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. | |
16 | ||
166f8a29 DM |
17 | */ |
18 | ||
a687059c LW |
19 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
20 | * confused with the original package (see point 3 below). Thanks, Henry! | |
21 | */ | |
22 | ||
23 | /* Additional note: this code is very heavily munged from Henry's version | |
24 | * in places. In some spots I've traded clarity for efficiency, so don't | |
25 | * blame Henry for some of the lack of readability. | |
26 | */ | |
27 | ||
e50aee73 AD |
28 | /* The names of the functions have been changed from regcomp and |
29 | * regexec to pregcomp and pregexec in order to avoid conflicts | |
30 | * with the POSIX routines of the same names. | |
31 | */ | |
32 | ||
b9d5759e AD |
33 | #ifdef PERL_EXT_RE_BUILD |
34 | /* need to replace pregcomp et al, so enable that */ | |
35 | # ifndef PERL_IN_XSUB_RE | |
36 | # define PERL_IN_XSUB_RE | |
37 | # endif | |
38 | /* need access to debugger hooks */ | |
cad2e5aa | 39 | # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING) |
b9d5759e AD |
40 | # define DEBUGGING |
41 | # endif | |
42 | #endif | |
43 | ||
44 | #ifdef PERL_IN_XSUB_RE | |
d06ea78c | 45 | /* We *really* need to overwrite these symbols: */ |
56953603 IZ |
46 | # define Perl_regexec_flags my_regexec |
47 | # define Perl_regdump my_regdump | |
48 | # define Perl_regprop my_regprop | |
cad2e5aa | 49 | # define Perl_re_intuit_start my_re_intuit_start |
d06ea78c GS |
50 | /* *These* symbols are masked to allow static link. */ |
51 | # define Perl_pregexec my_pregexec | |
9041c2e3 | 52 | # define Perl_reginitcolors my_reginitcolors |
490a3f88 | 53 | # define Perl_regclass_swash my_regclass_swash |
c5be433b GS |
54 | |
55 | # define PERL_NO_GET_CONTEXT | |
9041c2e3 | 56 | #endif |
56953603 | 57 | |
a687059c | 58 | /* |
e50aee73 | 59 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c LW |
60 | * |
61 | * Copyright (c) 1986 by University of Toronto. | |
62 | * Written by Henry Spencer. Not derived from licensed software. | |
63 | * | |
64 | * Permission is granted to anyone to use this software for any | |
65 | * purpose on any computer system, and to redistribute it freely, | |
66 | * subject to the following restrictions: | |
67 | * | |
68 | * 1. The author is not responsible for the consequences of use of | |
69 | * this software, no matter how awful, even if they arise | |
70 | * from defects in it. | |
71 | * | |
72 | * 2. The origin of this software must not be misrepresented, either | |
73 | * by explicit claim or by omission. | |
74 | * | |
75 | * 3. Altered versions must be plainly marked as such, and must not | |
76 | * be misrepresented as being the original software. | |
77 | * | |
78 | **** Alterations to Henry's code are... | |
79 | **** | |
4bb101f2 | 80 | **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
b94e2f88 | 81 | **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others |
a687059c | 82 | **** |
9ef589d8 LW |
83 | **** You may distribute under the terms of either the GNU General Public |
84 | **** License or the Artistic License, as specified in the README file. | |
a687059c LW |
85 | * |
86 | * Beware that some of this code is subtly aware of the way operator | |
87 | * precedence is structured in regular expressions. Serious changes in | |
88 | * regular-expression syntax might require a total rethink. | |
89 | */ | |
90 | #include "EXTERN.h" | |
864dbfa3 | 91 | #define PERL_IN_REGEXEC_C |
a687059c | 92 | #include "perl.h" |
0f5d15d6 | 93 | |
a687059c LW |
94 | #include "regcomp.h" |
95 | ||
c277df42 IZ |
96 | #define RF_tainted 1 /* tainted information used? */ |
97 | #define RF_warned 2 /* warned about big count? */ | |
ce862d02 | 98 | #define RF_evaled 4 /* Did an EVAL with setting? */ |
a0ed51b3 LW |
99 | #define RF_utf8 8 /* String contains multibyte chars? */ |
100 | ||
eb160463 | 101 | #define UTF ((PL_reg_flags & RF_utf8) != 0) |
ce862d02 IZ |
102 | |
103 | #define RS_init 1 /* eval environment created */ | |
104 | #define RS_set 2 /* replsv value is set */ | |
c277df42 | 105 | |
a687059c LW |
106 | #ifndef STATIC |
107 | #define STATIC static | |
108 | #endif | |
109 | ||
32fc9b6a | 110 | #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c))) |
7d3e948e | 111 | |
c277df42 IZ |
112 | /* |
113 | * Forwards. | |
114 | */ | |
115 | ||
33b8afdf | 116 | #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv)) |
53c4c00c | 117 | #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b) |
a0ed51b3 | 118 | |
52657f30 AL |
119 | #define HOPc(pos,off) ((char *)(PL_reg_match_utf8 \ |
120 | ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \ | |
121 | : (U8*)(pos + off))) | |
122 | #define HOPBACKc(pos, off) ((char*) \ | |
123 | ((PL_reg_match_utf8) \ | |
124 | ? reghopmaybe3((U8*)pos, -off, ((U8*)(off < 0 ? PL_regeol : PL_bostr))) \ | |
efb30f32 HS |
125 | : (pos - off >= PL_bostr) \ |
126 | ? (U8*)(pos - off) \ | |
52657f30 | 127 | : (U8*)NULL) \ |
efb30f32 | 128 | ) |
efb30f32 | 129 | |
1aa99e6b | 130 | #define reghopmaybe3_c(pos,off,lim) ((char*)reghopmaybe3((U8*)pos, off, (U8*)lim)) |
53c4c00c | 131 | #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)pos, off, (U8*)lim) : (U8*)(pos + off)) |
1aa99e6b | 132 | #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) |
1aa99e6b | 133 | |
1a4fad37 AL |
134 | #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \ |
135 | if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END | |
136 | #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a") | |
137 | #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0") | |
138 | #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ") | |
139 | #define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86") | |
51371543 | 140 | |
5f80c4cf | 141 | /* for use after a quantifier and before an EXACT-like node -- japhy */ |
e2d8ce26 JP |
142 | #define JUMPABLE(rn) ( \ |
143 | OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \ | |
cca55fe3 JP |
144 | OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ |
145 | OP(rn) == PLUS || OP(rn) == MINMOD || \ | |
146 | (PL_regkind[(U8)OP(rn)] == CURLY && ARG1(rn) > 0) \ | |
e2d8ce26 JP |
147 | ) |
148 | ||
cca55fe3 JP |
149 | #define HAS_TEXT(rn) ( \ |
150 | PL_regkind[(U8)OP(rn)] == EXACT || PL_regkind[(U8)OP(rn)] == REF \ | |
151 | ) | |
e2d8ce26 | 152 | |
a84d97b6 HS |
153 | /* |
154 | Search for mandatory following text node; for lookahead, the text must | |
155 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
156 | */ | |
cca55fe3 | 157 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
e2d8ce26 | 158 | while (JUMPABLE(rn)) \ |
a84d97b6 | 159 | if (OP(rn) == SUSPEND || PL_regkind[(U8)OP(rn)] == CURLY) \ |
e2d8ce26 | 160 | rn = NEXTOPER(NEXTOPER(rn)); \ |
cca55fe3 JP |
161 | else if (OP(rn) == PLUS) \ |
162 | rn = NEXTOPER(rn); \ | |
a84d97b6 HS |
163 | else if (OP(rn) == IFMATCH) \ |
164 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ | |
e2d8ce26 | 165 | else rn += NEXT_OFF(rn); \ |
5f80c4cf | 166 | } STMT_END |
74750237 | 167 | |
acfe0abc | 168 | static void restore_pos(pTHX_ void *arg); |
51371543 | 169 | |
76e3520e | 170 | STATIC CHECKPOINT |
cea2e8a9 | 171 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 172 | { |
97aff369 | 173 | dVAR; |
a3b680e6 | 174 | const int retval = PL_savestack_ix; |
b1ce53c5 | 175 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 176 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e LW |
177 | int p; |
178 | ||
e49a9654 IH |
179 | if (paren_elems_to_push < 0) |
180 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
181 | ||
a01268b5 | 182 | #define REGCP_OTHER_ELEMS 6 |
4b3c1a47 | 183 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
3280af22 | 184 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 185 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
186 | SSPUSHINT(PL_regendp[p]); |
187 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 188 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e LW |
189 | SSPUSHINT(p); |
190 | } | |
b1ce53c5 | 191 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
3280af22 NIS |
192 | SSPUSHINT(PL_regsize); |
193 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 194 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 195 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
196 | #define REGCP_FRAME_ELEMS 2 |
197 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
198 | * are needed for the regexp context stack bookkeeping. */ | |
199 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 200 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 201 | |
a0d0e21e LW |
202 | return retval; |
203 | } | |
204 | ||
c277df42 | 205 | /* These are needed since we do not localize EVAL nodes: */ |
a3621e74 | 206 | # define REGCP_SET(cp) DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, \ |
faccc32b | 207 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
02db2b7b | 208 | (IV)PL_savestack_ix)); cp = PL_savestack_ix |
c3464db5 | 209 | |
a3621e74 | 210 | # define REGCP_UNWIND(cp) DEBUG_EXECUTE_r(cp != PL_savestack_ix ? \ |
c3464db5 | 211 | PerlIO_printf(Perl_debug_log, \ |
faccc32b | 212 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ |
02db2b7b | 213 | (IV)(cp), (IV)PL_savestack_ix) : 0); regcpblow(cp) |
c277df42 | 214 | |
76e3520e | 215 | STATIC char * |
097eb12c | 216 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 217 | { |
97aff369 | 218 | dVAR; |
b1ce53c5 | 219 | I32 i; |
a0d0e21e | 220 | char *input; |
b1ce53c5 | 221 | |
a3621e74 YO |
222 | GET_RE_DEBUG_FLAGS_DECL; |
223 | ||
b1ce53c5 | 224 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 225 | i = SSPOPINT; |
b1ce53c5 JH |
226 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
227 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 228 | input = (char *) SSPOPPTR; |
a01268b5 | 229 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
230 | *PL_reglastparen = SSPOPINT; |
231 | PL_regsize = SSPOPINT; | |
b1ce53c5 JH |
232 | |
233 | /* Now restore the parentheses context. */ | |
41123dfd JH |
234 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
235 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 236 | I32 tmps; |
097eb12c | 237 | U32 paren = (U32)SSPOPINT; |
3280af22 | 238 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
239 | PL_regstartp[paren] = SSPOPINT; |
240 | tmps = SSPOPINT; | |
3280af22 NIS |
241 | if (paren <= *PL_reglastparen) |
242 | PL_regendp[paren] = tmps; | |
a3621e74 | 243 | DEBUG_EXECUTE_r( |
c3464db5 | 244 | PerlIO_printf(Perl_debug_log, |
b900a521 | 245 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 246 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 247 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 248 | (IV)PL_regendp[paren], |
3280af22 | 249 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 250 | ); |
a0d0e21e | 251 | } |
a3621e74 | 252 | DEBUG_EXECUTE_r( |
4f639d21 | 253 | if ((I32)(*PL_reglastparen + 1) <= rex->nparens) { |
c3464db5 | 254 | PerlIO_printf(Perl_debug_log, |
faccc32b | 255 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 256 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
257 | } |
258 | ); | |
daf18116 | 259 | #if 1 |
dafc8851 JH |
260 | /* It would seem that the similar code in regtry() |
261 | * already takes care of this, and in fact it is in | |
262 | * a better location to since this code can #if 0-ed out | |
263 | * but the code in regtry() is needed or otherwise tests | |
264 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
265 | * (as of patchlevel 7877) will fail. Then again, |
266 | * this code seems to be necessary or otherwise | |
267 | * building DynaLoader will fail: | |
268 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
269 | * --jhi */ | |
097eb12c AL |
270 | for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { |
271 | if (i > PL_regsize) | |
272 | PL_regstartp[i] = -1; | |
273 | PL_regendp[i] = -1; | |
a0d0e21e | 274 | } |
dafc8851 | 275 | #endif |
a0d0e21e LW |
276 | return input; |
277 | } | |
278 | ||
02db2b7b | 279 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 280 | |
95b24440 | 281 | #define TRYPAREN(paren, n, input, where) { \ |
29d1e993 HS |
282 | if (paren) { \ |
283 | if (n) { \ | |
284 | PL_regstartp[paren] = HOPc(input, -1) - PL_bostr; \ | |
285 | PL_regendp[paren] = input - PL_bostr; \ | |
286 | } \ | |
287 | else \ | |
288 | PL_regendp[paren] = -1; \ | |
289 | } \ | |
95b24440 DM |
290 | REGMATCH(next, where); \ |
291 | if (result) \ | |
29d1e993 HS |
292 | sayYES; \ |
293 | if (paren && n) \ | |
294 | PL_regendp[paren] = -1; \ | |
295 | } | |
296 | ||
297 | ||
a687059c | 298 | /* |
e50aee73 | 299 | * pregexec and friends |
a687059c LW |
300 | */ |
301 | ||
302 | /* | |
c277df42 | 303 | - pregexec - match a regexp against a string |
a687059c | 304 | */ |
c277df42 | 305 | I32 |
864dbfa3 | 306 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 307 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
308 | /* strend: pointer to null at end of string */ |
309 | /* strbeg: real beginning of string */ | |
310 | /* minend: end of match must be >=minend after stringarg. */ | |
311 | /* nosave: For optimizations. */ | |
312 | { | |
313 | return | |
9041c2e3 | 314 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
315 | nosave ? 0 : REXEC_COPY_STR); |
316 | } | |
0f5d15d6 | 317 | |
22e551b9 | 318 | |
9041c2e3 | 319 | /* |
cad2e5aa JH |
320 | * Need to implement the following flags for reg_anch: |
321 | * | |
322 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
323 | * USE_INTUIT_ML | |
324 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
325 | * INTUIT_AUTORITATIVE_ML | |
326 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
327 | * INTUIT_ONCE_ML | |
328 | * | |
329 | * Another flag for this function: SECOND_TIME (so that float substrs | |
330 | * with giant delta may be not rechecked). | |
331 | */ | |
332 | ||
333 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
334 | ||
3f7c398e | 335 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
336 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
337 | ||
338 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
339 | ||
6eb5f6b9 JH |
340 | /* XXXX Some places assume that there is a fixed substring. |
341 | An update may be needed if optimizer marks as "INTUITable" | |
342 | RExen without fixed substrings. Similarly, it is assumed that | |
343 | lengths of all the strings are no more than minlen, thus they | |
344 | cannot come from lookahead. | |
345 | (Or minlen should take into account lookahead.) */ | |
346 | ||
2c2d71f5 JH |
347 | /* A failure to find a constant substring means that there is no need to make |
348 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
349 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
350 | regtry() should be needed. |
351 | ||
352 | REx compiler's optimizer found 4 possible hints: | |
353 | a) Anchored substring; | |
354 | b) Fixed substring; | |
355 | c) Whether we are anchored (beginning-of-line or \G); | |
356 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 357 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
358 | string which does not contradict any of them. |
359 | */ | |
2c2d71f5 | 360 | |
6eb5f6b9 JH |
361 | /* Most of decisions we do here should have been done at compile time. |
362 | The nodes of the REx which we used for the search should have been | |
363 | deleted from the finite automaton. */ | |
364 | ||
cad2e5aa JH |
365 | char * |
366 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
367 | char *strend, U32 flags, re_scream_pos_data *data) | |
368 | { | |
97aff369 | 369 | dVAR; |
b7953727 | 370 | register I32 start_shift = 0; |
cad2e5aa | 371 | /* Should be nonnegative! */ |
b7953727 | 372 | register I32 end_shift = 0; |
2c2d71f5 JH |
373 | register char *s; |
374 | register SV *check; | |
a1933d95 | 375 | char *strbeg; |
cad2e5aa | 376 | char *t; |
a3b680e6 | 377 | const int do_utf8 = sv ? SvUTF8(sv) : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 378 | I32 ml_anch; |
bd61b366 SS |
379 | register char *other_last = NULL; /* other substr checked before this */ |
380 | char *check_at = NULL; /* check substr found at this pos */ | |
1df70142 | 381 | const I32 multiline = prog->reganch & PMf_MULTILINE; |
30944b6d | 382 | #ifdef DEBUGGING |
890ce7af AL |
383 | const char * const i_strpos = strpos; |
384 | SV * const dsv = PERL_DEBUG_PAD_ZERO(0); | |
30944b6d | 385 | #endif |
a3621e74 YO |
386 | |
387 | GET_RE_DEBUG_FLAGS_DECL; | |
388 | ||
a30b2f1f | 389 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 390 | |
b8d68ded | 391 | if (prog->reganch & ROPT_UTF8) { |
a3621e74 | 392 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b8d68ded JH |
