding any intermediate failure-only states. | |
2302 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2303 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2304 | it to free the inner regex. | |
2305 | ||
2306 | Note that failure backtracking rewinds the cursor position, while | |
2307 | success backtracking leaves it alone. | |
2308 | ||
2309 | A pattern is complete when the END op is executed, while a subpattern | |
2310 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2311 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2312 | behaviour. | |
2313 | ||
2314 | A common convention in this function is to use A and B to refer to the two | |
2315 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2316 | the subpattern to be matched possibly multiple times, while B is the entire | |
2317 | rest of the pattern. Variable and state names reflect this convention. | |
2318 | ||
2319 | The states in the main switch are the union of ops and failure/success of | |
2320 | substates associated with with that op. For example, IFMATCH is the op | |
2321 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2322 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2323 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2324 | just failed to match A. Resume states always come in pairs. The backtrack | |
2325 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2326 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2327 | on success or failure. | |
2328 | ||
2329 | The struct that holds a backtracking state is actually a big union, with | |
2330 | one variant for each major type of op. The variable st points to the | |
2331 | top-most backtrack struct. To make the code clearer, within each | |
2332 | block of code we #define ST to alias the relevant union. | |
2333 | ||
2334 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2335 | implementation: | |
2336 | ||
2337 | switch (state) { | |
2338 | .... | |
2339 | ||
2340 | #define ST st->u.ifmatch | |
2341 | ||
2342 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2343 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2344 | ... |
bf1f174e DM |
2345 | // push a yes backtrack state with a resume value of |
2346 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2347 | // first node of A: | |
2348 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2349 | // NOTREACHED | |
2350 | ||
2351 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2352 | next = B; | |
2353 | bar = ST.foo; // do something with the preserved value | |
2354 | break; | |
2355 | ||
2356 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2357 | ...; // do some housekeeping, then ... | |
2358 | sayNO; // propagate the failure | |
2359 | ||
2360 | #undef ST | |
95b24440 | 2361 | |
bf1f174e DM |
2362 | ... |
2363 | } | |
95b24440 | 2364 | |
bf1f174e DM |
2365 | For any old-timers reading this who are familiar with the old recursive |
2366 | approach, the code above is equivalent to: | |
95b24440 | 2367 | |
bf1f174e DM |
2368 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2369 | { | |
2370 | int foo = ... | |
95b24440 | 2371 | ... |
bf1f174e DM |
2372 | if (regmatch(A)) { |
2373 | next = B; | |
2374 | bar = foo; | |
2375 | break; | |
95b24440 | 2376 | } |
bf1f174e DM |
2377 | ...; // do some housekeeping, then ... |
2378 | sayNO; // propagate the failure | |
95b24440 | 2379 | } |
bf1f174e DM |
2380 | |
2381 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2382 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2383 | save, then do one of | |
2384 | ||
2385 | PUSH_STATE_GOTO(resume_state, node); | |
2386 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2387 | ||
2388 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2389 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2390 | On backtracking, the free slot is popped, and the saved state becomes the | |
2391 | new free state. An ST.foo field in this new top state can be temporarily | |
2392 | accessed to retrieve values, but once the main loop is re-entered, it | |
2393 | becomes available for reuse. | |
2394 | ||
2395 | Note that the depth of the backtrack stack constantly increases during the | |
2396 | left-to-right execution of the pattern, rather than going up and down with | |
2397 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2398 | end of the pattern, rather than at X in the following: | |
2399 | ||
2400 | /(((X)+)+)+....(Y)+....Z/ | |
2401 | ||
2402 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2403 | (?>A), which pop all the backtrack states associated with A before | |
2404 | continuing. | |
2405 | ||
2406 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2407 | PL_regmatch_state and st always point to the currently active state, | |
2408 | and PL_regmatch_slab points to the slab currently containing | |
2409 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2410 | allocated, and is never freed until interpreter destruction. When the slab | |
2411 | is full, a new one is allocated and chained to the end. At exit from | |
2412 | regmatch(), slabs allocated since entry are freed. | |
2413 | ||
2414 | */ | |
95b24440 | 2415 | |
40a82448 | 2416 | |
5bc10b2c | 2417 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2418 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2419 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2420 | PerlIO_printf(Perl_debug_log, \ | |
2421 | " %*s"pp" %s\n", \ | |
2422 | depth*2, "", \ | |
03363afd | 2423 | reg_name[st->resume_state] ); \ |
265c4333 | 2424 | }); |
5bc10b2c | 2425 | |
40a82448 | 2426 | |
3dab1dad | 2427 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2428 | |
3df15adc | 2429 | #ifdef DEBUGGING |
5bc10b2c | 2430 | |
ab3bbdeb YO |
2431 | STATIC void |
2432 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2433 | const char *start, const char *end, const char *blurb) | |
2434 | { | |
2435 | const bool utf8_pat= prog->reganch & ROPT_UTF8 ? 1 : 0; | |
2436 | if (!PL_colorset) | |
2437 | reginitcolors(); | |
2438 | { | |
2439 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2440 | prog->precomp, prog->prelen, 60); | |
2441 | ||
2442 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2443 | start, end - start, 60); | |
2444 | ||
2445 | PerlIO_printf(Perl_debug_log, | |
2446 | "%s%s REx%s %s against %s\n", | |
2447 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2448 | ||
2449 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2450 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2451 | utf8_pat ? "pattern" : "", | |
2452 | utf8_pat && do_utf8 ? " and " : "", | |
2453 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2454 | ); |
2455 | } | |
2456 | } | |
3df15adc YO |
2457 | |
2458 | STATIC void | |
786e8c11 YO |
2459 | S_dump_exec_pos(pTHX_ const char *locinput, |
2460 | const regnode *scan, | |
2461 | const char *loc_regeol, | |
2462 | const char *loc_bostr, | |
2463 | const char *loc_reg_starttry, | |
2464 | const bool do_utf8) | |
07be1b83 | 2465 | { |
786e8c11 | 2466 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2467 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2468 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2469 | /* The part of the string before starttry has one color |
2470 | (pref0_len chars), between starttry and current | |
2471 | position another one (pref_len - pref0_len chars), | |
2472 | after the current position the third one. | |
2473 | We assume that pref0_len <= pref_len, otherwise we | |
2474 | decrease pref0_len. */ | |
786e8c11 YO |
2475 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2476 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2477 | int pref0_len; |
2478 | ||
2479 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2480 | pref_len++; | |
786e8c11 YO |
2481 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2482 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2483 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2484 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2485 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2486 | l--; | |
2487 | if (pref0_len < 0) | |
2488 | pref0_len = 0; | |
2489 | if (pref0_len > pref_len) | |
2490 | pref0_len = pref_len; | |
2491 | { | |
3df15adc | 2492 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2493 | |
ab3bbdeb | 2494 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2495 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2496 | |
ab3bbdeb | 2497 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2498 | (locinput - pref_len + pref0_len), |
1de06328 | 2499 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2500 | |
ab3bbdeb | 2501 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2502 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2503 | |
1de06328 | 2504 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2505 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2506 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2507 | (IV)(locinput - loc_bostr), |
07be1b83 | 2508 | len0, s0, |
07be1b83 | 2509 | len1, s1, |
07be1b83 | 2510 | (docolor ? "" : "> <"), |
07be1b83 | 2511 | len2, s2, |
f9f4320a | 2512 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2513 | ""); |
2514 | } | |
2515 | } | |
3df15adc | 2516 | |
07be1b83 YO |
2517 | #endif |
2518 | ||
0a4db386 YO |
2519 | /* reg_check_named_buff_matched() |
2520 | * Checks to see if a named buffer has matched. The data array of | |
2521 | * buffer numbers corresponding to the buffer is expected to reside | |
2522 | * in the regexp->data->data array in the slot stored in the ARG() of | |
2523 | * node involved. Note that this routine doesn't actually care about the | |
2524 | * name, that information is not preserved from compilation to execution. | |
2525 | * Returns the index of the leftmost defined buffer with the given name | |
2526 | * or 0 if non of the buffers matched. | |
2527 | */ | |
2528 | STATIC I32 | |
2529 | S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) { | |
2530 | I32 n; | |
2531 | SV *sv_dat=(SV*)rex->data->data[ ARG( scan ) ]; | |
2532 | I32 *nums=(I32*)SvPVX(sv_dat); | |
2533 | for ( n=0; n<SvIVX(sv_dat); n++ ) { | |
2534 | if ((I32)*PL_reglastparen >= nums[n] && | |
2535 | PL_regendp[nums[n]] != -1) | |
2536 | { | |
2537 | return nums[n]; | |
2538 | } | |
2539 | } | |
2540 | return 0; | |
2541 | } | |
2542 | ||
d6a28714 | 2543 | STATIC I32 /* 0 failure, 1 success */ |
24b23f37 | 2544 | S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2545 | { |
54df2634 NC |
2546 | #if PERL_VERSION < 9 |
2547 | dMY_CXT; | |
2548 | #endif | |
27da23d5 | 2549 | dVAR; |
95b24440 | 2550 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2551 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2552 | |
3b0527fe DM |
2553 | regexp *rex = reginfo->prog; |
2554 | ||
5d9a96ca DM |
2555 | regmatch_slab *orig_slab; |
2556 | regmatch_state *orig_state; | |
a3621e74 | 2557 | |
5d9a96ca DM |
2558 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2559 | register regmatch_state *st; | |
95b24440 | 2560 | |
5d9a96ca DM |
2561 | /* cache heavy used fields of st in registers */ |
2562 | register regnode *scan; | |
2563 | register regnode *next; | |
24d3c4a9 DM |
2564 | register I32 n = 0; /* general value; init to avoid compiler warning */ |
2565 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ | |
5d9a96ca | 2566 | register char *locinput = PL_reginput; |
5d9a96ca | 2567 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2568 | |
b69b0499 | 2569 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2570 | int depth = 0; /* depth of backtrack stack */ |
1a147d38 | 2571 | int nochange_depth = 0; /* depth of GOSUB recursion with nochange*/ |
77cb431f DM |
2572 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2573 | subpattern */ | |
faec1544 | 2574 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2575 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2576 | U32 state_num; |
24b23f37 | 2577 | bool no_final = 0; |
10edeb5d | 2578 | |
24d3c4a9 DM |
2579 | /* these three flags are set by various ops to signal information to |
2580 | * the very next op. They have a useful lifetime of exactly one loop | |
2581 | * iteration, and are not preserved or restored by state pushes/pops | |
2582 | */ | |
2583 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2584 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2585 | int logical = 0; /* the following EVAL is: | |
2586 | 0: (?{...}) | |
2587 | 1: (?(?{...})X|Y) | |
2588 | 2: (??{...}) | |
2589 | or the following IFMATCH/UNLESSM is: | |
2590 | false: plain (?=foo) | |
2591 | true: used as a condition: (?(?=foo)) | |
2592 | */ | |
2593 | ||
95b24440 | 2594 | #ifdef DEBUGGING |
e68ec53f | 2595 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2596 | #endif |
2597 | ||
24b23f37 YO |
2598 | DEBUG_STACK_r( { |
2599 | PerlIO_printf(Perl_debug_log,"regmatch start\n"); | |
2600 | }); | |
5d9a96ca DM |
2601 | /* on first ever call to regmatch, allocate first slab */ |
2602 | if (!PL_regmatch_slab) { | |
2603 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2604 | PL_regmatch_slab->prev = NULL; | |
2605 | PL_regmatch_slab->next = NULL; | |
86545054 | 2606 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2607 | } |
2608 | ||
2609 | /* remember current high-water mark for exit */ | |
2610 | /* XXX this should be done with SAVE* instead */ | |
2611 | orig_slab = PL_regmatch_slab; | |
2612 | orig_state = PL_regmatch_state; | |
2613 | ||
2614 | /* grab next free state slot */ | |
2615 | st = ++PL_regmatch_state; | |
86545054 | 2616 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2617 | st = PL_regmatch_state = S_push_slab(aTHX); |
2618 | ||
d6a28714 JH |
2619 | /* Note that nextchr is a byte even in UTF */ |
2620 | nextchr = UCHARAT(locinput); | |
2621 | scan = prog; | |
2622 | while (scan != NULL) { | |
8ba1375e | 2623 | |
a3621e74 | 2624 | DEBUG_EXECUTE_r( { |
6136c704 | 2625 | SV * const prop = sv_newmortal(); |
1de06328 | 2626 | regnode *rnext=regnext(scan); |
786e8c11 | 2627 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2628 | regprop(rex, prop, scan); |
07be1b83 YO |
2629 | |
2630 | PerlIO_printf(Perl_debug_log, | |
2631 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
5bc10b2c | 2632 | (IV)(scan - rex->program), depth*2, "", |
07be1b83 | 2633 | SvPVX_const(prop), |
1de06328 YO |
2634 | (PL_regkind[OP(scan)] == END || !rnext) ? |
2635 | 0 : (IV)(rnext - rex->program)); | |
2a782b5b | 2636 | }); |
d6a28714 JH |
2637 | |
2638 | next = scan + NEXT_OFF(scan); | |
2639 | if (next == scan) | |
2640 | next = NULL; | |
40a82448 | 2641 | state_num = OP(scan); |
d6a28714 | 2642 | |
40a82448 DM |
2643 | reenter_switch: |
2644 | switch (state_num) { | |
d6a28714 | 2645 | case BOL: |
7fba1cd6 | 2646 | if (locinput == PL_bostr) |
d6a28714 | 2647 | { |
3b0527fe | 2648 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2649 | break; |
