Commit | Line | Data |
a0d0e21e |
1 | /* regexec.c |
2 | */ |
3 | |
4 | /* |
5 | * "One Ring to rule them all, One Ring to find them..." |
6 | */ |
7 | |
a687059c |
8 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
9 | * confused with the original package (see point 3 below). Thanks, Henry! |
10 | */ |
11 | |
12 | /* Additional note: this code is very heavily munged from Henry's version |
13 | * in places. In some spots I've traded clarity for efficiency, so don't |
14 | * blame Henry for some of the lack of readability. |
15 | */ |
16 | |
e50aee73 |
17 | /* The names of the functions have been changed from regcomp and |
18 | * regexec to pregcomp and pregexec in order to avoid conflicts |
19 | * with the POSIX routines of the same names. |
20 | */ |
21 | |
b9d5759e |
22 | #ifdef PERL_EXT_RE_BUILD |
23 | /* need to replace pregcomp et al, so enable that */ |
24 | # ifndef PERL_IN_XSUB_RE |
25 | # define PERL_IN_XSUB_RE |
26 | # endif |
27 | /* need access to debugger hooks */ |
28 | # ifndef DEBUGGING |
29 | # define DEBUGGING |
30 | # endif |
31 | #endif |
32 | |
33 | #ifdef PERL_IN_XSUB_RE |
d06ea78c |
34 | /* We *really* need to overwrite these symbols: */ |
56953603 |
35 | # define Perl_regexec_flags my_regexec |
36 | # define Perl_regdump my_regdump |
37 | # define Perl_regprop my_regprop |
d06ea78c |
38 | /* *These* symbols are masked to allow static link. */ |
39 | # define Perl_pregexec my_pregexec |
d88dccdf |
40 | # define Perl_reginitcolors my_reginitcolors |
56953603 |
41 | #endif |
42 | |
f0fcb552 |
43 | /*SUPPRESS 112*/ |
a687059c |
44 | /* |
e50aee73 |
45 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
46 | * |
47 | * Copyright (c) 1986 by University of Toronto. |
48 | * Written by Henry Spencer. Not derived from licensed software. |
49 | * |
50 | * Permission is granted to anyone to use this software for any |
51 | * purpose on any computer system, and to redistribute it freely, |
52 | * subject to the following restrictions: |
53 | * |
54 | * 1. The author is not responsible for the consequences of use of |
55 | * this software, no matter how awful, even if they arise |
56 | * from defects in it. |
57 | * |
58 | * 2. The origin of this software must not be misrepresented, either |
59 | * by explicit claim or by omission. |
60 | * |
61 | * 3. Altered versions must be plainly marked as such, and must not |
62 | * be misrepresented as being the original software. |
63 | * |
64 | **** Alterations to Henry's code are... |
65 | **** |
4eb8286e |
66 | **** Copyright (c) 1991-1999, Larry Wall |
a687059c |
67 | **** |
9ef589d8 |
68 | **** You may distribute under the terms of either the GNU General Public |
69 | **** License or the Artistic License, as specified in the README file. |
a687059c |
70 | * |
71 | * Beware that some of this code is subtly aware of the way operator |
72 | * precedence is structured in regular expressions. Serious changes in |
73 | * regular-expression syntax might require a total rethink. |
74 | */ |
75 | #include "EXTERN.h" |
864dbfa3 |
76 | #define PERL_IN_REGEXEC_C |
a687059c |
77 | #include "perl.h" |
0f5d15d6 |
78 | |
a687059c |
79 | #include "regcomp.h" |
80 | |
c277df42 |
81 | #define RF_tainted 1 /* tainted information used? */ |
82 | #define RF_warned 2 /* warned about big count? */ |
ce862d02 |
83 | #define RF_evaled 4 /* Did an EVAL with setting? */ |
a0ed51b3 |
84 | #define RF_utf8 8 /* String contains multibyte chars? */ |
85 | |
86 | #define UTF (PL_reg_flags & RF_utf8) |
ce862d02 |
87 | |
88 | #define RS_init 1 /* eval environment created */ |
89 | #define RS_set 2 /* replsv value is set */ |
c277df42 |
90 | |
a687059c |
91 | #ifndef STATIC |
92 | #define STATIC static |
93 | #endif |
94 | |
c277df42 |
95 | /* |
96 | * Forwards. |
97 | */ |
98 | |
ae5c130c |
99 | #define REGINCLASS(p,c) (*(p) ? reginclass(p,c) : ANYOF_TEST(p,c)) |
a0ed51b3 |
100 | #define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p)) |
101 | |
102 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
103 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) |
104 | |
dfe13c55 |
105 | #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) |
106 | #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) |
107 | #define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) |
108 | #define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) |
109 | #define HOPc(pos,off) ((char*)HOP(pos,off)) |
110 | #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) |
a0d0e21e |
111 | |
76e3520e |
112 | STATIC CHECKPOINT |
cea2e8a9 |
113 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e |
114 | { |
11343788 |
115 | dTHR; |
3280af22 |
116 | int retval = PL_savestack_ix; |
117 | int i = (PL_regsize - parenfloor) * 4; |
a0d0e21e |
118 | int p; |
119 | |
120 | SSCHECK(i + 5); |
3280af22 |
121 | for (p = PL_regsize; p > parenfloor; p--) { |
cf93c79d |
122 | SSPUSHINT(PL_regendp[p]); |
123 | SSPUSHINT(PL_regstartp[p]); |
3280af22 |
124 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e |
125 | SSPUSHINT(p); |
126 | } |
3280af22 |
127 | SSPUSHINT(PL_regsize); |
128 | SSPUSHINT(*PL_reglastparen); |
129 | SSPUSHPTR(PL_reginput); |
a0d0e21e |
130 | SSPUSHINT(i + 3); |
131 | SSPUSHINT(SAVEt_REGCONTEXT); |
132 | return retval; |
133 | } |
134 | |
c277df42 |
135 | /* These are needed since we do not localize EVAL nodes: */ |
c3464db5 |
136 | # define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \ |
137 | " Setting an EVAL scope, savestack=%i\n", \ |
3280af22 |
138 | PL_savestack_ix)); lastcp = PL_savestack_ix |
c3464db5 |
139 | |
3280af22 |
140 | # define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \ |
c3464db5 |
141 | PerlIO_printf(Perl_debug_log, \ |
142 | " Clearing an EVAL scope, savestack=%i..%i\n", \ |
3280af22 |
143 | lastcp, PL_savestack_ix) : 0); regcpblow(lastcp) |
c277df42 |
144 | |
76e3520e |
145 | STATIC char * |