393 | "UTF-8 regex...\n")); |
394 | PL_reg_flags |= RF_utf8; | |
395 | } | |
396 | ||
a3621e74 | 397 | DEBUG_EXECUTE_r({ |
1df70142 | 398 | const char *s = PL_reg_match_utf8 ? |
c728cb41 JH |
399 | sv_uni_display(dsv, sv, 60, UNI_DISPLAY_REGEX) : |
400 | strpos; | |
1df70142 | 401 | const int len = PL_reg_match_utf8 ? |
b8d68ded | 402 | strlen(s) : strend - strpos; |
2a782b5b JH |
403 | if (!PL_colorset) |
404 | reginitcolors(); | |
b8d68ded | 405 | if (PL_reg_match_utf8) |
a3621e74 | 406 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b8d68ded | 407 | "UTF-8 target...\n")); |
2a782b5b | 408 | PerlIO_printf(Perl_debug_log, |
a0288114 | 409 | "%sGuessing start of match, REx%s \"%s%.60s%s%s\" against \"%s%.*s%s%s\"...\n", |
e4584336 | 410 | PL_colors[4], PL_colors[5], PL_colors[0], |
2a782b5b JH |
411 | prog->precomp, |
412 | PL_colors[1], | |
413 | (strlen(prog->precomp) > 60 ? "..." : ""), | |
414 | PL_colors[0], | |
415 | (int)(len > 60 ? 60 : len), | |
416 | s, PL_colors[1], | |
417 | (len > 60 ? "..." : "") | |
418 | ); | |
419 | }); | |
cad2e5aa | 420 | |
c344f387 JH |
421 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
422 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 423 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 424 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 425 | goto fail; |
2c2d71f5 | 426 | } |
a1933d95 | 427 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 428 | PL_regeol = strend; |
33b8afdf JH |
429 | if (do_utf8) { |
430 | if (!prog->check_utf8 && prog->check_substr) | |
431 | to_utf8_substr(prog); | |
432 | check = prog->check_utf8; | |
433 | } else { | |
434 | if (!prog->check_substr && prog->check_utf8) | |
435 | to_byte_substr(prog); | |
436 | check = prog->check_substr; | |
437 | } | |
438 | if (check == &PL_sv_undef) { | |
a3621e74 | 439 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
440 | "Non-utf string cannot match utf check string\n")); |
441 | goto fail; | |
442 | } | |
2c2d71f5 | 443 | if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ |
cad2e5aa JH |
444 | ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) |
445 | || ( (prog->reganch & ROPT_ANCH_BOL) | |
7fba1cd6 | 446 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 447 | |
7e25d62c JH |
448 | if (!ml_anch) { |
449 | if ( !(prog->reganch & (ROPT_ANCH_GPOS /* Checked by the caller */ | |
450 | | ROPT_IMPLICIT)) /* not a real BOL */ | |
3f7c398e | 451 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
452 | && sv && !SvROK(sv) |
453 | && (strpos != strbeg)) { | |
a3621e74 | 454 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
455 | goto fail; |
456 | } | |
457 | if (prog->check_offset_min == prog->check_offset_max && | |
cce850e4 | 458 | !(prog->reganch & ROPT_CANY_SEEN)) { |
2c2d71f5 | 459 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
460 | I32 slen; |
461 | ||
1aa99e6b | 462 | s = HOP3c(strpos, prog->check_offset_min, strend); |
653099ff GS |
463 | if (SvTAIL(check)) { |
464 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 465 | |
9041c2e3 | 466 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 467 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 468 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 469 | goto fail_finish; |
cad2e5aa JH |
470 | } |
471 | /* Now should match s[0..slen-2] */ | |
472 | slen--; | |
3f7c398e | 473 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 474 | || (slen > 1 |
3f7c398e | 475 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 476 | report_neq: |
a3621e74 | 477 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
478 | goto fail_finish; |
479 | } | |
cad2e5aa | 480 | } |
3f7c398e | 481 | else if (*SvPVX_const(check) != *s |
653099ff | 482 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 483 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 484 | goto report_neq; |
c315bfe8 | 485 | check_at = s; |
2c2d71f5 | 486 | goto success_at_start; |
7e25d62c | 487 | } |
cad2e5aa | 488 | } |
2c2d71f5 | 489 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 490 | s = strpos; |
2c2d71f5 | 491 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
2c2d71f5 | 492 | end_shift = prog->minlen - start_shift - |
653099ff | 493 | CHR_SVLEN(check) + (SvTAIL(check) != 0); |
2c2d71f5 | 494 | if (!ml_anch) { |
a3b680e6 | 495 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 496 | - (SvTAIL(check) != 0); |
a3b680e6 | 497 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
498 | |
499 | if (end_shift < eshift) | |
500 | end_shift = eshift; | |
501 | } | |
cad2e5aa | 502 | } |
2c2d71f5 | 503 | else { /* Can match at random position */ |
cad2e5aa JH |
504 | ml_anch = 0; |
505 | s = strpos; | |
2c2d71f5 JH |
506 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
507 | /* Should be nonnegative! */ | |
508 | end_shift = prog->minlen - start_shift - | |
653099ff | 509 | CHR_SVLEN(check) + (SvTAIL(check) != 0); |
cad2e5aa JH |
510 | } |
511 | ||
2c2d71f5 | 512 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 513 | if (end_shift < 0) |
6bbae5e6 | 514 | Perl_croak(aTHX_ "panic: end_shift"); |
2c2d71f5 JH |
515 | #endif |
516 | ||
2c2d71f5 JH |
517 | restart: |
518 | /* Find a possible match in the region s..strend by looking for | |
519 | the "check" substring in the region corrected by start/end_shift. */ | |
cad2e5aa | 520 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 521 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 522 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 523 | |
2c2d71f5 JH |
524 | if (PL_screamfirst[BmRARE(check)] >= 0 |
525 | || ( BmRARE(check) == '\n' | |
526 | && (BmPREVIOUS(check) == SvCUR(check) - 1) | |
527 | && SvTAIL(check) )) | |
9041c2e3 | 528 | s = screaminstr(sv, check, |
2c2d71f5 | 529 | start_shift + (s - strbeg), end_shift, pp, 0); |
cad2e5aa | 530 | else |
2c2d71f5 | 531 | goto fail_finish; |
4addbd3b HS |
532 | /* we may be pointing at the wrong string */ |
533 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 534 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
535 | if (data) |
536 | *data->scream_olds = s; | |
537 | } | |
f33976b4 | 538 | else if (prog->reganch & ROPT_CANY_SEEN) |
3baa4c62 JH |
539 | s = fbm_instr((U8*)(s + start_shift), |
540 | (U8*)(strend - end_shift), | |
7fba1cd6 | 541 | check, multiline ? FBMrf_MULTILINE : 0); |
cad2e5aa | 542 | else |
1aa99e6b IH |
543 | s = fbm_instr(HOP3(s, start_shift, strend), |
544 | HOP3(strend, -end_shift, strbeg), | |
7fba1cd6 | 545 | check, multiline ? FBMrf_MULTILINE : 0); |
cad2e5aa JH |
546 | |
547 | /* Update the count-of-usability, remove useless subpatterns, | |
548 | unshift s. */ | |
2c2d71f5 | 549 | |
a0288114 | 550 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s %s substr \"%s%.*s%s\"%s%s", |
2c2d71f5 | 551 | (s ? "Found" : "Did not find"), |
33b8afdf | 552 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) ? "anchored" : "floating"), |
2c2d71f5 | 553 | PL_colors[0], |
7b0972df | 554 | (int)(SvCUR(check) - (SvTAIL(check)!=0)), |
3f7c398e | 555 | SvPVX_const(check), |
2c2d71f5 JH |
556 | PL_colors[1], (SvTAIL(check) ? "$" : ""), |
557 | (s ? " at offset " : "...\n") ) ); | |
558 | ||
559 | if (!s) | |
560 | goto fail_finish; | |
561 | ||
6eb5f6b9 JH |
562 | check_at = s; |
563 | ||
2c2d71f5 | 564 | /* Finish the diagnostic message */ |
a3621e74 | 565 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 JH |
566 | |
567 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. | |
568 | Start with the other substr. | |
569 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 570 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
571 | *always* match. Probably should be marked during compile... |
572 | Probably it is right to do no SCREAM here... | |
573 | */ | |
574 | ||
33b8afdf | 575 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) : (prog->float_substr && prog->anchored_substr)) { |
30944b6d | 576 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
577 | /* XXXX May be hopelessly wrong for UTF... */ |
578 | if (!other_last) | |
6eb5f6b9 | 579 | other_last = strpos; |
33b8afdf | 580 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
581 | do_other_anchored: |
582 | { | |
890ce7af AL |
583 | char * const last = HOP3c(s, -start_shift, strbeg); |
584 | char *last1, *last2; | |
2c2d71f5 | 585 | char *s1 = s; |
33b8afdf | 586 | SV* must; |
2c2d71f5 | 587 | |
2c2d71f5 JH |
588 | t = s - prog->check_offset_max; |
589 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 590 | && (!do_utf8 |
1aa99e6b | 591 | || ((t = reghopmaybe3_c(s, -(prog->check_offset_max), strpos)) |
2c2d71f5 | 592 | && t > strpos))) |
30944b6d | 593 | /* EMPTY */; |
2c2d71f5 JH |
594 | else |
595 | t = strpos; | |
1aa99e6b | 596 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
597 | if (t < other_last) /* These positions already checked */ |
598 | t = other_last; | |
1aa99e6b | 599 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
600 | if (last < last1) |
601 | last1 = last; | |
602 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ | |
603 | /* On end-of-str: see comment below. */ | |
33b8afdf JH |
604 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
605 | if (must == &PL_sv_undef) { | |
606 | s = (char*)NULL; | |
a3621e74 | 607 | DEBUG_EXECUTE_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
608 | } |
609 | else | |
610 | s = fbm_instr( | |
611 | (unsigned char*)t, | |
612 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
613 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
614 | must, | |
7fba1cd6 | 615 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 616 | ); |
a3621e74 | 617 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a0288114 | 618 | "%s anchored substr \"%s%.*s%s\"%s", |
2c2d71f5 JH |
619 | (s ? "Found" : "Contradicts"), |
620 | PL_colors[0], | |
33b8afdf JH |
621 | (int)(SvCUR(must) |
622 | - (SvTAIL(must)!=0)), | |
3f7c398e | 623 | SvPVX_const(must), |
33b8afdf | 624 | PL_colors[1], (SvTAIL(must) ? "$" : ""))); |
2c2d71f5 JH |
625 | if (!s) { |
626 | if (last1 >= last2) { | |
a3621e74 | 627 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
628 | ", giving up...\n")); |
629 | goto fail_finish; | |
630 | } | |
a3621e74 | 631 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 632 | ", trying floating at offset %ld...\n", |
1aa99e6b IH |
633 | (long)(HOP3c(s1, 1, strend) - i_strpos))); |
634 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); | |
635 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
636 | goto restart; |
637 | } | |
638 | else { | |
a3621e74 | 639 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 640 | (long)(s - i_strpos))); |
1aa99e6b IH |
641 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
642 | other_last = HOP3c(s, 1, strend); | |
30944b6d | 643 | s = s1; |
2c2d71f5 JH |
644 | if (t == strpos) |
645 | goto try_at_start; | |
2c2d71f5 JH |
646 | goto try_at_offset; |
647 | } | |
30944b6d | 648 | } |
2c2d71f5 JH |
649 | } |
650 | else { /* Take into account the floating substring. */ | |
33b8afdf JH |
651 | char *last, *last1; |
652 | char *s1 = s; | |
653 | SV* must; | |
654 | ||
655 | t = HOP3c(s, -start_shift, strbeg); | |
656 | last1 = last = | |
657 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
658 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
659 | last = HOP3c(t, prog->float_max_offset, strend); | |
660 | s = HOP3c(t, prog->float_min_offset, strend); | |
661 | if (s < other_last) | |
662 | s = other_last; | |
2c2d71f5 | 663 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
664 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
665 | /* fbm_instr() takes into account exact value of end-of-str | |
666 | if the check is SvTAIL(ed). Since false positives are OK, | |
667 | and end-of-str is not later than strend we are OK. */ | |
668 | if (must == &PL_sv_undef) { | |
669 | s = (char*)NULL; | |
a3621e74 | 670 | DEBUG_EXECUTE_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
671 | } |
672 | else | |
2c2d71f5 | 673 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
674 | (unsigned char*)last + SvCUR(must) |
675 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 676 | must, multiline ? FBMrf_MULTILINE : 0); |
a0288114 | 677 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s floating substr \"%s%.*s%s\"%s", |
33b8afdf JH |
678 | (s ? "Found" : "Contradicts"), |
679 | PL_colors[0], | |
680 | (int)(SvCUR(must) - (SvTAIL(must)!=0)), | |
3f7c398e | 681 | SvPVX_const(must), |
33b8afdf JH |
682 | PL_colors[1], (SvTAIL(must) ? "$" : ""))); |
683 | if (!s) { | |
684 | if (last1 == last) { | |
a3621e74 | 685 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
686 | ", giving up...\n")); |
687 | goto fail_finish; | |
2c2d71f5 | 688 | } |
a3621e74 | 689 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
690 | ", trying anchored starting at offset %ld...\n", |
691 | (long)(s1 + 1 - i_strpos))); | |
692 | other_last = last; | |
693 | s = HOP3c(t, 1, strend); | |
694 | goto restart; | |
695 | } | |
696 | else { | |
a3621e74 | 697 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
698 | (long)(s - i_strpos))); |
699 | other_last = s; /* Fix this later. --Hugo */ | |
700 | s = s1; | |
701 | if (t == strpos) | |
702 | goto try_at_start; | |
703 | goto try_at_offset; | |
704 | } | |
2c2d71f5 | 705 | } |
cad2e5aa | 706 | } |
2c2d71f5 JH |
707 | |
708 | t = s - prog->check_offset_max; | |
2c2d71f5 | 709 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 710 | && (!do_utf8 |
1aa99e6b IH |
711 | || ((t = reghopmaybe3_c(s, -prog->check_offset_max, strpos)) |
712 | && t > strpos))) { | |
2c2d71f5 JH |
713 | /* Fixed substring is found far enough so that the match |
714 | cannot start at strpos. */ | |
715 | try_at_offset: | |
cad2e5aa | 716 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
717 | /* Eventually fbm_*() should handle this, but often |
718 | anchored_offset is not 0, so this check will not be wasted. */ | |
719 | /* XXXX In the code below we prefer to look for "^" even in | |
720 | presence of anchored substrings. And we search even | |
721 | beyond the found float position. These pessimizations | |
722 | are historical artefacts only. */ | |
723 | find_anchor: | |
2c2d71f5 | 724 | while (t < strend - prog->minlen) { |
cad2e5aa | 725 | if (*t == '\n') { |
4ee3650e | 726 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 727 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
728 | /* Since we moved from the found position, |
729 | we definitely contradict the found anchored | |
30944b6d IZ |
730 | substr. Due to the above check we do not |
731 | contradict "check" substr. | |
732 | Thus we can arrive here only if check substr | |
733 | is float. Redo checking for "other"=="fixed". | |
734 | */ | |
9041c2e3 | 735 | strpos = t + 1; |
a3621e74 | 736 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 737 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
738 | goto do_other_anchored; |
739 | } | |
4ee3650e GS |
740 | /* We don't contradict the found floating substring. */ |
741 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 742 | s = t + 1; |
a3621e74 | 743 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 744 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
745 | goto set_useful; |
746 | } | |
4ee3650e GS |
747 | /* Position contradicts check-string */ |
748 | /* XXXX probably better to look for check-string | |
749 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 750 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 751 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 752 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
753 | goto restart; |
754 | } | |
755 | t++; | |
756 | } | |
a3621e74 | 757 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 758 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 759 | goto fail_finish; |
cad2e5aa | 760 | } |
f5952150 | 761 | else { |
a3621e74 | 762 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 763 | PL_colors[0], PL_colors[1])); |
f5952150 | 764 | } |
cad2e5aa JH |
765 | s = t; |
766 | set_useful: | |
33b8afdf | 767 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
768 | } |
769 | else { | |
f5952150 | 770 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 771 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
772 | unless it was an MBOL and we are not after MBOL, |
773 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
774 | try_at_start: |
775 | /* Even in this situation we may use MBOL flag if strpos is offset | |
776 | wrt the start of the string. */ | |
05b4157f | 777 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 778 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d IZ |