2650 | } | |
d6a28714 JH |
2651 | sayNO; |
2652 | case MBOL: | |
12d33761 HS |
2653 | if (locinput == PL_bostr || |
2654 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2655 | { |
b8c5462f JH |
2656 | break; |
2657 | } | |
d6a28714 JH |
2658 | sayNO; |
2659 | case SBOL: | |
c2a73568 | 2660 | if (locinput == PL_bostr) |
b8c5462f | 2661 | break; |
d6a28714 JH |
2662 | sayNO; |
2663 | case GPOS: | |
3b0527fe | 2664 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2665 | break; |
2666 | sayNO; | |
2667 | case EOL: | |
d6a28714 JH |
2668 | goto seol; |
2669 | case MEOL: | |
d6a28714 | 2670 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2671 | sayNO; |
b8c5462f | 2672 | break; |
d6a28714 JH |
2673 | case SEOL: |
2674 | seol: | |
2675 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2676 | sayNO; |
d6a28714 | 2677 | if (PL_regeol - locinput > 1) |
b8c5462f | 2678 | sayNO; |
b8c5462f | 2679 | break; |
d6a28714 JH |
2680 | case EOS: |
2681 | if (PL_regeol != locinput) | |
b8c5462f | 2682 | sayNO; |
d6a28714 | 2683 | break; |
ffc61ed2 | 2684 | case SANY: |
d6a28714 | 2685 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2686 | sayNO; |
f33976b4 DB |
2687 | if (do_utf8) { |
2688 | locinput += PL_utf8skip[nextchr]; | |
2689 | if (locinput > PL_regeol) | |
2690 | sayNO; | |
2691 | nextchr = UCHARAT(locinput); | |
2692 | } | |
2693 | else | |
2694 | nextchr = UCHARAT(++locinput); | |
2695 | break; | |
2696 | case CANY: | |
2697 | if (!nextchr && locinput >= PL_regeol) | |
2698 | sayNO; | |
b8c5462f | 2699 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2700 | break; |
ffc61ed2 | 2701 | case REG_ANY: |
1aa99e6b IH |
2702 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2703 | sayNO; | |
2704 | if (do_utf8) { | |
b8c5462f | 2705 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2706 | if (locinput > PL_regeol) |
2707 | sayNO; | |
a0ed51b3 | 2708 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2709 | } |
1aa99e6b IH |
2710 | else |
2711 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2712 | break; |
166ba7cd DM |
2713 | |
2714 | #undef ST | |
2715 | #define ST st->u.trie | |
786e8c11 YO |
2716 | case TRIEC: |
2717 | /* In this case the charclass data is available inline so | |
2718 | we can fail fast without a lot of extra overhead. | |
2719 | */ | |
2720 | if (scan->flags == EXACT || !do_utf8) { | |
2721 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2722 | DEBUG_EXECUTE_r( | |
2723 | PerlIO_printf(Perl_debug_log, | |
2724 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2725 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2726 | ); |
2727 | sayNO_SILENT; | |
2728 | /* NOTREACHED */ | |
2729 | } | |
2730 | } | |
2731 | /* FALL THROUGH */ | |
5b47454d | 2732 | case TRIE: |
3dab1dad | 2733 | { |
07be1b83 | 2734 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2735 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2736 | trie_type = do_utf8 ? | |
2737 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2738 | : trie_plain; | |
2739 | ||
2740 | /* what trie are we using right now */ | |
be8e71aa | 2741 | reg_trie_data * const trie |
3dab1dad YO |
2742 | = (reg_trie_data*)rex->data->data[ ARG( scan ) ]; |
2743 | U32 state = trie->startstate; | |
166ba7cd | 2744 | |
3dab1dad YO |
2745 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2746 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2747 | ) { | |
2748 | if (trie->states[ state ].wordnum) { | |
2749 | DEBUG_EXECUTE_r( | |
2750 | PerlIO_printf(Perl_debug_log, | |
2751 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2752 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2753 | ); |
2754 | break; | |
2755 | } else { | |
2756 | DEBUG_EXECUTE_r( | |
2757 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2758 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2759 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2760 | ); |
2761 | sayNO_SILENT; | |
2762 | } | |
2763 | } | |
166ba7cd | 2764 | |
786e8c11 YO |
2765 | { |
2766 | U8 *uc = ( U8* )locinput; | |
2767 | ||
2768 | STRLEN len = 0; | |
2769 | STRLEN foldlen = 0; | |
2770 | U8 *uscan = (U8*)NULL; | |
2771 | STRLEN bufflen=0; | |
2772 | SV *sv_accept_buff = NULL; | |
2773 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2774 | ||
2775 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2776 | ST.B = next; | |
2777 | ST.jump = trie->jump; | |
786e8c11 | 2778 | ST.me = scan; |
786e8c11 | 2779 | |
07be1b83 YO |
2780 | /* |
2781 | traverse the TRIE keeping track of all accepting states | |
2782 | we transition through until we get to a failing node. | |
2783 | */ | |
2784 | ||
a3621e74 | 2785 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 2786 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 2787 | UV uvc = 0; |
786e8c11 YO |
2788 | U16 charid; |
2789 | /* We use charid to hold the wordnum as we don't use it | |
2790 | for charid until after we have done the wordnum logic. | |
2791 | We define an alias just so that the wordnum logic reads | |
2792 | more naturally. */ | |
2793 | ||
2794 | #define got_wordnum charid | |
2795 | got_wordnum = trie->states[ state ].wordnum; | |
2796 | ||
2797 | if ( got_wordnum ) { | |
2798 | if ( ! ST.accepted ) { | |
5b47454d DM |
2799 | ENTER; |
2800 | SAVETMPS; | |
2801 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2802 | sv_accept_buff=newSV(bufflen * | |
2803 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2804 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2805 | SvPOK_on(sv_accept_buff); |
2806 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2807 | SAVETMPS; |
2808 | ST.accept_buff = | |
5b47454d DM |
2809 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2810 | } | |
786e8c11 | 2811 | do { |
166ba7cd | 2812 | if (ST.accepted >= bufflen) { |
5b47454d | 2813 | bufflen *= 2; |
166ba7cd | 2814 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2815 | SvGROW(sv_accept_buff, |
2816 | bufflen * sizeof(reg_trie_accepted)); | |
2817 | } | |
2818 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2819 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2820 | |
786e8c11 YO |
2821 | |
2822 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2823 | ST.accept_buff[ST.accepted].endpos = uc; | |
2824 | ++ST.accepted; | |
2825 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2826 | } | |
2827 | #undef got_wordnum | |
a3621e74 | 2828 | |
07be1b83 | 2829 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2830 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2831 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2832 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2833 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2834 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2835 | }); |
a3621e74 YO |
2836 | |
2837 | if ( base ) { | |
4cadc6a9 YO |
2838 | REXEC_TRIE_READ_CHAR(trie_type, trie, uc, uscan, len, |
2839 | uvc, charid, foldlen, foldbuf, uniflags); | |
a3621e74 | 2840 | |
5b47454d DM |
2841 | if (charid && |
2842 | (base + charid > trie->uniquecharcount ) | |
2843 | && (base + charid - 1 - trie->uniquecharcount | |
2844 | < trie->lasttrans) | |
2845 | && trie->trans[base + charid - 1 - | |
2846 | trie->uniquecharcount].check == state) | |
2847 | { | |
2848 | state = trie->trans[base + charid - 1 - | |
2849 | trie->uniquecharcount ].next; | |
2850 | } | |
2851 | else { | |
2852 | state = 0; | |
2853 | } | |
2854 | uc += len; | |
2855 | ||
2856 | } | |
2857 | else { | |
a3621e74 YO |
2858 | state = 0; |
2859 | } | |
2860 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 2861 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2862 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 2863 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
2864 | ); |
2865 | } | |
166ba7cd | 2866 | if (!ST.accepted ) |
a3621e74 | 2867 | sayNO; |
a3621e74 | 2868 | |
166ba7cd DM |
2869 | DEBUG_EXECUTE_r( |
2870 | PerlIO_printf( Perl_debug_log, | |
2871 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 2872 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
2873 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
2874 | ); | |
786e8c11 | 2875 | }} |
166ba7cd DM |
2876 | |
2877 | /* FALL THROUGH */ | |
2878 | ||
2879 | case TRIE_next_fail: /* we failed - try next alterative */ | |
2880 | ||
2881 | if ( ST.accepted == 1 ) { | |
2882 | /* only one choice left - just continue */ | |
2883 | DEBUG_EXECUTE_r({ | |
2884 | reg_trie_data * const trie | |
2885 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2886 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2887 | ? av_fetch( trie->words, ST.accept_buff[ 0 ].wordnum-1, 0 ) | |
2888 | : NULL; | |
2889 | PerlIO_printf( Perl_debug_log, | |
2890 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 2891 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd DM |
2892 | ST.accept_buff[ 0 ].wordnum, |
2893 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", | |
2894 | PL_colors[5] ); | |
2895 | }); | |
2896 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
2897 | /* in this case we free tmps/leave before we call regmatch | |
2898 | as we wont be using accept_buff again. */ | |
2899 | FREETMPS; | |
2900 | LEAVE; | |
2901 | locinput = PL_reginput; | |
2902 | nextchr = UCHARAT(locinput); | |
786e8c11 | 2903 | |
7f69552c | 2904 | if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum]) |
786e8c11 YO |
2905 | scan = ST.B; |
2906 | else | |
7f69552c | 2907 | scan = ST.me + ST.jump[ST.accept_buff[0].wordnum]; |
786e8c11 | 2908 | |
166ba7cd DM |
2909 | continue; /* execute rest of RE */ |
2910 | } | |
2911 | ||
2912 | if (!ST.accepted-- ) { | |
2913 | FREETMPS; | |
2914 | LEAVE; | |
2915 | sayNO; | |
2916 | } | |
2917 | ||
a3621e74 | 2918 | /* |
166ba7cd DM |
2919 | There are at least two accepting states left. Presumably |
2920 | the number of accepting states is going to be low, | |
2921 | typically two. So we simply scan through to find the one | |
2922 | with lowest wordnum. Once we find it, we swap the last | |
2923 | state into its place and decrement the size. We then try to | |
2924 | match the rest of the pattern at the point where the word | |
2925 | ends. If we succeed, control just continues along the | |
2926 | regex; if we fail we return here to try the next accepting | |
2927 | state | |
2928 | */ | |
a3621e74 | 2929 | |
166ba7cd DM |
2930 | { |
2931 | U32 best = 0; | |
2932 | U32 cur; | |
2933 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
2934 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 2935 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 2936 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 2937 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
2938 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
2939 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
2940 | ); | |
2941 | ||
2942 | if (ST.accept_buff[cur].wordnum < | |
2943 | ST.accept_buff[best].wordnum) | |
2944 | best = cur; | |
a3621e74 | 2945 | } |
166ba7cd DM |
2946 | |
2947 | DEBUG_EXECUTE_r({ | |
2948 | reg_trie_data * const trie | |
2949 | = (reg_trie_data*)rex->data->data[ ARG(ST.me) ]; | |
2950 | SV ** const tmp = RX_DEBUG(reginfo->prog) | |
2951 | ? av_fetch( trie->words, ST.accept_buff[ best ].wordnum - 1, 0 ) | |
2952 | : NULL; | |
7f69552c | 2953 | regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ? |
265c4333 | 2954 | ST.B : |
7f69552c | 2955 | ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
265c4333 YO |
2956 | PerlIO_printf( Perl_debug_log, |
2957 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 2958 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 2959 | ST.accept_buff[best].wordnum, |
265c4333 YO |
2960 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", |
2961 | REG_NODE_NUM(nextop), | |
166ba7cd DM |
2962 | PL_colors[5] ); |
2963 | }); | |
2964 | ||
2965 | if ( best<ST.accepted ) { | |
2966 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
2967 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
2968 | ST.accept_buff[ ST.accepted ] = tmp; | |
2969 | best = ST.accepted; | |
a3621e74 | 2970 | } |
166ba7cd | 2971 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
7f69552c | 2972 | if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) { |
786e8c11 YO |
2973 | PUSH_STATE_GOTO(TRIE_next, ST.B); |
2974 | /* NOTREACHED */ | |
2975 | } else { | |
7f69552c | 2976 | PUSH_STATE_GOTO(TRIE_next, ST.me + ST.jump[ST.accept_buff[best].wordnum]); |
786e8c11 YO |
2977 | /* NOTREACHED */ |
2978 | } | |
2979 | /* NOTREACHED */ | |
166ba7cd | 2980 | } |
166ba7cd DM |
2981 | /* NOTREACHED */ |
2982 | ||
2983 | #undef ST | |
2984 | ||
95b24440 DM |
2985 | case EXACT: { |
2986 | char *s = STRING(scan); | |
24d3c4a9 | 2987 | ln = STR_LEN(scan); |
eb160463 | 2988 | if (do_utf8 != UTF) { |
bc517b45 | 2989 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 2990 | char *l = locinput; |
24d3c4a9 | 2991 | const char * const e = s + ln; |
a72c7584 | 2992 | |
5ff6fc6d JH |
2993 | if (do_utf8) { |
2994 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 2995 | while (s < e) { |
a3b680e6 | 2996 | STRLEN ulen; |
1aa99e6b | 2997 | if (l >= PL_regeol) |
5ff6fc6d JH |
2998 | sayNO; |
2999 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 3000 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 3001 | uniflags)) |
5ff6fc6d | 3002 | sayNO; |
bc517b45 | 3003 | l += ulen; |
5ff6fc6d | 3004 | s ++; |
1aa99e6b | 3005 | } |
5ff6fc6d JH |
3006 | } |
3007 | else { | |
3008 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 3009 | while (s < e) { |
a3b680e6 | 3010 | STRLEN ulen; |
1aa99e6b IH |
3011 | if (l >= PL_regeol) |
3012 | sayNO; | |
5ff6fc6d | 3013 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 3014 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 3015 | uniflags)) |
1aa99e6b | 3016 | sayNO; |
bc517b45 | 3017 | s += ulen; |
a72c7584 | 3018 | l ++; |
1aa99e6b | 3019 | } |
5ff6fc6d | 3020 | } |
1aa99e6b IH |
3021 | locinput = l; |
3022 | nextchr = UCHARAT(locinput); | |
3023 | break; | |
3024 | } | |
bc517b45 | 3025 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
3026 | /* Inline the first character, for speed. */ |
3027 | if (UCHARAT(s) != nextchr) | |
3028 | sayNO; | |
24d3c4a9 | 3029 | if (PL_regeol - locinput < ln) |
d6a28714 | 3030 | sayNO; |
24d3c4a9 | 3031 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3032 | sayNO; |
24d3c4a9 | 3033 | locinput += ln; |
d6a28714 JH |
3034 | nextchr = UCHARAT(locinput); |
3035 | break; | |
95b24440 | 3036 | } |
d6a28714 | 3037 | case EXACTFL: |
b8c5462f JH |
3038 | PL_reg_flags |= RF_tainted; |
3039 | /* FALL THROUGH */ | |
95b24440 | 3040 | case EXACTF: { |
be8e71aa | 3041 | char * const s = STRING(scan); |
24d3c4a9 | 3042 | ln = STR_LEN(scan); |
d6a28714 | 3043 | |
d07ddd77 JH |
3044 | if (do_utf8 || UTF) { |