cea2e8a9 |
146 | S_regcppop(pTHX) |
a0d0e21e |
147 | { |
11343788 |
148 | dTHR; |
a0d0e21e |
149 | I32 i = SSPOPINT; |
150 | U32 paren = 0; |
151 | char *input; |
cf93c79d |
152 | I32 tmps; |
a0d0e21e |
153 | assert(i == SAVEt_REGCONTEXT); |
154 | i = SSPOPINT; |
155 | input = (char *) SSPOPPTR; |
3280af22 |
156 | *PL_reglastparen = SSPOPINT; |
157 | PL_regsize = SSPOPINT; |
c277df42 |
158 | for (i -= 3; i > 0; i -= 4) { |
a0d0e21e |
159 | paren = (U32)SSPOPINT; |
3280af22 |
160 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d |
161 | PL_regstartp[paren] = SSPOPINT; |
162 | tmps = SSPOPINT; |
3280af22 |
163 | if (paren <= *PL_reglastparen) |
164 | PL_regendp[paren] = tmps; |
c277df42 |
165 | DEBUG_r( |
c3464db5 |
166 | PerlIO_printf(Perl_debug_log, |
167 | " restoring \\%d to %d(%d)..%d%s\n", |
cf93c79d |
168 | paren, PL_regstartp[paren], |
169 | PL_reg_start_tmp[paren] - PL_bostr, |
170 | PL_regendp[paren], |
3280af22 |
171 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 |
172 | ); |
a0d0e21e |
173 | } |
c277df42 |
174 | DEBUG_r( |
3280af22 |
175 | if (*PL_reglastparen + 1 <= PL_regnpar) { |
c3464db5 |
176 | PerlIO_printf(Perl_debug_log, |
177 | " restoring \\%d..\\%d to undef\n", |
3280af22 |
178 | *PL_reglastparen + 1, PL_regnpar); |
c277df42 |
179 | } |
180 | ); |
3280af22 |
181 | for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { |
182 | if (paren > PL_regsize) |
cf93c79d |
183 | PL_regstartp[paren] = -1; |
184 | PL_regendp[paren] = -1; |
a0d0e21e |
185 | } |
186 | return input; |
187 | } |
188 | |
0f5d15d6 |
189 | STATIC char * |
cea2e8a9 |
190 | S_regcp_set_to(pTHX_ I32 ss) |
0f5d15d6 |
191 | { |
46124e9e |
192 | dTHR; |
0f5d15d6 |
193 | I32 tmp = PL_savestack_ix; |
194 | |
195 | PL_savestack_ix = ss; |
196 | regcppop(); |
197 | PL_savestack_ix = tmp; |
942e002e |
198 | return Nullch; |
0f5d15d6 |
199 | } |
200 | |
201 | typedef struct re_cc_state |
202 | { |
203 | I32 ss; |
204 | regnode *node; |
205 | struct re_cc_state *prev; |
206 | CURCUR *cc; |
207 | regexp *re; |
208 | } re_cc_state; |
209 | |
c277df42 |
210 | #define regcpblow(cp) LEAVE_SCOPE(cp) |
a0d0e21e |
211 | |
a687059c |
212 | /* |
e50aee73 |
213 | * pregexec and friends |
a687059c |
214 | */ |
215 | |
216 | /* |
c277df42 |
217 | - pregexec - match a regexp against a string |
a687059c |
218 | */ |
c277df42 |
219 | I32 |
864dbfa3 |
220 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 |
221 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 |
222 | /* strend: pointer to null at end of string */ |
223 | /* strbeg: real beginning of string */ |
224 | /* minend: end of match must be >=minend after stringarg. */ |
225 | /* nosave: For optimizations. */ |
226 | { |
227 | return |
228 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
229 | nosave ? 0 : REXEC_COPY_STR); |
230 | } |
0f5d15d6 |
231 | |
232 | STATIC void |
cea2e8a9 |
233 | S_cache_re(pTHX_ regexp *prog) |
0f5d15d6 |
234 | { |
46124e9e |
235 | dTHR; |
0f5d15d6 |
236 | PL_regprecomp = prog->precomp; /* Needed for FAIL. */ |
237 | #ifdef DEBUGGING |
238 | PL_regprogram = prog->program; |
239 | #endif |
240 | PL_regnpar = prog->nparens; |
241 | PL_regdata = prog->data; |
242 | PL_reg_re = prog; |
243 | } |
22e551b9 |
244 | |
9661b544 |
245 | STATIC void |
cea2e8a9 |
246 | S_restore_pos(pTHX_ void *arg) |
35ef4773 |
247 | { |
248 | dTHR; |
cf93c79d |
249 | if (PL_reg_eval_set) { |
250 | if (PL_reg_oldsaved) { |
251 | PL_reg_re->subbeg = PL_reg_oldsaved; |
252 | PL_reg_re->sublen = PL_reg_oldsavedlen; |
253 | RX_MATCH_COPIED_on(PL_reg_re); |
254 | } |
9661b544 |
255 | PL_reg_magic->mg_len = PL_reg_oldpos; |
256 | PL_reg_eval_set = 0; |
5c5e4c24 |
257 | PL_curpm = PL_reg_oldcurpm; |
9661b544 |
258 | } |
259 | } |
260 | |
261 | |
a687059c |
262 | /* |
c277df42 |
263 | - regexec_flags - match a regexp against a string |
a687059c |
264 | */ |
79072805 |
265 | I32 |
864dbfa3 |
266 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
22e551b9 |
267 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) |
c277df42 |
268 | /* strend: pointer to null at end of string */ |
269 | /* strbeg: real beginning of string */ |
270 | /* minend: end of match must be >=minend after stringarg. */ |
271 | /* data: May be used for some additional optimizations. */ |
272 | /* nosave: For optimizations. */ |
a687059c |
273 | { |
5c0ca799 |
274 | dTHR; |
a0d0e21e |
275 | register char *s; |
c277df42 |
276 | register regnode *c; |
a0d0e21e |
277 | register char *startpos = stringarg; |
278 | register I32 tmp; |
c277df42 |
279 | I32 minlen; /* must match at least this many chars */ |
a0d0e21e |
280 | I32 dontbother = 0; /* how many characters not to try at end */ |
281 | CURCUR cc; |
c277df42 |
282 | I32 start_shift = 0; /* Offset of the start to find |
a0ed51b3 |
283 | constant substr. */ /* CC */ |
284 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
c277df42 |
285 | I32 scream_pos = -1; /* Internal iterator of scream. */ |
286 | char *scream_olds; |
3280af22 |
287 | SV* oreplsv = GvSV(PL_replgv); |
a687059c |
288 | |
a0d0e21e |
289 | cc.cur = 0; |
4633a7c4 |
290 | cc.oldcc = 0; |
3280af22 |
291 | PL_regcc = &cc; |
a0d0e21e |
292 | |
0f5d15d6 |
293 | cache_re(prog); |
a0d0e21e |
294 | #ifdef DEBUGGING |
3280af22 |
295 | PL_regnarrate = PL_debug & 512; |
a0d0e21e |
296 | #endif |
297 | |
298 | /* Be paranoid... */ |
299 | if (prog == NULL || startpos == NULL) { |
cea2e8a9 |
300 | Perl_croak(aTHX_ "NULL regexp parameter"); |
a0d0e21e |
301 | return 0; |
302 | } |
303 | |
c277df42 |
304 | minlen = prog->minlen; |
305 | if (strend - startpos < minlen) goto phooey; |
306 | |
a0d0e21e |
307 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
3280af22 |
308 | PL_regprev = '\n'; |
a0d0e21e |
309 | else { |
a0ed51b3 |
310 | PL_regprev = (U32)stringarg[-1]; |
3280af22 |
311 | if (!PL_multiline && PL_regprev == '\n') |