779 | /* May be due to an implicit anchor of m{.*foo} */ |
780 | && !(prog->reganch & ROPT_IMPLICIT)) | |
781 | { | |
cad2e5aa JH |
782 | t = strpos; |
783 | goto find_anchor; | |
784 | } | |
a3621e74 | 785 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 786 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
e4584336 | 787 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 788 | ); |
2c2d71f5 | 789 | success_at_start: |
30944b6d | 790 | if (!(prog->reganch & ROPT_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
791 | && (do_utf8 ? ( |
792 | prog->check_utf8 /* Could be deleted already */ | |
793 | && --BmUSEFUL(prog->check_utf8) < 0 | |
794 | && (prog->check_utf8 == prog->float_utf8) | |
795 | ) : ( | |
796 | prog->check_substr /* Could be deleted already */ | |
797 | && --BmUSEFUL(prog->check_substr) < 0 | |
798 | && (prog->check_substr == prog->float_substr) | |
799 | ))) | |
66e933ab | 800 | { |
cad2e5aa | 801 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 802 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
803 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
804 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
805 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
806 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
807 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
808 | check = NULL; /* abort */ | |
cad2e5aa | 809 | s = strpos; |
3cf5c195 IZ |
810 | /* XXXX This is a remnant of the old implementation. It |
811 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 812 | other heuristics. */ |
cad2e5aa JH |
813 | prog->reganch &= ~RE_USE_INTUIT; |
814 | } | |
815 | else | |
816 | s = strpos; | |
817 | } | |
818 | ||
6eb5f6b9 JH |
819 | /* Last resort... */ |
820 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
821 | if (prog->regstclass) { | |
822 | /* minlen == 0 is possible if regstclass is \b or \B, | |
823 | and the fixed substr is ''$. | |
824 | Since minlen is already taken into account, s+1 is before strend; | |
825 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
826 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
827 | regstclass does not come from lookahead... */ | |
828 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
829 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
890ce7af | 830 | const U8* const str = (U8*)STRING(prog->regstclass); |
06b5626a | 831 | const int cl_l = (PL_regkind[(U8)OP(prog->regstclass)] == EXACT |
1aa99e6b | 832 | ? CHR_DIST(str+STR_LEN(prog->regstclass), str) |
66e933ab | 833 | : 1); |
a3b680e6 | 834 | const char * const endpos = (prog->anchored_substr || prog->anchored_utf8 || ml_anch) |
1aa99e6b | 835 | ? HOP3c(s, (prog->minlen ? cl_l : 0), strend) |
33b8afdf | 836 | : (prog->float_substr || prog->float_utf8 |
1aa99e6b IH |
837 | ? HOP3c(HOP3c(check_at, -start_shift, strbeg), |
838 | cl_l, strend) | |
839 | : strend); | |
6eb5f6b9 JH |
840 | |
841 | t = s; | |
3b0527fe | 842 | s = find_byclass(prog, prog->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
843 | if (!s) { |
844 | #ifdef DEBUGGING | |
cbbf8932 | 845 | const char *what = NULL; |
6eb5f6b9 JH |
846 | #endif |
847 | if (endpos == strend) { | |
a3621e74 | 848 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
849 | "Could not match STCLASS...\n") ); |
850 | goto fail; | |
851 | } | |
a3621e74 | 852 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 853 | "This position contradicts STCLASS...\n") ); |
653099ff GS |
854 | if ((prog->reganch & ROPT_ANCH) && !ml_anch) |
855 | goto fail; | |
6eb5f6b9 | 856 | /* Contradict one of substrings */ |
33b8afdf JH |
857 | if (prog->anchored_substr || prog->anchored_utf8) { |
858 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 859 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 860 | hop_and_restart: |
1aa99e6b | 861 | s = HOP3c(t, 1, strend); |
66e933ab GS |
862 | if (s + start_shift + end_shift > strend) { |
863 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 864 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
865 | "Could not match STCLASS...\n") ); |
866 | goto fail; | |
867 | } | |
5e39e1e5 HS |
868 | if (!check) |
869 | goto giveup; | |
a3621e74 | 870 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 871 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
872 | what, (long)(s + start_shift - i_strpos)) ); |
873 | goto restart; | |
874 | } | |
66e933ab | 875 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
876 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
877 | goto retry_floating_check; | |
878 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 879 | s = check_at; |
5e39e1e5 HS |
880 | if (!check) |
881 | goto giveup; | |
a3621e74 | 882 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 883 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
884 | (long)(other_last - i_strpos)) ); |
885 | goto do_other_anchored; | |
886 | } | |
60e71179 GS |
887 | /* Another way we could have checked stclass at the |
888 | current position only: */ | |
889 | if (ml_anch) { | |
890 | s = t = t + 1; | |
5e39e1e5 HS |
891 | if (!check) |
892 | goto giveup; | |
a3621e74 | 893 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 894 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 895 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 896 | goto try_at_offset; |
66e933ab | 897 | } |
33b8afdf | 898 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 899 | goto fail; |
6eb5f6b9 JH |
900 | /* Check is floating subtring. */ |
901 | retry_floating_check: | |
902 | t = check_at - start_shift; | |
a3621e74 | 903 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
904 | goto hop_and_restart; |
905 | } | |
b7953727 | 906 | if (t != s) { |
a3621e74 | 907 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 908 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
909 | (long)(t - i_strpos), (long)(s - i_strpos)) |
910 | ); | |
911 | } | |
912 | else { | |
a3621e74 | 913 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
914 | "Does not contradict STCLASS...\n"); |
915 | ); | |
916 | } | |
6eb5f6b9 | 917 | } |
5e39e1e5 | 918 | giveup: |
a3621e74 | 919 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
920 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
921 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 922 | return s; |
2c2d71f5 JH |
923 | |
924 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
925 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
926 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 927 | fail: |
a3621e74 | 928 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 929 | PL_colors[4], PL_colors[5])); |
bd61b366 | 930 | return NULL; |
cad2e5aa | 931 | } |
9661b544 | 932 | |
6eb5f6b9 | 933 | /* We know what class REx starts with. Try to find this position... */ |
3b0527fe DM |
934 | /* if reginfo is NULL, its a dryrun */ |
935 | ||
3c3eec57 | 936 | STATIC char * |
3b0527fe DM |
937 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, const char |
938 | *strend, const regmatch_info *reginfo) | |
a687059c | 939 | { |
27da23d5 | 940 | dVAR; |
1df70142 | 941 | const I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
6eb5f6b9 | 942 | char *m; |
d8093b23 | 943 | STRLEN ln; |
5dab1207 | 944 | STRLEN lnc; |
078c425b | 945 | register STRLEN uskip; |
d8093b23 G |
946 | unsigned int c1; |
947 | unsigned int c2; | |
6eb5f6b9 JH |
948 | char *e; |
949 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 950 | register const bool do_utf8 = PL_reg_match_utf8; |
cad2e5aa | 951 | |
6eb5f6b9 JH |
952 | /* We know what class it must start with. */ |
953 | switch (OP(c)) { | |
6eb5f6b9 | 954 | case ANYOF: |
388cc4de | 955 | if (do_utf8) { |
078c425b | 956 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
388cc4de HS |
957 | if ((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
958 | !UTF8_IS_INVARIANT((U8)s[0]) ? | |
32fc9b6a DM |
959 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
960 | REGINCLASS(prog, c, (U8*)s)) { | |
3b0527fe | 961 | if (tmp && (!reginfo || regtry(reginfo, s))) |
388cc4de HS |
962 | goto got_it; |
963 | else | |
964 | tmp = doevery; | |
965 | } | |
966 | else | |
967 | tmp = 1; | |
078c425b | 968 | s += uskip; |
388cc4de HS |
969 | } |
970 | } | |
971 | else { | |
972 | while (s < strend) { | |
973 | STRLEN skip = 1; | |
974 | ||
32fc9b6a | 975 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
976 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
977 | /* The assignment of 2 is intentional: | |
978 | * for the folded sharp s, the skip is 2. */ | |
979 | (skip = SHARP_S_SKIP))) { | |
3b0527fe | 980 | if (tmp && (!reginfo || regtry(reginfo, s))) |
388cc4de HS |
981 | goto got_it; |
982 | else | |
983 | tmp = doevery; | |
984 | } | |
985 | else | |
986 | tmp = 1; | |
987 | s += skip; | |
988 | } | |
a0d0e21e | 989 | } |
6eb5f6b9 | 990 | break; |
f33976b4 DB |
991 | case CANY: |
992 | while (s < strend) { | |
3b0527fe | 993 | if (tmp && (!reginfo || regtry(reginfo, s))) |
f33976b4 DB |
994 | goto got_it; |
995 | else | |
996 | tmp = doevery; | |
997 | s++; | |
998 | } | |
999 | break; | |
6eb5f6b9 | 1000 | case EXACTF: |
5dab1207 NIS |
1001 | m = STRING(c); |
1002 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1003 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1004 | if (UTF) { |
a2a2844f | 1005 | STRLEN ulen1, ulen2; |
5dab1207 | 1006 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1007 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1008 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1009 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1010 | |
1011 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1012 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1013 | ||
89ebb4a3 | 1014 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1015 | 0, uniflags); |
89ebb4a3 | 1016 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1017 | 0, uniflags); |
5dab1207 NIS |
1018 | lnc = 0; |
1019 | while (sm < ((U8 *) m + ln)) { | |
1020 | lnc++; | |
1021 | sm += UTF8SKIP(sm); | |
1022 | } | |
1aa99e6b IH |
1023 | } |
1024 | else { | |
1025 | c1 = *(U8*)m; | |
1026 | c2 = PL_fold[c1]; | |
1027 | } | |
6eb5f6b9 JH |
1028 | goto do_exactf; |
1029 | case EXACTFL: | |
5dab1207 NIS |
1030 | m = STRING(c); |
1031 | ln = STR_LEN(c); | |
1032 | lnc = (I32) ln; | |
d8093b23 | 1033 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1034 | c2 = PL_fold_locale[c1]; |
1035 | do_exactf: | |
db12adc6 | 1036 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1037 | |
3b0527fe | 1038 | if (!reginfo && e < s) |
6eb5f6b9 | 1039 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1040 | |
60a8b682 JH |
1041 | /* The idea in the EXACTF* cases is to first find the |
1042 | * first character of the EXACTF* node and then, if | |
1043 | * necessary, case-insensitively compare the full | |
1044 | * text of the node. The c1 and c2 are the first | |
1045 | * characters (though in Unicode it gets a bit | |
1046 | * more complicated because there are more cases | |
7f16dd3d JH |
1047 | * than just upper and lower: one needs to use |
1048 | * the so-called folding case for case-insensitive | |
1049 | * matching (called "loose matching" in Unicode). | |
1050 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1051 | |
1aa99e6b | 1052 | if (do_utf8) { |
575cac57 | 1053 | UV c, f; |
89ebb4a3 | 1054 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1055 | STRLEN len, foldlen; |
4ad0818d | 1056 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1057 | if (c1 == c2) { |
5dab1207 NIS |
1058 | /* Upper and lower of 1st char are equal - |
1059 | * probably not a "letter". */ | |
1aa99e6b | 1060 | while (s <= e) { |
89ebb4a3 | 1061 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1062 | uniflags); |
80aecb99 JH |
1063 | if ( c == c1 |
1064 | && (ln == len || | |
66423254 | 1065 | ibcmp_utf8(s, (char **)0, 0, do_utf8, |
eb160463 | 1066 | m, (char **)0, ln, (bool)UTF)) |
3b0527fe | 1067 | && (!reginfo || regtry(reginfo, s)) ) |
1aa99e6b | 1068 | goto got_it; |
80aecb99 | 1069 | else { |
1df70142 | 1070 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; |
80aecb99 | 1071 | uvchr_to_utf8(tmpbuf, c); |
ac7e0132 | 1072 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); |
80aecb99 JH |
1073 | if ( f != c |
1074 | && (f == c1 || f == c2) | |
1075 | && (ln == foldlen || | |
66423254 JH |
1076 | !ibcmp_utf8((char *) foldbuf, |
1077 | (char **)0, foldlen, do_utf8, | |
d07ddd77 | 1078 | m, |
eb160463 | 1079 | (char **)0, ln, (bool)UTF)) |
3b0527fe | 1080 | && (!reginfo || regtry(reginfo, s)) ) |
80aecb99 JH |
1081 | goto got_it; |
1082 | } | |
1aa99e6b IH |
1083 | s += len; |
1084 | } | |
09091399 JH |
1085 | } |
1086 | else { | |
1aa99e6b | 1087 | while (s <= e) { |
89ebb4a3 | 1088 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1089 | uniflags); |
80aecb99 | 1090 | |
60a8b682 | 1091 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1092 | * Note that not all the possible combinations |
1093 | * are handled here: some of them are handled | |
1094 | * by the standard folding rules, and some of | |
1095 | * them (the character class or ANYOF cases) | |
1096 | * are handled during compiletime in | |
1097 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1098 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1099 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1100 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 JH |
1101 | |
1102 | if ( (c == c1 || c == c2) | |
1103 | && (ln == len || | |
66423254 | 1104 | ibcmp_utf8(s, (char **)0, 0, do_utf8, |
eb160463 | 1105 | m, (char **)0, ln, (bool)UTF)) |
3b0527fe | 1106 | && (!reginfo || regtry(reginfo, s)) ) |
1aa99e6b | 1107 | goto got_it; |
80aecb99 | 1108 | else { |
1df70142 | 1109 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; |
80aecb99 | 1110 | uvchr_to_utf8(tmpbuf, c); |
ac7e0132 | 1111 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); |
80aecb99 JH |
1112 | if ( f != c |
1113 | && (f == c1 || f == c2) | |
1114 | && (ln == foldlen || | |
a6872d42 | 1115 | !ibcmp_utf8((char *) foldbuf, |
66423254 | 1116 | (char **)0, foldlen, do_utf8, |
d07ddd77 | 1117 | m, |
eb160463 | 1118 | (char **)0, ln, (bool)UTF)) |
3b0527fe | 1119 | && (!reginfo || regtry(reginfo, s)) ) |
80aecb99 JH |
1120 | goto got_it; |
1121 | } | |
1aa99e6b IH |
1122 | s += len; |
1123 | } | |
09091399 | 1124 | } |
1aa99e6b IH |
1125 | } |
1126 | else { | |
1127 | if (c1 == c2) | |
1128 | while (s <= e) { | |
1129 | if ( *(U8*)s == c1 | |
1130 | && (ln == 1 || !(OP(c) == EXACTF | |
1131 | ? ibcmp(s, m, ln) | |
1132 | : ibcmp_locale(s, m, ln))) | |
3b0527fe | 1133 | && (!reginfo || regtry(reginfo, s)) ) |
1aa99e6b IH |
1134 | goto got_it; |
1135 | s++; | |
1136 | } | |
1137 | else | |
1138 | while (s <= e) { | |
1139 | if ( (*(U8*)s == c1 || *(U8*)s == c2) | |
1140 | && (ln == 1 || !(OP(c) == EXACTF | |
1141 | ? ibcmp(s, m, ln) | |
1142 | : ibcmp_locale(s, m, ln))) | |
3b0527fe | 1143 | && (!reginfo || regtry(reginfo, s)) ) |
1aa99e6b IH |
1144 | goto got_it; |
1145 | s++; | |
1146 | } | |
b3c9acc1 IZ |
1147 | } |
1148 | break; | |
bbce6d69 | 1149 | case BOUNDL: |
3280af22 | 1150 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1151 | /* FALL THROUGH */ |
a0d0e21e | 1152 | case BOUND: |
ffc61ed2 | 1153 | if (do_utf8) { |
12d33761 | 1154 | if (s == PL_bostr) |
ffc61ed2 JH |
1155 | tmp = '\n'; |
1156 | else { | |
6136c704 | 1157 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1158 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1159 | } |
1160 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1161 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1162 | LOAD_UTF8_CHARCLASS_ALNUM(); |
078c425b | 1163 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1164 | if (tmp == !(OP(c) == BOUND ? |
3568d838 | 1165 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1166 | isALNUM_LC_utf8((U8*)s))) |
1167 | { | |
1168 | tmp = !tmp; | |
3b0527fe | 1169 | if ((!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1170 | goto got_it; |
1171 | } | |
078c425b | 1172 | s += uskip; |
a687059c | 1173 | } |
a0d0e21e | 1174 | } |
667bb95a | 1175 | else { |
12d33761 | 1176 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1177 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
1178 | while (s < strend) { | |
1179 | if (tmp == | |
1180 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1181 | tmp = !tmp; | |
3b0527fe | 1182 | if ((!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1183 | goto got_it; |
1184 | } | |
1185 | s++; | |
a0ed51b3 | 1186 | } |
a0ed51b3 | 1187 | } |
3b0527fe | 1188 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1189 | goto got_it; |