3045 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3046 | const char * const l = locinput; |
d07ddd77 | 3047 | char *e = PL_regeol; |
bc517b45 | 3048 | |
24d3c4a9 | 3049 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3050 | l, &e, 0, do_utf8)) { |
5486206c JH |
3051 | /* One more case for the sharp s: |
3052 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3053 | * the 0xC3 0x9F are the UTF-8 | |
3054 | * byte sequence for the U+00DF. */ | |
3055 | if (!(do_utf8 && | |
3056 | toLOWER(s[0]) == 's' && | |
24d3c4a9 | 3057 | ln >= 2 && |
5486206c JH |
3058 | toLOWER(s[1]) == 's' && |
3059 | (U8)l[0] == 0xC3 && | |
3060 | e - l >= 2 && | |
3061 | (U8)l[1] == 0x9F)) | |
3062 | sayNO; | |
3063 | } | |
d07ddd77 JH |
3064 | locinput = e; |
3065 | nextchr = UCHARAT(locinput); | |
3066 | break; | |
a0ed51b3 | 3067 | } |
d6a28714 | 3068 | |
bc517b45 JH |
3069 | /* Neither the target and the pattern are utf8. */ |
3070 | ||
d6a28714 JH |
3071 | /* Inline the first character, for speed. */ |
3072 | if (UCHARAT(s) != nextchr && | |
3073 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3074 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3075 | sayNO; |
24d3c4a9 | 3076 | if (PL_regeol - locinput < ln) |
b8c5462f | 3077 | sayNO; |
24d3c4a9 DM |
3078 | if (ln > 1 && (OP(scan) == EXACTF |
3079 | ? ibcmp(s, locinput, ln) | |
3080 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3081 | sayNO; |
24d3c4a9 | 3082 | locinput += ln; |
d6a28714 | 3083 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3084 | break; |
95b24440 | 3085 | } |
d6a28714 | 3086 | case ANYOF: |
ffc61ed2 | 3087 | if (do_utf8) { |
9e55ce06 JH |
3088 | STRLEN inclasslen = PL_regeol - locinput; |
3089 | ||
32fc9b6a | 3090 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3091 | goto anyof_fail; |
ffc61ed2 JH |
3092 | if (locinput >= PL_regeol) |
3093 | sayNO; | |
0f0076b4 | 3094 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3095 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3096 | break; |
ffc61ed2 JH |
3097 | } |
3098 | else { | |
3099 | if (nextchr < 0) | |
3100 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3101 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3102 | goto anyof_fail; |
ffc61ed2 JH |
3103 | if (!nextchr && locinput >= PL_regeol) |
3104 | sayNO; | |
3105 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3106 | break; |
3107 | } | |
262b90c4 | 3108 | anyof_fail: |
e0f9d4a8 JH |
3109 | /* If we might have the case of the German sharp s |
3110 | * in a casefolding Unicode character class. */ | |
3111 | ||
ebc501f0 JH |
3112 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3113 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3114 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3115 | } |
e0f9d4a8 JH |
3116 | else |
3117 | sayNO; | |
b8c5462f | 3118 | break; |
d6a28714 | 3119 | case ALNUML: |
b8c5462f JH |
3120 | PL_reg_flags |= RF_tainted; |
3121 | /* FALL THROUGH */ | |
d6a28714 | 3122 | case ALNUM: |
b8c5462f | 3123 | if (!nextchr) |
4633a7c4 | 3124 | sayNO; |
ffc61ed2 | 3125 | if (do_utf8) { |
1a4fad37 | 3126 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3127 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3128 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3129 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3130 | { |
3131 | sayNO; | |
a0ed51b3 | 3132 | } |
b8c5462f | 3133 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3134 | nextchr = UCHARAT(locinput); |
3135 | break; | |
3136 | } | |
ffc61ed2 | 3137 | if (!(OP(scan) == ALNUM |
d6a28714 | 3138 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3139 | sayNO; |
b8c5462f | 3140 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3141 | break; |
d6a28714 | 3142 | case NALNUML: |
b8c5462f JH |
3143 | PL_reg_flags |= RF_tainted; |
3144 | /* FALL THROUGH */ | |
d6a28714 JH |
3145 | case NALNUM: |
3146 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3147 | sayNO; |
ffc61ed2 | 3148 | if (do_utf8) { |
1a4fad37 | 3149 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3150 | if (OP(scan) == NALNUM |
bb7a0f54 | 3151 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3152 | : isALNUM_LC_utf8((U8*)locinput)) |
3153 | { | |
b8c5462f | 3154 | sayNO; |
d6a28714 | 3155 | } |
b8c5462f JH |
3156 | locinput += PL_utf8skip[nextchr]; |
3157 | nextchr = UCHARAT(locinput); | |
3158 | break; | |
3159 | } | |
ffc61ed2 | 3160 | if (OP(scan) == NALNUM |
d6a28714 | 3161 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3162 | sayNO; |
76e3520e | 3163 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3164 | break; |
d6a28714 JH |
3165 | case BOUNDL: |
3166 | case NBOUNDL: | |
3280af22 | 3167 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3168 | /* FALL THROUGH */ |
d6a28714 JH |
3169 | case BOUND: |
3170 | case NBOUND: | |
3171 | /* was last char in word? */ | |
ffc61ed2 | 3172 | if (do_utf8) { |
12d33761 | 3173 | if (locinput == PL_bostr) |
24d3c4a9 | 3174 | ln = '\n'; |
ffc61ed2 | 3175 | else { |
a3b680e6 | 3176 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3177 | |
24d3c4a9 | 3178 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3179 | } |
3180 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3181 | ln = isALNUM_uni(ln); |
1a4fad37 | 3182 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3183 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3184 | } |
3185 | else { | |
24d3c4a9 | 3186 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3187 | n = isALNUM_LC_utf8((U8*)locinput); |
3188 | } | |
a0ed51b3 | 3189 | } |
d6a28714 | 3190 | else { |
24d3c4a9 | 3191 | ln = (locinput != PL_bostr) ? |
12d33761 | 3192 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3193 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3194 | ln = isALNUM(ln); |
ffc61ed2 JH |
3195 | n = isALNUM(nextchr); |
3196 | } | |
3197 | else { | |
24d3c4a9 | 3198 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3199 | n = isALNUM_LC(nextchr); |
3200 | } | |
d6a28714 | 3201 | } |
24d3c4a9 | 3202 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3203 | OP(scan) == BOUNDL)) |
3204 | sayNO; | |
a0ed51b3 | 3205 | break; |
d6a28714 | 3206 | case SPACEL: |
3280af22 | 3207 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3208 | /* FALL THROUGH */ |
d6a28714 | 3209 | case SPACE: |
9442cb0e | 3210 | if (!nextchr) |
4633a7c4 | 3211 | sayNO; |
1aa99e6b | 3212 | if (do_utf8) { |
fd400ab9 | 3213 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3214 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3215 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3216 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3217 | : isSPACE_LC_utf8((U8*)locinput))) |
3218 | { | |
3219 | sayNO; | |
3220 | } | |
3221 | locinput += PL_utf8skip[nextchr]; | |
3222 | nextchr = UCHARAT(locinput); | |
3223 | break; | |
d6a28714 | 3224 | } |
ffc61ed2 JH |
3225 | if (!(OP(scan) == SPACE |
3226 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3227 | sayNO; | |
3228 | nextchr = UCHARAT(++locinput); | |
3229 | } | |
3230 | else { | |
3231 | if (!(OP(scan) == SPACE | |
3232 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3233 | sayNO; | |
3234 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3235 | } |
a0ed51b3 | 3236 | break; |
d6a28714 | 3237 | case NSPACEL: |
3280af22 | 3238 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3239 | /* FALL THROUGH */ |
d6a28714 | 3240 | case NSPACE: |
9442cb0e | 3241 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3242 | sayNO; |
1aa99e6b | 3243 | if (do_utf8) { |
1a4fad37 | 3244 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3245 | if (OP(scan) == NSPACE |
bb7a0f54 | 3246 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3247 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3248 | { |
3249 | sayNO; | |
3250 | } | |
3251 | locinput += PL_utf8skip[nextchr]; | |
3252 | nextchr = UCHARAT(locinput); | |
3253 | break; | |
a0ed51b3 | 3254 | } |
ffc61ed2 | 3255 | if (OP(scan) == NSPACE |
d6a28714 | 3256 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3257 | sayNO; |
76e3520e | 3258 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3259 | break; |
d6a28714 | 3260 | case DIGITL: |
a0ed51b3 LW |
3261 | PL_reg_flags |= RF_tainted; |
3262 | /* FALL THROUGH */ | |
d6a28714 | 3263 | case DIGIT: |
9442cb0e | 3264 | if (!nextchr) |
a0ed51b3 | 3265 | sayNO; |
1aa99e6b | 3266 | if (do_utf8) { |
1a4fad37 | 3267 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3268 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3269 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3270 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3271 | { |
a0ed51b3 | 3272 | sayNO; |
dfe13c55 | 3273 | } |
6f06b55f | 3274 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3275 | nextchr = UCHARAT(locinput); |
3276 | break; | |
3277 | } | |
ffc61ed2 | 3278 | if (!(OP(scan) == DIGIT |
9442cb0e | 3279 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3280 | sayNO; |
3281 | nextchr = UCHARAT(++locinput); | |
3282 | break; | |
d6a28714 | 3283 | case NDIGITL: |
b8c5462f JH |
3284 | PL_reg_flags |= RF_tainted; |
3285 | /* FALL THROUGH */ | |
d6a28714 | 3286 | case NDIGIT: |
9442cb0e | 3287 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3288 | sayNO; |
1aa99e6b | 3289 | if (do_utf8) { |
1a4fad37 | 3290 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3291 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3292 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3293 | : isDIGIT_LC_utf8((U8*)locinput)) |
3294 | { | |
a0ed51b3 | 3295 | sayNO; |
9442cb0e | 3296 | } |
6f06b55f | 3297 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3298 | nextchr = UCHARAT(locinput); |
3299 | break; | |
3300 | } | |
ffc61ed2 | 3301 | if (OP(scan) == NDIGIT |
9442cb0e | 3302 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3303 | sayNO; |
3304 | nextchr = UCHARAT(++locinput); | |
3305 | break; | |
3306 | case CLUMP: | |
b7c83a7e | 3307 | if (locinput >= PL_regeol) |
a0ed51b3 | 3308 | sayNO; |
b7c83a7e | 3309 | if (do_utf8) { |
1a4fad37 | 3310 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3311 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3312 | sayNO; | |
3313 | locinput += PL_utf8skip[nextchr]; | |
3314 | while (locinput < PL_regeol && | |
3315 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3316 | locinput += UTF8SKIP(locinput); | |
3317 | if (locinput > PL_regeol) | |
3318 | sayNO; | |
eb08e2da JH |
3319 | } |
3320 | else | |
3321 | locinput++; | |
a0ed51b3 LW |
3322 | nextchr = UCHARAT(locinput); |
3323 | break; | |
81714fb9 YO |
3324 | |
3325 | case NREFFL: | |
3326 | { | |
3327 | char *s; | |
ff1157ca | 3328 | char type; |
81714fb9 YO |
3329 | PL_reg_flags |= RF_tainted; |
3330 | /* FALL THROUGH */ | |
3331 | case NREF: | |
3332 | case NREFF: | |
ff1157ca | 3333 | type = OP(scan); |
0a4db386 YO |
3334 | n = reg_check_named_buff_matched(rex,scan); |
3335 | ||
3336 | if ( n ) { | |
3337 | type = REF + ( type - NREF ); | |
3338 | goto do_ref; | |
3339 | } else { | |
81714fb9 | 3340 | sayNO; |
0a4db386 YO |
3341 | } |
3342 | /* unreached */ | |
c8756f30 | 3343 | case REFFL: |
3280af22 | 3344 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3345 | /* FALL THROUGH */ |
c277df42 | 3346 | case REF: |
81714fb9 | 3347 | case REFF: |
c277df42 | 3348 | n = ARG(scan); /* which paren pair */ |
81714fb9 YO |
3349 | type = OP(scan); |
3350 | do_ref: | |
24d3c4a9 | 3351 | ln = PL_regstartp[n]; |
2c2d71f5 | 3352 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3353 | if ((I32)*PL_reglastparen < n || ln == -1) |
af3f8c16 | 3354 | sayNO; /* Do not match unless seen CLOSEn. */ |
24d3c4a9 | 3355 | if (ln == PL_regendp[n]) |
a0d0e21e | 3356 | break; |
a0ed51b3 | 3357 | |
24d3c4a9 | 3358 | s = PL_bostr + ln; |
81714fb9 | 3359 | if (do_utf8 && type != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3360 | char *l = locinput; |
a3b680e6 | 3361 | const char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 LW |
3362 | /* |
3363 | * Note that we can't do the "other character" lookup trick as | |
3364 | * in the 8-bit case (no pun intended) because in Unicode we | |
3365 | * have to map both upper and title case to lower case. | |
3366 | */ | |
81714fb9 | 3367 | if (type == REFF) { |
a0ed51b3 | 3368 | while (s < e) { |
a3b680e6 AL |
3369 | STRLEN ulen1, ulen2; |
3370 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3371 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3372 | ||
a0ed51b3 LW |
3373 | if (l >= PL_regeol) |
3374 | sayNO; | |
a2a2844f JH |
3375 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3376 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3377 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3378 | sayNO; |
a2a2844f JH |
3379 | s += ulen1; |
3380 | l += ulen2; | |
a0ed51b3 LW |
3381 | } |
3382 | } | |
3383 | locinput = l; | |
3384 | nextchr = UCHARAT(locinput); | |
3385 | break; | |
3386 | } | |
3387 | ||
a0d0e21e | 3388 | /* Inline the first character, for speed. */ |
76e3520e | 3389 | if (UCHARAT(s) != nextchr && |
81714fb9 YO |
3390 | (type == REF || |
3391 | (UCHARAT(s) != (type == REFF | |
3392 | ? PL_fold : PL_fold_locale)[nextchr]))) | |
4633a7c4 | 3393 | sayNO; |
24d3c4a9 DM |
3394 | ln = PL_regendp[n] - ln; |
3395 | if (locinput + ln > PL_regeol) | |
4633a7c4 | 3396 | sayNO; |
81714fb9 | 3397 | if (ln > 1 && (type == REF |
24d3c4a9 | 3398 | ? memNE(s, locinput, ln) |
81714fb9 | 3399 | : (type == REFF |
24d3c4a9 DM |
3400 | ? ibcmp(s, locinput, ln) |
3401 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 3402 | sayNO; |
24d3c4a9 | 3403 | locinput += ln; |
76e3520e | 3404 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3405 | break; |
81714fb9 | 3406 | } |
a0d0e21e | 3407 | case NOTHING: |
c277df42 | 3408 | case TAIL: |
a0d0e21e LW |
3409 | break; |
3410 | case BACK: | |
3411 | break; | |
40a82448 DM |
3412 | |
3413 | #undef ST | |
3414 | #define ST st->u.eval | |
c277df42 | 3415 | { |
c277df42 | 3416 | SV *ret; |
6bda09f9 | 3417 | regexp *re; |
1a147d38 YO |
3418 | regnode *startpoint; |
3419 | ||
3420 | case GOSTART: | |
3421 | case GOSUB: /* /(...(?1))/ */ | |
6bda09f9 | 3422 | if (cur_eval && cur_eval->locinput==locinput) { |
24b23f37 | 3423 | if (cur_eval->u.eval.close_paren == (U32)ARG(scan)) |
1a147d38 | 3424 | Perl_croak(aTHX_ "Infinite recursion in regex"); |
6bda09f9 | 3425 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) |
1a147d38 YO |
3426 | Perl_croak(aTHX_ |
3427 | "Pattern subroutine nesting without pos change" | |
3428 | " exceeded limit in regex"); | |
6bda09f9 YO |
3429 | } else { |
3430 | nochange_depth = 0; | |
1a147d38 | 3431 | } |
6bda09f9 YO |
3432 | re = rex; |
3433 | (void)ReREFCNT_inc(rex); | |
1a147d38 | 3434 | if (OP(scan)==GOSUB) { |
6bda09f9 YO |
3435 | startpoint = scan + ARG2L(scan); |
3436 | ST.close_paren = ARG(scan); | |
3437 | } else { | |
3438 | startpoint = re->program+1; | |