312 | PL_regprev = '\0'; /* force ^ to NOT match */ |
a0d0e21e |
313 | } |
bbce6d69 |
314 | |
a0d0e21e |
315 | /* Check validity of program. */ |
22c35a8c |
316 | if (UCHARAT(prog->program) != REG_MAGIC) { |
cea2e8a9 |
317 | Perl_croak(aTHX_ "corrupted regexp program"); |
a0d0e21e |
318 | } |
319 | |
3280af22 |
320 | PL_reg_flags = 0; |
321 | PL_reg_eval_set = 0; |
a0d0e21e |
322 | |
a0ed51b3 |
323 | if (prog->reganch & ROPT_UTF8) |
324 | PL_reg_flags |= RF_utf8; |
325 | |
326 | /* Mark beginning of line for ^ and lookbehind. */ |
327 | PL_regbol = startpos; |
328 | PL_bostr = strbeg; |
9661b544 |
329 | PL_reg_sv = sv; |
a0ed51b3 |
330 | |
331 | /* Mark end of line for $ (and such) */ |
332 | PL_regeol = strend; |
333 | |
334 | /* see how far we have to get to not match where we matched before */ |
335 | PL_regtill = startpos+minend; |
336 | |
0f5d15d6 |
337 | /* We start without call_cc context. */ |
338 | PL_reg_call_cc = 0; |
339 | |
a0d0e21e |
340 | /* If there is a "must appear" string, look for it. */ |
341 | s = startpos; |
c277df42 |
342 | if (!(flags & REXEC_CHECKED) |
343 | && prog->check_substr != Nullsv && |
774d564b |
344 | !(prog->reganch & ROPT_ANCH_GPOS) && |
c277df42 |
345 | (!(prog->reganch & (ROPT_ANCH_BOL | ROPT_ANCH_MBOL)) |
3280af22 |
346 | || (PL_multiline && prog->check_substr == prog->anchored_substr)) ) |
a0d0e21e |
347 | { |
a0ed51b3 |
348 | char *t; |
349 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
c277df42 |
350 | /* Should be nonnegative! */ |
cf93c79d |
351 | end_shift = minlen - start_shift - |
352 | CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0); |
22e551b9 |
353 | if (flags & REXEC_SCREAM) { |
cf93c79d |
354 | SV *c = prog->check_substr; |
355 | |
356 | if (PL_screamfirst[BmRARE(c)] >= 0 |
357 | || ( BmRARE(c) == '\n' |
358 | && (BmPREVIOUS(c) == SvCUR(c) - 1) |
359 | && SvTAIL(c) )) |
22e551b9 |
360 | s = screaminstr(sv, prog->check_substr, |
c277df42 |
361 | start_shift + (stringarg - strbeg), |
362 | end_shift, &scream_pos, 0); |
a0d0e21e |
363 | else |
364 | s = Nullch; |
c277df42 |
365 | scream_olds = s; |
0a12ae7d |
366 | } |
a0d0e21e |
367 | else |
c277df42 |
368 | s = fbm_instr((unsigned char*)s + start_shift, |
369 | (unsigned char*)strend - end_shift, |
cf93c79d |
370 | prog->check_substr, PL_multiline ? FBMrf_MULTILINE : 0); |
a0d0e21e |
371 | if (!s) { |
c277df42 |
372 | ++BmUSEFUL(prog->check_substr); /* hooray */ |
a0d0e21e |
373 | goto phooey; /* not present */ |
a0ed51b3 |
374 | } |
375 | else if (s - stringarg > prog->check_offset_max && |
376 | (UTF |
dfe13c55 |
377 | ? ((t = reghopmaybe_c(s, -(prog->check_offset_max))) && t >= stringarg) |
a0ed51b3 |
378 | : (t = s - prog->check_offset_max) != 0 |
379 | ) |
380 | ) |
381 | { |
c277df42 |
382 | ++BmUSEFUL(prog->check_substr); /* hooray/2 */ |
a0ed51b3 |
383 | s = t; |
384 | } |
385 | else if (!(prog->reganch & ROPT_NAUGHTY) |
c277df42 |
386 | && --BmUSEFUL(prog->check_substr) < 0 |
387 | && prog->check_substr == prog->float_substr) { /* boo */ |
388 | SvREFCNT_dec(prog->check_substr); |
389 | prog->check_substr = Nullsv; /* disable */ |
390 | prog->float_substr = Nullsv; /* clear */ |
a0d0e21e |
391 | s = startpos; |
a0ed51b3 |
392 | } |
393 | else |
394 | s = startpos; |
a0d0e21e |
395 | } |
a687059c |
396 | |
35ef4773 |
397 | DEBUG_r(if (!PL_colorset) reginitcolors()); |
398 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
8d300b32 |
399 | "%sMatching%s `%s%.60s%s%s' against `%s%.*s%s%s'\n", |
400 | PL_colors[4],PL_colors[5],PL_colors[0], |
401 | prog->precomp, |
402 | PL_colors[1], |
c277df42 |
403 | (strlen(prog->precomp) > 60 ? "..." : ""), |
8d300b32 |
404 | PL_colors[0], |
c277df42 |
405 | (strend - startpos > 60 ? 60 : strend - startpos), |
8d300b32 |
406 | startpos, PL_colors[1], |
c277df42 |
407 | (strend - startpos > 60 ? "..." : "")) |
408 | ); |
409 | |
22e551b9 |
410 | if (prog->reganch & ROPT_GPOS_SEEN) { |
411 | MAGIC *mg; |
22e551b9 |
412 | |
ad94a511 |
413 | if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG |
414 | && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) |
415 | PL_reg_ganch = strbeg + mg->mg_len; |
416 | else |
417 | PL_reg_ganch = startpos; |
22e551b9 |
418 | } |
419 | |
a0d0e21e |
420 | /* Simplest case: anchored match need be tried only once. */ |
774d564b |
421 | /* [unless only anchor is BOL and multiline is set] */ |
22e551b9 |
422 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { |
a0d0e21e |
423 | if (regtry(prog, startpos)) |
424 | goto got_it; |
22e551b9 |
425 | else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) |
426 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
774d564b |
427 | { |
a0d0e21e |
428 | if (minlen) |
429 | dontbother = minlen - 1; |
dfe13c55 |
430 | strend = HOPc(strend, -dontbother); |
a0d0e21e |
431 | /* for multiline we only have to try after newlines */ |
432 | if (s > startpos) |
433 | s--; |
434 | while (s < strend) { |
6f06b55f |
435 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ |
a0d0e21e |
436 | if (s < strend && regtry(prog, s)) |
437 | goto got_it; |
438 | } |
35c8bce7 |
439 | } |
35c8bce7 |
440 | } |
a0d0e21e |
441 | goto phooey; |
22e551b9 |
442 | } else if (prog->reganch & ROPT_ANCH_GPOS) { |
443 | if (regtry(prog, PL_reg_ganch)) |
444 | goto got_it; |
445 | goto phooey; |
a0d0e21e |
446 | } |
35c8bce7 |
447 | |
a0d0e21e |
448 | /* Messy cases: unanchored match. */ |
c277df42 |
449 | if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { |
450 | /* we have /x+whatever/ */ |
451 | /* it must be a one character string */ |
452 | char ch = SvPVX(prog->anchored_substr)[0]; |
a0ed51b3 |
453 | if (UTF) { |
454 | while (s < strend) { |
455 | if (*s == ch) { |
456 | if (regtry(prog, s)) goto got_it; |
457 | s += UTF8SKIP(s); |
458 | while (s < strend && *s == ch) |
459 | s += UTF8SKIP(s); |
460 | } |
461 | s += UTF8SKIP(s); |