1190 | break; | |
bbce6d69 | 1191 | case NBOUNDL: |
3280af22 | 1192 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1193 | /* FALL THROUGH */ |
a0d0e21e | 1194 | case NBOUND: |
ffc61ed2 | 1195 | if (do_utf8) { |
12d33761 | 1196 | if (s == PL_bostr) |
ffc61ed2 JH |
1197 | tmp = '\n'; |
1198 | else { | |
6136c704 | 1199 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1200 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1201 | } |
1202 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1203 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1204 | LOAD_UTF8_CHARCLASS_ALNUM(); |
078c425b | 1205 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1206 | if (tmp == !(OP(c) == NBOUND ? |
3568d838 | 1207 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1208 | isALNUM_LC_utf8((U8*)s))) |
1209 | tmp = !tmp; | |
3b0527fe | 1210 | else if ((!reginfo || regtry(reginfo, s))) |
ffc61ed2 | 1211 | goto got_it; |
078c425b | 1212 | s += uskip; |
ffc61ed2 | 1213 | } |
a0d0e21e | 1214 | } |
667bb95a | 1215 | else { |
12d33761 | 1216 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1217 | tmp = ((OP(c) == NBOUND ? |
1218 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
1219 | while (s < strend) { | |
1220 | if (tmp == | |
1221 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1222 | tmp = !tmp; | |
3b0527fe | 1223 | else if ((!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1224 | goto got_it; |
1225 | s++; | |
1226 | } | |
a0ed51b3 | 1227 | } |
3b0527fe | 1228 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, s))) |
a0ed51b3 LW |
1229 | goto got_it; |
1230 | break; | |
a0d0e21e | 1231 | case ALNUM: |
ffc61ed2 | 1232 | if (do_utf8) { |
1a4fad37 | 1233 | LOAD_UTF8_CHARCLASS_ALNUM(); |
078c425b | 1234 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1235 | if (swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { |
3b0527fe | 1236 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1237 | goto got_it; |
1238 | else | |
1239 | tmp = doevery; | |
1240 | } | |
bbce6d69 | 1241 | else |
ffc61ed2 | 1242 | tmp = 1; |
078c425b | 1243 | s += uskip; |
bbce6d69 | 1244 | } |
bbce6d69 | 1245 | } |
ffc61ed2 JH |
1246 | else { |
1247 | while (s < strend) { | |
1248 | if (isALNUM(*s)) { | |
3b0527fe | 1249 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1250 | goto got_it; |
1251 | else | |
1252 | tmp = doevery; | |
1253 | } | |
a0ed51b3 | 1254 | else |
ffc61ed2 JH |
1255 | tmp = 1; |
1256 | s++; | |
a0ed51b3 | 1257 | } |
a0ed51b3 LW |
1258 | } |
1259 | break; | |
bbce6d69 | 1260 | case ALNUML: |
3280af22 | 1261 | PL_reg_flags |= RF_tainted; |
ffc61ed2 | 1262 | if (do_utf8) { |
078c425b | 1263 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1264 | if (isALNUM_LC_utf8((U8*)s)) { |
3b0527fe | 1265 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1266 | goto got_it; |
1267 | else | |
1268 | tmp = doevery; | |
1269 | } | |
a687059c | 1270 | else |
ffc61ed2 | 1271 | tmp = 1; |
078c425b | 1272 | s += uskip; |
a0d0e21e | 1273 | } |
a0d0e21e | 1274 | } |
ffc61ed2 JH |
1275 | else { |
1276 | while (s < strend) { | |
1277 | if (isALNUM_LC(*s)) { | |
3b0527fe | 1278 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1279 | goto got_it; |
1280 | else | |
1281 | tmp = doevery; | |
1282 | } | |
a0ed51b3 | 1283 | else |
ffc61ed2 JH |
1284 | tmp = 1; |
1285 | s++; | |
a0ed51b3 | 1286 | } |
a0ed51b3 LW |
1287 | } |
1288 | break; | |
a0d0e21e | 1289 | case NALNUM: |
ffc61ed2 | 1290 | if (do_utf8) { |
1a4fad37 | 1291 | LOAD_UTF8_CHARCLASS_ALNUM(); |
078c425b | 1292 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1293 | if (!swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8)) { |
3b0527fe | 1294 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1295 | goto got_it; |
1296 | else | |
1297 | tmp = doevery; | |
1298 | } | |
bbce6d69 | 1299 | else |
ffc61ed2 | 1300 | tmp = 1; |
078c425b | 1301 | s += uskip; |
bbce6d69 | 1302 | } |
bbce6d69 | 1303 | } |
ffc61ed2 JH |
1304 | else { |
1305 | while (s < strend) { | |
1306 | if (!isALNUM(*s)) { | |
3b0527fe | 1307 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1308 | goto got_it; |
1309 | else | |
1310 | tmp = doevery; | |
1311 | } | |
a0ed51b3 | 1312 | else |
ffc61ed2 JH |
1313 | tmp = 1; |
1314 | s++; | |
a0ed51b3 | 1315 | } |
a0ed51b3 LW |
1316 | } |
1317 | break; | |
bbce6d69 | 1318 | case NALNUML: |
3280af22 | 1319 | PL_reg_flags |= RF_tainted; |
ffc61ed2 | 1320 | if (do_utf8) { |
078c425b | 1321 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1322 | if (!isALNUM_LC_utf8((U8*)s)) { |
3b0527fe | 1323 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1324 | goto got_it; |
1325 | else | |
1326 | tmp = doevery; | |
1327 | } | |
a687059c | 1328 | else |
ffc61ed2 | 1329 | tmp = 1; |
078c425b | 1330 | s += uskip; |
a687059c | 1331 | } |
a0d0e21e | 1332 | } |
ffc61ed2 JH |
1333 | else { |
1334 | while (s < strend) { | |
1335 | if (!isALNUM_LC(*s)) { | |
3b0527fe | 1336 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1337 | goto got_it; |
1338 | else | |
1339 | tmp = doevery; | |
1340 | } | |
a0ed51b3 | 1341 | else |
ffc61ed2 JH |
1342 | tmp = 1; |
1343 | s++; | |
a0ed51b3 | 1344 | } |
a0ed51b3 LW |
1345 | } |
1346 | break; | |
a0d0e21e | 1347 | case SPACE: |
ffc61ed2 | 1348 | if (do_utf8) { |
1a4fad37 | 1349 | LOAD_UTF8_CHARCLASS_SPACE(); |
078c425b | 1350 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1351 | if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)) { |
3b0527fe | 1352 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1353 | goto got_it; |
1354 | else | |
1355 | tmp = doevery; | |
1356 | } | |
a0d0e21e | 1357 | else |
ffc61ed2 | 1358 | tmp = 1; |
078c425b | 1359 | s += uskip; |
2304df62 | 1360 | } |
a0d0e21e | 1361 | } |
ffc61ed2 JH |
1362 | else { |
1363 | while (s < strend) { | |
1364 | if (isSPACE(*s)) { | |
3b0527fe | 1365 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1366 | goto got_it; |
1367 | else | |
1368 | tmp = doevery; | |
1369 | } | |
a0ed51b3 | 1370 | else |
ffc61ed2 JH |
1371 | tmp = 1; |
1372 | s++; | |
a0ed51b3 | 1373 | } |
a0ed51b3 LW |
1374 | } |
1375 | break; | |
bbce6d69 | 1376 | case SPACEL: |
3280af22 | 1377 | PL_reg_flags |= RF_tainted; |
ffc61ed2 | 1378 | if (do_utf8) { |
078c425b | 1379 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1380 | if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { |
3b0527fe | 1381 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1382 | goto got_it; |
1383 | else | |
1384 | tmp = doevery; | |
1385 | } | |
bbce6d69 | 1386 | else |
ffc61ed2 | 1387 | tmp = 1; |
078c425b | 1388 | s += uskip; |
bbce6d69 | 1389 | } |
bbce6d69 | 1390 | } |
ffc61ed2 JH |
1391 | else { |
1392 | while (s < strend) { | |
1393 | if (isSPACE_LC(*s)) { | |
3b0527fe | 1394 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1395 | goto got_it; |
1396 | else | |
1397 | tmp = doevery; | |
1398 | } | |
a0ed51b3 | 1399 | else |
ffc61ed2 JH |
1400 | tmp = 1; |
1401 | s++; | |
a0ed51b3 | 1402 | } |
a0ed51b3 LW |
1403 | } |
1404 | break; | |
a0d0e21e | 1405 | case NSPACE: |
ffc61ed2 | 1406 | if (do_utf8) { |
1a4fad37 | 1407 | LOAD_UTF8_CHARCLASS_SPACE(); |
078c425b | 1408 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1409 | if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8))) { |
3b0527fe | 1410 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1411 | goto got_it; |
1412 | else | |
1413 | tmp = doevery; | |
1414 | } | |
a0d0e21e | 1415 | else |
ffc61ed2 | 1416 | tmp = 1; |
078c425b | 1417 | s += uskip; |
a687059c | 1418 | } |
a0d0e21e | 1419 | } |
ffc61ed2 JH |
1420 | else { |
1421 | while (s < strend) { | |
1422 | if (!isSPACE(*s)) { | |
3b0527fe | 1423 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1424 | goto got_it; |
1425 | else | |
1426 | tmp = doevery; | |
1427 | } | |
a0ed51b3 | 1428 | else |
ffc61ed2 JH |
1429 | tmp = 1; |
1430 | s++; | |
a0ed51b3 | 1431 | } |
a0ed51b3 LW |
1432 | } |
1433 | break; | |
bbce6d69 | 1434 | case NSPACEL: |
3280af22 | 1435 | PL_reg_flags |= RF_tainted; |
ffc61ed2 | 1436 | if (do_utf8) { |
078c425b | 1437 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1438 | if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { |
3b0527fe | 1439 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1440 | goto got_it; |
1441 | else | |
1442 | tmp = doevery; | |
1443 | } | |
bbce6d69 | 1444 | else |
ffc61ed2 | 1445 | tmp = 1; |
078c425b | 1446 | s += uskip; |
bbce6d69 | 1447 | } |
bbce6d69 | 1448 | } |
ffc61ed2 JH |
1449 | else { |
1450 | while (s < strend) { | |
1451 | if (!isSPACE_LC(*s)) { | |
3b0527fe | 1452 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1453 | goto got_it; |
1454 | else | |
1455 | tmp = doevery; | |
1456 | } | |
a0ed51b3 | 1457 | else |
ffc61ed2 JH |
1458 | tmp = 1; |
1459 | s++; | |
a0ed51b3 | 1460 | } |
a0ed51b3 LW |
1461 | } |
1462 | break; | |
a0d0e21e | 1463 | case DIGIT: |
ffc61ed2 | 1464 | if (do_utf8) { |
1a4fad37 | 1465 | LOAD_UTF8_CHARCLASS_DIGIT(); |
078c425b | 1466 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1467 | if (swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { |
3b0527fe | 1468 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1469 | goto got_it; |
1470 | else | |
1471 | tmp = doevery; | |
1472 | } | |
a0d0e21e | 1473 | else |
ffc61ed2 | 1474 | tmp = 1; |
078c425b | 1475 | s += uskip; |
2b69d0c2 | 1476 | } |
a0d0e21e | 1477 | } |
ffc61ed2 JH |
1478 | else { |
1479 | while (s < strend) { | |
1480 | if (isDIGIT(*s)) { | |
3b0527fe | 1481 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1482 | goto got_it; |
1483 | else | |
1484 | tmp = doevery; | |
1485 | } | |
a0ed51b3 | 1486 | else |
ffc61ed2 JH |
1487 | tmp = 1; |
1488 | s++; | |
a0ed51b3 | 1489 | } |
a0ed51b3 LW |
1490 | } |
1491 | break; | |
b8c5462f JH |
1492 | case DIGITL: |
1493 | PL_reg_flags |= RF_tainted; | |
ffc61ed2 | 1494 | if (do_utf8) { |
078c425b | 1495 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1496 | if (isDIGIT_LC_utf8((U8*)s)) { |
3b0527fe | 1497 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1498 | goto got_it; |
1499 | else | |
1500 | tmp = doevery; | |
1501 | } | |
b8c5462f | 1502 | else |
ffc61ed2 | 1503 | tmp = 1; |
078c425b | 1504 | s += uskip; |
b8c5462f | 1505 | } |
b8c5462f | 1506 | } |
ffc61ed2 JH |
1507 | else { |
1508 | while (s < strend) { | |
1509 | if (isDIGIT_LC(*s)) { | |
3b0527fe | 1510 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1511 | goto got_it; |
1512 | else | |
1513 | tmp = doevery; | |
1514 | } | |
b8c5462f | 1515 | else |
ffc61ed2 JH |
1516 | tmp = 1; |
1517 | s++; | |
b8c5462f | 1518 | } |
b8c5462f JH |
1519 | } |
1520 | break; | |
a0d0e21e | 1521 | case NDIGIT: |
ffc61ed2 | 1522 | if (do_utf8) { |
1a4fad37 | 1523 | LOAD_UTF8_CHARCLASS_DIGIT(); |
078c425b | 1524 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
3568d838 | 1525 | if (!swash_fetch(PL_utf8_digit,(U8*)s, do_utf8)) { |
3b0527fe | 1526 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1527 | goto got_it; |
1528 | else | |
1529 | tmp = doevery; | |
1530 | } | |
a0d0e21e | 1531 | else |
ffc61ed2 | 1532 | tmp = 1; |
078c425b | 1533 | s += uskip; |
a687059c | 1534 | } |
a0d0e21e | 1535 | } |
ffc61ed2 JH |
1536 | else { |
1537 | while (s < strend) { | |
1538 | if (!isDIGIT(*s)) { | |
3b0527fe | 1539 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1540 | goto got_it; |
1541 | else | |
1542 | tmp = doevery; | |
1543 | } | |
a0ed51b3 | 1544 | else |
ffc61ed2 JH |
1545 | tmp = 1; |
1546 | s++; | |
a0ed51b3 | 1547 | } |
a0ed51b3 LW |
1548 | } |
1549 | break; | |
b8c5462f JH |
1550 | case NDIGITL: |
1551 | PL_reg_flags |= RF_tainted; | |
ffc61ed2 | 1552 | if (do_utf8) { |
078c425b | 1553 | while (s + (uskip = UTF8SKIP(s)) <= strend) { |
ffc61ed2 | 1554 | if (!isDIGIT_LC_utf8((U8*)s)) { |
3b0527fe | 1555 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1556 | goto got_it; |
1557 | else | |
1558 | tmp = doevery; | |
1559 | } | |
b8c5462f | 1560 | else |
ffc61ed2 | 1561 | tmp = 1; |
078c425b | 1562 | s += uskip; |
b8c5462f | 1563 | } |
a0ed51b3 | 1564 | } |
ffc61ed2 JH |
1565 | else { |
1566 | while (s < strend) { | |
1567 | if (!isDIGIT_LC(*s)) { | |
3b0527fe | 1568 | if (tmp && (!reginfo || regtry(reginfo, s))) |
ffc61ed2 JH |
1569 | goto got_it; |
1570 | else | |
1571 | tmp = doevery; | |
1572 | } | |
cf93c79d | 1573 | else |
ffc61ed2 JH |
1574 | tmp = 1; |
1575 | s++; | |
b8c5462f | 1576 | } |
b8c5462f JH |
1577 | } |
1578 | break; | |
b3c9acc1 | 1579 | default: |
3c3eec57 GS |
1580 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1581 | break; | |
d6a28714 | 1582 | } |
6eb5f6b9 JH |
1583 | return 0; |
1584 | got_it: | |
1585 | return s; | |
1586 | } | |
1587 | ||
1588 | /* | |
1589 | - regexec_flags - match a regexp against a string | |
1590 | */ | |
1591 | I32 | |
1592 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1593 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1594 | /* strend: pointer to null at end of string */ | |
1595 | /* strbeg: real beginning of string */ | |
1596 | /* minend: end of match must be >=minend after stringarg. */ | |
1597 | /* data: May be used for some additional optimizations. */ | |
1598 | /* nosave: For optimizations. */ | |
1599 | { | |
97aff369 | 1600 | dVAR; |
6eb5f6b9 JH |
1601 | register char *s; |
1602 | register regnode *c; | |
1603 | register char *startpos = stringarg; | |
6eb5f6b9 JH |
1604 | I32 minlen; /* must match at least this many chars */ |
1605 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1606 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1607 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1608 | char *scream_olds = NULL; |
6eb5f6b9 | 1609 | SV* oreplsv = GvSV(PL_replgv); |
1df70142 | 1610 | const bool do_utf8 = DO_UTF8(sv); |
2757e526 | 1611 | I32 multiline; |
2a782b5b | 1612 | #ifdef DEBUGGING |
2757e526 JH |
1613 | SV* dsv0; |
1614 | SV* dsv1; | |
2a782b5b | 1615 | #endif |
3b0527fe | 1616 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1617 | |
1618 | GET_RE_DEBUG_FLAGS_DECL; | |
1619 | ||
9d4ba2ae | 1620 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1621 | |
1622 | /* Be paranoid... */ | |
1623 | if (prog == NULL || startpos == NULL) { | |
1624 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1625 | return 0; | |
1626 | } | |
1627 | ||
2757e526 | 1628 | multiline = prog->reganch & PMf_MULTILINE; |
3b0527fe | 1629 | reginfo.prog = prog; |
2757e526 JH |
1630 | |
1631 | #ifdef DEBUGGING | |
1632 | dsv0 = PERL_DEBUG_PAD_ZERO(0); | |
1633 | dsv1 = PERL_DEBUG_PAD_ZERO(1); | |
1634 | #endif | |
1635 | ||
bac06658 JH |
1636 | RX_MATCH_UTF8_set(prog, do_utf8); |
1637 | ||
6eb5f6b9 | 1638 | minlen = prog->minlen; |
61a36c01 | 1639 | if (strend - startpos < minlen) { |
a3621e74 | 1640 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1641 | "String too short [regexec_flags]...\n")); |
1642 | goto phooey; | |
1aa99e6b | 1643 | } |
6eb5f6b9 | 1644 | |
6eb5f6b9 JH |
1645 | /* Check validity of program. */ |
1646 | if (UCHARAT(prog->program) != REG_MAGIC) { | |
1647 | Perl_croak(aTHX_ "corrupted regexp program"); | |
1648 | } | |
1649 | ||
1650 | PL_reg_flags = 0; | |
1651 | PL_reg_eval_set = 0; | |
1652 | PL_reg_maxiter = 0; | |
1653 | ||
1654 | if (prog->reganch & ROPT_UTF8) | |
1655 | PL_reg_flags |= RF_utf8; | |
1656 | ||
1657 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1658 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1659 | PL_bostr = strbeg; |
3b0527fe | 1660 | reginfo.sv = sv; |
6eb5f6b9 JH |
1661 | |
1662 | /* Mark end of line for $ (and such) */ | |
1663 | PL_regeol = strend; | |
1664 | ||
1665 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1666 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1667 | |
6eb5f6b9 JH |
1668 | /* If there is a "must appear" string, look for it. */ |
1669 | s = startpos; | |
1670 | ||
3b0527fe | 1671 | if (prog->reganch & ROPT_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1672 | MAGIC *mg; |
1673 | ||
1674 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
3b0527fe | 1675 | reginfo.ganch = startpos; |
6eb5f6b9 JH |
1676 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1677 | && SvMAGIC(sv) | |
14befaf4 DM |
1678 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1679 | && mg->mg_len >= 0) { | |
3b0527fe | 1680 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
6eb5f6b9 | 1681 | if (prog->reganch & ROPT_ANCH_GPOS) { |
3b0527fe | 1682 | if (s > reginfo.ganch) |
6eb5f6b9 | 1683 | goto phooey; |
3b0527fe | 1684 | s = reginfo.ganch; |
6eb5f6b9 JH |
1685 | } |
1686 | } | |
1687 | else /* pos() not defined */ | |
3b0527fe | 1688 | reginfo.ganch = strbeg; |
6eb5f6b9 JH |
1689 | } |
1690 | ||