3439 | ST.close_paren = 0; | |
3440 | } | |
3441 | goto eval_recurse_doit; | |
3442 | /* NOTREACHED */ | |
3443 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
3444 | if (cur_eval && cur_eval->locinput==locinput) { | |
3445 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) | |
1a147d38 | 3446 | Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex"); |
6bda09f9 YO |
3447 | } else { |
3448 | nochange_depth = 0; | |
3449 | } | |
8e5e9ebe | 3450 | { |
4aabdb9b DM |
3451 | /* execute the code in the {...} */ |
3452 | dSP; | |
6136c704 | 3453 | SV ** const before = SP; |
4aabdb9b DM |
3454 | OP_4tree * const oop = PL_op; |
3455 | COP * const ocurcop = PL_curcop; | |
3456 | PAD *old_comppad; | |
4aabdb9b DM |
3457 | |
3458 | n = ARG(scan); | |
32fc9b6a | 3459 | PL_op = (OP_4tree*)rex->data->data[n]; |
24b23f37 YO |
3460 | DEBUG_STATE_r( PerlIO_printf(Perl_debug_log, |
3461 | " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); | |
32fc9b6a | 3462 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]); |
4aabdb9b DM |
3463 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
3464 | ||
8e5e9ebe RGS |
3465 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3466 | SPAGAIN; | |
3467 | if (SP == before) | |
075aa684 | 3468 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3469 | else { |
3470 | ret = POPs; | |
3471 | PUTBACK; | |
3472 | } | |
4aabdb9b DM |
3473 | |
3474 | PL_op = oop; | |
3475 | PAD_RESTORE_LOCAL(old_comppad); | |
3476 | PL_curcop = ocurcop; | |
24d3c4a9 | 3477 | if (!logical) { |
4aabdb9b DM |
3478 | /* /(?{...})/ */ |
3479 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3480 | break; |
3481 | } | |
8e5e9ebe | 3482 | } |
24d3c4a9 DM |
3483 | if (logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3484 | logical = 0; | |
4aabdb9b | 3485 | { |
4f639d21 DM |
3486 | /* extract RE object from returned value; compiling if |
3487 | * necessary */ | |
3488 | ||
6136c704 | 3489 | MAGIC *mg = NULL; |
be8e71aa | 3490 | const SV *sv; |
faf82a0b AE |
3491 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3492 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3493 | else if (SvSMAGICAL(ret)) { | |
3494 | if (SvGMAGICAL(ret)) | |
3495 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3496 | else | |
3497 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3498 | } |
faf82a0b | 3499 | |
0f5d15d6 IZ |
3500 | if (mg) { |
3501 | re = (regexp *)mg->mg_obj; | |
df0003d4 | 3502 | (void)ReREFCNT_inc(re); |
0f5d15d6 IZ |
3503 | } |
3504 | else { | |
3505 | STRLEN len; | |
6136c704 | 3506 | const char * const t = SvPV_const(ret, len); |
0f5d15d6 | 3507 | PMOP pm; |
a3b680e6 | 3508 | const I32 osize = PL_regsize; |
0f5d15d6 | 3509 | |
5fcd1c1b | 3510 | Zero(&pm, 1, PMOP); |
4aabdb9b | 3511 | if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; |
f9f4320a | 3512 | re = CALLREGCOMP((char*)t, (char*)t + len, &pm); |
9041c2e3 | 3513 | if (!(SvFLAGS(ret) |
faf82a0b AE |
3514 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3515 | | SVs_GMG))) | |
14befaf4 DM |
3516 | sv_magic(ret,(SV*)ReREFCNT_inc(re), |
3517 | PERL_MAGIC_qr,0,0); | |
0f5d15d6 | 3518 | PL_regsize = osize; |
0f5d15d6 | 3519 | } |
4aabdb9b | 3520 | } |
6bda09f9 YO |
3521 | DEBUG_EXECUTE_r( |
3522 | debug_start_match(re, do_utf8, locinput, PL_regeol, | |
3523 | "Matching embedded"); | |
3524 | ); | |
3525 | startpoint = re->program + 1; | |
1a147d38 | 3526 | ST.close_paren = 0; /* only used for GOSUB */ |
6bda09f9 YO |
3527 | /* borrowed from regtry */ |
3528 | if (PL_reg_start_tmpl <= re->nparens) { | |
3529 | PL_reg_start_tmpl = re->nparens*3/2 + 3; | |
3530 | if(PL_reg_start_tmp) | |
3531 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3532 | else | |
3533 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3534 | } | |
aa283a38 | 3535 | |
1a147d38 | 3536 | eval_recurse_doit: /* Share code with GOSUB below this line */ |
aa283a38 | 3537 | /* run the pattern returned from (??{...}) */ |
40a82448 DM |
3538 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3539 | REGCP_SET(ST.lastcp); | |
6bda09f9 | 3540 | |
1a147d38 YO |
3541 | PL_regstartp = re->startp; /* essentially NOOP on GOSUB */ |
3542 | PL_regendp = re->endp; /* essentially NOOP on GOSUB */ | |
6bda09f9 | 3543 | |
4aabdb9b DM |
3544 | *PL_reglastparen = 0; |
3545 | *PL_reglastcloseparen = 0; | |
4aabdb9b | 3546 | PL_reginput = locinput; |
ae0beba1 | 3547 | PL_regsize = 0; |
4aabdb9b DM |
3548 | |
3549 | /* XXXX This is too dramatic a measure... */ | |
3550 | PL_reg_maxiter = 0; | |
3551 | ||
faec1544 DM |
3552 | ST.toggle_reg_flags = PL_reg_flags; |
3553 | if (re->reganch & ROPT_UTF8) | |
3554 | PL_reg_flags |= RF_utf8; | |
3555 | else | |
3556 | PL_reg_flags &= ~RF_utf8; | |
3557 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3558 | ||
40a82448 | 3559 | ST.prev_rex = rex; |
faec1544 | 3560 | ST.prev_curlyx = cur_curlyx; |
aa283a38 | 3561 | rex = re; |
faec1544 | 3562 | cur_curlyx = NULL; |
40a82448 | 3563 | ST.B = next; |
faec1544 DM |
3564 | ST.prev_eval = cur_eval; |
3565 | cur_eval = st; | |
faec1544 | 3566 | /* now continue from first node in postoned RE */ |
6bda09f9 | 3567 | PUSH_YES_STATE_GOTO(EVAL_AB, startpoint); |
4aabdb9b | 3568 | /* NOTREACHED */ |
a0ed51b3 | 3569 | } |
24d3c4a9 DM |
3570 | /* logical is 1, /(?(?{...})X|Y)/ */ |
3571 | sw = (bool)SvTRUE(ret); | |
3572 | logical = 0; | |
c277df42 IZ |
3573 | break; |
3574 | } | |
40a82448 | 3575 | |
faec1544 DM |
3576 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3577 | /* note: this is called twice; first after popping B, then A */ | |
3578 | PL_reg_flags ^= ST.toggle_reg_flags; | |
40a82448 DM |
3579 | ReREFCNT_dec(rex); |
3580 | rex = ST.prev_rex; | |
faec1544 DM |
3581 | regcpblow(ST.cp); |
3582 | cur_eval = ST.prev_eval; | |
3583 | cur_curlyx = ST.prev_curlyx; | |
40a82448 DM |
3584 | /* XXXX This is too dramatic a measure... */ |
3585 | PL_reg_maxiter = 0; | |
262b90c4 | 3586 | sayYES; |
40a82448 | 3587 | |
40a82448 | 3588 | |
faec1544 DM |
3589 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3590 | /* note: this is called twice; first after popping B, then A */ | |
3591 | PL_reg_flags ^= ST.toggle_reg_flags; | |
3592 | ReREFCNT_dec(rex); | |
3593 | rex = ST.prev_rex; | |
40a82448 DM |
3594 | PL_reginput = locinput; |
3595 | REGCP_UNWIND(ST.lastcp); | |
3596 | regcppop(rex); | |
faec1544 DM |
3597 | cur_eval = ST.prev_eval; |
3598 | cur_curlyx = ST.prev_curlyx; | |
3599 | /* XXXX This is too dramatic a measure... */ | |
3600 | PL_reg_maxiter = 0; | |
40a82448 | 3601 | sayNO_SILENT; |
40a82448 DM |
3602 | #undef ST |
3603 | ||
a0d0e21e | 3604 | case OPEN: |
c277df42 | 3605 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3606 | PL_reg_start_tmp[n] = locinput; |
3607 | if (n > PL_regsize) | |
3608 | PL_regsize = n; | |
a0d0e21e LW |
3609 | break; |
3610 | case CLOSE: | |
c277df42 | 3611 | n = ARG(scan); /* which paren pair */ |
cf93c79d IZ |
3612 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
3613 | PL_regendp[n] = locinput - PL_bostr; | |
7f69552c YO |
3614 | /*if (n > PL_regsize) |
3615 | PL_regsize = n;*/ | |
eb160463 | 3616 | if (n > (I32)*PL_reglastparen) |
3280af22 | 3617 | *PL_reglastparen = n; |
a01268b5 | 3618 | *PL_reglastcloseparen = n; |
6bda09f9 YO |
3619 | if (cur_eval && cur_eval->u.eval.close_paren == (U32)n) { |
3620 | goto fake_end; | |
3621 | } | |
a0d0e21e | 3622 | break; |
c277df42 IZ |
3623 | case GROUPP: |
3624 | n = ARG(scan); /* which paren pair */ | |
24d3c4a9 | 3625 | sw = (bool)((I32)*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 | 3626 | break; |
0a4db386 YO |
3627 | case NGROUPP: |
3628 | /* reg_check_named_buff_matched returns 0 for no match */ | |
3629 | sw = (bool)(0 < reg_check_named_buff_matched(rex,scan)); | |
3630 | break; | |
1a147d38 | 3631 | case INSUBP: |
0a4db386 | 3632 | n = ARG(scan); |
86413ec0 | 3633 | sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == (U32)n)); |
0a4db386 YO |
3634 | break; |
3635 | case DEFINEP: | |
3636 | sw = 0; | |
3637 | break; | |
c277df42 | 3638 | case IFTHEN: |
2c2d71f5 | 3639 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3640 | if (sw) |
c277df42 IZ |
3641 | next = NEXTOPER(NEXTOPER(scan)); |
3642 | else { | |
3643 | next = scan + ARG(scan); | |
3644 | if (OP(next) == IFTHEN) /* Fake one. */ | |
3645 | next = NEXTOPER(NEXTOPER(next)); | |
3646 | } | |
3647 | break; | |
3648 | case LOGICAL: | |
24d3c4a9 | 3649 | logical = scan->flags; |
c277df42 | 3650 | break; |
c476f425 | 3651 | |
2ab05381 | 3652 | /******************************************************************* |
2ab05381 | 3653 | |
c476f425 DM |
3654 | The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/ |
3655 | pattern, where A and B are subpatterns. (For simple A, CURLYM or | |
3656 | STAR/PLUS/CURLY/CURLYN are used instead.) | |
2ab05381 | 3657 | |
c476f425 | 3658 | A*B is compiled as <CURLYX><A><WHILEM><B> |
2ab05381 | 3659 | |
c476f425 DM |
3660 | On entry to the subpattern, CURLYX is called. This pushes a CURLYX |
3661 | state, which contains the current count, initialised to -1. It also sets | |
3662 | cur_curlyx to point to this state, with any previous value saved in the | |
3663 | state block. | |
2ab05381 | 3664 | |
c476f425 DM |
3665 | CURLYX then jumps straight to the WHILEM op, rather than executing A, |
3666 | since the pattern may possibly match zero times (i.e. it's a while {} loop | |
3667 | rather than a do {} while loop). | |
2ab05381 | 3668 | |
c476f425 DM |
3669 | Each entry to WHILEM represents a successful match of A. The count in the |
3670 | CURLYX block is incremented, another WHILEM state is pushed, and execution | |
3671 | passes to A or B depending on greediness and the current count. | |
2ab05381 | 3672 | |
c476f425 DM |
3673 | For example, if matching against the string a1a2a3b (where the aN are |
3674 | substrings that match /A/), then the match progresses as follows: (the | |
3675 | pushed states are interspersed with the bits of strings matched so far): | |
2ab05381 | 3676 | |
c476f425 DM |
3677 | <CURLYX cnt=-1> |
3678 | <CURLYX cnt=0><WHILEM> | |
3679 | <CURLYX cnt=1><WHILEM> a1 <WHILEM> | |
3680 | <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM> | |
3681 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> | |
3682 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b | |
2ab05381 | 3683 | |
c476f425 DM |
3684 | (Contrast this with something like CURLYM, which maintains only a single |
3685 | backtrack state: | |
2ab05381 | 3686 | |
c476f425 DM |
3687 | <CURLYM cnt=0> a1 |
3688 | a1 <CURLYM cnt=1> a2 | |
3689 | a1 a2 <CURLYM cnt=2> a3 | |
3690 | a1 a2 a3 <CURLYM cnt=3> b | |
3691 | ) | |
2ab05381 | 3692 | |
c476f425 DM |
3693 | Each WHILEM state block marks a point to backtrack to upon partial failure |
3694 | of A or B, and also contains some minor state data related to that | |
3695 | iteration. The CURLYX block, pointed to by cur_curlyx, contains the | |
3696 | overall state, such as the count, and pointers to the A and B ops. | |
2ab05381 | 3697 | |
c476f425 DM |
3698 | This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx |
3699 | must always point to the *current* CURLYX block, the rules are: | |
2ab05381 | 3700 | |
c476f425 DM |
3701 | When executing CURLYX, save the old cur_curlyx in the CURLYX state block, |
3702 | and set cur_curlyx to point the new block. | |
2ab05381 | 3703 | |
c476f425 DM |
3704 | When popping the CURLYX block after a successful or unsuccessful match, |
3705 | restore the previous cur_curlyx. | |
2ab05381 | 3706 | |
c476f425 DM |
3707 | When WHILEM is about to execute B, save the current cur_curlyx, and set it |
3708 | to the outer one saved in the CURLYX block. | |
2ab05381 | 3709 | |
c476f425 DM |
3710 | When popping the WHILEM block after a successful or unsuccessful B match, |
3711 | restore the previous cur_curlyx. | |
2ab05381 | 3712 | |
c476f425 DM |
3713 | Here's an example for the pattern (AI* BI)*BO |
3714 | I and O refer to inner and outer, C and W refer to CURLYX and WHILEM: | |
2ab05381 | 3715 | |
c476f425 DM |
3716 | cur_ |
3717 | curlyx backtrack stack | |
3718 | ------ --------------- | |
3719 | NULL | |
3720 | CO <CO prev=NULL> <WO> | |
3721 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3722 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi | |
3723 | NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo | |
2ab05381 | 3724 | |
c476f425 DM |
3725 | At this point the pattern succeeds, and we work back down the stack to |
3726 | clean up, restoring as we go: | |
95b24440 | 3727 | |
c476f425 DM |
3728 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi |
3729 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3730 | CO <CO prev=NULL> <WO> | |
3731 | NULL | |
a0374537 | 3732 | |
c476f425 DM |
3733 | *******************************************************************/ |
3734 | ||
3735 | #define ST st->u.curlyx | |
3736 | ||
3737 | case CURLYX: /* start of /A*B/ (for complex A) */ | |
3738 | { | |
3739 | /* No need to save/restore up to this paren */ | |
3740 | I32 parenfloor = scan->flags; | |
3741 | ||
3742 | assert(next); /* keep Coverity happy */ | |
3743 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
3744 | next += ARG(next); | |
3745 | ||
3746 | /* XXXX Probably it is better to teach regpush to support | |
3747 | parenfloor > PL_regsize... */ | |
3748 | if (parenfloor > (I32)*PL_reglastparen) | |
3749 | parenfloor = *PL_reglastparen; /* Pessimization... */ | |
3750 | ||
3751 | ST.prev_curlyx= cur_curlyx; | |
3752 | cur_curlyx = st; | |
3753 | ST.cp = PL_savestack_ix; | |
3754 | ||
3755 | /* these fields contain the state of the current curly. | |
3756 | * they are accessed by subsequent WHILEMs */ | |
3757 | ST.parenfloor = parenfloor; | |
3758 | ST.min = ARG1(scan); | |
3759 | ST.max = ARG2(scan); | |
3760 | ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
3761 | ST.B = next; | |
24d3c4a9 DM |
3762 | ST.minmod = minmod; |
3763 | minmod = 0; | |
c476f425 DM |
3764 | ST.count = -1; /* this will be updated by WHILEM */ |
3765 | ST.lastloc = NULL; /* this will be updated by WHILEM */ | |
3766 | ||
3767 | PL_reginput = locinput; | |
3768 | PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next)); | |
5f66b61c | 3769 | /* NOTREACHED */ |
c476f425 | 3770 | } |
a0d0e21e | 3771 | |
c476f425 DM |
3772 | case CURLYX_end: /* just finished matching all of A*B */ |
3773 | regcpblow(ST.cp); | |
3774 | cur_curlyx = ST.prev_curlyx; | |
3775 | sayYES; | |
3776 | /* NOTREACHED */ | |
a0d0e21e | 3777 | |
c476f425 DM |
3778 | case CURLYX_end_fail: /* just failed to match all of A*B */ |
3779 | regcpblow(ST.cp); | |
3780 | cur_curlyx = ST.prev_curlyx; | |
3781 | sayNO; | |
3782 | /* NOTREACHED */ | |
4633a7c4 | 3783 | |
a0d0e21e | 3784 | |
c476f425 DM |
3785 | #undef ST |
3786 | #define ST st->u.whilem | |
3787 | ||
3788 | case WHILEM: /* just matched an A in /A*B/ (for complex A) */ | |