462 | } |
463 | } |
464 | else { |
465 | while (s < strend) { |
466 | if (*s == ch) { |
467 | if (regtry(prog, s)) goto got_it; |
c277df42 |
468 | s++; |
a0ed51b3 |
469 | while (s < strend && *s == ch) |
470 | s++; |
471 | } |
472 | s++; |
a0d0e21e |
473 | } |
a687059c |
474 | } |
c277df42 |
475 | } |
476 | /*SUPPRESS 560*/ |
477 | else if (prog->anchored_substr != Nullsv |
478 | || (prog->float_substr != Nullsv |
479 | && prog->float_max_offset < strend - s)) { |
480 | SV *must = prog->anchored_substr |
481 | ? prog->anchored_substr : prog->float_substr; |
482 | I32 back_max = |
483 | prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; |
484 | I32 back_min = |
485 | prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; |
486 | I32 delta = back_max - back_min; |
cf93c79d |
487 | char *last = HOPc(strend, /* Cannot start after this */ |
488 | -(CHR_SVLEN(must) - (SvTAIL(must) != 0) + back_min)); |
a0ed51b3 |
489 | char *last1; /* Last position checked before */ |
490 | |
491 | if (s > PL_bostr) |
dfe13c55 |
492 | last1 = HOPc(s, -1); |
a0ed51b3 |
493 | else |
494 | last1 = s - 1; /* bogus */ |
c277df42 |
495 | |
496 | /* XXXX check_substr already used to find `s', can optimize if |
497 | check_substr==must. */ |
498 | scream_pos = -1; |
499 | dontbother = end_shift; |
dfe13c55 |
500 | strend = HOPc(strend, -dontbother); |
c277df42 |
501 | while ( (s <= last) && |
22e551b9 |
502 | ((flags & REXEC_SCREAM) |
503 | ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, |
c277df42 |
504 | end_shift, &scream_pos, 0)) |
a0ed51b3 |
505 | : (s = fbm_instr((unsigned char*)HOP(s, back_min), |
cf93c79d |
506 | (unsigned char*)strend, must, |
507 | PL_multiline ? FBMrf_MULTILINE : 0))) ) { |
dfe13c55 |
508 | if (HOPc(s, -back_max) > last1) { |
509 | last1 = HOPc(s, -back_min); |
510 | s = HOPc(s, -back_max); |
a0ed51b3 |
511 | } |
512 | else { |
dfe13c55 |
513 | char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
c277df42 |
514 | |
dfe13c55 |
515 | last1 = HOPc(s, -back_min); |
c277df42 |
516 | s = t; |
a0d0e21e |
517 | } |
a0ed51b3 |
518 | if (UTF) { |
519 | while (s <= last1) { |
520 | if (regtry(prog, s)) |
521 | goto got_it; |
522 | s += UTF8SKIP(s); |
523 | } |
524 | } |
525 | else { |
526 | while (s <= last1) { |
527 | if (regtry(prog, s)) |
528 | goto got_it; |
529 | s++; |
530 | } |
a0d0e21e |
531 | } |
532 | } |
533 | goto phooey; |
a0ed51b3 |
534 | } |
535 | else if (c = prog->regstclass) { |
a0d0e21e |
536 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
a0ed51b3 |
537 | char *cc; |
a687059c |
538 | |
a0d0e21e |
539 | if (minlen) |
540 | dontbother = minlen - 1; |
dfe13c55 |
541 | strend = HOPc(strend, -dontbother); /* don't bother with what can't match */ |
a0d0e21e |
542 | tmp = 1; |
543 | /* We know what class it must start with. */ |
544 | switch (OP(c)) { |
a0ed51b3 |
545 | case ANYOFUTF8: |
546 | cc = (char *) OPERAND(c); |
547 | while (s < strend) { |
548 | if (REGINCLASSUTF8(c, (U8*)s)) { |
549 | if (tmp && regtry(prog, s)) |
550 | goto got_it; |
551 | else |
552 | tmp = doevery; |
553 | } |
554 | else |
555 | tmp = 1; |
556 | s += UTF8SKIP(s); |
557 | } |
558 | break; |
a0d0e21e |
559 | case ANYOF: |
a0ed51b3 |
560 | cc = (char *) OPERAND(c); |
a0d0e21e |
561 | while (s < strend) { |
a0ed51b3 |
562 | if (REGINCLASS(cc, *s)) { |
a0d0e21e |
563 | if (tmp && regtry(prog, s)) |
564 | goto got_it; |
565 | else |
566 | tmp = doevery; |
a687059c |
567 | } |
a0d0e21e |
568 | else |
569 | tmp = 1; |
570 | s++; |
571 | } |
572 | break; |
bbce6d69 |
573 | case BOUNDL: |
3280af22 |
574 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
575 | /* FALL THROUGH */ |
a0d0e21e |
576 | case BOUND: |
a0ed51b3 |
577 | if (minlen) { |
578 | dontbother++; |
579 | strend -= 1; |
580 | } |
3280af22 |
581 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
582 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
583 | while (s < strend) { |
95bac841 |
584 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e |
585 | tmp = !tmp; |
586 | if (regtry(prog, s)) |
587 | goto got_it; |
a687059c |
588 | } |
a0d0e21e |
589 | s++; |
590 | } |
591 | if ((minlen || tmp) && regtry(prog,s)) |
592 | goto got_it; |
593 | break; |
a0ed51b3 |
594 | case BOUNDLUTF8: |
595 | PL_reg_flags |= RF_tainted; |
596 | /* FALL THROUGH */ |
597 | case BOUNDUTF8: |
598 | if (minlen) { |
599 | dontbother++; |
dfe13c55 |
600 | strend = reghop_c(strend, -1); |
a0ed51b3 |
601 | } |
dfe13c55 |
602 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
603 | tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
604 | while (s < strend) { |
dfe13c55 |
605 | if (tmp == !(OP(c) == BOUND ? |
606 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
607 | isALNUM_LC_utf8((U8*)s))) |
608 | { |
a0ed51b3 |
609 | tmp = !tmp; |
610 | if (regtry(prog, s)) |
611 | goto got_it; |
612 | } |
613 | s += UTF8SKIP(s); |
614 | } |
615 | if ((minlen || tmp) && regtry(prog,s)) |
616 | goto got_it; |
617 | break; |
bbce6d69 |
618 | case NBOUNDL: |
3280af22 |
619 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
620 | /* FALL THROUGH */ |
a0d0e21e |
621 | case NBOUND: |
a0ed51b3 |
622 | if (minlen) { |
623 | dontbother++; |
624 | strend -= 1; |
625 | } |
3280af22 |
626 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
627 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
628 | while (s < strend) { |
95bac841 |
629 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e |
630 | tmp = !tmp; |
631 | else if (regtry(prog, s)) |
632 | goto got_it; |
633 | s++; |
634 | } |
635 | if ((minlen || !tmp) && regtry(prog,s)) |
636 | goto got_it; |
637 | break; |
a0ed51b3 |
638 | case NBOUNDLUTF8: |
639 | PL_reg_flags |= RF_tainted; |
640 | /* FALL THROUGH */ |
641 | case NBOUNDUTF8: |
642 | if (minlen) { |
643 | dontbother++; |
dfe13c55 |
644 | strend = reghop_c(strend, -1); |
a0ed51b3 |
645 | } |
dfe13c55 |