a0714e2c | 1691 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1692 | re_scream_pos_data d; |
1693 | ||
1694 | d.scream_olds = &scream_olds; | |
1695 | d.scream_pos = &scream_pos; | |
1696 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1697 | if (!s) { |
a3621e74 | 1698 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1699 | goto phooey; /* not present */ |
3fa9c3d7 | 1700 | } |
6eb5f6b9 JH |
1701 | } |
1702 | ||
a3621e74 | 1703 | DEBUG_EXECUTE_r({ |
1df70142 AL |
1704 | const char * const s0 = UTF |
1705 | ? pv_uni_display(dsv0, (U8*)prog->precomp, prog->prelen, 60, | |
1706 | UNI_DISPLAY_REGEX) | |
1707 | : prog->precomp; | |
1708 | const int len0 = UTF ? SvCUR(dsv0) : prog->prelen; | |
1709 | const char * const s1 = do_utf8 ? sv_uni_display(dsv1, sv, 60, | |
c728cb41 | 1710 | UNI_DISPLAY_REGEX) : startpos; |
1df70142 | 1711 | const int len1 = do_utf8 ? SvCUR(dsv1) : strend - startpos; |
2a782b5b JH |
1712 | if (!PL_colorset) |
1713 | reginitcolors(); | |
1714 | PerlIO_printf(Perl_debug_log, | |
a0288114 | 1715 | "%sMatching REx%s \"%s%*.*s%s%s\" against \"%s%.*s%s%s\"\n", |
e4584336 | 1716 | PL_colors[4], PL_colors[5], PL_colors[0], |
9e55ce06 | 1717 | len0, len0, s0, |
2a782b5b | 1718 | PL_colors[1], |
9e55ce06 | 1719 | len0 > 60 ? "..." : "", |
2a782b5b | 1720 | PL_colors[0], |
9e55ce06 JH |
1721 | (int)(len1 > 60 ? 60 : len1), |
1722 | s1, PL_colors[1], | |
1723 | (len1 > 60 ? "..." : "") | |
2a782b5b JH |
1724 | ); |
1725 | }); | |
6eb5f6b9 JH |
1726 | |
1727 | /* Simplest case: anchored match need be tried only once. */ | |
1728 | /* [unless only anchor is BOL and multiline is set] */ | |
1729 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { | |
3b0527fe | 1730 | if (s == startpos && regtry(®info, startpos)) |
6eb5f6b9 | 1731 | goto got_it; |
7fba1cd6 | 1732 | else if (multiline || (prog->reganch & ROPT_IMPLICIT) |
6eb5f6b9 JH |
1733 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
1734 | { | |
1735 | char *end; | |
1736 | ||
1737 | if (minlen) | |
1738 | dontbother = minlen - 1; | |
1aa99e6b | 1739 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1740 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1741 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1742 | if (s == startpos) |
1743 | goto after_try; | |
1744 | while (1) { | |
3b0527fe | 1745 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1746 | goto got_it; |
1747 | after_try: | |
1748 | if (s >= end) | |
1749 | goto phooey; | |
1750 | if (prog->reganch & RE_USE_INTUIT) { | |
1751 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); | |
1752 | if (!s) | |
1753 | goto phooey; | |
1754 | } | |
1755 | else | |
1756 | s++; | |
1757 | } | |
1758 | } else { | |
1759 | if (s > startpos) | |
1760 | s--; | |
1761 | while (s < end) { | |
1762 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
3b0527fe | 1763 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1764 | goto got_it; |
1765 | } | |
1766 | } | |
1767 | } | |
1768 | } | |
1769 | goto phooey; | |
1770 | } else if (prog->reganch & ROPT_ANCH_GPOS) { | |
3b0527fe | 1771 | if (regtry(®info, reginfo.ganch)) |
6eb5f6b9 JH |
1772 | goto got_it; |
1773 | goto phooey; | |
1774 | } | |
1775 | ||
1776 | /* Messy cases: unanchored match. */ | |
33b8afdf | 1777 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->reganch & ROPT_SKIP) { |
6eb5f6b9 JH |
1778 | /* we have /x+whatever/ */ |
1779 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1780 | char ch; |
bf93d4cc GS |
1781 | #ifdef DEBUGGING |
1782 | int did_match = 0; | |
1783 | #endif | |
33b8afdf JH |
1784 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1785 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1786 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1787 | |
1aa99e6b | 1788 | if (do_utf8) { |
6eb5f6b9 JH |
1789 | while (s < strend) { |
1790 | if (*s == ch) { | |
a3621e74 | 1791 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1792 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1793 | s += UTF8SKIP(s); |
1794 | while (s < strend && *s == ch) | |
1795 | s += UTF8SKIP(s); | |
1796 | } | |
1797 | s += UTF8SKIP(s); | |
1798 | } | |
1799 | } | |
1800 | else { | |
1801 | while (s < strend) { | |
1802 | if (*s == ch) { | |
a3621e74 | 1803 | DEBUG_EXECUTE_r( did_match = 1 ); |
3b0527fe | 1804 | if (regtry(®info, s)) goto got_it; |
6eb5f6b9 JH |
1805 | s++; |
1806 | while (s < strend && *s == ch) | |
1807 | s++; | |
1808 | } | |
1809 | s++; | |
1810 | } | |
1811 | } | |
a3621e74 | 1812 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1813 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1814 | "Did not find anchored character...\n") |
1815 | ); | |
6eb5f6b9 | 1816 | } |
a0714e2c SS |
1817 | else if (prog->anchored_substr != NULL |
1818 | || prog->anchored_utf8 != NULL | |
1819 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1820 | && prog->float_max_offset < strend - s)) { |
1821 | SV *must; | |
1822 | I32 back_max; | |
1823 | I32 back_min; | |
1824 | char *last; | |
6eb5f6b9 | 1825 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1826 | #ifdef DEBUGGING |
1827 | int did_match = 0; | |
1828 | #endif | |
33b8afdf JH |
1829 | if (prog->anchored_substr || prog->anchored_utf8) { |
1830 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1831 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1832 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1833 | back_max = back_min = prog->anchored_offset; | |
1834 | } else { | |
1835 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1836 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1837 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1838 | back_max = prog->float_max_offset; | |
1839 | back_min = prog->float_min_offset; | |
1840 | } | |
1841 | if (must == &PL_sv_undef) | |
1842 | /* could not downgrade utf8 check substring, so must fail */ | |
1843 | goto phooey; | |
1844 | ||
1845 | last = HOP3c(strend, /* Cannot start after this */ | |
1846 | -(I32)(CHR_SVLEN(must) | |
1847 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
6eb5f6b9 JH |
1848 | |
1849 | if (s > PL_bostr) | |
1850 | last1 = HOPc(s, -1); | |
1851 | else | |
1852 | last1 = s - 1; /* bogus */ | |
1853 | ||
a0288114 | 1854 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1855 | check_substr==must. */ |
1856 | scream_pos = -1; | |
1857 | dontbother = end_shift; | |
1858 | strend = HOPc(strend, -dontbother); | |
1859 | while ( (s <= last) && | |
9041c2e3 | 1860 | ((flags & REXEC_SCREAM) |
1aa99e6b | 1861 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, strend) - strbeg, |
6eb5f6b9 | 1862 | end_shift, &scream_pos, 0)) |
1aa99e6b | 1863 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, strend), |
9041c2e3 | 1864 | (unsigned char*)strend, must, |
7fba1cd6 | 1865 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1866 | /* we may be pointing at the wrong string */ |
1867 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1868 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1869 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1870 | if (HOPc(s, -back_max) > last1) { |
1871 | last1 = HOPc(s, -back_min); | |
1872 | s = HOPc(s, -back_max); | |
1873 | } | |
1874 | else { | |
52657f30 | 1875 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1876 | |
1877 | last1 = HOPc(s, -back_min); | |
52657f30 | 1878 | s = t; |
6eb5f6b9 | 1879 | } |
1aa99e6b | 1880 | if (do_utf8) { |
6eb5f6b9 | 1881 | while (s <= last1) { |
3b0527fe | 1882 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1883 | goto got_it; |
1884 | s += UTF8SKIP(s); | |
1885 | } | |
1886 | } | |
1887 | else { | |
1888 | while (s <= last1) { | |
3b0527fe | 1889 | if (regtry(®info, s)) |
6eb5f6b9 JH |
1890 | goto got_it; |
1891 | s++; | |
1892 | } | |
1893 | } | |
1894 | } | |
a3621e74 | 1895 | DEBUG_EXECUTE_r(if (!did_match) |
b7953727 | 1896 | PerlIO_printf(Perl_debug_log, |
a0288114 | 1897 | "Did not find %s substr \"%s%.*s%s\"%s...\n", |
33b8afdf | 1898 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc GS |
1899 | ? "anchored" : "floating"), |
1900 | PL_colors[0], | |
1901 | (int)(SvCUR(must) - (SvTAIL(must)!=0)), | |
3f7c398e | 1902 | SvPVX_const(must), |
b7953727 JH |
1903 | PL_colors[1], (SvTAIL(must) ? "$" : "")) |
1904 | ); | |
6eb5f6b9 JH |
1905 | goto phooey; |
1906 | } | |
155aba94 | 1907 | else if ((c = prog->regstclass)) { |
f14c76ed RGS |
1908 | if (minlen) { |
1909 | I32 op = (U8)OP(prog->regstclass); | |
66e933ab | 1910 | /* don't bother with what can't match */ |
f14c76ed RGS |
1911 | if (PL_regkind[op] != EXACT && op != CANY) |
1912 | strend = HOPc(strend, -(minlen - 1)); | |
1913 | } | |
a3621e74 | 1914 | DEBUG_EXECUTE_r({ |
ffc61ed2 | 1915 | SV *prop = sv_newmortal(); |
cfd0369c NC |
1916 | const char *s0; |
1917 | const char *s1; | |
9e55ce06 JH |
1918 | int len0; |
1919 | int len1; | |
1920 | ||
32fc9b6a | 1921 | regprop(prog, prop, c); |
9e55ce06 | 1922 | s0 = UTF ? |
3f7c398e | 1923 | pv_uni_display(dsv0, (U8*)SvPVX_const(prop), SvCUR(prop), 60, |
c728cb41 | 1924 | UNI_DISPLAY_REGEX) : |
cfd0369c | 1925 | SvPVX_const(prop); |
9e55ce06 JH |
1926 | len0 = UTF ? SvCUR(dsv0) : SvCUR(prop); |
1927 | s1 = UTF ? | |
c728cb41 | 1928 | sv_uni_display(dsv1, sv, 60, UNI_DISPLAY_REGEX) : s; |
9e55ce06 JH |
1929 | len1 = UTF ? SvCUR(dsv1) : strend - s; |
1930 | PerlIO_printf(Perl_debug_log, | |
a0288114 | 1931 | "Matching stclass \"%*.*s\" against \"%*.*s\"\n", |
9e55ce06 JH |
1932 | len0, len0, s0, |
1933 | len1, len1, s1); | |
ffc61ed2 | 1934 | }); |
3b0527fe | 1935 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1936 | goto got_it; |
a3621e74 | 1937 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass...\n")); |
d6a28714 JH |
1938 | } |
1939 | else { | |
1940 | dontbother = 0; | |
a0714e2c | 1941 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 1942 | /* Trim the end. */ |
d6a28714 | 1943 | char *last; |
33b8afdf JH |
1944 | SV* float_real; |
1945 | ||
1946 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1947 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1948 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
1949 | |
1950 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 1951 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
1952 | end_shift, &scream_pos, 1); /* last one */ |
1953 | if (!last) | |
ffc61ed2 | 1954 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
1955 | /* we may be pointing at the wrong string */ |
1956 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 1957 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 1958 | } |
d6a28714 JH |
1959 | else { |
1960 | STRLEN len; | |
cfd0369c | 1961 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 1962 | |
33b8afdf | 1963 | if (SvTAIL(float_real)) { |
d6a28714 JH |
1964 | if (memEQ(strend - len + 1, little, len - 1)) |
1965 | last = strend - len + 1; | |
7fba1cd6 | 1966 | else if (!multiline) |
9041c2e3 | 1967 | last = memEQ(strend - len, little, len) |
bd61b366 | 1968 | ? strend - len : NULL; |
b8c5462f | 1969 | else |
d6a28714 JH |
1970 | goto find_last; |
1971 | } else { | |
1972 | find_last: | |
9041c2e3 | 1973 | if (len) |
d6a28714 | 1974 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 1975 | else |
a0288114 | 1976 | last = strend; /* matching "$" */ |
b8c5462f | 1977 | } |
b8c5462f | 1978 | } |
bf93d4cc | 1979 | if (last == NULL) { |
a3621e74 | 1980 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
bf93d4cc | 1981 | "%sCan't trim the tail, match fails (should not happen)%s\n", |
e4584336 | 1982 | PL_colors[4], PL_colors[5])); |
bf93d4cc GS |
1983 | goto phooey; /* Should not happen! */ |
1984 | } | |
d6a28714 JH |
1985 | dontbother = strend - last + prog->float_min_offset; |
1986 | } | |
1987 | if (minlen && (dontbother < minlen)) | |
1988 | dontbother = minlen - 1; | |
1989 | strend -= dontbother; /* this one's always in bytes! */ | |
1990 | /* We don't know much -- general case. */ | |
1aa99e6b | 1991 | if (do_utf8) { |
d6a28714 | 1992 | for (;;) { |
3b0527fe | 1993 | if (regtry(®info, s)) |
d6a28714 JH |
1994 | goto got_it; |
1995 | if (s >= strend) | |
1996 | break; | |
b8c5462f | 1997 | s += UTF8SKIP(s); |
d6a28714 JH |
1998 | }; |
1999 | } | |
2000 | else { | |
2001 | do { | |
3b0527fe | 2002 | if (regtry(®info, s)) |
d6a28714 JH |
2003 | goto got_it; |
2004 | } while (s++ < strend); | |
2005 | } | |
2006 | } | |
2007 | ||
2008 | /* Failure. */ | |
2009 | goto phooey; | |
2010 | ||
2011 | got_it: | |
2012 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2013 | ||
2014 | if (PL_reg_eval_set) { | |
2015 | /* Preserve the current value of $^R */ | |
2016 | if (oreplsv != GvSV(PL_replgv)) | |
2017 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2018 | restored, the value remains | |
2019 | the same. */ | |
4f639d21 | 2020 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2021 | } |
2022 | ||
2023 | /* make sure $`, $&, $', and $digit will work later */ | |
2024 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2025 | RX_MATCH_COPY_FREE(prog); |
d6a28714 JH |
2026 | if (flags & REXEC_COPY_STR) { |
2027 | I32 i = PL_regeol - startpos + (stringarg - strbeg); | |
f8c7b90f | 2028 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2029 | if ((SvIsCOW(sv) |
2030 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2031 | if (DEBUG_C_TEST) { | |
2032 | PerlIO_printf(Perl_debug_log, | |
2033 | "Copy on write: regexp capture, type %d\n", | |
2034 | (int) SvTYPE(sv)); | |
2035 | } | |
2036 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2037 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2038 | assert (SvPOKp(prog->saved_copy)); |
2039 | } else | |
2040 | #endif | |
2041 | { | |
2042 | RX_MATCH_COPIED_on(prog); | |
2043 | s = savepvn(strbeg, i); | |
2044 | prog->subbeg = s; | |
2045 | } | |
d6a28714 | 2046 | prog->sublen = i; |
d6a28714 JH |
2047 | } |
2048 | else { | |
2049 | prog->subbeg = strbeg; | |
2050 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2051 | } | |
2052 | } | |
9041c2e3 | 2053 | |
d6a28714 JH |
2054 | return 1; |
2055 | ||
2056 | phooey: | |
a3621e74 | 2057 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2058 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2059 | if (PL_reg_eval_set) |
4f639d21 | 2060 | restore_pos(aTHX_ prog); |
d6a28714 JH |
2061 | return 0; |
2062 | } | |
2063 | ||
2064 | /* | |
2065 | - regtry - try match at specific point | |
2066 | */ | |
2067 | STATIC I32 /* 0 failure, 1 success */ | |
3b0527fe | 2068 | S_regtry(pTHX_ const regmatch_info *reginfo, char *startpos) |
d6a28714 | 2069 | { |
97aff369 | 2070 | dVAR; |
d6a28714 JH |
2071 | register I32 *sp; |
2072 | register I32 *ep; | |
2073 | CHECKPOINT lastcp; | |
3b0527fe | 2074 | regexp *prog = reginfo->prog; |
a3621e74 | 2075 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 | 2076 | |
02db2b7b IZ |
2077 | #ifdef DEBUGGING |
2078 | PL_regindent = 0; /* XXXX Not good when matches are reenterable... */ | |
2079 | #endif | |
d6a28714 JH |
2080 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { |
2081 | MAGIC *mg; | |
2082 | ||
2083 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2084 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2085 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2086 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2087 | )); |
e8347627 | 2088 | SAVEI32(cxstack[cxstack_ix].blk_oldsp); |
d6a28714 JH |
2089 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2090 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2091 | SAVETMPS; | |
2092 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2093 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2094 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2095 | ||
3b0527fe | 2096 | if (reginfo->sv) { |
d6a28714 | 2097 | /* Make $_ available to executed code. */ |
3b0527fe | 2098 | if (reginfo->sv != DEFSV) { |
59f00321 | 2099 | SAVE_DEFSV; |
3b0527fe | 2100 | DEFSV = reginfo->sv; |
b8c5462f | 2101 | } |
d6a28714 | 2102 | |
3b0527fe DM |
2103 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2104 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2105 | /* prepare for quick setting of pos */ |
d300d9fa NC |
2106 | #ifdef PERL_OLD_COPY_ON_WRITE |
2107 | if (SvIsCOW(sv)) | |
2108 | sv_force_normal_flags(sv, 0); | |
2109 | #endif | |
3b0527fe | 2110 | mg = sv_magicext(reginfo->sv, (SV*)0, PERL_MAGIC_regex_global, |
d300d9fa | 2111 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2112 | mg->mg_len = -1; |
b8c5462f | 2113 | } |
d6a28714 JH |
2114 | PL_reg_magic = mg; |
2115 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2116 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2117 | } |
09687e5a | 2118 | if (!PL_reg_curpm) { |