3789 | { | |
3790 | /* see the discussion above about CURLYX/WHILEM */ | |
c476f425 DM |
3791 | I32 n; |
3792 | assert(cur_curlyx); /* keep Coverity happy */ | |
3793 | n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */ | |
3794 | ST.save_lastloc = cur_curlyx->u.curlyx.lastloc; | |
3795 | ST.cache_offset = 0; | |
3796 | ST.cache_mask = 0; | |
3797 | ||
3798 | PL_reginput = locinput; | |
3799 | ||
3800 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3801 | "%*s whilem: matched %ld out of %ld..%ld\n", | |
3802 | REPORT_CODE_OFF+depth*2, "", (long)n, | |
3803 | (long)cur_curlyx->u.curlyx.min, | |
3804 | (long)cur_curlyx->u.curlyx.max) | |
3805 | ); | |
a0d0e21e | 3806 | |
c476f425 | 3807 | /* First just match a string of min A's. */ |
a0d0e21e | 3808 | |
c476f425 DM |
3809 | if (n < cur_curlyx->u.curlyx.min) { |
3810 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3811 | PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A); | |
3812 | /* NOTREACHED */ | |
3813 | } | |
3814 | ||
3815 | /* If degenerate A matches "", assume A done. */ | |
3816 | ||
3817 | if (locinput == cur_curlyx->u.curlyx.lastloc) { | |
3818 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3819 | "%*s whilem: empty match detected, trying continuation...\n", | |
3820 | REPORT_CODE_OFF+depth*2, "") | |
3821 | ); | |
3822 | goto do_whilem_B_max; | |
3823 | } | |
3824 | ||
3825 | /* super-linear cache processing */ | |
3826 | ||
3827 | if (scan->flags) { | |
a0d0e21e | 3828 | |
2c2d71f5 | 3829 | if (!PL_reg_maxiter) { |
c476f425 DM |
3830 | /* start the countdown: Postpone detection until we |
3831 | * know the match is not *that* much linear. */ | |
2c2d71f5 | 3832 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
66bf836d DM |
3833 | /* possible overflow for long strings and many CURLYX's */ |
3834 | if (PL_reg_maxiter < 0) | |
3835 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
3836 | PL_reg_leftiter = PL_reg_maxiter; |
3837 | } | |
c476f425 | 3838 | |
2c2d71f5 | 3839 | if (PL_reg_leftiter-- == 0) { |
c476f425 | 3840 | /* initialise cache */ |
3298f257 | 3841 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 3842 | if (PL_reg_poscache) { |
eb160463 | 3843 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
3844 | Renew(PL_reg_poscache, size, char); |
3845 | PL_reg_poscache_size = size; | |
3846 | } | |
3847 | Zero(PL_reg_poscache, size, char); | |
3848 | } | |
3849 | else { | |
3850 | PL_reg_poscache_size = size; | |
a02a5408 | 3851 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 3852 | } |
c476f425 DM |
3853 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
3854 | "%swhilem: Detected a super-linear match, switching on caching%s...\n", | |
3855 | PL_colors[4], PL_colors[5]) | |
3856 | ); | |
2c2d71f5 | 3857 | } |
c476f425 | 3858 | |
2c2d71f5 | 3859 | if (PL_reg_leftiter < 0) { |
c476f425 DM |
3860 | /* have we already failed at this position? */ |
3861 | I32 offset, mask; | |
3862 | offset = (scan->flags & 0xf) - 1 | |
3863 | + (locinput - PL_bostr) * (scan->flags>>4); | |
3864 | mask = 1 << (offset % 8); | |
3865 | offset /= 8; | |
3866 | if (PL_reg_poscache[offset] & mask) { | |
3867 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3868 | "%*s whilem: (cache) already tried at this position...\n", | |
3869 | REPORT_CODE_OFF+depth*2, "") | |
2c2d71f5 | 3870 | ); |
3298f257 | 3871 | sayNO; /* cache records failure */ |
2c2d71f5 | 3872 | } |
c476f425 DM |
3873 | ST.cache_offset = offset; |
3874 | ST.cache_mask = mask; | |
2c2d71f5 | 3875 | } |
c476f425 | 3876 | } |
2c2d71f5 | 3877 | |
c476f425 | 3878 | /* Prefer B over A for minimal matching. */ |
a687059c | 3879 | |
c476f425 DM |
3880 | if (cur_curlyx->u.curlyx.minmod) { |
3881 | ST.save_curlyx = cur_curlyx; | |
3882 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
3883 | ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor); | |
3884 | REGCP_SET(ST.lastcp); | |
3885 | PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B); | |
3886 | /* NOTREACHED */ | |
3887 | } | |
a0d0e21e | 3888 | |
c476f425 DM |
3889 | /* Prefer A over B for maximal matching. */ |
3890 | ||
3891 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ | |
3892 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
3893 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3894 | REGCP_SET(ST.lastcp); | |
3895 | PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A); | |
3896 | /* NOTREACHED */ | |
3897 | } | |
3898 | goto do_whilem_B_max; | |
3899 | } | |
3900 | /* NOTREACHED */ | |
3901 | ||
3902 | case WHILEM_B_min: /* just matched B in a minimal match */ | |
3903 | case WHILEM_B_max: /* just matched B in a maximal match */ | |
3904 | cur_curlyx = ST.save_curlyx; | |
3905 | sayYES; | |
3906 | /* NOTREACHED */ | |
3907 | ||
3908 | case WHILEM_B_max_fail: /* just failed to match B in a maximal match */ | |
3909 | cur_curlyx = ST.save_curlyx; | |
3910 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
3911 | cur_curlyx->u.curlyx.count--; | |
3912 | CACHEsayNO; | |
3913 | /* NOTREACHED */ | |
3914 | ||
3915 | case WHILEM_A_min_fail: /* just failed to match A in a minimal match */ | |
3916 | REGCP_UNWIND(ST.lastcp); | |
3917 | regcppop(rex); | |
3918 | /* FALL THROUGH */ | |
3919 | case WHILEM_A_pre_fail: /* just failed to match even minimal A */ | |
3920 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
3921 | cur_curlyx->u.curlyx.count--; | |
3922 | CACHEsayNO; | |
3923 | /* NOTREACHED */ | |
3924 | ||
3925 | case WHILEM_A_max_fail: /* just failed to match A in a maximal match */ | |
3926 | REGCP_UNWIND(ST.lastcp); | |
3927 | regcppop(rex); /* Restore some previous $<digit>s? */ | |
3928 | PL_reginput = locinput; | |
3929 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
3930 | "%*s whilem: failed, trying continuation...\n", | |
3931 | REPORT_CODE_OFF+depth*2, "") | |
3932 | ); | |
3933 | do_whilem_B_max: | |
3934 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
3935 | && ckWARN(WARN_REGEXP) | |
3936 | && !(PL_reg_flags & RF_warned)) | |
3937 | { | |
3938 | PL_reg_flags |= RF_warned; | |
3939 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", | |
3940 | "Complex regular subexpression recursion", | |
3941 | REG_INFTY - 1); | |
3942 | } | |
3943 | ||
3944 | /* now try B */ | |
3945 | ST.save_curlyx = cur_curlyx; | |
3946 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
3947 | PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B); | |
3948 | /* NOTREACHED */ | |
3949 | ||
3950 | case WHILEM_B_min_fail: /* just failed to match B in a minimal match */ | |
3951 | cur_curlyx = ST.save_curlyx; | |
3952 | REGCP_UNWIND(ST.lastcp); | |
3953 | regcppop(rex); | |
3954 | ||
3955 | if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) { | |
3956 | /* Maximum greed exceeded */ | |
3957 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
3958 | && ckWARN(WARN_REGEXP) | |
3959 | && !(PL_reg_flags & RF_warned)) | |
3960 | { | |
3280af22 | 3961 | PL_reg_flags |= RF_warned; |
c476f425 DM |
3962 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), |
3963 | "%s limit (%d) exceeded", | |
3964 | "Complex regular subexpression recursion", | |
3965 | REG_INFTY - 1); | |
a0d0e21e | 3966 | } |
c476f425 | 3967 | cur_curlyx->u.curlyx.count--; |
3ab3c9b4 | 3968 | CACHEsayNO; |
a0d0e21e | 3969 | } |
c476f425 DM |
3970 | |
3971 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
3972 | "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "") | |
3973 | ); | |
3974 | /* Try grabbing another A and see if it helps. */ | |
3975 | PL_reginput = locinput; | |
3976 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3977 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
3978 | REGCP_SET(ST.lastcp); | |
3979 | PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A); | |
5f66b61c | 3980 | /* NOTREACHED */ |
40a82448 DM |
3981 | |
3982 | #undef ST | |
3983 | #define ST st->u.branch | |
3984 | ||
3985 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
3986 | next = scan + ARG(scan); |
3987 | if (next == scan) | |
3988 | next = NULL; | |
40a82448 DM |
3989 | scan = NEXTOPER(scan); |
3990 | /* FALL THROUGH */ | |
c277df42 | 3991 | |
40a82448 DM |
3992 | case BRANCH: /* /(...|A|...)/ */ |
3993 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
3994 | if (!next || (OP(next) != BRANCH && OP(next) != BRANCHJ)) | |
3995 | /* last branch; skip state push and jump direct to node */ | |
3996 | continue; | |
3997 | ST.lastparen = *PL_reglastparen; | |
3998 | ST.next_branch = next; | |
3999 | REGCP_SET(ST.cp); | |
4000 | PL_reginput = locinput; | |
02db2b7b | 4001 | |
40a82448 DM |
4002 | /* Now go into the branch */ |
4003 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
4004 | /* NOTREACHED */ | |
4005 | ||
4006 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ | |
4007 | REGCP_UNWIND(ST.cp); | |
4008 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
4009 | PL_regendp[n] = -1; | |
4010 | *PL_reglastparen = n; | |
0a4db386 | 4011 | /*dmq: *PL_reglastcloseparen = n; */ |
40a82448 DM |
4012 | scan = ST.next_branch; |
4013 | /* no more branches? */ | |
4014 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) | |
4015 | sayNO; | |
4016 | continue; /* execute next BRANCH[J] op */ | |
4017 | /* NOTREACHED */ | |
4018 | ||
a0d0e21e | 4019 | case MINMOD: |
24d3c4a9 | 4020 | minmod = 1; |
a0d0e21e | 4021 | break; |
40a82448 DM |
4022 | |
4023 | #undef ST | |
4024 | #define ST st->u.curlym | |
4025 | ||
4026 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
4027 | ||
4028 | /* This is an optimisation of CURLYX that enables us to push | |
4029 | * only a single backtracking state, no matter now many matches | |
4030 | * there are in {m,n}. It relies on the pattern being constant | |
4031 | * length, with no parens to influence future backrefs | |
4032 | */ | |
4033 | ||
4034 | ST.me = scan; | |
dc45a647 | 4035 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
4036 | |
4037 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
4038 | if (ST.me->flags) { | |
4039 | I32 paren = ST.me->flags; | |
4040 | if (paren > PL_regsize) | |
4041 | PL_regsize = paren; | |
4042 | if (paren > (I32)*PL_reglastparen) | |
4043 | *PL_reglastparen = paren; | |
c277df42 | 4044 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 4045 | } |
40a82448 DM |
4046 | ST.A = scan; |
4047 | ST.B = next; | |
4048 | ST.alen = 0; | |
4049 | ST.count = 0; | |
24d3c4a9 DM |
4050 | ST.minmod = minmod; |
4051 | minmod = 0; | |
40a82448 DM |
4052 | ST.c1 = CHRTEST_UNINIT; |
4053 | REGCP_SET(ST.cp); | |
6407bf3b | 4054 | |
40a82448 DM |
4055 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
4056 | goto curlym_do_B; | |
4057 | ||
4058 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 4059 | PL_reginput = locinput; |
40a82448 DM |
4060 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
4061 | /* NOTREACHED */ | |
5f80c4cf | 4062 | |
40a82448 DM |
4063 | case CURLYM_A: /* we've just matched an A */ |
4064 | locinput = st->locinput; | |
4065 | nextchr = UCHARAT(locinput); | |
4066 | ||
4067 | ST.count++; | |
4068 | /* after first match, determine A's length: u.curlym.alen */ | |
4069 | if (ST.count == 1) { | |
4070 | if (PL_reg_match_utf8) { | |
4071 | char *s = locinput; | |
4072 | while (s < PL_reginput) { | |
4073 | ST.alen++; | |
4074 | s += UTF8SKIP(s); | |
4075 | } | |
4076 | } | |
4077 | else { | |
4078 | ST.alen = PL_reginput - locinput; | |
4079 | } | |
4080 | if (ST.alen == 0) | |
4081 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
4082 | } | |
0cadcf80 DM |
4083 | DEBUG_EXECUTE_r( |
4084 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 4085 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 4086 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 4087 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
4088 | ); |
4089 | ||
40a82448 | 4090 | locinput = PL_reginput; |
0a4db386 YO |
4091 | |
4092 | if (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4093 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4094 | goto fake_end; |
4095 | ||
4096 | if ( ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)) ) | |
40a82448 DM |
4097 | goto curlym_do_A; /* try to match another A */ |
4098 | goto curlym_do_B; /* try to match B */ | |
5f80c4cf | 4099 | |
40a82448 DM |
4100 | case CURLYM_A_fail: /* just failed to match an A */ |
4101 | REGCP_UNWIND(ST.cp); | |
0a4db386 YO |
4102 | |
4103 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ | |
4104 | || (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4105 | cur_eval->u.eval.close_paren == (U32)ST.me->flags)) |
40a82448 | 4106 | sayNO; |
0cadcf80 | 4107 | |
40a82448 DM |
4108 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
4109 | PL_reginput = locinput; | |
4110 | if (ST.c1 == CHRTEST_UNINIT) { | |
4111 | /* calculate c1 and c2 for possible match of 1st char | |
4112 | * following curly */ | |
4113 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
4114 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
4115 | regnode *text_node = ST.B; | |
4116 | if (! HAS_TEXT(text_node)) | |
4117 | FIND_NEXT_IMPT(text_node); | |
4118 | if (HAS_TEXT(text_node) | |
4119 | && PL_regkind[OP(text_node)] != REF) | |
4120 | { | |
4121 | ST.c1 = (U8)*STRING(text_node); | |
4122 | ST.c2 = | |
4123 | (OP(text_node) == EXACTF || OP(text_node) == REFF) | |
4124 | ? PL_fold[ST.c1] | |
4125 | : (OP(text_node) == EXACTFL || OP(text_node) == REFFL) | |
4126 | ? PL_fold_locale[ST.c1] | |
4127 | : ST.c1; | |
c277df42 | 4128 | } |
c277df42 | 4129 | } |
40a82448 DM |
4130 | } |
4131 | ||
4132 | DEBUG_EXECUTE_r( | |
4133 | PerlIO_printf(Perl_debug_log, | |
4134 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 4135 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
4136 | "", (IV)ST.count) |
4137 | ); | |
4138 | if (ST.c1 != CHRTEST_VOID | |
4139 | && UCHARAT(PL_reginput) != ST.c1 | |
4140 | && UCHARAT(PL_reginput) != ST.c2) | |
4141 | { | |
4142 | /* simulate B failing */ | |
4143 | state_num = CURLYM_B_fail; | |
4144 | goto reenter_switch; | |
4145 | } | |
4146 | ||
4147 | if (ST.me->flags) { | |
4148 | /* mark current A as captured */ | |
4149 | I32 paren = ST.me->flags; | |
4150 | if (ST.count) { | |
4151 | PL_regstartp[paren] | |
4152 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; | |
4153 | PL_regendp[paren] = PL_reginput - PL_bostr; | |
0a4db386 | 4154 | /*dmq: *PL_reglastcloseparen = paren; */ |
c277df42 | 4155 | } |
40a82448 DM |
4156 | else |
4157 | PL_regendp[paren] = -1; | |
0a4db386 | 4158 | if (cur_eval && cur_eval->u.eval.close_paren && |
24b23f37 | 4159 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4160 | { |
4161 | if (ST.count) | |
4162 | goto fake_end; | |
4163 | else | |
4164 | sayNO; | |
4165 | } | |
c277df42 | 4166 | } |
0a4db386 | 4167 | |
40a82448 | 4168 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4169 | /* NOTREACHED */ |