646 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
647 | tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
648 | while (s < strend) { |
dfe13c55 |
649 | if (tmp == !(OP(c) == NBOUND ? |
650 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
651 | isALNUM_LC_utf8((U8*)s))) |
a0ed51b3 |
652 | tmp = !tmp; |
653 | else if (regtry(prog, s)) |
654 | goto got_it; |
655 | s += UTF8SKIP(s); |
656 | } |
657 | if ((minlen || !tmp) && regtry(prog,s)) |
658 | goto got_it; |
659 | break; |
a0d0e21e |
660 | case ALNUM: |
661 | while (s < strend) { |
bbce6d69 |
662 | if (isALNUM(*s)) { |
663 | if (tmp && regtry(prog, s)) |
664 | goto got_it; |
665 | else |
666 | tmp = doevery; |
667 | } |
668 | else |
669 | tmp = 1; |
670 | s++; |
671 | } |
672 | break; |
a0ed51b3 |
673 | case ALNUMUTF8: |
674 | while (s < strend) { |
dfe13c55 |
675 | if (swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
676 | if (tmp && regtry(prog, s)) |
677 | goto got_it; |
678 | else |
679 | tmp = doevery; |
680 | } |
681 | else |
682 | tmp = 1; |
683 | s += UTF8SKIP(s); |
684 | } |
685 | break; |
bbce6d69 |
686 | case ALNUML: |
3280af22 |
687 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
688 | while (s < strend) { |
689 | if (isALNUM_LC(*s)) { |
a0d0e21e |
690 | if (tmp && regtry(prog, s)) |
691 | goto got_it; |
a687059c |
692 | else |
a0d0e21e |
693 | tmp = doevery; |
694 | } |
695 | else |
696 | tmp = 1; |
697 | s++; |
698 | } |
699 | break; |
a0ed51b3 |
700 | case ALNUMLUTF8: |
701 | PL_reg_flags |= RF_tainted; |
702 | while (s < strend) { |
dfe13c55 |
703 | if (isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
704 | if (tmp && regtry(prog, s)) |
705 | goto got_it; |
706 | else |
707 | tmp = doevery; |
708 | } |
709 | else |
710 | tmp = 1; |
711 | s += UTF8SKIP(s); |
712 | } |
713 | break; |
a0d0e21e |
714 | case NALNUM: |
715 | while (s < strend) { |
bbce6d69 |
716 | if (!isALNUM(*s)) { |
717 | if (tmp && regtry(prog, s)) |
718 | goto got_it; |
719 | else |
720 | tmp = doevery; |
721 | } |
722 | else |
723 | tmp = 1; |
724 | s++; |
725 | } |
726 | break; |
a0ed51b3 |
727 | case NALNUMUTF8: |
728 | while (s < strend) { |
dfe13c55 |
729 | if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
730 | if (tmp && regtry(prog, s)) |
731 | goto got_it; |
732 | else |
733 | tmp = doevery; |
734 | } |
735 | else |
736 | tmp = 1; |
737 | s += UTF8SKIP(s); |
738 | } |
739 | break; |
bbce6d69 |
740 | case NALNUML: |
3280af22 |
741 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
742 | while (s < strend) { |
743 | if (!isALNUM_LC(*s)) { |
a0d0e21e |
744 | if (tmp && regtry(prog, s)) |
745 | goto got_it; |
a687059c |
746 | else |
a0d0e21e |
747 | tmp = doevery; |
a687059c |
748 | } |
a0d0e21e |
749 | else |
750 | tmp = 1; |
751 | s++; |
752 | } |
753 | break; |
a0ed51b3 |
754 | case NALNUMLUTF8: |
755 | PL_reg_flags |= RF_tainted; |
756 | while (s < strend) { |
dfe13c55 |
757 | if (!isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
758 | if (tmp && regtry(prog, s)) |
759 | goto got_it; |
760 | else |
761 | tmp = doevery; |
762 | } |
763 | else |
764 | tmp = 1; |
765 | s += UTF8SKIP(s); |
766 | } |
767 | break; |
a0d0e21e |
768 | case SPACE: |
769 | while (s < strend) { |
770 | if (isSPACE(*s)) { |
771 | if (tmp && regtry(prog, s)) |
772 | goto got_it; |
773 | else |
774 | tmp = doevery; |
2304df62 |
775 | } |
a0d0e21e |
776 | else |
777 | tmp = 1; |
778 | s++; |
779 | } |
780 | break; |
a0ed51b3 |
781 | case SPACEUTF8: |
782 | while (s < strend) { |
dfe13c55 |
783 | if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { |
a0ed51b3 |
784 | if (tmp && regtry(prog, s)) |
785 | goto got_it; |
786 | else |
787 | tmp = doevery; |
788 | } |
789 | else |
790 | tmp = 1; |
791 | s += UTF8SKIP(s); |
792 | } |
793 | break; |
bbce6d69 |
794 | case SPACEL: |
3280af22 |
795 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
796 | while (s < strend) { |
797 | if (isSPACE_LC(*s)) { |
798 | if (tmp && regtry(prog, s)) |
799 | goto got_it; |
800 | else |
801 | tmp = doevery; |
802 | } |
803 | else |
804 | tmp = 1; |
805 | s++; |
806 | } |
807 | break; |
a0ed51b3 |
808 | case SPACELUTF8: |
809 | PL_reg_flags |= RF_tainted; |
810 | while (s < strend) { |
dfe13c55 |
811 | if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { |
a0ed51b3 |
812 | if (tmp && regtry(prog, s)) |
813 | goto got_it; |
814 | else |
815 | tmp = doevery; |
816 | } |
817 | else |
818 | tmp = 1; |
819 | s += UTF8SKIP(s); |
820 | } |
821 | break; |
a0d0e21e |
822 | case NSPACE: |
823 | while (s < strend) { |
824 | if (!isSPACE(*s)) { |
825 | if (tmp && regtry(prog, s)) |
826 | goto got_it; |
827 | else |
828 | tmp = doevery; |
a687059c |
829 | } |
a0d0e21e |
830 | else |
831 | tmp = 1; |
832 | s++; |
833 | } |
834 | break; |
a0ed51b3 |
835 | case NSPACEUTF8: |
836 | while (s < strend) { |
dfe13c55 |
837 | if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { |
a0ed51b3 |
838 | if (tmp && regtry(prog, s)) |
839 | goto got_it; |
840 | else |
841 | tmp = doevery; |
842 | } |
843 | else |
844 | tmp = 1; |
845 | s += UTF8SKIP(s); |
846 | } |
847 | break; |
bbce6d69 |
848 | case NSPACEL: |
3280af22 |
849 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
850 | while (s < strend) { |
851 | if (!isSPACE_LC(*s)) { |
852 | if (tmp && regtry(prog, s)) |
853 | goto got_it; |
854 | else |
855 | tmp = doevery; |
856 | } |
857 | else |
858 | tmp = 1; |
859 | s++; |
860 | } |
861 | break; |
a0ed51b3 |
862 | case NSPACELUTF8: |
863 | PL_reg_flags |= RF_tainted; |
864 | while (s < strend) { |
dfe13c55 |
865 | if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { |
a0ed51b3 |
866 | if (tmp && regtry(prog, s)) |
867 | goto got_it; |
868 | else |
869 | tmp = doevery; |
870 | } |
871 | else |
872 | tmp = 1; |
873 | s += UTF8SKIP(s); |
874 | } |
875 | break; |
a0d0e21e |
876 | case DIGIT: |
877 | while (s < strend) { |