a02a5408 | 2119 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2120 | #ifdef USE_ITHREADS |
2121 | { | |
2122 | SV* repointer = newSViv(0); | |
577e12cc | 2123 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2124 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2125 | av_push(PL_regex_padav,repointer); |
2126 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2127 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2128 | } | |
2129 | #endif | |
2130 | } | |
aaa362c4 | 2131 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2132 | PL_reg_oldcurpm = PL_curpm; |
2133 | PL_curpm = PL_reg_curpm; | |
2134 | if (RX_MATCH_COPIED(prog)) { | |
2135 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2136 | since it may be needed if this match fails. Thus | |
2137 | $` inside (?{}) could fail... */ | |
2138 | PL_reg_oldsaved = prog->subbeg; | |
2139 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2140 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2141 | PL_nrs = prog->saved_copy; |
2142 | #endif | |
d6a28714 JH |
2143 | RX_MATCH_COPIED_off(prog); |
2144 | } | |
2145 | else | |
bd61b366 | 2146 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2147 | prog->subbeg = PL_bostr; |
2148 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2149 | } | |
973dddac | 2150 | prog->startp[0] = startpos - PL_bostr; |
d6a28714 JH |
2151 | PL_reginput = startpos; |
2152 | PL_regstartp = prog->startp; | |
2153 | PL_regendp = prog->endp; | |
2154 | PL_reglastparen = &prog->lastparen; | |
a01268b5 | 2155 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2156 | prog->lastparen = 0; |
03994de8 | 2157 | prog->lastcloseparen = 0; |
d6a28714 | 2158 | PL_regsize = 0; |
a3621e74 | 2159 | DEBUG_EXECUTE_r(PL_reg_starttry = startpos); |
d6a28714 JH |
2160 | if (PL_reg_start_tmpl <= prog->nparens) { |
2161 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2162 | if(PL_reg_start_tmp) | |
2163 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2164 | else | |
a02a5408 | 2165 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2166 | } |
2167 | ||
2168 | /* XXXX What this code is doing here?!!! There should be no need | |
2169 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2170 | this! --ilya*/ |
dafc8851 JH |
2171 | |
2172 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2173 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2174 | * PL_reglastparen), is not needed at all by the test suite |
2175 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2176 | * enough, for building DynaLoader, or otherwise this | |
2177 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2178 | * will happen. Meanwhile, this code *is* needed for the | |
2179 | * above-mentioned test suite tests to succeed. The common theme | |
2180 | * on those tests seems to be returning null fields from matches. | |
2181 | * --jhi */ | |
dafc8851 | 2182 | #if 1 |
d6a28714 JH |
2183 | sp = prog->startp; |
2184 | ep = prog->endp; | |
2185 | if (prog->nparens) { | |
097eb12c | 2186 | register I32 i; |
eb160463 | 2187 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2188 | *++sp = -1; |
2189 | *++ep = -1; | |
2190 | } | |
2191 | } | |
dafc8851 | 2192 | #endif |
02db2b7b | 2193 | REGCP_SET(lastcp); |
3b0527fe | 2194 | if (regmatch(reginfo, prog->program + 1)) { |
d6a28714 JH |
2195 | prog->endp[0] = PL_reginput - PL_bostr; |
2196 | return 1; | |
2197 | } | |
02db2b7b | 2198 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2199 | return 0; |
2200 | } | |
2201 | ||
02db2b7b IZ |
2202 | #define RE_UNWIND_BRANCH 1 |
2203 | #define RE_UNWIND_BRANCHJ 2 | |
2204 | ||
2205 | union re_unwind_t; | |
2206 | ||
2207 | typedef struct { /* XX: makes sense to enlarge it... */ | |
2208 | I32 type; | |
2209 | I32 prev; | |
2210 | CHECKPOINT lastcp; | |
2211 | } re_unwind_generic_t; | |
2212 | ||
2213 | typedef struct { | |
2214 | I32 type; | |
2215 | I32 prev; | |
2216 | CHECKPOINT lastcp; | |
2217 | I32 lastparen; | |
2218 | regnode *next; | |
2219 | char *locinput; | |
2220 | I32 nextchr; | |
3a2830be | 2221 | int minmod; |
02db2b7b IZ |
2222 | #ifdef DEBUGGING |
2223 | int regindent; | |
2224 | #endif | |
2225 | } re_unwind_branch_t; | |
2226 | ||
2227 | typedef union re_unwind_t { | |
2228 | I32 type; | |
2229 | re_unwind_generic_t generic; | |
2230 | re_unwind_branch_t branch; | |
2231 | } re_unwind_t; | |
2232 | ||
8ba1375e MJD |
2233 | #define sayYES goto yes |
2234 | #define sayNO goto no | |
e0f9d4a8 | 2235 | #define sayNO_ANYOF goto no_anyof |
8ba1375e | 2236 | #define sayYES_FINAL goto yes_final |
8ba1375e MJD |
2237 | #define sayNO_FINAL goto no_final |
2238 | #define sayNO_SILENT goto do_no | |
2239 | #define saySAME(x) if (x) goto yes; else goto no | |
2240 | ||
3ab3c9b4 HS |
2241 | #define POSCACHE_SUCCESS 0 /* caching success rather than failure */ |
2242 | #define POSCACHE_SEEN 1 /* we know what we're caching */ | |
2243 | #define POSCACHE_START 2 /* the real cache: this bit maps to pos 0 */ | |
7409bbd3 | 2244 | |
3ab3c9b4 | 2245 | #define CACHEsayYES STMT_START { \ |
d8319b27 | 2246 | if (st->u.whilem.cache_offset | st->u.whilem.cache_bit) { \ |
7409bbd3 DM |
2247 | if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) { \ |
2248 | PL_reg_poscache[0] |= (1<<POSCACHE_SUCCESS) | (1<<POSCACHE_SEEN); \ | |
2249 | PL_reg_poscache[st->u.whilem.cache_offset] |= (1<<st->u.whilem.cache_bit); \ | |
2250 | } \ | |
2251 | else if (PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS)) { \ | |
2252 | PL_reg_poscache[st->u.whilem.cache_offset] |= (1<<st->u.whilem.cache_bit); \ | |
2253 | } \ | |
2254 | else { \ | |
3ab3c9b4 HS |
2255 | /* cache records failure, but this is success */ \ |
2256 | DEBUG_r( \ | |
2257 | PerlIO_printf(Perl_debug_log, \ | |
2258 | "%*s (remove success from failure cache)\n", \ | |
2259 | REPORT_CODE_OFF+PL_regindent*2, "") \ | |
2260 | ); \ | |
d8319b27 | 2261 | PL_reg_poscache[st->u.whilem.cache_offset] &= ~(1<<st->u.whilem.cache_bit); \ |
3ab3c9b4 HS |
2262 | } \ |
2263 | } \ | |
2264 | sayYES; \ | |
2265 | } STMT_END | |
7409bbd3 | 2266 | |
3ab3c9b4 | 2267 | #define CACHEsayNO STMT_START { \ |
d8319b27 | 2268 | if (st->u.whilem.cache_offset | st->u.whilem.cache_bit) { \ |
7409bbd3 | 2269 | if (!(PL_reg_poscache[0] & (1<<POSCACHE_SEEN))) { \ |
3ab3c9b4 | 2270 | PL_reg_poscache[0] |= (1<<POSCACHE_SEEN); \ |
7409bbd3 DM |
2271 | PL_reg_poscache[st->u.whilem.cache_offset] |= (1<<st->u.whilem.cache_bit); \ |
2272 | } \ | |
2273 | else if (!(PL_reg_poscache[0] & (1<<POSCACHE_SUCCESS))) { \ | |
2274 | PL_reg_poscache[st->u.whilem.cache_offset] |= (1<<st->u.whilem.cache_bit); \ | |
2275 | } \ | |
2276 | else { \ | |
3ab3c9b4 HS |
2277 | /* cache records success, but this is failure */ \ |
2278 | DEBUG_r( \ | |
2279 | PerlIO_printf(Perl_debug_log, \ | |
2280 | "%*s (remove failure from success cache)\n", \ | |
2281 | REPORT_CODE_OFF+PL_regindent*2, "") \ | |
2282 | ); \ | |
d8319b27 | 2283 | PL_reg_poscache[st->u.whilem.cache_offset] &= ~(1<<st->u.whilem.cache_bit); \ |
3ab3c9b4 HS |
2284 | } \ |
2285 | } \ | |
2286 | sayNO; \ | |
2287 | } STMT_END | |
2288 | ||
a3621e74 YO |
2289 | /* this is used to determine how far from the left messages like |
2290 | 'failed...' are printed. Currently 29 makes these messages line | |
2291 | up with the opcode they refer to. Earlier perls used 25 which | |
2292 | left these messages outdented making reviewing a debug output | |
2293 | quite difficult. | |
2294 | */ | |
2295 | #define REPORT_CODE_OFF 29 | |
2296 | ||
2297 | ||
2298 | /* Make sure there is a test for this +1 options in re_tests */ | |
2299 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2300 | ||
86545054 DM |
2301 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2302 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2303 | ||
5d9a96ca DM |
2304 | /* grab a new slab and return the first slot in it */ |
2305 | ||
2306 | STATIC regmatch_state * | |
2307 | S_push_slab(pTHX) | |
2308 | { | |
2309 | regmatch_slab *s = PL_regmatch_slab->next; | |
2310 | if (!s) { | |
2311 | Newx(s, 1, regmatch_slab); | |
2312 | s->prev = PL_regmatch_slab; | |
2313 | s->next = NULL; | |
2314 | PL_regmatch_slab->next = s; | |
2315 | } | |
2316 | PL_regmatch_slab = s; | |
86545054 | 2317 | return SLAB_FIRST(s); |
5d9a96ca | 2318 | } |
5b47454d | 2319 | |
95b24440 DM |
2320 | /* simulate a recursive call to regmatch */ |
2321 | ||
2322 | #define REGMATCH(ns, where) \ | |
5d9a96ca DM |
2323 | st->scan = scan; \ |
2324 | scan = (ns); \ | |
2325 | st->resume_state = resume_##where; \ | |
95b24440 DM |
2326 | goto start_recurse; \ |
2327 | resume_point_##where: | |
2328 | ||
aa283a38 DM |
2329 | |
2330 | /* push a new regex state. Set newst to point to it */ | |
2331 | ||
2332 | #define PUSH_STATE(newst, resume) \ | |
2333 | depth++; \ | |
2334 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "PUSH STATE(%d)\n", depth)); \ | |
2335 | st->scan = scan; \ | |
2336 | st->next = next; \ | |
2337 | st->n = n; \ | |
2338 | st->locinput = locinput; \ | |
2339 | st->resume_state = resume; \ | |
2340 | newst = st+1; \ | |
86545054 | 2341 | if (newst > SLAB_LAST(PL_regmatch_slab)) \ |
aa283a38 DM |
2342 | newst = S_push_slab(aTHX); \ |
2343 | PL_regmatch_state = newst; \ | |
2344 | newst->cc = 0; \ | |
2345 | newst->minmod = 0; \ | |
2346 | newst->sw = 0; \ | |
2347 | newst->logical = 0; \ | |
2348 | newst->unwind = 0; \ | |
2349 | locinput = PL_reginput; \ | |
2350 | nextchr = UCHARAT(locinput); | |
2351 | ||
2352 | #define POP_STATE \ | |
2353 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "POP STATE(%d)\n", depth)); \ | |
2354 | depth--; \ | |
2355 | st--; \ | |
86545054 | 2356 | if (st < SLAB_FIRST(PL_regmatch_slab)) { \ |
aa283a38 | 2357 | PL_regmatch_slab = PL_regmatch_slab->prev; \ |
86545054 | 2358 | st = SLAB_LAST(PL_regmatch_slab); \ |
aa283a38 DM |
2359 | } \ |
2360 | PL_regmatch_state = st; \ | |
2361 | scan = st->scan; \ | |
2362 | next = st->next; \ | |
2363 | n = st->n; \ | |
2364 | locinput = st->locinput; \ | |
2365 | nextchr = UCHARAT(locinput); | |
2366 | ||
d6a28714 JH |
2367 | /* |
2368 | - regmatch - main matching routine | |
2369 | * | |
2370 | * Conceptually the strategy is simple: check to see whether the current | |
2371 | * node matches, call self recursively to see whether the rest matches, | |
2372 | * and then act accordingly. In practice we make some effort to avoid | |
2373 | * recursion, in particular by going through "ordinary" nodes (that don't | |
2374 | * need to know whether the rest of the match failed) by a loop instead of | |
2375 | * by recursion. | |
2376 | */ | |
2377 | /* [lwall] I've hoisted the register declarations to the outer block in order to | |
2378 | * maybe save a little bit of pushing and popping on the stack. It also takes | |
2379 | * advantage of machines that use a register save mask on subroutine entry. | |
95b24440 DM |
2380 | * |
2381 | * This function used to be heavily recursive, but since this had the | |
2382 | * effect of blowing the CPU stack on complex regexes, it has been | |
2383 | * restructured to be iterative, and to save state onto the heap rather | |
2384 | * than the stack. Essentially whereever regmatch() used to be called, it | |
2385 | * pushes the current state, notes where to return, then jumps back into | |
2386 | * the main loop. | |
2387 | * | |
2388 | * Originally the structure of this function used to look something like | |
2389 | ||
2390 | S_regmatch() { | |
2391 | int a = 1, b = 2; | |
2392 | ... | |
2393 | while (scan != NULL) { | |
5d9a96ca | 2394 | a++; // do stuff with a and b |
95b24440 DM |
2395 | ... |
2396 | switch (OP(scan)) { | |
2397 | case FOO: { | |
2398 | int local = 3; | |
2399 | ... | |
2400 | if (regmatch(...)) // recurse | |
2401 | goto yes; | |
2402 | } | |
2403 | ... | |
2404 | } | |
2405 | } | |
2406 | yes: | |
2407 | return 1; | |
2408 | } | |
2409 | ||
2410 | * Now it looks something like this: | |
2411 | ||
5d9a96ca | 2412 | typedef struct { |
95b24440 DM |
2413 | int a, b, local; |
2414 | int resume_state; | |
5d9a96ca | 2415 | } regmatch_state; |
95b24440 DM |
2416 | |
2417 | S_regmatch() { | |
5d9a96ca DM |
2418 | regmatch_state *st = new(); |
2419 | int depth=0; | |
2420 | st->a++; // do stuff with a and b | |
95b24440 DM |
2421 | ... |
2422 | while (scan != NULL) { | |
2423 | ... | |
2424 | switch (OP(scan)) { | |
2425 | case FOO: { | |
5d9a96ca | 2426 | st->local = 3; |
95b24440 | 2427 | ... |
5d9a96ca DM |
2428 | st->scan = scan; |
2429 | scan = ...; | |
2430 | st->resume_state = resume_FOO; | |
2431 | goto start_recurse; // recurse | |
95b24440 | 2432 | |
5d9a96ca DM |
2433 | resume_point_FOO: |
2434 | if (result) | |
95b24440 DM |
2435 | goto yes; |
2436 | } | |
2437 | ... | |
2438 | } | |
5d9a96ca DM |
2439 | start_recurse: |
2440 | st = new(); push a new state | |
2441 | st->a = 1; st->b = 2; | |
2442 | depth++; | |
95b24440 | 2443 | } |
5d9a96ca | 2444 | yes: |
95b24440 | 2445 | result = 1; |
5d9a96ca DM |
2446 | if (depth--) { |
2447 | st = pop(); | |
95b24440 DM |
2448 | switch (resume_state) { |
2449 | case resume_FOO: | |
2450 | goto resume_point_FOO; | |
2451 | ... | |
2452 | } | |
2453 | } | |
2454 | return result | |
2455 | } | |
2456 | ||
2457 | * WARNING: this means that any line in this function that contains a | |
2458 | * REGMATCH() or TRYPAREN() is actually simulating a recursive call to | |
2459 | * regmatch() using gotos instead. Thus the values of any local variables | |
2460 | * not saved in the regmatch_state structure will have been lost when | |
2461 | * execution resumes on the next line . | |
5d9a96ca DM |
2462 | * |
2463 | * States (ie the st pointer) are allocated in slabs of about 4K in size. | |
2464 | * PL_regmatch_state always points to the currently active state, and | |
2465 | * PL_regmatch_slab points to the slab currently containing PL_regmatch_state. | |
2466 | * The first time regmatch is called, the first slab is allocated, and is | |
2467 | * never freed until interpreter desctruction. When the slab is full, | |
2468 | * a new one is allocated chained to the end. At exit from regmatch, slabs | |
2469 | * allocated since entry are freed. | |
d6a28714 | 2470 | */ |
95b24440 DM |
2471 | |
2472 | ||
d6a28714 | 2473 | STATIC I32 /* 0 failure, 1 success */ |
3b0527fe | 2474 | S_regmatch(pTHX_ const regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2475 | { |
27da23d5 | 2476 | dVAR; |
95b24440 | 2477 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2478 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2479 | |
3b0527fe DM |
2480 | regexp *rex = reginfo->prog; |
2481 | ||
5d9a96ca DM |
2482 | regmatch_slab *orig_slab; |
2483 | regmatch_state *orig_state; | |
a3621e74 | 2484 | |
5d9a96ca DM |
2485 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2486 | register regmatch_state *st; | |
95b24440 | 2487 | |
5d9a96ca DM |
2488 | /* cache heavy used fields of st in registers */ |
2489 | register regnode *scan; | |
2490 | register regnode *next; | |
2491 | register I32 n = 0; /* initialize to shut up compiler warning */ | |
2492 | register char *locinput = PL_reginput; | |
95b24440 | 2493 | |
5d9a96ca DM |
2494 | /* these variables are NOT saved during a recusive RFEGMATCH: */ |
2495 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ | |
2496 | bool result; /* return value of S_regmatch */ | |
2497 | regnode *inner; /* Next node in internal branch. */ | |
2498 | int depth = 0; /* depth of recursion */ | |
aa283a38 | 2499 | regmatch_state *newst; /* when pushing a state, this is the new one */ |
77cb431f DM |
2500 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2501 | subpattern */ | |
95b24440 DM |
2502 | |
2503 | #ifdef DEBUGGING | |
ab74612d | 2504 | SV *re_debug_flags = NULL; |
a3621e74 | 2505 | GET_RE_DEBUG_FLAGS; |
d6a28714 JH |
2506 | PL_regindent++; |
2507 | #endif | |
2508 | ||
5d9a96ca DM |
2509 | /* on first ever call to regmatch, allocate first slab */ |
2510 | if (!PL_regmatch_slab) { | |
2511 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2512 | PL_regmatch_slab->prev = NULL; | |
2513 | PL_regmatch_slab->next = NULL; | |
86545054 | 2514 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2515 | } |
2516 | ||
2517 | /* remember current high-water mark for exit */ | |
2518 | /* XXX this should be done with SAVE* instead */ | |
2519 | orig_slab = PL_regmatch_slab; | |
2520 | orig_state = PL_regmatch_state; | |
2521 | ||
2522 | /* grab next free state slot */ | |
2523 | st = ++PL_regmatch_state; | |
86545054 | 2524 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2525 | st = PL_regmatch_state = S_push_slab(aTHX); |
2526 | ||
2527 | st->minmod = 0; | |
2528 | st->sw = 0; | |
2529 | st->logical = 0; | |