40a82448 DM |
4170 | |
4171 | case CURLYM_B_fail: /* just failed to match a B */ | |
4172 | REGCP_UNWIND(ST.cp); | |
4173 | if (ST.minmod) { | |
4174 | if (ST.count == ARG2(ST.me) /* max */) | |
4175 | sayNO; | |
4176 | goto curlym_do_A; /* try to match a further A */ | |
4177 | } | |
4178 | /* backtrack one A */ | |
4179 | if (ST.count == ARG1(ST.me) /* min */) | |
4180 | sayNO; | |
4181 | ST.count--; | |
4182 | locinput = HOPc(locinput, -ST.alen); | |
4183 | goto curlym_do_B; /* try to match B */ | |
4184 | ||
c255a977 DM |
4185 | #undef ST |
4186 | #define ST st->u.curly | |
40a82448 | 4187 | |
c255a977 DM |
4188 | #define CURLY_SETPAREN(paren, success) \ |
4189 | if (paren) { \ | |
4190 | if (success) { \ | |
4191 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; \ | |
4192 | PL_regendp[paren] = locinput - PL_bostr; \ | |
0a4db386 | 4193 | *PL_reglastcloseparen = paren; \ |
c255a977 DM |
4194 | } \ |
4195 | else \ | |
4196 | PL_regendp[paren] = -1; \ | |
4197 | } | |
4198 | ||
4199 | case STAR: /* /A*B/ where A is width 1 */ | |
4200 | ST.paren = 0; | |
4201 | ST.min = 0; | |
4202 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4203 | scan = NEXTOPER(scan); |
4204 | goto repeat; | |
c255a977 DM |
4205 | case PLUS: /* /A+B/ where A is width 1 */ |
4206 | ST.paren = 0; | |
4207 | ST.min = 1; | |
4208 | ST.max = REG_INFTY; | |
c277df42 | 4209 | scan = NEXTOPER(scan); |
c255a977 DM |
4210 | goto repeat; |
4211 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4212 | ST.paren = scan->flags; /* Which paren to set */ | |
4213 | if (ST.paren > PL_regsize) | |
4214 | PL_regsize = ST.paren; | |
4215 | if (ST.paren > (I32)*PL_reglastparen) | |
4216 | *PL_reglastparen = ST.paren; | |
4217 | ST.min = ARG1(scan); /* min to match */ | |
4218 | ST.max = ARG2(scan); /* max to match */ | |
0a4db386 | 4219 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4220 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4221 | ST.min=1; |
4222 | ST.max=1; | |
4223 | } | |
c255a977 DM |
4224 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
4225 | goto repeat; | |
4226 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4227 | ST.paren = 0; | |
4228 | ST.min = ARG1(scan); /* min to match */ | |
4229 | ST.max = ARG2(scan); /* max to match */ | |
4230 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4231 | repeat: |
a0d0e21e LW |
4232 | /* |
4233 | * Lookahead to avoid useless match attempts | |
4234 | * when we know what character comes next. | |
c255a977 | 4235 | * |
5f80c4cf JP |
4236 | * Used to only do .*x and .*?x, but now it allows |
4237 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4238 | * of the quantifier and the EXACT-like node. -- japhy | |
4239 | */ | |
4240 | ||
c255a977 DM |
4241 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4242 | sayNO; | |
cca55fe3 | 4243 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4244 | U8 *s; |
4245 | regnode *text_node = next; | |
4246 | ||
3dab1dad YO |
4247 | if (! HAS_TEXT(text_node)) |
4248 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4249 | |
9e137952 | 4250 | if (! HAS_TEXT(text_node)) |
c255a977 | 4251 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4252 | else { |
3dab1dad | 4253 | if (PL_regkind[OP(text_node)] == REF) { |
c255a977 | 4254 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4255 | goto assume_ok_easy; |
cca55fe3 | 4256 | } |
be8e71aa YO |
4257 | else |
4258 | s = (U8*)STRING(text_node); | |
5f80c4cf JP |
4259 | |
4260 | if (!UTF) { | |
c255a977 | 4261 | ST.c2 = ST.c1 = *s; |
f65d3ee7 | 4262 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) |
c255a977 | 4263 | ST.c2 = PL_fold[ST.c1]; |
f65d3ee7 | 4264 | else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) |
c255a977 | 4265 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4266 | } |
5f80c4cf | 4267 | else { /* UTF */ |
f65d3ee7 | 4268 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) { |
a2a2844f | 4269 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4270 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4271 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4272 | |
4273 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4274 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4275 | #ifdef EBCDIC |
4276 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4277 | ckWARN(WARN_UTF8) ? | |
4278 | 0 : UTF8_ALLOW_ANY); | |
4279 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4280 | ckWARN(WARN_UTF8) ? | |
4281 | 0 : UTF8_ALLOW_ANY); | |
4282 | #else | |
c255a977 | 4283 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4284 | uniflags); |
c255a977 | 4285 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4286 | uniflags); |
4287 | #endif | |
5f80c4cf JP |
4288 | } |
4289 | else { | |
c255a977 | 4290 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4291 | uniflags); |
5f80c4cf | 4292 | } |
1aa99e6b IH |
4293 | } |
4294 | } | |
bbce6d69 | 4295 | } |
a0d0e21e | 4296 | else |
c255a977 | 4297 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4298 | assume_ok_easy: |
c255a977 DM |
4299 | |
4300 | ST.A = scan; | |
4301 | ST.B = next; | |
3280af22 | 4302 | PL_reginput = locinput; |
24d3c4a9 DM |
4303 | if (minmod) { |
4304 | minmod = 0; | |
c255a977 | 4305 | if (ST.min && regrepeat(rex, ST.A, ST.min) < ST.min) |
4633a7c4 | 4306 | sayNO; |
c255a977 | 4307 | ST.count = ST.min; |
a0ed51b3 | 4308 | locinput = PL_reginput; |
c255a977 DM |
4309 | REGCP_SET(ST.cp); |
4310 | if (ST.c1 == CHRTEST_VOID) | |
4311 | goto curly_try_B_min; | |
4312 | ||
4313 | ST.oldloc = locinput; | |
4314 | ||
4315 | /* set ST.maxpos to the furthest point along the | |
4316 | * string that could possibly match */ | |
4317 | if (ST.max == REG_INFTY) { | |
4318 | ST.maxpos = PL_regeol - 1; | |
4319 | if (do_utf8) | |
4320 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4321 | ST.maxpos--; | |
4322 | } | |
4323 | else if (do_utf8) { | |
4324 | int m = ST.max - ST.min; | |
4325 | for (ST.maxpos = locinput; | |
4326 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4327 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4328 | } | |
4329 | else { | |
4330 | ST.maxpos = locinput + ST.max - ST.min; | |
4331 | if (ST.maxpos >= PL_regeol) | |
4332 | ST.maxpos = PL_regeol - 1; | |
4333 | } | |
4334 | goto curly_try_B_min_known; | |
4335 | ||
4336 | } | |
4337 | else { | |
4338 | ST.count = regrepeat(rex, ST.A, ST.max); | |
4339 | locinput = PL_reginput; | |
4340 | if (ST.count < ST.min) | |
4341 | sayNO; | |
4342 | if ((ST.count > ST.min) | |
4343 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4344 | { | |
4345 | /* A{m,n} must come at the end of the string, there's | |
4346 | * no point in backing off ... */ | |
4347 | ST.min = ST.count; | |
4348 | /* ...except that $ and \Z can match before *and* after | |
4349 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4350 | We may back off by one in this case. */ | |
4351 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4352 | ST.min--; | |
4353 | } | |
4354 | REGCP_SET(ST.cp); | |
4355 | goto curly_try_B_max; | |
4356 | } | |
4357 | /* NOTREACHED */ | |
4358 | ||
4359 | ||
4360 | case CURLY_B_min_known_fail: | |
4361 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4362 | if (ST.paren && ST.count) | |
4363 | PL_regendp[ST.paren] = -1; | |
4364 | ||
4365 | PL_reginput = locinput; /* Could be reset... */ | |
4366 | REGCP_UNWIND(ST.cp); | |
4367 | /* Couldn't or didn't -- move forward. */ | |
4368 | ST.oldloc = locinput; | |
4369 | if (do_utf8) | |
4370 | locinput += UTF8SKIP(locinput); | |
4371 | else | |
4372 | locinput++; | |
4373 | ST.count++; | |
4374 | curly_try_B_min_known: | |
4375 | /* find the next place where 'B' could work, then call B */ | |
4376 | { | |
4377 | int n; | |
4378 | if (do_utf8) { | |
4379 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4380 | if (ST.c1 == ST.c2) { | |
4381 | STRLEN len; | |
4382 | /* set n to utf8_distance(oldloc, locinput) */ | |
4383 | while (locinput <= ST.maxpos && | |
4384 | utf8n_to_uvchr((U8*)locinput, | |
4385 | UTF8_MAXBYTES, &len, | |
4386 | uniflags) != (UV)ST.c1) { | |
4387 | locinput += len; | |
4388 | n++; | |
4389 | } | |
1aa99e6b IH |
4390 | } |
4391 | else { | |
c255a977 DM |
4392 | /* set n to utf8_distance(oldloc, locinput) */ |
4393 | while (locinput <= ST.maxpos) { | |
4394 | STRLEN len; | |
4395 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4396 | UTF8_MAXBYTES, &len, | |
4397 | uniflags); | |
4398 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4399 | break; | |
4400 | locinput += len; | |
4401 | n++; | |
1aa99e6b | 4402 | } |
0fe9bf95 IZ |
4403 | } |
4404 | } | |
c255a977 DM |
4405 | else { |
4406 | if (ST.c1 == ST.c2) { | |
4407 | while (locinput <= ST.maxpos && | |
4408 | UCHARAT(locinput) != ST.c1) | |
4409 | locinput++; | |
bbce6d69 | 4410 | } |
c255a977 DM |
4411 | else { |
4412 | while (locinput <= ST.maxpos | |
4413 | && UCHARAT(locinput) != ST.c1 | |
4414 | && UCHARAT(locinput) != ST.c2) | |
4415 | locinput++; | |
a0ed51b3 | 4416 | } |
c255a977 DM |
4417 | n = locinput - ST.oldloc; |
4418 | } | |
4419 | if (locinput > ST.maxpos) | |
4420 | sayNO; | |
4421 | /* PL_reginput == oldloc now */ | |
4422 | if (n) { | |
4423 | ST.count += n; | |
4424 | if (regrepeat(rex, ST.A, n) < n) | |
4633a7c4 | 4425 | sayNO; |
a0d0e21e | 4426 | } |
c255a977 DM |
4427 | PL_reginput = locinput; |
4428 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4429 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4430 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4431 | goto fake_end; |
4432 | } | |
c255a977 | 4433 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); |
a0d0e21e | 4434 | } |
c255a977 DM |
4435 | /* NOTREACHED */ |
4436 | ||
4437 | ||
4438 | case CURLY_B_min_fail: | |
4439 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4440 | if (ST.paren && ST.count) | |
4441 | PL_regendp[ST.paren] = -1; | |
4442 | ||
4443 | REGCP_UNWIND(ST.cp); | |
4444 | /* failed -- move forward one */ | |
4445 | PL_reginput = locinput; | |
4446 | if (regrepeat(rex, ST.A, 1)) { | |
4447 | ST.count++; | |
a0ed51b3 | 4448 | locinput = PL_reginput; |
c255a977 DM |
4449 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4450 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4451 | { |
c255a977 DM |
4452 | curly_try_B_min: |
4453 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4454 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4455 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4456 | goto fake_end; |
4457 | } | |
c255a977 | 4458 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); |
a0d0e21e LW |
4459 | } |
4460 | } | |
4633a7c4 | 4461 | sayNO; |
c255a977 DM |
4462 | /* NOTREACHED */ |
4463 | ||
4464 | ||
4465 | curly_try_B_max: | |
4466 | /* a successful greedy match: now try to match B */ | |
40d049e4 | 4467 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4468 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
40d049e4 YO |
4469 | goto fake_end; |
4470 | } | |
c255a977 DM |
4471 | { |
4472 | UV c = 0; | |
4473 | if (ST.c1 != CHRTEST_VOID) | |
4474 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4475 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4476 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4477 | /* If it could work, try it. */ |
4478 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4479 | CURLY_SETPAREN(ST.paren, ST.count); | |
4480 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4481 | /* NOTREACHED */ | |
4482 | } | |
4483 | } | |
4484 | /* FALL THROUGH */ | |
4485 | case CURLY_B_max_fail: | |
4486 | /* failed to find B in a greedy match */ | |
4487 | if (ST.paren && ST.count) | |
4488 | PL_regendp[ST.paren] = -1; | |
4489 | ||
4490 | REGCP_UNWIND(ST.cp); | |
4491 | /* back up. */ | |
4492 | if (--ST.count < ST.min) | |
4493 | sayNO; | |
4494 | PL_reginput = locinput = HOPc(locinput, -1); | |
4495 | goto curly_try_B_max; | |
4496 | ||
4497 | #undef ST | |
4498 | ||
a0d0e21e | 4499 | case END: |
6bda09f9 | 4500 | fake_end: |
faec1544 DM |
4501 | if (cur_eval) { |
4502 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4503 | I32 tmpix; | |
4504 | ||
4505 | ||
4506 | st->u.eval.toggle_reg_flags | |
4507 | = cur_eval->u.eval.toggle_reg_flags; | |
4508 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4509 | ||
4510 | st->u.eval.prev_rex = rex; /* inner */ | |
4511 | rex = cur_eval->u.eval.prev_rex; /* outer */ | |
4512 | cur_curlyx = cur_eval->u.eval.prev_curlyx; | |
4513 | ReREFCNT_inc(rex); | |
4514 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ | |
4515 | REGCP_SET(st->u.eval.lastcp); | |
4516 | PL_reginput = locinput; | |
4517 | ||
4518 | /* Restore parens of the outer rex without popping the | |
4519 | * savestack */ | |
4520 | tmpix = PL_savestack_ix; | |
4521 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4522 | regcppop(rex); | |
4523 | PL_savestack_ix = tmpix; | |
4524 | ||
4525 | st->u.eval.prev_eval = cur_eval; | |
4526 | cur_eval = cur_eval->u.eval.prev_eval; | |
4527 | DEBUG_EXECUTE_r( | |
6bda09f9 YO |
4528 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %x\n", |
4529 | REPORT_CODE_OFF+depth*2, "",(int)cur_eval);); | |
faec1544 DM |
4530 | PUSH_YES_STATE_GOTO(EVAL_AB, |
4531 | st->u.eval.prev_eval->u.eval.B); /* match B */ | |
4532 | } | |
4533 | ||
3b0527fe | 4534 | if (locinput < reginfo->till) { |
a3621e74 | 4535 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4536 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4537 | PL_colors[4], | |
4538 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4539 | (long)(reginfo->till - PL_reg_starttry), |
7821416a | 4540 | PL_colors[5])); |
262b90c4 | 4541 | sayNO_SILENT; /* Cannot match: too short. */ |
7821416a IZ |
4542 | } |
4543 | PL_reginput = locinput; /* put where regtry can find it */ | |
262b90c4 | 4544 | sayYES; /* Success! */ |
dad79028 DM |
4545 | |
4546 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4547 | DEBUG_EXECUTE_r( | |
4548 | PerlIO_printf(Perl_debug_log, | |
4549 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4550 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4551 | PL_reginput = locinput; /* put where regtry can find it */ |
262b90c4 | 4552 | sayYES; /* Success! */ |
dad79028 | 4553 | |
40a82448 DM |
4554 | #undef ST |
4555 | #define ST st->u.ifmatch | |
4556 | ||
4557 | case SUSPEND: /* (?>A) */ | |
4558 | ST.wanted = 1; | |
9fe1d20c | 4559 | PL_reginput = locinput; |
9041c2e3 | 4560 | goto do_ifmatch; |
dad79028 | 4561 | |
40a82448 DM |
4562 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4563 | ST.wanted = 0; | |
dad79028 DM |
4564 | goto ifmatch_trivial_fail_test; |
4565 | ||
40a82448 DM |
4566 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4567 | ST.wanted = 1; | |
dad79028 | 4568 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4569 | if (scan->flags) { |