878 | if (isDIGIT(*s)) { |
879 | if (tmp && regtry(prog, s)) |
880 | goto got_it; |
881 | else |
882 | tmp = doevery; |
2b69d0c2 |
883 | } |
a0d0e21e |
884 | else |
885 | tmp = 1; |
886 | s++; |
887 | } |
888 | break; |
a0ed51b3 |
889 | case DIGITUTF8: |
890 | while (s < strend) { |
dfe13c55 |
891 | if (swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
892 | if (tmp && regtry(prog, s)) |
893 | goto got_it; |
894 | else |
895 | tmp = doevery; |
896 | } |
897 | else |
898 | tmp = 1; |
899 | s += UTF8SKIP(s); |
900 | } |
901 | break; |
a0d0e21e |
902 | case NDIGIT: |
903 | while (s < strend) { |
904 | if (!isDIGIT(*s)) { |
905 | if (tmp && regtry(prog, s)) |
906 | goto got_it; |
907 | else |
908 | tmp = doevery; |
a687059c |
909 | } |
a0d0e21e |
910 | else |
911 | tmp = 1; |
912 | s++; |
913 | } |
914 | break; |
a0ed51b3 |
915 | case NDIGITUTF8: |
916 | while (s < strend) { |
dfe13c55 |
917 | if (!swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
918 | if (tmp && regtry(prog, s)) |
919 | goto got_it; |
920 | else |
921 | tmp = doevery; |
922 | } |
923 | else |
924 | tmp = 1; |
925 | s += UTF8SKIP(s); |
926 | } |
927 | break; |
a687059c |
928 | } |
a0d0e21e |
929 | } |
930 | else { |
c277df42 |
931 | dontbother = 0; |
932 | if (prog->float_substr != Nullsv) { /* Trim the end. */ |
933 | char *last; |
934 | I32 oldpos = scream_pos; |
935 | |
22e551b9 |
936 | if (flags & REXEC_SCREAM) { |
937 | last = screaminstr(sv, prog->float_substr, s - strbeg, |
c277df42 |
938 | end_shift, &scream_pos, 1); /* last one */ |
cf93c79d |
939 | if (!last) |
c277df42 |
940 | last = scream_olds; /* Only one occurence. */ |
a0ed51b3 |
941 | } |
942 | else { |
c277df42 |
943 | STRLEN len; |
944 | char *little = SvPV(prog->float_substr, len); |
cf93c79d |
945 | |
946 | if (SvTAIL(prog->float_substr)) { |
947 | if (memEQ(strend - len + 1, little, len - 1)) |
948 | last = strend - len + 1; |
949 | else if (!PL_multiline) |
950 | last = memEQ(strend - len, little, len) |
951 | ? strend - len : Nullch; |
952 | else |
953 | goto find_last; |
954 | } else { |
955 | find_last: |
956 | if (len) |
957 | last = rninstr(s, strend, little, little + len); |
958 | else |
959 | last = strend; /* matching `$' */ |
960 | } |
c277df42 |
961 | } |
962 | if (last == NULL) goto phooey; /* Should not happen! */ |
19b4f81a |
963 | dontbother = strend - last + prog->float_min_offset; |
c277df42 |
964 | } |
965 | if (minlen && (dontbother < minlen)) |
a0d0e21e |
966 | dontbother = minlen - 1; |
a0ed51b3 |
967 | strend -= dontbother; /* this one's always in bytes! */ |
a0d0e21e |
968 | /* We don't know much -- general case. */ |
a0ed51b3 |
969 | if (UTF) { |
970 | for (;;) { |
84df6dba |
971 | if (regtry(prog, s)) |
a0ed51b3 |
972 | goto got_it; |
a0ed51b3 |
973 | if (s >= strend) |
974 | break; |
975 | s += UTF8SKIP(s); |
976 | }; |
977 | } |
978 | else { |
979 | do { |
980 | if (regtry(prog, s)) |
981 | goto got_it; |
982 | } while (s++ < strend); |
983 | } |
a0d0e21e |
984 | } |
985 | |
986 | /* Failure. */ |
987 | goto phooey; |
a687059c |
988 | |
a0d0e21e |
989 | got_it: |
3280af22 |
990 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); |
5f05dabc |
991 | |
5c5e4c24 |
992 | if (PL_reg_eval_set) { |
993 | /* Preserve the current value of $^R */ |
994 | if (oreplsv != GvSV(PL_replgv)) |
995 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is |
996 | restored, the value remains |
997 | the same. */ |
9661b544 |
998 | restore_pos(0); |
5c5e4c24 |
999 | } |
cf93c79d |
1000 | |
1001 | /* make sure $`, $&, $', and $digit will work later */ |
1002 | if ( !(flags & REXEC_NOT_FIRST) ) { |
1003 | if (RX_MATCH_COPIED(prog)) { |
1004 | Safefree(prog->subbeg); |
1005 | RX_MATCH_COPIED_off(prog); |
1006 | } |
1007 | if (flags & REXEC_COPY_STR) { |
1008 | I32 i = PL_regeol - startpos + (stringarg - strbeg); |
1009 | |
1010 | s = savepvn(strbeg, i); |
1011 | prog->subbeg = s; |
1012 | prog->sublen = i; |
1013 | RX_MATCH_COPIED_on(prog); |
1014 | } |
1015 | else { |
1016 | prog->subbeg = strbeg; |
1017 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ |
1018 | } |
1019 | } |
5c5e4c24 |
1020 | |
a0d0e21e |
1021 | return 1; |
1022 | |
1023 | phooey: |
9661b544 |
1024 | if (PL_reg_eval_set) |
1025 | restore_pos(0); |
a0d0e21e |
1026 | return 0; |
a687059c |
1027 | } |
1028 | |
1029 | /* |
1030 | - regtry - try match at specific point |
1031 | */ |
76e3520e |
1032 | STATIC I32 /* 0 failure, 1 success */ |
cea2e8a9 |
1033 | S_regtry(pTHX_ regexp *prog, char *startpos) |
a687059c |
1034 | { |
c277df42 |
1035 | dTHR; |
a0d0e21e |
1036 | register I32 i; |
cf93c79d |
1037 | register I32 *sp; |
1038 | register I32 *ep; |
c277df42 |
1039 | CHECKPOINT lastcp; |
a0d0e21e |
1040 | |
3280af22 |
1041 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { |
9661b544 |
1042 | MAGIC *mg; |
1043 | |
3280af22 |
1044 | PL_reg_eval_set = RS_init; |
ce862d02 |
1045 | DEBUG_r(DEBUG_s( |
c3464db5 |
1046 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", |
3280af22 |
1047 | PL_stack_sp - PL_stack_base); |
ce862d02 |
1048 | )); |
1049 | SAVEINT(cxstack[cxstack_ix].blk_oldsp); |
3280af22 |
1050 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
ce862d02 |
1051 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ |
1052 | SAVETMPS; |
1053 | /* Apparently this is not needed, judging by wantarray. */ |
1054 | /* SAVEINT(cxstack[cxstack_ix].blk_gimme); |
1055 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
9661b544 |
1056 | |
1057 | if (PL_reg_sv) { |
1058 | /* Make $_ available to executed code. */ |
127ad2b7 |
1059 | if (PL_reg_sv != DEFSV) { |
1060 | /* SAVE_DEFSV does *not* suffice here for USE_THREADS */ |
1061 | SAVESPTR(DEFSV); |
1062 | DEFSV = PL_reg_sv; |
9661b544 |
1063 | } |
1064 | |
1065 | if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) |
1066 | && (mg = mg_find(PL_reg_sv, 'g')))) { |