2530 | st->unwind = 0; | |
2531 | st->cc = NULL; | |
d6a28714 JH |
2532 | /* Note that nextchr is a byte even in UTF */ |
2533 | nextchr = UCHARAT(locinput); | |
2534 | scan = prog; | |
2535 | while (scan != NULL) { | |
8ba1375e | 2536 | |
a3621e74 | 2537 | DEBUG_EXECUTE_r( { |
6136c704 | 2538 | SV * const prop = sv_newmortal(); |
1df70142 AL |
2539 | const int docolor = *PL_colors[0]; |
2540 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ | |
1aa99e6b | 2541 | int l = (PL_regeol - locinput) > taill ? taill : (PL_regeol - locinput); |
d6a28714 JH |
2542 | /* The part of the string before starttry has one color |
2543 | (pref0_len chars), between starttry and current | |
2544 | position another one (pref_len - pref0_len chars), | |
2545 | after the current position the third one. | |
2546 | We assume that pref0_len <= pref_len, otherwise we | |
2547 | decrease pref0_len. */ | |
9041c2e3 | 2548 | int pref_len = (locinput - PL_bostr) > (5 + taill) - l |
1aa99e6b IH |
2549 | ? (5 + taill) - l : locinput - PL_bostr; |
2550 | int pref0_len; | |
d6a28714 | 2551 | |
df1ffd02 | 2552 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) |
1aa99e6b IH |
2553 | pref_len++; |
2554 | pref0_len = pref_len - (locinput - PL_reg_starttry); | |
d6a28714 | 2555 | if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) |
9041c2e3 | 2556 | l = ( PL_regeol - locinput > (5 + taill) - pref_len |
d6a28714 | 2557 | ? (5 + taill) - pref_len : PL_regeol - locinput); |
df1ffd02 | 2558 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
1aa99e6b | 2559 | l--; |
d6a28714 JH |
2560 | if (pref0_len < 0) |
2561 | pref0_len = 0; | |
2562 | if (pref0_len > pref_len) | |
2563 | pref0_len = pref_len; | |
32fc9b6a | 2564 | regprop(rex, prop, scan); |
2a782b5b | 2565 | { |
1df70142 | 2566 | const char * const s0 = |
f14c76ed | 2567 | do_utf8 && OP(scan) != CANY ? |
95b24440 | 2568 | pv_uni_display(PERL_DEBUG_PAD(0), (U8*)(locinput - pref_len), |
c728cb41 | 2569 | pref0_len, 60, UNI_DISPLAY_REGEX) : |
2a782b5b | 2570 | locinput - pref_len; |
1df70142 AL |
2571 | const int len0 = do_utf8 ? strlen(s0) : pref0_len; |
2572 | const char * const s1 = do_utf8 && OP(scan) != CANY ? | |
95b24440 DM |
2573 | pv_uni_display(PERL_DEBUG_PAD(1), |
2574 | (U8*)(locinput - pref_len + pref0_len), | |
c728cb41 | 2575 | pref_len - pref0_len, 60, UNI_DISPLAY_REGEX) : |
2a782b5b | 2576 | locinput - pref_len + pref0_len; |
1df70142 AL |
2577 | const int len1 = do_utf8 ? strlen(s1) : pref_len - pref0_len; |
2578 | const char * const s2 = do_utf8 && OP(scan) != CANY ? | |
95b24440 | 2579 | pv_uni_display(PERL_DEBUG_PAD(2), (U8*)locinput, |
c728cb41 | 2580 | PL_regeol - locinput, 60, UNI_DISPLAY_REGEX) : |
2a782b5b | 2581 | locinput; |
1df70142 | 2582 | const int len2 = do_utf8 ? strlen(s2) : l; |
2a782b5b JH |
2583 | PerlIO_printf(Perl_debug_log, |
2584 | "%4"IVdf" <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3"IVdf":%*s%s\n", | |
2585 | (IV)(locinput - PL_bostr), | |
2586 | PL_colors[4], | |
2587 | len0, s0, | |
2588 | PL_colors[5], | |
2589 | PL_colors[2], | |
2590 | len1, s1, | |
2591 | PL_colors[3], | |
2592 | (docolor ? "" : "> <"), | |
2593 | PL_colors[0], | |
2594 | len2, s2, | |
2595 | PL_colors[1], | |
2596 | 15 - l - pref_len + 1, | |
2597 | "", | |
4f639d21 | 2598 | (IV)(scan - rex->program), PL_regindent*2, "", |
3f7c398e | 2599 | SvPVX_const(prop)); |
2a782b5b JH |
2600 | } |
2601 | }); | |
d6a28714 JH |
2602 | |
2603 | next = scan + NEXT_OFF(scan); | |
2604 | if (next == scan) | |
2605 | next = NULL; | |
2606 | ||
2607 | switch (OP(scan)) { | |
2608 | case BOL: | |
7fba1cd6 | 2609 | if (locinput == PL_bostr) |
d6a28714 | 2610 | { |
3b0527fe | 2611 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2612 | break; |
2613 | } | |
d6a28714 JH |
2614 | sayNO; |
2615 | case MBOL: | |
12d33761 HS |
2616 | if (locinput == PL_bostr || |
2617 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2618 | { |
b8c5462f JH |
2619 | break; |
2620 | } | |
d6a28714 JH |
2621 | sayNO; |
2622 | case SBOL: | |
c2a73568 | 2623 | if (locinput == PL_bostr) |
b8c5462f | 2624 | break; |
d6a28714 JH |
2625 | sayNO; |
2626 | case GPOS: | |
3b0527fe | 2627 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2628 | break; |
2629 | sayNO; | |
2630 | case EOL: | |
d6a28714 JH |
2631 | goto seol; |
2632 | case MEOL: | |
d6a28714 | 2633 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2634 | sayNO; |
b8c5462f | 2635 | break; |
d6a28714 JH |
2636 | case SEOL: |
2637 | seol: | |
2638 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2639 | sayNO; |
d6a28714 | 2640 | if (PL_regeol - locinput > 1) |
b8c5462f | 2641 | sayNO; |
b8c5462f | 2642 | break; |
d6a28714 JH |
2643 | case EOS: |
2644 | if (PL_regeol != locinput) | |
b8c5462f | 2645 | sayNO; |
d6a28714 | 2646 | break; |
ffc61ed2 | 2647 | case SANY: |
d6a28714 | 2648 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2649 | sayNO; |
f33976b4 DB |
2650 | if (do_utf8) { |
2651 | locinput += PL_utf8skip[nextchr]; | |
2652 | if (locinput > PL_regeol) | |
2653 | sayNO; | |
2654 | nextchr = UCHARAT(locinput); | |
2655 | } | |
2656 | else | |
2657 | nextchr = UCHARAT(++locinput); | |
2658 | break; | |
2659 | case CANY: | |
2660 | if (!nextchr && locinput >= PL_regeol) | |
2661 | sayNO; | |
b8c5462f | 2662 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2663 | break; |
ffc61ed2 | 2664 | case REG_ANY: |
1aa99e6b IH |
2665 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2666 | sayNO; | |
2667 | if (do_utf8) { | |
b8c5462f | 2668 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2669 | if (locinput > PL_regeol) |
2670 | sayNO; | |
a0ed51b3 | 2671 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2672 | } |
1aa99e6b IH |
2673 | else |
2674 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2675 | break; |
a3621e74 YO |
2676 | |
2677 | ||
2678 | ||
2679 | /* | |
2680 | traverse the TRIE keeping track of all accepting states | |
2681 | we transition through until we get to a failing node. | |
2682 | ||
a3621e74 YO |
2683 | |
2684 | */ | |
5b47454d | 2685 | case TRIE: |
a3621e74 YO |
2686 | case TRIEF: |
2687 | case TRIEFL: | |
2688 | { | |
a3621e74 YO |
2689 | U8 *uc = ( U8* )locinput; |
2690 | U32 state = 1; | |
2691 | U16 charid = 0; | |
2692 | U32 base = 0; | |
2693 | UV uvc = 0; | |
2694 | STRLEN len = 0; | |
2695 | STRLEN foldlen = 0; | |
a3621e74 YO |
2696 | U8 *uscan = (U8*)NULL; |
2697 | STRLEN bufflen=0; | |
95b24440 | 2698 | SV *sv_accept_buff = NULL; |
5b47454d DM |
2699 | const enum { trie_plain, trie_utf8, trie_uft8_fold } |
2700 | trie_type = do_utf8 ? | |
2701 | (OP(scan) == TRIE ? trie_utf8 : trie_uft8_fold) | |
2702 | : trie_plain; | |
2703 | ||
7087a21c NC |
2704 | /* what trie are we using right now */ |
2705 | reg_trie_data *trie | |
32fc9b6a | 2706 | = (reg_trie_data*)rex->data->data[ ARG( scan ) ]; |
d8319b27 | 2707 | st->u.trie.accepted = 0; /* how many accepting states we have seen */ |
95b24440 | 2708 | result = 0; |
a3621e74 YO |
2709 | |
2710 | while ( state && uc <= (U8*)PL_regeol ) { | |
2711 | ||
5b47454d | 2712 | if (trie->states[ state ].wordnum) { |
d8319b27 | 2713 | if (!st->u.trie.accepted ) { |
5b47454d DM |
2714 | ENTER; |
2715 | SAVETMPS; | |
2716 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2717 | sv_accept_buff=newSV(bufflen * | |
2718 | sizeof(reg_trie_accepted) - 1); | |
2719 | SvCUR_set(sv_accept_buff, | |
2720 | sizeof(reg_trie_accepted)); | |
2721 | SvPOK_on(sv_accept_buff); | |
2722 | sv_2mortal(sv_accept_buff); | |
d8319b27 | 2723 | st->u.trie.accept_buff = |
5b47454d DM |
2724 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2725 | } | |
2726 | else { | |
d8319b27 | 2727 | if (st->u.trie.accepted >= bufflen) { |
5b47454d | 2728 | bufflen *= 2; |
d8319b27 | 2729 | st->u.trie.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2730 | SvGROW(sv_accept_buff, |
2731 | bufflen * sizeof(reg_trie_accepted)); | |
2732 | } | |
2733 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2734 | + sizeof(reg_trie_accepted)); | |
2735 | } | |
d8319b27 DM |
2736 | st->u.trie.accept_buff[st->u.trie.accepted].wordnum = trie->states[state].wordnum; |
2737 | st->u.trie.accept_buff[st->u.trie.accepted].endpos = uc; | |
2738 | ++st->u.trie.accepted; | |
5b47454d | 2739 | } |
a3621e74 YO |
2740 | |
2741 | base = trie->states[ state ].trans.base; | |
2742 | ||
2743 | DEBUG_TRIE_EXECUTE_r( | |
2744 | PerlIO_printf( Perl_debug_log, | |
e4584336 | 2745 | "%*s %sState: %4"UVxf", Base: %4"UVxf", Accepted: %4"UVxf" ", |
a3621e74 | 2746 | REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], |
d8319b27 | 2747 | (UV)state, (UV)base, (UV)st->u.trie.accepted ); |
a3621e74 YO |
2748 | ); |
2749 | ||
2750 | if ( base ) { | |
5b47454d DM |
2751 | switch (trie_type) { |
2752 | case trie_uft8_fold: | |
a3621e74 YO |
2753 | if ( foldlen>0 ) { |
2754 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); | |
2755 | foldlen -= len; | |
2756 | uscan += len; | |
2757 | len=0; | |
2758 | } else { | |
1df70142 | 2759 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
a3621e74 YO |
2760 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); |
2761 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); | |
2762 | foldlen -= UNISKIP( uvc ); | |
2763 | uscan = foldbuf + UNISKIP( uvc ); | |
2764 | } | |
5b47454d DM |
2765 | break; |
2766 | case trie_utf8: | |
2767 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, | |
2768 | &len, uniflags ); | |
2769 | break; | |
2770 | case trie_plain: | |
e4584336 | 2771 | uvc = (UV)*uc; |
a3621e74 YO |
2772 | len = 1; |
2773 | } | |
2774 | ||
5b47454d DM |
2775 | if (uvc < 256) { |
2776 | charid = trie->charmap[ uvc ]; | |
2777 | } | |
2778 | else { | |
2779 | charid = 0; | |
2780 | if (trie->widecharmap) { | |
2781 | SV** svpp = (SV**)NULL; | |
2782 | svpp = hv_fetch(trie->widecharmap, | |
2783 | (char*)&uvc, sizeof(UV), 0); | |
2784 | if (svpp) | |
2785 | charid = (U16)SvIV(*svpp); | |
2786 | } | |
2787 | } | |
a3621e74 | 2788 | |
5b47454d DM |
2789 | if (charid && |
2790 | (base + charid > trie->uniquecharcount ) | |
2791 | && (base + charid - 1 - trie->uniquecharcount | |
2792 | < trie->lasttrans) | |
2793 | && trie->trans[base + charid - 1 - | |
2794 | trie->uniquecharcount].check == state) | |
2795 | { | |
2796 | state = trie->trans[base + charid - 1 - | |
2797 | trie->uniquecharcount ].next; | |
2798 | } | |
2799 | else { | |
2800 | state = 0; | |
2801 | } | |
2802 | uc += len; | |
2803 | ||
2804 | } | |
2805 | else { | |
a3621e74 YO |
2806 | state = 0; |
2807 | } | |
2808 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 RB |
2809 | PerlIO_printf( Perl_debug_log, |
2810 | "Charid:%3x CV:%4"UVxf" After State: %4"UVxf"%s\n", | |
2811 | charid, uvc, (UV)state, PL_colors[5] ); | |
a3621e74 YO |
2812 | ); |
2813 | } | |
d8319b27 | 2814 | if (!st->u.trie.accepted ) |
a3621e74 | 2815 | sayNO; |
a3621e74 YO |
2816 | |
2817 | /* | |
2818 | There was at least one accepting state that we | |
2819 | transitioned through. Presumably the number of accepting | |
2820 | states is going to be low, typically one or two. So we | |
2821 | simply scan through to find the one with lowest wordnum. | |
2822 | Once we find it, we swap the last state into its place | |
2823 | and decrement the size. We then try to match the rest of | |
2824 | the pattern at the point where the word ends, if we | |
2825 | succeed then we end the loop, otherwise the loop | |
2826 | eventually terminates once all of the accepting states | |
2827 | have been tried. | |
2828 | */ | |
a3621e74 | 2829 | |
d8319b27 | 2830 | if ( st->u.trie.accepted == 1 ) { |
a3621e74 | 2831 | DEBUG_EXECUTE_r({ |
097eb12c | 2832 | SV ** const tmp = av_fetch( trie->words, st->u.trie.accept_buff[ 0 ].wordnum-1, 0 ); |
a3621e74 YO |
2833 | PerlIO_printf( Perl_debug_log, |
2834 | "%*s %sonly one match : #%d <%s>%s\n", | |
e4584336 | 2835 | REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], |
d8319b27 | 2836 | st->u.trie.accept_buff[ 0 ].wordnum, |
cfd0369c | 2837 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr", |
a3621e74 YO |
2838 | PL_colors[5] ); |
2839 | }); | |
d8319b27 | 2840 | PL_reginput = (char *)st->u.trie.accept_buff[ 0 ].endpos; |
a3621e74 YO |
2841 | /* in this case we free tmps/leave before we call regmatch |
2842 | as we wont be using accept_buff again. */ | |
2843 | FREETMPS; | |
2844 | LEAVE; | |
95b24440 DM |
2845 | REGMATCH(scan + NEXT_OFF(scan), TRIE1); |
2846 | /*** all unsaved local vars undefined at this point */ | |
a3621e74 YO |
2847 | } else { |
2848 | DEBUG_EXECUTE_r( | |
e4584336 | 2849 | PerlIO_printf( Perl_debug_log,"%*s %sgot %"IVdf" possible matches%s\n", |
d8319b27 | 2850 | REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], (IV)st->u.trie.accepted, |
a3621e74 YO |
2851 | PL_colors[5] ); |
2852 | ); | |
d8319b27 | 2853 | while ( !result && st->u.trie.accepted-- ) { |
a3621e74 YO |
2854 | U32 best = 0; |
2855 | U32 cur; | |
d8319b27 | 2856 | for( cur = 1 ; cur <= st->u.trie.accepted ; cur++ ) { |
e4584336 RB |
2857 | DEBUG_TRIE_EXECUTE_r( |
2858 | PerlIO_printf( Perl_debug_log, | |
2859 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", | |
2860 | REPORT_CODE_OFF + PL_regindent * 2, "", PL_colors[4], | |
d8319b27 DM |
2861 | (IV)best, st->u.trie.accept_buff[ best ].wordnum, (IV)cur, |
2862 | st->u.trie.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
e4584336 | 2863 | ); |
a3621e74 | 2864 | |
d8319b27 DM |
2865 | if (st->u.trie.accept_buff[cur].wordnum < |
2866 | st->u.trie.accept_buff[best].wordnum) | |
e822a8b4 | 2867 | best = cur; |
a3621e74 YO |
2868 | } |
2869 | DEBUG_EXECUTE_r({ | |
87830502 | 2870 | reg_trie_data * const trie = (reg_trie_data*) |
32fc9b6a | 2871 | rex->data->data[ARG(scan)]; |
d8319b27 | 2872 | SV ** const tmp = av_fetch( trie->words, st->u.trie.accept_buff[ best ].wordnum - 1, 0 ); |
a3621e74 | 2873 | PerlIO_printf( Perl_debug_log, "%*s %strying alternation #%d <%s> at 0x%p%s\n", |
e4584336 | 2874 | REPORT_CODE_OFF+PL_regindent*2, "", PL_colors[4], |
d8319b27 | 2875 | st->u.trie.accept_buff[best].wordnum, |
cfd0369c | 2876 | tmp ? SvPV_nolen_const( *tmp ) : "not compiled under -Dr",scan, |
a3621e74 YO |
2877 | PL_colors[5] ); |
2878 | }); | |
d8319b27 DM |
2879 | if ( best<st->u.trie.accepted ) { |
2880 | reg_trie_accepted tmp = st->u.trie.accept_buff[ best ]; | |
2881 | st->u.trie.accept_buff[ best ] = st->u.trie.accept_buff[ st->u.trie.accepted ]; | |
2882 | st->u.trie.accept_buff[ st->u.trie.accepted ] = tmp; | |
2883 | best = st->u.trie.accepted; | |
a3621e74 | 2884 | } |
d8319b27 | 2885 | PL_reginput = (char *)st->u.trie.accept_buff[ best ].endpos; |
a3621e74 YO |
2886 | |
2887 | /* | |
2888 | as far as I can tell we only need the SAVETMPS/FREETMPS | |
2889 | for re's with EVAL in them but I'm leaving them in for | |
2890 | all until I can be sure. | |
2891 | */ | |
2892 | SAVETMPS; | |
95b24440 DM |
2893 | REGMATCH(scan + NEXT_OFF(scan), TRIE2); |
2894 | /*** all unsaved local vars undefined at this point */ | |
a3621e74 YO |
2895 | FREETMPS; |
2896 | } | |
2897 | FREETMPS; | |
2898 | LEAVE; | |
2899 | } | |
2900 | ||
95b24440 | 2901 | if (result) { |
a3621e74 YO |
2902 | sayYES; |
2903 | } else { | |
2904 | sayNO; | |
2905 | } | |
2906 | } | |
2907 | /* unreached codepoint */ | |
95b24440 DM |
2908 | case EXACT: { |
2909 | char *s = STRING(scan); | |
5d9a96ca | 2910 | st->ln = STR_LEN(scan); |
eb160463 | 2911 | if (do_utf8 != UTF) { |
bc517b45 | 2912 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 2913 | char *l = locinput; |
5d9a96ca | 2914 | const char *e = s + st->ln; |
a72c7584 | 2915 | |
5ff6fc6d JH |
2916 | if (do_utf8) { |
2917 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 2918 | while (s < e) { |
a3b680e6 | 2919 | STRLEN ulen; |
1aa99e6b | 2920 | if (l >= PL_regeol) |
5ff6fc6d JH |
2921 | sayNO; |
2922 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 2923 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 2924 | uniflags)) |
5ff6fc6d | 2925 | sayNO; |
bc517b45 | 2926 | l += ulen; |
5ff6fc6d | 2927 | s ++; |
1aa99e6b | 2928 | } |
5ff6fc6d JH |
2929 | } |
2930 | else { | |
2931 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 2932 | while (s < e) { |
a3b680e6 | 2933 | STRLEN ulen; |
1aa99e6b IH |
2934 | if (l >= PL_regeol) |
2935 | sayNO; | |
5ff6fc6d | 2936 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 2937 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 2938 | uniflags)) |
1aa99e6b | 2939 | sayNO; |
bc517b45 | 2940 | s += ulen; |
a72c7584 | 2941 | l ++; |
1aa99e6b | 2942 | } |
5ff6fc6d | 2943 | } |
1aa99e6b IH |
2944 | locinput = l; |
2945 | nextchr = UCHARAT(locinput); | |
2946 | break; | |
2947 | } | |