52657f30 | 4570 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4571 | if (!s) { |
4572 | /* trivial fail */ | |
24d3c4a9 DM |
4573 | if (logical) { |
4574 | logical = 0; | |
4575 | sw = 1 - (bool)ST.wanted; | |
dad79028 | 4576 | } |
40a82448 | 4577 | else if (ST.wanted) |
dad79028 DM |
4578 | sayNO; |
4579 | next = scan + ARG(scan); | |
4580 | if (next == scan) | |
4581 | next = NULL; | |
4582 | break; | |
4583 | } | |
efb30f32 | 4584 | PL_reginput = s; |
a0ed51b3 LW |
4585 | } |
4586 | else | |
4587 | PL_reginput = locinput; | |
4588 | ||
c277df42 | 4589 | do_ifmatch: |
40a82448 | 4590 | ST.me = scan; |
24d3c4a9 | 4591 | ST.logical = logical; |
40a82448 DM |
4592 | /* execute body of (?...A) */ |
4593 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
4594 | /* NOTREACHED */ | |
4595 | ||
4596 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
4597 | ST.wanted = !ST.wanted; | |
4598 | /* FALL THROUGH */ | |
4599 | ||
4600 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
24d3c4a9 DM |
4601 | if (ST.logical) { |
4602 | sw = (bool)ST.wanted; | |
40a82448 DM |
4603 | } |
4604 | else if (!ST.wanted) | |
4605 | sayNO; | |
4606 | ||
4607 | if (OP(ST.me) == SUSPEND) | |
4608 | locinput = PL_reginput; | |
4609 | else { | |
4610 | locinput = PL_reginput = st->locinput; | |
4611 | nextchr = UCHARAT(locinput); | |
4612 | } | |
4613 | scan = ST.me + ARG(ST.me); | |
4614 | if (scan == ST.me) | |
4615 | scan = NULL; | |
4616 | continue; /* execute B */ | |
4617 | ||
4618 | #undef ST | |
dad79028 | 4619 | |
c277df42 | 4620 | case LONGJMP: |
c277df42 IZ |
4621 | next = scan + ARG(scan); |
4622 | if (next == scan) | |
4623 | next = NULL; | |
a0d0e21e | 4624 | break; |
24b23f37 YO |
4625 | case OPERROR: |
4626 | reginfo->cutpoint=PL_regeol; | |
4627 | goto do_commit; | |
4628 | /* NOTREACHED */ | |
4629 | case CUT: | |
4630 | if ( locinput > reginfo->bol ) | |
4631 | reginfo->cutpoint = HOPBACKc(locinput, 1); | |
4632 | /* FALLTHROUGH */ | |
54612592 | 4633 | case COMMIT: |
24b23f37 YO |
4634 | do_commit: |
4635 | PL_reginput = locinput; | |
54612592 YO |
4636 | PUSH_STATE_GOTO(COMMIT_next,next); |
4637 | /* NOTREACHED */ | |
4638 | case COMMIT_next_fail: | |
4639 | no_final = 1; | |
4640 | /* FALLTHROUGH */ | |
7f69552c YO |
4641 | case OPFAIL: |
4642 | sayNO; | |
a0d0e21e | 4643 | default: |
b900a521 | 4644 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 4645 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 4646 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c | 4647 | } |
95b24440 | 4648 | |
a0d0e21e | 4649 | scan = next; |
95b24440 DM |
4650 | continue; |
4651 | /* NOTREACHED */ | |
4652 | ||
40a82448 DM |
4653 | push_yes_state: |
4654 | /* push a state that backtracks on success */ | |
4655 | st->u.yes.prev_yes_state = yes_state; | |
4656 | yes_state = st; | |
4657 | /* FALL THROUGH */ | |
4658 | push_state: | |
4659 | /* push a new regex state, then continue at scan */ | |
4660 | { | |
4661 | regmatch_state *newst; | |
4662 | ||
24b23f37 YO |
4663 | DEBUG_STACK_r({ |
4664 | regmatch_state *cur = st; | |
4665 | regmatch_state *curyes = yes_state; | |
4666 | int curd = depth; | |
4667 | regmatch_slab *slab = PL_regmatch_slab; | |
4668 | for (;curd > -1;cur--,curd--) { | |
4669 | if (cur < SLAB_FIRST(slab)) { | |
4670 | slab = slab->prev; | |
4671 | cur = SLAB_LAST(slab); | |
4672 | } | |
4673 | PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n", | |
4674 | REPORT_CODE_OFF + 2 + depth * 2,"", | |
4675 | curd, reg_name[cur->resume_state], | |
4676 | (curyes == cur) ? "yes" : "" | |
4677 | ); | |
4678 | if (curyes == cur) | |
4679 | curyes = cur->u.yes.prev_yes_state; | |
4680 | } | |
4681 | } else | |
4682 | DEBUG_STATE_pp("push") | |
4683 | ); | |
40a82448 | 4684 | depth++; |
40a82448 DM |
4685 | st->locinput = locinput; |
4686 | newst = st+1; | |
4687 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
4688 | newst = S_push_slab(aTHX); | |
4689 | PL_regmatch_state = newst; | |
786e8c11 | 4690 | |
40a82448 DM |
4691 | locinput = PL_reginput; |
4692 | nextchr = UCHARAT(locinput); | |
4693 | st = newst; | |
4694 | continue; | |
4695 | /* NOTREACHED */ | |
4696 | } | |
a0d0e21e | 4697 | } |
a687059c | 4698 | |
a0d0e21e LW |
4699 | /* |
4700 | * We get here only if there's trouble -- normally "case END" is | |
4701 | * the terminating point. | |
4702 | */ | |
cea2e8a9 | 4703 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 4704 | /*NOTREACHED*/ |
4633a7c4 LW |
4705 | sayNO; |
4706 | ||
262b90c4 | 4707 | yes: |
77cb431f DM |
4708 | if (yes_state) { |
4709 | /* we have successfully completed a subexpression, but we must now | |
4710 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 4711 | assert(st != yes_state); |
5bc10b2c DM |
4712 | #ifdef DEBUGGING |
4713 | while (st != yes_state) { | |
4714 | st--; | |
4715 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4716 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4717 | st = SLAB_LAST(PL_regmatch_slab); | |
4718 | } | |
f9c6ae86 | 4719 | DEBUG_STATE_r({ |
54612592 YO |
4720 | if (no_final) { |
4721 | DEBUG_STATE_pp("pop (no final)"); | |
4722 | } else { | |
4723 | DEBUG_STATE_pp("pop (yes)"); | |
4724 | } | |
f9c6ae86 | 4725 | }); |
5bc10b2c DM |
4726 | depth--; |
4727 | } | |
4728 | #else | |
77cb431f DM |
4729 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
4730 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
4731 | { | |
4732 | /* not in this slab, pop slab */ | |
4733 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
4734 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4735 | st = SLAB_LAST(PL_regmatch_slab); | |
4736 | } | |
4737 | depth -= (st - yes_state); | |
5bc10b2c | 4738 | #endif |
77cb431f DM |
4739 | st = yes_state; |
4740 | yes_state = st->u.yes.prev_yes_state; | |
4741 | PL_regmatch_state = st; | |
24b23f37 | 4742 | |
77cb431f | 4743 | |
54612592 | 4744 | state_num = st->resume_state + no_final; |
24d3c4a9 | 4745 | goto reenter_switch; |
77cb431f DM |
4746 | } |
4747 | ||
a3621e74 | 4748 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 4749 | PL_colors[4], PL_colors[5])); |
02db2b7b | 4750 | |
95b24440 | 4751 | result = 1; |
aa283a38 | 4752 | goto final_exit; |
4633a7c4 LW |
4753 | |
4754 | no: | |
a3621e74 | 4755 | DEBUG_EXECUTE_r( |
7821416a | 4756 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 4757 | "%*s %sfailed...%s\n", |
5bc10b2c | 4758 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 4759 | PL_colors[4], PL_colors[5]) |
7821416a | 4760 | ); |
aa283a38 | 4761 | |
262b90c4 | 4762 | no_silent: |
54612592 YO |
4763 | if (no_final) { |
4764 | if (yes_state) { | |
4765 | goto yes; | |
4766 | } else { | |
4767 | goto final_exit; | |
4768 | } | |
4769 | } | |
aa283a38 DM |
4770 | if (depth) { |
4771 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
4772 | st--; |
4773 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4774 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4775 | st = SLAB_LAST(PL_regmatch_slab); | |
4776 | } | |
4777 | PL_regmatch_state = st; | |
40a82448 DM |
4778 | locinput= st->locinput; |
4779 | nextchr = UCHARAT(locinput); | |
4780 | ||
5bc10b2c DM |
4781 | DEBUG_STATE_pp("pop"); |
4782 | depth--; | |
262b90c4 DM |
4783 | if (yes_state == st) |
4784 | yes_state = st->u.yes.prev_yes_state; | |
5bc10b2c | 4785 | |
24d3c4a9 DM |
4786 | state_num = st->resume_state + 1; /* failure = success + 1 */ |
4787 | goto reenter_switch; | |
95b24440 | 4788 | } |
24d3c4a9 | 4789 | result = 0; |
aa283a38 | 4790 | |
262b90c4 | 4791 | final_exit: |
aa283a38 | 4792 | |
5d9a96ca DM |
4793 | /* restore original high-water mark */ |
4794 | PL_regmatch_slab = orig_slab; | |
4795 | PL_regmatch_state = orig_state; | |
4796 | ||
4797 | /* free all slabs above current one */ | |
4798 | if (orig_slab->next) { | |
c4fd8992 | 4799 | regmatch_slab *sl = orig_slab->next; |
5d9a96ca DM |
4800 | orig_slab->next = NULL; |
4801 | while (sl) { | |
c4fd8992 | 4802 | regmatch_slab * const osl = sl; |
5d9a96ca | 4803 | sl = sl->next; |
ad65c075 | 4804 | Safefree(osl); |
5d9a96ca DM |
4805 | } |
4806 | } | |
4807 | ||
95b24440 | 4808 | return result; |
a687059c LW |
4809 | } |
4810 | ||
4811 | /* | |
4812 | - regrepeat - repeatedly match something simple, report how many | |
4813 | */ | |
4814 | /* | |
4815 | * [This routine now assumes that it will only match on things of length 1. | |
4816 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 4817 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 4818 | */ |
76e3520e | 4819 | STATIC I32 |
32fc9b6a | 4820 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max) |
a687059c | 4821 | { |
27da23d5 | 4822 | dVAR; |
a0d0e21e | 4823 | register char *scan; |
a0d0e21e | 4824 | register I32 c; |
3280af22 | 4825 | register char *loceol = PL_regeol; |
a0ed51b3 | 4826 | register I32 hardcount = 0; |
53c4c00c | 4827 | register bool do_utf8 = PL_reg_match_utf8; |
a0d0e21e | 4828 | |
3280af22 | 4829 | scan = PL_reginput; |
faf11cac HS |
4830 | if (max == REG_INFTY) |
4831 | max = I32_MAX; | |
4832 | else if (max < loceol - scan) | |
7f596f4c | 4833 | loceol = scan + max; |
a0d0e21e | 4834 | switch (OP(p)) { |
22c35a8c | 4835 | case REG_ANY: |
1aa99e6b | 4836 | if (do_utf8) { |
ffc61ed2 | 4837 | loceol = PL_regeol; |
1aa99e6b | 4838 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
4839 | scan += UTF8SKIP(scan); |
4840 | hardcount++; | |
4841 | } | |
4842 | } else { | |
4843 | while (scan < loceol && *scan != '\n') | |
4844 | scan++; | |
a0ed51b3 LW |
4845 | } |
4846 | break; | |
ffc61ed2 | 4847 | case SANY: |
def8e4ea JH |
4848 | if (do_utf8) { |
4849 | loceol = PL_regeol; | |
a0804c9e | 4850 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
4851 | scan += UTF8SKIP(scan); |
4852 | hardcount++; | |
4853 | } | |
4854 | } | |
4855 | else | |
4856 | scan = loceol; | |
a0ed51b3 | 4857 | break; |
f33976b4 DB |
4858 | case CANY: |
4859 | scan = loceol; | |
4860 | break; | |
090f7165 JH |
4861 | case EXACT: /* length of string is 1 */ |
4862 | c = (U8)*STRING(p); | |
4863 | while (scan < loceol && UCHARAT(scan) == c) | |
4864 | scan++; | |
bbce6d69 | 4865 | break; |
4866 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 4867 | c = (U8)*STRING(p); |
bbce6d69 | 4868 | while (scan < loceol && |
22c35a8c | 4869 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 4870 | scan++; |
4871 | break; | |
4872 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 4873 | PL_reg_flags |= RF_tainted; |
cd439c50 | 4874 | c = (U8)*STRING(p); |
bbce6d69 | 4875 | while (scan < loceol && |
22c35a8c | 4876 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
4877 | scan++; |
4878 | break; | |
4879 | case ANYOF: | |
ffc61ed2 JH |
4880 | if (do_utf8) { |
4881 | loceol = PL_regeol; | |
cfc92286 | 4882 | while (hardcount < max && scan < loceol && |
32fc9b6a | 4883 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
4884 | scan += UTF8SKIP(scan); |
4885 | hardcount++; | |
4886 | } | |
4887 | } else { | |
32fc9b6a | 4888 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
4889 | scan++; |
4890 | } | |
a0d0e21e LW |
4891 | break; |
4892 | case ALNUM: | |
1aa99e6b | 4893 | if (do_utf8) { |
ffc61ed2 | 4894 | loceol = PL_regeol; |
1a4fad37 | 4895 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4896 | while (hardcount < max && scan < loceol && |
3568d838 | 4897 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4898 | scan += UTF8SKIP(scan); |
4899 | hardcount++; | |
4900 | } | |
4901 | } else { | |
4902 | while (scan < loceol && isALNUM(*scan)) | |
4903 | scan++; | |
a0ed51b3 LW |
4904 | } |
4905 | break; | |
bbce6d69 | 4906 | case ALNUML: |
3280af22 | 4907 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4908 | if (do_utf8) { |
ffc61ed2 | 4909 | loceol = PL_regeol; |
1aa99e6b IH |
4910 | while (hardcount < max && scan < loceol && |
4911 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4912 | scan += UTF8SKIP(scan); |
4913 | hardcount++; | |
4914 | } | |
4915 | } else { | |
4916 | while (scan < loceol && isALNUM_LC(*scan)) | |
4917 | scan++; | |
a0ed51b3 LW |
4918 | } |
4919 | break; | |
a0d0e21e | 4920 | case NALNUM: |
1aa99e6b | 4921 | if (do_utf8) { |
ffc61ed2 | 4922 | loceol = PL_regeol; |
1a4fad37 | 4923 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 4924 | while (hardcount < max && scan < loceol && |
3568d838 | 4925 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
4926 | scan += UTF8SKIP(scan); |
4927 | hardcount++; | |
4928 | } | |
4929 | } else { | |
4930 | while (scan < loceol && !isALNUM(*scan)) | |
4931 | scan++; | |
a0ed51b3 LW |
4932 | } |
4933 | break; | |
bbce6d69 | 4934 | case NALNUML: |
3280af22 | 4935 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4936 | if (do_utf8) { |
ffc61ed2 | 4937 | loceol = PL_regeol; |
1aa99e6b IH |
4938 | while (hardcount < max && scan < loceol && |
4939 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
4940 | scan += UTF8SKIP(scan); |
4941 | hardcount++; | |
4942 | } | |
4943 | } else { | |
4944 | while (scan < loceol && !isALNUM_LC(*scan)) | |
4945 | scan++; | |
a0ed51b3 LW |
4946 | } |
4947 | break; | |
a0d0e21e | 4948 | case SPACE: |
1aa99e6b | 4949 | if (do_utf8) { |
ffc61ed2 | 4950 | loceol = PL_regeol; |
1a4fad37 | 4951 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4952 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4953 | (*scan == ' ' || |
4954 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4955 | scan += UTF8SKIP(scan); |
4956 | hardcount++; | |
4957 | } | |
4958 | } else { | |
4959 | while (scan < loceol && isSPACE(*scan)) | |
4960 | scan++; | |
a0ed51b3 LW |
4961 | } |
4962 | break; | |
bbce6d69 | 4963 | case SPACEL: |
3280af22 | 4964 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4965 | if (do_utf8) { |
ffc61ed2 | 4966 | loceol = PL_regeol; |
1aa99e6b | 4967 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4968 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4969 | scan += UTF8SKIP(scan); | |
4970 | hardcount++; | |
4971 | } | |
4972 | } else { | |
4973 | while (scan < loceol && isSPACE_LC(*scan)) | |
4974 | scan++; | |
a0ed51b3 LW |
4975 | } |
4976 | break; | |
a0d0e21e | 4977 | case NSPACE: |
1aa99e6b | 4978 | if (do_utf8) { |
ffc61ed2 | 4979 | loceol = PL_regeol; |
1a4fad37 | 4980 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 4981 | while (hardcount < max && scan < loceol && |
3568d838 JH |
4982 | !(*scan == ' ' || |
4983 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
4984 | scan += UTF8SKIP(scan); |
4985 | hardcount++; | |
4986 | } | |
4987 | } else { | |
4988 | while (scan < loceol && !isSPACE(*scan)) | |
4989 | scan++; | |
4990 | break; | |
a0ed51b3 | 4991 | } |
bbce6d69 | 4992 | case NSPACEL: |
3280af22 | 4993 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 4994 | if (do_utf8) { |
ffc61ed2 | 4995 | loceol = PL_regeol; |
1aa99e6b | 4996 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