1067 | /* prepare for quick setting of pos */ |
1068 | sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); |
1069 | mg = mg_find(PL_reg_sv, 'g'); |
1070 | mg->mg_len = -1; |
1071 | } |
1072 | PL_reg_magic = mg; |
1073 | PL_reg_oldpos = mg->mg_len; |
cea2e8a9 |
1074 | SAVEDESTRUCTOR(S_restore_pos, 0); |
9661b544 |
1075 | } |
5c5e4c24 |
1076 | if (!PL_reg_curpm) |
1077 | New(22,PL_reg_curpm, 1, PMOP); |
1078 | PL_reg_curpm->op_pmregexp = prog; |
1079 | PL_reg_oldcurpm = PL_curpm; |
1080 | PL_curpm = PL_reg_curpm; |
cf93c79d |
1081 | if (RX_MATCH_COPIED(prog)) { |
1082 | /* Here is a serious problem: we cannot rewrite subbeg, |
1083 | since it may be needed if this match fails. Thus |
1084 | $` inside (?{}) could fail... */ |
1085 | PL_reg_oldsaved = prog->subbeg; |
1086 | PL_reg_oldsavedlen = prog->sublen; |
1087 | RX_MATCH_COPIED_off(prog); |
1088 | } |
1089 | else |
1090 | PL_reg_oldsaved = Nullch; |
5c5e4c24 |
1091 | prog->subbeg = PL_bostr; |
cf93c79d |
1092 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ |
ce862d02 |
1093 | } |
cf93c79d |
1094 | prog->startp[0] = startpos - PL_bostr; |
3280af22 |
1095 | PL_reginput = startpos; |
1096 | PL_regstartp = prog->startp; |
1097 | PL_regendp = prog->endp; |
1098 | PL_reglastparen = &prog->lastparen; |
a0d0e21e |
1099 | prog->lastparen = 0; |
3280af22 |
1100 | PL_regsize = 0; |
364723c2 |
1101 | DEBUG_r(PL_reg_starttry = startpos); |
3280af22 |
1102 | if (PL_reg_start_tmpl <= prog->nparens) { |
1103 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; |
1104 | if(PL_reg_start_tmp) |
1105 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
c277df42 |
1106 | else |
3280af22 |
1107 | New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
c277df42 |
1108 | } |
a0d0e21e |
1109 | |
5c5e4c24 |
1110 | /* XXXX What this code is doing here?!!! There should be no need |
1111 | to do this again and again, PL_reglastparen should take care of |
1112 | this! */ |
a0d0e21e |
1113 | sp = prog->startp; |
1114 | ep = prog->endp; |
1115 | if (prog->nparens) { |
5c5e4c24 |
1116 | for (i = prog->nparens; i >= 1; i--) { |
cf93c79d |
1117 | *++sp = -1; |
1118 | *++ep = -1; |
a687059c |
1119 | } |
a0d0e21e |
1120 | } |
c277df42 |
1121 | REGCP_SET; |
7e5428c5 |
1122 | if (regmatch(prog->program + 1)) { |
cf93c79d |
1123 | prog->endp[0] = PL_reginput - PL_bostr; |
a0d0e21e |
1124 | return 1; |
1125 | } |
c277df42 |
1126 | REGCP_UNWIND; |
1127 | return 0; |
a687059c |
1128 | } |
1129 | |
1130 | /* |
1131 | - regmatch - main matching routine |
1132 | * |
1133 | * Conceptually the strategy is simple: check to see whether the current |
1134 | * node matches, call self recursively to see whether the rest matches, |
1135 | * and then act accordingly. In practice we make some effort to avoid |
1136 | * recursion, in particular by going through "ordinary" nodes (that don't |
1137 | * need to know whether the rest of the match failed) by a loop instead of |
1138 | * by recursion. |
1139 | */ |
1140 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
1141 | * maybe save a little bit of pushing and popping on the stack. It also takes |
1142 | * advantage of machines that use a register save mask on subroutine entry. |
1143 | */ |
76e3520e |
1144 | STATIC I32 /* 0 failure, 1 success */ |
cea2e8a9 |
1145 | S_regmatch(pTHX_ regnode *prog) |
a687059c |
1146 | { |
c277df42 |
1147 | dTHR; |
1148 | register regnode *scan; /* Current node. */ |
1149 | regnode *next; /* Next node. */ |
1150 | regnode *inner; /* Next node in internal branch. */ |
c3464db5 |
1151 | register I32 nextchr; /* renamed nextchr - nextchar colides with |
1152 | function of same name */ |
a0d0e21e |
1153 | register I32 n; /* no or next */ |
1154 | register I32 ln; /* len or last */ |
1155 | register char *s; /* operand or save */ |
3280af22 |
1156 | register char *locinput = PL_reginput; |
c277df42 |
1157 | register I32 c1, c2, paren; /* case fold search, parenth */ |
1158 | int minmod = 0, sw = 0, logical = 0; |
4633a7c4 |
1159 | #ifdef DEBUGGING |
3280af22 |
1160 | PL_regindent++; |
4633a7c4 |
1161 | #endif |
a0d0e21e |
1162 | |
a0ed51b3 |
1163 | /* Note that nextchr is a byte even in UTF */ |
76e3520e |
1164 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1165 | scan = prog; |
1166 | while (scan != NULL) { |
c277df42 |
1167 | #define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) |
a687059c |
1168 | #ifdef DEBUGGING |
c277df42 |
1169 | # define sayYES goto yes |
1170 | # define sayNO goto no |
1171 | # define saySAME(x) if (x) goto yes; else goto no |
1172 | # define REPORT_CODE_OFF 24 |
4633a7c4 |
1173 | #else |
c277df42 |
1174 | # define sayYES return 1 |
1175 | # define sayNO return 0 |
1176 | # define saySAME(x) return x |
a687059c |
1177 | #endif |
c277df42 |
1178 | DEBUG_r( { |
1179 | SV *prop = sv_newmortal(); |
3280af22 |
1180 | int docolor = *PL_colors[0]; |
c277df42 |
1181 | int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
3280af22 |
1182 | int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput); |
beed8111 |
1183 | /* The part of the string before starttry has one color |
1184 | (pref0_len chars), between starttry and current |
1185 | position another one (pref_len - pref0_len chars), |
1186 | after the current position the third one. |
1187 | We assume that pref0_len <= pref_len, otherwise we |
1188 | decrease pref0_len. */ |
3280af22 |
1189 | int pref_len = (locinput - PL_bostr > (5 + taill) - l |
1190 | ? (5 + taill) - l : locinput - PL_bostr); |
364723c2 |
1191 | int pref0_len = pref_len - (locinput - PL_reg_starttry); |
c277df42 |
1192 | |
3280af22 |
1193 | if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) |
1194 | l = ( PL_regeol - locinput > (5 + taill) - pref_len |
1195 | ? (5 + taill) - pref_len : PL_regeol - locinput); |
8d300b32 |
1196 | if (pref0_len < 0) |
1197 | pref0_len = 0; |
beed8111 |