bc517b45 | 2948 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
2949 | /* Inline the first character, for speed. */ |
2950 | if (UCHARAT(s) != nextchr) | |
2951 | sayNO; | |
5d9a96ca | 2952 | if (PL_regeol - locinput < st->ln) |
d6a28714 | 2953 | sayNO; |
5d9a96ca | 2954 | if (st->ln > 1 && memNE(s, locinput, st->ln)) |
d6a28714 | 2955 | sayNO; |
5d9a96ca | 2956 | locinput += st->ln; |
d6a28714 JH |
2957 | nextchr = UCHARAT(locinput); |
2958 | break; | |
95b24440 | 2959 | } |
d6a28714 | 2960 | case EXACTFL: |
b8c5462f JH |
2961 | PL_reg_flags |= RF_tainted; |
2962 | /* FALL THROUGH */ | |
95b24440 DM |
2963 | case EXACTF: { |
2964 | char *s = STRING(scan); | |
5d9a96ca | 2965 | st->ln = STR_LEN(scan); |
d6a28714 | 2966 | |
d07ddd77 JH |
2967 | if (do_utf8 || UTF) { |
2968 | /* Either target or the pattern are utf8. */ | |
d6a28714 | 2969 | char *l = locinput; |
d07ddd77 | 2970 | char *e = PL_regeol; |
bc517b45 | 2971 | |
5d9a96ca | 2972 | if (ibcmp_utf8(s, 0, st->ln, (bool)UTF, |
1feea2c7 | 2973 | l, &e, 0, do_utf8)) { |
5486206c JH |
2974 | /* One more case for the sharp s: |
2975 | * pack("U0U*", 0xDF) =~ /ss/i, | |
2976 | * the 0xC3 0x9F are the UTF-8 | |
2977 | * byte sequence for the U+00DF. */ | |
2978 | if (!(do_utf8 && | |
2979 | toLOWER(s[0]) == 's' && | |
5d9a96ca | 2980 | st->ln >= 2 && |
5486206c JH |
2981 | toLOWER(s[1]) == 's' && |
2982 | (U8)l[0] == 0xC3 && | |
2983 | e - l >= 2 && | |
2984 | (U8)l[1] == 0x9F)) | |
2985 | sayNO; | |
2986 | } | |
d07ddd77 JH |
2987 | locinput = e; |
2988 | nextchr = UCHARAT(locinput); | |
2989 | break; | |
a0ed51b3 | 2990 | } |
d6a28714 | 2991 | |
bc517b45 JH |
2992 | /* Neither the target and the pattern are utf8. */ |
2993 | ||
d6a28714 JH |
2994 | /* Inline the first character, for speed. */ |
2995 | if (UCHARAT(s) != nextchr && | |
2996 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
2997 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 2998 | sayNO; |
5d9a96ca | 2999 | if (PL_regeol - locinput < st->ln) |
b8c5462f | 3000 | sayNO; |
5d9a96ca DM |
3001 | if (st->ln > 1 && (OP(scan) == EXACTF |
3002 | ? ibcmp(s, locinput, st->ln) | |
3003 | : ibcmp_locale(s, locinput, st->ln))) | |
4633a7c4 | 3004 | sayNO; |
5d9a96ca | 3005 | locinput += st->ln; |
d6a28714 | 3006 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3007 | break; |
95b24440 | 3008 | } |
d6a28714 | 3009 | case ANYOF: |
ffc61ed2 | 3010 | if (do_utf8) { |
9e55ce06 JH |
3011 | STRLEN inclasslen = PL_regeol - locinput; |
3012 | ||
32fc9b6a | 3013 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
e0f9d4a8 | 3014 | sayNO_ANYOF; |
ffc61ed2 JH |
3015 | if (locinput >= PL_regeol) |
3016 | sayNO; | |
0f0076b4 | 3017 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3018 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3019 | break; |
ffc61ed2 JH |
3020 | } |
3021 | else { | |
3022 | if (nextchr < 0) | |
3023 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3024 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
e0f9d4a8 | 3025 | sayNO_ANYOF; |
ffc61ed2 JH |
3026 | if (!nextchr && locinput >= PL_regeol) |
3027 | sayNO; | |
3028 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3029 | break; |
3030 | } | |
3031 | no_anyof: | |
3032 | /* If we might have the case of the German sharp s | |
3033 | * in a casefolding Unicode character class. */ | |
3034 | ||
ebc501f0 JH |
3035 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3036 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3037 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3038 | } |
e0f9d4a8 JH |
3039 | else |
3040 | sayNO; | |
b8c5462f | 3041 | break; |
d6a28714 | 3042 | case ALNUML: |
b8c5462f JH |
3043 | PL_reg_flags |= RF_tainted; |
3044 | /* FALL THROUGH */ | |
d6a28714 | 3045 | case ALNUM: |
b8c5462f | 3046 | if (!nextchr) |
4633a7c4 | 3047 | sayNO; |
ffc61ed2 | 3048 | if (do_utf8) { |
1a4fad37 | 3049 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3050 | if (!(OP(scan) == ALNUM |
3568d838 | 3051 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3052 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3053 | { |
3054 | sayNO; | |
a0ed51b3 | 3055 | } |
b8c5462f | 3056 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3057 | nextchr = UCHARAT(locinput); |
3058 | break; | |
3059 | } | |
ffc61ed2 | 3060 | if (!(OP(scan) == ALNUM |
d6a28714 | 3061 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3062 | sayNO; |
b8c5462f | 3063 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3064 | break; |
d6a28714 | 3065 | case NALNUML: |
b8c5462f JH |
3066 | PL_reg_flags |= RF_tainted; |
3067 | /* FALL THROUGH */ | |
d6a28714 JH |
3068 | case NALNUM: |
3069 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3070 | sayNO; |
ffc61ed2 | 3071 | if (do_utf8) { |
1a4fad37 | 3072 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3073 | if (OP(scan) == NALNUM |
3568d838 | 3074 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3075 | : isALNUM_LC_utf8((U8*)locinput)) |
3076 | { | |
b8c5462f | 3077 | sayNO; |
d6a28714 | 3078 | } |
b8c5462f JH |
3079 | locinput += PL_utf8skip[nextchr]; |
3080 | nextchr = UCHARAT(locinput); | |
3081 | break; | |
3082 | } | |
ffc61ed2 | 3083 | if (OP(scan) == NALNUM |
d6a28714 | 3084 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3085 | sayNO; |
76e3520e | 3086 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3087 | break; |
d6a28714 JH |
3088 | case BOUNDL: |
3089 | case NBOUNDL: | |
3280af22 | 3090 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3091 | /* FALL THROUGH */ |
d6a28714 JH |
3092 | case BOUND: |
3093 | case NBOUND: | |
3094 | /* was last char in word? */ | |
ffc61ed2 | 3095 | if (do_utf8) { |
12d33761 | 3096 | if (locinput == PL_bostr) |
5d9a96ca | 3097 | st->ln = '\n'; |
ffc61ed2 | 3098 | else { |
a3b680e6 | 3099 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3100 | |
4ad0818d | 3101 | st->ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3102 | } |
3103 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
5d9a96ca | 3104 | st->ln = isALNUM_uni(st->ln); |
1a4fad37 | 3105 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3106 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3107 | } |
3108 | else { | |
5d9a96ca | 3109 | st->ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(st->ln)); |
ffc61ed2 JH |
3110 | n = isALNUM_LC_utf8((U8*)locinput); |
3111 | } | |
a0ed51b3 | 3112 | } |
d6a28714 | 3113 | else { |
5d9a96ca | 3114 | st->ln = (locinput != PL_bostr) ? |
12d33761 | 3115 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3116 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
5d9a96ca | 3117 | st->ln = isALNUM(st->ln); |
ffc61ed2 JH |
3118 | n = isALNUM(nextchr); |
3119 | } | |
3120 | else { | |
5d9a96ca | 3121 | st->ln = isALNUM_LC(st->ln); |
ffc61ed2 JH |
3122 | n = isALNUM_LC(nextchr); |
3123 | } | |
d6a28714 | 3124 | } |
5d9a96ca | 3125 | if (((!st->ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3126 | OP(scan) == BOUNDL)) |
3127 | sayNO; | |
a0ed51b3 | 3128 | break; |
d6a28714 | 3129 | case SPACEL: |
3280af22 | 3130 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3131 | /* FALL THROUGH */ |
d6a28714 | 3132 | case SPACE: |
9442cb0e | 3133 | if (!nextchr) |
4633a7c4 | 3134 | sayNO; |
1aa99e6b | 3135 | if (do_utf8) { |
fd400ab9 | 3136 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3137 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3138 | if (!(OP(scan) == SPACE |
3568d838 | 3139 | ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3140 | : isSPACE_LC_utf8((U8*)locinput))) |
3141 | { | |
3142 | sayNO; | |
3143 | } | |
3144 | locinput += PL_utf8skip[nextchr]; | |
3145 | nextchr = UCHARAT(locinput); | |
3146 | break; | |
d6a28714 | 3147 | } |
ffc61ed2 JH |
3148 | if (!(OP(scan) == SPACE |
3149 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3150 | sayNO; | |
3151 | nextchr = UCHARAT(++locinput); | |
3152 | } | |
3153 | else { | |
3154 | if (!(OP(scan) == SPACE | |
3155 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3156 | sayNO; | |
3157 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3158 | } |
a0ed51b3 | 3159 | break; |
d6a28714 | 3160 | case NSPACEL: |
3280af22 | 3161 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3162 | /* FALL THROUGH */ |
d6a28714 | 3163 | case NSPACE: |
9442cb0e | 3164 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3165 | sayNO; |
1aa99e6b | 3166 | if (do_utf8) { |
1a4fad37 | 3167 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3168 | if (OP(scan) == NSPACE |
3568d838 | 3169 | ? swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3170 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3171 | { |
3172 | sayNO; | |
3173 | } | |
3174 | locinput += PL_utf8skip[nextchr]; | |
3175 | nextchr = UCHARAT(locinput); | |
3176 | break; | |
a0ed51b3 | 3177 | } |
ffc61ed2 | 3178 | if (OP(scan) == NSPACE |
d6a28714 | 3179 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3180 | sayNO; |
76e3520e | 3181 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3182 | break; |
d6a28714 | 3183 | case DIGITL: |
a0ed51b3 LW |
3184 | PL_reg_flags |= RF_tainted; |
3185 | /* FALL THROUGH */ | |
d6a28714 | 3186 | case DIGIT: |
9442cb0e | 3187 | if (!nextchr) |
a0ed51b3 | 3188 | sayNO; |
1aa99e6b | 3189 | if (do_utf8) { |
1a4fad37 | 3190 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3191 | if (!(OP(scan) == DIGIT |
3568d838 | 3192 | ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3193 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3194 | { |
a0ed51b3 | 3195 | sayNO; |
dfe13c55 | 3196 | } |
6f06b55f | 3197 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3198 | nextchr = UCHARAT(locinput); |
3199 | break; | |
3200 | } | |
ffc61ed2 | 3201 | if (!(OP(scan) == DIGIT |
9442cb0e | 3202 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3203 | sayNO; |
3204 | nextchr = UCHARAT(++locinput); | |
3205 | break; | |
d6a28714 | 3206 | case NDIGITL: |
b8c5462f JH |
3207 | PL_reg_flags |= RF_tainted; |
3208 | /* FALL THROUGH */ | |
d6a28714 | 3209 | case NDIGIT: |
9442cb0e | 3210 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3211 | sayNO; |
1aa99e6b | 3212 | if (do_utf8) { |
1a4fad37 | 3213 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3214 | if (OP(scan) == NDIGIT |
3568d838 | 3215 | ? swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3216 | : isDIGIT_LC_utf8((U8*)locinput)) |
3217 | { | |
a0ed51b3 | 3218 | sayNO; |
9442cb0e | 3219 | } |
6f06b55f | 3220 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3221 | nextchr = UCHARAT(locinput); |
3222 | break; | |
3223 | } | |
ffc61ed2 | 3224 | if (OP(scan) == NDIGIT |
9442cb0e | 3225 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3226 | sayNO; |
3227 | nextchr = UCHARAT(++locinput); | |
3228 | break; | |
3229 | case CLUMP: | |
b7c83a7e | 3230 | if (locinput >= PL_regeol) |
a0ed51b3 | 3231 | sayNO; |
b7c83a7e | 3232 | if (do_utf8) { |
1a4fad37 | 3233 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3234 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3235 | sayNO; | |
3236 | locinput += PL_utf8skip[nextchr]; | |
3237 | while (locinput < PL_regeol && | |
3238 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3239 | locinput += UTF8SKIP(locinput); | |
3240 | if (locinput > PL_regeol) | |
3241 | sayNO; | |
eb08e2da JH |
3242 | } |
3243 | else | |
3244 | locinput++; | |
a0ed51b3 LW |
3245 | nextchr = UCHARAT(locinput); |
3246 | break; | |
c8756f30 | 3247 | case REFFL: |
3280af22 | 3248 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3249 | /* FALL THROUGH */ |
c277df42 | 3250 | case REF: |
95b24440 DM |
3251 | case REFF: { |
3252 | char *s; | |
c277df42 | 3253 | n = ARG(scan); /* which paren pair */ |
5d9a96ca | 3254 | st->ln = PL_regstartp[n]; |
2c2d71f5 | 3255 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
5d9a96ca | 3256 | if ((I32)*PL_reglastparen < n || st->ln == -1) |
af3f8c16 | 3257 | sayNO; /* Do not match unless seen CLOSEn. */ |
5d9a96ca | 3258 | if (st->ln == PL_regendp[n]) |
a0d0e21e | 3259 | break; |
a0ed51b3 | 3260 | |
5d9a96ca | 3261 | s = PL_bostr + st->ln; |
1aa99e6b | 3262 | if (do_utf8 && OP(scan) != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3263 | char *l = locinput; |
a3b680e6 | 3264 | const char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 LW |
3265 | /* |
3266 | * Note that we can't do the "other character" lookup trick as | |
3267 | * in the 8-bit case (no pun intended) because in Unicode we | |
3268 | * have to map both upper and title case to lower case. | |
3269 | */ | |
3270 | if (OP(scan) == REFF) { | |
3271 | while (s < e) { | |
a3b680e6 AL |
3272 | STRLEN ulen1, ulen2; |
3273 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3274 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3275 | ||
a0ed51b3 LW |
3276 | if (l >= PL_regeol) |
3277 | sayNO; | |
a2a2844f JH |
3278 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3279 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3280 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3281 | sayNO; |
a2a2844f JH |
3282 | s += ulen1; |
3283 | l += ulen2; | |
a0ed51b3 LW |
3284 | } |
3285 | } | |
3286 | locinput = l; | |
3287 | nextchr = UCHARAT(locinput); | |
3288 | break; | |
3289 | } | |
3290 | ||
a0d0e21e | 3291 | /* Inline the first character, for speed. */ |
76e3520e | 3292 | if (UCHARAT(s) != nextchr && |
c8756f30 AK |
3293 | (OP(scan) == REF || |
3294 | (UCHARAT(s) != ((OP(scan) == REFF | |
22c35a8c | 3295 | ? PL_fold : PL_fold_locale)[nextchr])))) |
4633a7c4 | 3296 | sayNO; |
5d9a96ca DM |
3297 | st->ln = PL_regendp[n] - st->ln; |
3298 | if (locinput + st->ln > PL_regeol) | |
4633a7c4 | 3299 | sayNO; |
5d9a96ca DM |
3300 | if (st->ln > 1 && (OP(scan) == REF |
3301 | ? memNE(s, locinput, st->ln) | |
c8756f30 | 3302 | : (OP(scan) == REFF |
5d9a96ca DM |
3303 | ? ibcmp(s, locinput, st->ln) |
3304 | : ibcmp_locale(s, locinput, st->ln)))) | |
4633a7c4 | 3305 | sayNO; |
5d9a96ca | 3306 | locinput += st->ln; |
76e3520e | 3307 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3308 | break; |
95b24440 | 3309 | } |
a0d0e21e LW |
3310 | |
3311 | case NOTHING: | |
c277df42 | 3312 | case TAIL: |
a0d0e21e LW |
3313 | break; |
3314 | case BACK: | |
3315 | break; | |
c277df42 IZ |
3316 | case EVAL: |
3317 | { | |
c277df42 | 3318 | SV *ret; |
8e5e9ebe | 3319 | { |
4aabdb9b DM |
3320 | /* execute the code in the {...} */ |
3321 | dSP; | |
6136c704 | 3322 | SV ** const before = SP; |
4aabdb9b DM |
3323 | OP_4tree * const oop = PL_op; |
3324 | COP * const ocurcop = PL_curcop; | |
3325 | PAD *old_comppad; | |
4aabdb9b DM |
3326 | |
3327 | n = ARG(scan); | |
32fc9b6a | 3328 | PL_op = (OP_4tree*)rex->data->data[n]; |
4aabdb9b | 3329 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); |
32fc9b6a | 3330 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rex->data->data[n + 2]); |
4aabdb9b DM |
3331 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
3332 | ||
8e5e9ebe RGS |
3333 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3334 | SPAGAIN; | |
3335 | if (SP == before) | |
075aa684 | 3336 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3337 | else { |
3338 | ret = POPs; | |
3339 | PUTBACK; | |
3340 | } | |
4aabdb9b DM |
3341 | |
3342 | PL_op = oop; | |
3343 | PAD_RESTORE_LOCAL(old_comppad); | |
3344 | PL_curcop = ocurcop; | |
3345 | if (!st->logical) { | |
3346 | /* /(?{...})/ */ | |
3347 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3348 | break; |
3349 | } | |
8e5e9ebe | 3350 | } |
4aabdb9b DM |
3351 | if (st->logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3352 | regexp *re; | |
4aabdb9b | 3353 | { |
4f639d21 DM |
3354 | /* extract RE object from returned value; compiling if |
3355 | * necessary */ | |
3356 | ||
6136c704 | 3357 | MAGIC *mg = NULL; |
4aabdb9b | 3358 | SV *sv; |
faf82a0b AE |
3359 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3360 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3361 | else if (SvSMAGICAL(ret)) { | |
3362 | if (SvGMAGICAL(ret)) | |
3363 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3364 | else | |
3365 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3366 | } |
faf82a0b | 3367 | |
0f5d15d6 IZ |
3368 | if (mg) { |
3369 | re = (regexp *)mg->mg_obj; | |
df0003d4 | 3370 | (void)ReREFCNT_inc(re); |
0f5d15d6 IZ |
3371 | } |
3372 | else { | |
3373 | STRLEN len; | |
6136c704 | 3374 | const char * const t = SvPV_const(ret, len); |
0f5d15d6 | 3375 | PMOP pm; |
a3b680e6 | 3376 | const I32 osize = PL_regsize; |
0f5d15d6 | 3377 | |
5fcd1c1b | 3378 | Zero(&pm, 1, PMOP); |
4aabdb9b | 3379 | if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; |
83003860 | 3380 | re = CALLREGCOMP(aTHX_ (char*)t, (char*)t + len, &pm); |
9041c2e3 |