4997 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
4998 | scan += UTF8SKIP(scan); | |
4999 | hardcount++; | |
5000 | } | |
5001 | } else { | |
5002 | while (scan < loceol && !isSPACE_LC(*scan)) | |
5003 | scan++; | |
a0ed51b3 LW |
5004 | } |
5005 | break; | |
a0d0e21e | 5006 | case DIGIT: |
1aa99e6b | 5007 | if (do_utf8) { |
ffc61ed2 | 5008 | loceol = PL_regeol; |
1a4fad37 | 5009 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5010 | while (hardcount < max && scan < loceol && |
3568d838 | 5011 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5012 | scan += UTF8SKIP(scan); |
5013 | hardcount++; | |
5014 | } | |
5015 | } else { | |
5016 | while (scan < loceol && isDIGIT(*scan)) | |
5017 | scan++; | |
a0ed51b3 LW |
5018 | } |
5019 | break; | |
a0d0e21e | 5020 | case NDIGIT: |
1aa99e6b | 5021 | if (do_utf8) { |
ffc61ed2 | 5022 | loceol = PL_regeol; |
1a4fad37 | 5023 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5024 | while (hardcount < max && scan < loceol && |
3568d838 | 5025 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5026 | scan += UTF8SKIP(scan); |
5027 | hardcount++; | |
5028 | } | |
5029 | } else { | |
5030 | while (scan < loceol && !isDIGIT(*scan)) | |
5031 | scan++; | |
a0ed51b3 LW |
5032 | } |
5033 | break; | |
a0d0e21e LW |
5034 | default: /* Called on something of 0 width. */ |
5035 | break; /* So match right here or not at all. */ | |
5036 | } | |
a687059c | 5037 | |
a0ed51b3 LW |
5038 | if (hardcount) |
5039 | c = hardcount; | |
5040 | else | |
5041 | c = scan - PL_reginput; | |
3280af22 | 5042 | PL_reginput = scan; |
a687059c | 5043 | |
a3621e74 | 5044 | DEBUG_r({ |
e68ec53f | 5045 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 5046 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
5047 | SV * const prop = sv_newmortal(); |
5048 | regprop(prog, prop, p); | |
5049 | PerlIO_printf(Perl_debug_log, | |
be8e71aa YO |
5050 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
5051 | REPORT_CODE_OFF+1, "", SvPVX_const(prop),(IV)c,(IV)max); | |
a3621e74 | 5052 | }); |
be8e71aa | 5053 | }); |
9041c2e3 | 5054 | |
a0d0e21e | 5055 | return(c); |
a687059c LW |
5056 | } |
5057 | ||
c277df42 | 5058 | |
be8e71aa | 5059 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 5060 | /* |
ffc61ed2 JH |
5061 | - regclass_swash - prepare the utf8 swash |
5062 | */ | |
5063 | ||
5064 | SV * | |
32fc9b6a | 5065 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 5066 | { |
97aff369 | 5067 | dVAR; |
9e55ce06 JH |
5068 | SV *sw = NULL; |
5069 | SV *si = NULL; | |
5070 | SV *alt = NULL; | |
3dab1dad | 5071 | const struct reg_data * const data = prog ? prog->data : NULL; |
ffc61ed2 | 5072 | |
4f639d21 | 5073 | if (data && data->count) { |
a3b680e6 | 5074 | const U32 n = ARG(node); |
ffc61ed2 | 5075 | |
4f639d21 DM |
5076 | if (data->what[n] == 's') { |
5077 | SV * const rv = (SV*)data->data[n]; | |
890ce7af | 5078 | AV * const av = (AV*)SvRV((SV*)rv); |
2d03de9c | 5079 | SV **const ary = AvARRAY(av); |
9e55ce06 | 5080 | SV **a, **b; |
9041c2e3 | 5081 | |
711a919c | 5082 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
5083 | * documentation of these array elements. */ |
5084 | ||
b11f357e | 5085 | si = *ary; |
8f7f7219 | 5086 | a = SvROK(ary[1]) ? &ary[1] : 0; |
b11f357e JH |
5087 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0; |
5088 | ||
ffc61ed2 JH |
5089 | if (a) |
5090 | sw = *a; | |
5091 | else if (si && doinit) { | |
5092 | sw = swash_init("utf8", "", si, 1, 0); | |
5093 | (void)av_store(av, 1, sw); | |
5094 | } | |
9e55ce06 JH |
5095 | if (b) |
5096 | alt = *b; | |
ffc61ed2 JH |
5097 | } |
5098 | } | |
5099 | ||
9e55ce06 JH |
5100 | if (listsvp) |
5101 | *listsvp = si; | |
5102 | if (altsvp) | |
5103 | *altsvp = alt; | |
ffc61ed2 JH |
5104 | |
5105 | return sw; | |
5106 | } | |
76234dfb | 5107 | #endif |
ffc61ed2 JH |
5108 | |
5109 | /* | |
ba7b4546 | 5110 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
5111 | |
5112 | The n is the ANYOF regnode, the p is the target string, lenp | |
5113 | is pointer to the maximum length of how far to go in the p | |
5114 | (if the lenp is zero, UTF8SKIP(p) is used), | |
5115 | do_utf8 tells whether the target string is in UTF-8. | |
5116 | ||
bbce6d69 | 5117 | */ |
5118 | ||
76e3520e | 5119 | STATIC bool |
32fc9b6a | 5120 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 5121 | { |
27da23d5 | 5122 | dVAR; |
a3b680e6 | 5123 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 5124 | bool match = FALSE; |
cc07378b | 5125 | UV c = *p; |
ae9ddab8 | 5126 | STRLEN len = 0; |
9e55ce06 | 5127 | STRLEN plen; |
1aa99e6b | 5128 | |
19f67299 TS |
5129 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5130 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5131 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5132 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5133 | if (len == (STRLEN)-1) |
5134 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5135 | } |
bbce6d69 | 5136 | |
0f0076b4 | 5137 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5138 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5139 | if (lenp) |
5140 | *lenp = 0; | |
ffc61ed2 | 5141 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5142 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5143 | match = TRUE; | |
bbce6d69 | 5144 | } |
3568d838 | 5145 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5146 | match = TRUE; |
ffc61ed2 | 5147 | if (!match) { |
9e55ce06 | 5148 | AV *av; |
32fc9b6a | 5149 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5150 | |
5151 | if (sw) { | |
3568d838 | 5152 | if (swash_fetch(sw, p, do_utf8)) |
ffc61ed2 JH |
5153 | match = TRUE; |
5154 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5155 | if (!match && lenp && av) { |
5156 | I32 i; | |
9e55ce06 | 5157 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5158 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5159 | STRLEN len; |
890ce7af | 5160 | const char * const s = SvPV_const(sv, len); |
9e55ce06 | 5161 | |
061b10df | 5162 | if (len <= plen && memEQ(s, (char*)p, len)) { |
9e55ce06 JH |
5163 | *lenp = len; |
5164 | match = TRUE; | |
5165 | break; | |
5166 | } | |
5167 | } | |
5168 | } | |
5169 | if (!match) { | |
89ebb4a3 | 5170 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 JH |
5171 | STRLEN tmplen; |
5172 | ||
9e55ce06 JH |
5173 | to_utf8_fold(p, tmpbuf, &tmplen); |
5174 | if (swash_fetch(sw, tmpbuf, do_utf8)) | |
5175 | match = TRUE; | |
5176 | } | |
ffc61ed2 JH |
5177 | } |
5178 | } | |
bbce6d69 | 5179 | } |
9e55ce06 | 5180 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5181 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5182 | } |
1aa99e6b | 5183 | if (!match && c < 256) { |
ffc61ed2 JH |
5184 | if (ANYOF_BITMAP_TEST(n, c)) |
5185 | match = TRUE; | |
5186 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5187 | U8 f; |
a0ed51b3 | 5188 | |
ffc61ed2 JH |
5189 | if (flags & ANYOF_LOCALE) { |
5190 | PL_reg_flags |= RF_tainted; | |
5191 | f = PL_fold_locale[c]; | |
5192 | } | |
5193 | else | |
5194 | f = PL_fold[c]; | |
5195 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5196 | match = TRUE; | |
5197 | } | |
5198 | ||
5199 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5200 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5201 | if ( |
5202 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5203 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5204 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5205 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5206 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5207 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5208 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5209 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5210 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5211 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5212 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5213 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5214 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5215 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5216 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5217 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5218 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5219 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5220 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5221 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5222 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5223 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5224 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5225 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5226 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5227 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5228 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5229 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5230 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5231 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5232 | ) /* How's that for a conditional? */ | |
5233 | { | |
5234 | match = TRUE; | |
5235 | } | |
a0ed51b3 | 5236 | } |
a0ed51b3 LW |
5237 | } |
5238 | ||
a0ed51b3 LW |
5239 | return (flags & ANYOF_INVERT) ? !match : match; |
5240 | } | |
161b471a | 5241 | |
dfe13c55 | 5242 | STATIC U8 * |
0ce71af7 | 5243 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5244 | { |
97aff369 | 5245 | dVAR; |
a0ed51b3 | 5246 | if (off >= 0) { |
1aa99e6b | 5247 | while (off-- && s < lim) { |
ffc61ed2 | 5248 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5249 | s += UTF8SKIP(s); |
ffc61ed2 | 5250 | } |
a0ed51b3 LW |
5251 | } |
5252 | else { | |
1de06328 YO |
5253 | while (off++ && s > lim) { |
5254 | s--; | |
5255 | if (UTF8_IS_CONTINUED(*s)) { | |
5256 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5257 | s--; | |
a0ed51b3 | 5258 | } |
1de06328 | 5259 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5260 | } |
5261 | } | |
5262 | return s; | |
5263 | } | |
161b471a | 5264 | |
f9f4320a YO |
5265 | #ifdef XXX_dmq |
5266 | /* there are a bunch of places where we use two reghop3's that should | |
5267 | be replaced with this routine. but since thats not done yet | |
5268 | we ifdef it out - dmq | |
5269 | */ | |
dfe13c55 | 5270 | STATIC U8 * |
1de06328 YO |
5271 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5272 | { | |
5273 | dVAR; | |
5274 | if (off >= 0) { | |
5275 | while (off-- && s < rlim) { | |
5276 | /* XXX could check well-formedness here */ | |
5277 | s += UTF8SKIP(s); | |
5278 | } | |
5279 | } | |
5280 | else { | |
5281 | while (off++ && s > llim) { | |
5282 | s--; | |
5283 | if (UTF8_IS_CONTINUED(*s)) { | |
5284 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5285 | s--; | |
5286 | } | |
5287 | /* XXX could check well-formedness here */ | |
5288 | } | |
5289 | } | |
5290 | return s; | |
5291 | } | |
f9f4320a | 5292 | #endif |
1de06328 YO |
5293 | |
5294 | STATIC U8 * | |
0ce71af7 | 5295 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5296 | { |
97aff369 | 5297 | dVAR; |
a0ed51b3 | 5298 | if (off >= 0) { |
1aa99e6b | 5299 | while (off-- && s < lim) { |
ffc61ed2 | 5300 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5301 | s += UTF8SKIP(s); |
ffc61ed2 | 5302 | } |
a0ed51b3 | 5303 | if (off >= 0) |
3dab1dad | 5304 | return NULL; |
a0ed51b3 LW |
5305 | } |
5306 | else { | |
1de06328 YO |
5307 | while (off++ && s > lim) { |
5308 | s--; | |
5309 | if (UTF8_IS_CONTINUED(*s)) { | |
5310 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5311 | s--; | |
a0ed51b3 | 5312 | } |
1de06328 | 5313 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5314 | } |
5315 | if (off <= 0) | |
3dab1dad | 5316 | return NULL; |
a0ed51b3 LW |
5317 | } |
5318 | return s; | |
5319 | } | |
51371543 | 5320 | |
51371543 | 5321 | static void |
acfe0abc | 5322 | restore_pos(pTHX_ void *arg) |
51371543 | 5323 | { |
97aff369 | 5324 | dVAR; |
097eb12c | 5325 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5326 | if (PL_reg_eval_set) { |
5327 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5328 | rex->subbeg = PL_reg_oldsaved; |
5329 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5330 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5331 | rex->saved_copy = PL_nrs; |
ed252734 | 5332 | #endif |
4f639d21 | 5333 | RX_MATCH_COPIED_on(rex); |
51371543 GS |
5334 | } |
5335 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
5336 | PL_reg_eval_set = 0; | |
5337 | PL_curpm = PL_reg_oldcurpm; | |
5338 | } | |
5339 | } | |
33b8afdf JH |
5340 | |
5341 | STATIC void | |
5342 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
5343 | { | |
33b8afdf | 5344 | if (prog->float_substr && !prog->float_utf8) { |
097eb12c AL |
5345 | SV* const sv = newSVsv(prog->float_substr); |
5346 | prog->float_utf8 = sv; | |
33b8afdf JH |
5347 | sv_utf8_upgrade(sv); |
5348 | if (SvTAIL(prog->float_substr)) | |
5349 | SvTAIL_on(sv); | |
5350 | if (prog->float_substr == prog->check_substr) | |
5351 | prog->check_utf8 = sv; | |
5352 | } | |
5353 | if (prog->anchored_substr && !prog->anchored_utf8) { | |
097eb12c AL |
5354 | SV* const sv = newSVsv(prog->anchored_substr); |
5355 | prog->anchored_utf8 = sv; | |
33b8afdf JH |
5356 | sv_utf8_upgrade(sv); |
5357 | if (SvTAIL(prog->anchored_substr)) | |
5358 | SvTAIL_on(sv); | |
5359 | if (prog->anchored_substr == prog->check_substr) | |
5360 | prog->check_utf8 = sv; | |
5361 | } | |
5362 | } | |
5363 | ||
5364 | STATIC void | |
5365 | S_to_byte_substr(pTHX_ register regexp *prog) | |
5366 | { | |
97aff369 | 5367 | dVAR; |
33b8afdf | 5368 | if (prog->float_utf8 && !prog->float_substr) { |
097eb12c AL |
5369 | SV* sv = newSVsv(prog->float_utf8); |
5370 | prog->float_substr = sv; | |
33b8afdf JH |
5371 | if (sv_utf8_downgrade(sv, TRUE)) { |
5372 | if (SvTAIL(prog->float_utf8)) | |
5373 | SvTAIL_on(sv); | |
5374 | } else { | |
5375 | SvREFCNT_dec(sv); | |
5376 | prog->float_substr = sv = &PL_sv_undef; | |
5377 | } | |
5378 | if (prog->float_utf8 == prog->check_utf8) | |
5379 | prog->check_substr = sv; | |
5380 | } | |
5381 | if (prog->anchored_utf8 && !prog->anchored_substr) { | |
097eb12c AL |
5382 | SV* sv = newSVsv(prog->anchored_utf8); |
5383 | prog->anchored_substr = sv; | |
33b8afdf JH |
5384 | if (sv_utf8_downgrade(sv, TRUE)) { |
5385 | if (SvTAIL(prog->anchored_utf8)) | |
5386 | SvTAIL_on(sv); | |
5387 | } else { | |
5388 | SvREFCNT_dec(sv); | |
5389 | prog->anchored_substr = sv = &PL_sv_undef; | |
5390 | } | |
5391 | if (prog->anchored_utf8 == prog->check_utf8) | |
5392 | prog->check_substr = sv; | |
5393 | } | |
5394 | } | |
66610fdd RGS |
5395 | |
5396 | /* | |
5397 | * Local variables: | |
5398 | * c-indentation-style: bsd | |
5399 | * c-basic-offset: 4 | |
5400 | * indent-tabs-mode: t | |
5401 | * End: | |
5402 | * | |
37442d52 RGS |
5403 | * ex: set ts=8 sts=4 sw=4 noet: |
5404 | */ |