1198 | if (pref0_len > pref_len) |
1199 | pref0_len = pref_len; |
c277df42 |
1200 | regprop(prop, scan); |
1201 | PerlIO_printf(Perl_debug_log, |
8d300b32 |
1202 | "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n", |
3280af22 |
1203 | locinput - PL_bostr, |
8d300b32 |
1204 | PL_colors[4], pref0_len, |
1205 | locinput - pref_len, PL_colors[5], |
1206 | PL_colors[2], pref_len - pref0_len, |
1207 | locinput - pref_len + pref0_len, PL_colors[3], |
c277df42 |
1208 | (docolor ? "" : "> <"), |
3280af22 |
1209 | PL_colors[0], l, locinput, PL_colors[1], |
c277df42 |
1210 | 15 - l - pref_len + 1, |
1211 | "", |
3280af22 |
1212 | scan - PL_regprogram, PL_regindent*2, "", |
c277df42 |
1213 | SvPVX(prop)); |
1214 | } ); |
a687059c |
1215 | |
c277df42 |
1216 | next = scan + NEXT_OFF(scan); |
a0d0e21e |
1217 | if (next == scan) |
1218 | next = NULL; |
a687059c |
1219 | |
a0d0e21e |
1220 | switch (OP(scan)) { |
1221 | case BOL: |
3280af22 |
1222 | if (locinput == PL_bostr |
1223 | ? PL_regprev == '\n' |
1224 | : (PL_multiline && |
1225 | (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
1226 | { |
a0ed51b3 |
1227 | /* regtill = regbol; */ |
a0d0e21e |
1228 | break; |
1229 | } |
4633a7c4 |
1230 | sayNO; |
a0d0e21e |
1231 | case MBOL: |
3280af22 |
1232 | if (locinput == PL_bostr |
1233 | ? PL_regprev == '\n' |
1234 | : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
a0d0e21e |
1235 | { |
1236 | break; |
1237 | } |
4633a7c4 |
1238 | sayNO; |
a0d0e21e |
1239 | case SBOL: |
3280af22 |
1240 | if (locinput == PL_regbol && PL_regprev == '\n') |
a0d0e21e |
1241 | break; |
4633a7c4 |
1242 | sayNO; |
774d564b |
1243 | case GPOS: |
22e551b9 |
1244 | if (locinput == PL_reg_ganch) |
a0d0e21e |
1245 | break; |
4633a7c4 |
1246 | sayNO; |
a0d0e21e |
1247 | case EOL: |
3280af22 |
1248 | if (PL_multiline) |
a0d0e21e |
1249 | goto meol; |
1250 | else |
1251 | goto seol; |
1252 | case MEOL: |
1253 | meol: |
3280af22 |
1254 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
4633a7c4 |
1255 | sayNO; |
a0d0e21e |
1256 | break; |
1257 | case SEOL: |
1258 | seol: |
3280af22 |
1259 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
4633a7c4 |
1260 | sayNO; |
3280af22 |
1261 | if (PL_regeol - locinput > 1) |
4633a7c4 |
1262 | sayNO; |
a0d0e21e |
1263 | break; |
b85d18e9 |
1264 | case EOS: |
3280af22 |
1265 | if (PL_regeol != locinput) |
b85d18e9 |
1266 | sayNO; |
1267 | break; |
a0ed51b3 |
1268 | case SANYUTF8: |
1269 | if (nextchr & 0x80) { |
6f06b55f |
1270 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1271 | if (locinput > PL_regeol) |
1272 | sayNO; |
1273 | nextchr = UCHARAT(locinput); |
1274 | break; |
1275 | } |
1276 | if (!nextchr && locinput >= PL_regeol) |
1277 | sayNO; |
1278 | nextchr = UCHARAT(++locinput); |
1279 | break; |
a0d0e21e |
1280 | case SANY: |
3280af22 |
1281 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1282 | sayNO; |
76e3520e |
1283 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1284 | break; |
a0ed51b3 |
1285 | case ANYUTF8: |
1286 | if (nextchr & 0x80) { |
6f06b55f |
1287 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1288 | if (locinput > PL_regeol) |
1289 | sayNO; |
1290 | nextchr = UCHARAT(locinput); |
1291 | break; |
1292 | } |
1293 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
1294 | sayNO; |
1295 | nextchr = UCHARAT(++locinput); |
1296 | break; |
22c35a8c |
1297 | case REG_ANY: |
3280af22 |
1298 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
4633a7c4 |
1299 | sayNO; |
76e3520e |
1300 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1301 | break; |
bbce6d69 |
1302 | case EXACT: |
161b471a |
1303 | s = (char *) OPERAND(scan); |
c277df42 |
1304 | ln = UCHARAT(s++); |
a0d0e21e |
1305 | /* Inline the first character, for speed. */ |
76e3520e |
1306 | if (UCHARAT(s) != nextchr) |
4633a7c4 |
1307 | sayNO; |
3280af22 |
1308 | if (PL_regeol - locinput < ln) |
4633a7c4 |
1309 | sayNO; |
36477c24 |
1310 | if (ln > 1 && memNE(s, locinput, ln)) |
4633a7c4 |
1311 | sayNO; |
a0d0e21e |
1312 | locinput += ln; |
76e3520e |
1313 | nextchr = UCHARAT(locinput); |
bbce6d69 |
1314 | break; |
1315 | case EXACTFL: |
3280af22 |
1316 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1317 | /* FALL THROUGH */ |
1318 | case EXACTF: |
161b471a |
1319 | s = (char *) OPERAND(scan); |
c277df42 |
1320 | ln = UCHARAT(s++); |
a0ed51b3 |
1321 | |
1322 | if (UTF) { |
1323 | char *l = locinput; |
1324 | char *e = s + ln; |
1325 | c1 = OP(scan) == EXACTF; |
1326 | while (s < e) { |
1327 | if (l >= PL_regeol) |
1328 | sayNO; |
dfe13c55 |
1329 | if (utf8_to_uv((U8*)s, 0) != (c1 ? |
1330 | toLOWER_utf8((U8*)l) : |
1331 | toLOWER_LC_utf8((U8*)l))) |
1332 | { |
a0ed51b3 |
1333 | sayNO; |
dfe13c55 |
1334 | } |
a0ed51b3 |
1335 | s += UTF8SKIP(s); |
1336 | l += UTF8SKIP(l); |
1337 | } |
1338 | locinput = l; |
1339 | nextchr = UCHARAT(locinput); |
1340 | break; |
1341 | } |
1342 | |
bbce6d69 |
1343 | /* Inline the first character, for speed. */ |
76e3520e |
1344 | if (UCHARAT(s) != nextchr && |
bbce6d69 |
1345 | UCHARAT(s) != ((OP(scan) == EXACTF) |
22c35a8c |
1346 | ? PL_fold : PL_fold_locale)[nextchr]) |
bbce6d69 |
1347 | sayNO; |
3280af22 |
1348 | if (PL_regeol - locinput < ln) |
bbce6d69 |
1349 | sayNO; |
5f05dabc |
1350 | if (ln > 1 && (OP(scan) == EXACTF |
1351 | ? ibcmp(s, locinput, ln) |
1352 | : ibcmp_locale(s, locinput, ln))) |
bbce6d69 |
1353 | sayNO; |
1354 | locinput += ln; |
76e3520e |
1355 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1356 | break; |
a0ed51b3 |
1357 | case ANYOFUTF8: |
1358 | s = (char *) OPERAND(scan); |
1359 | if (!REGINCLASSUTF8(scan, (U8*)locinput)) |
1360 | sayNO; |
1361 | if (locinput >= PL_regeol) |
1362 | sayNO; |
6f06b55f |
1363 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1364 | nextchr = UCHARAT(locinput); |
1365 | break; |
a0d0e21e |
1366 | case ANYOF: |
161b471a |
1367 | s = (char *) OPERAND(scan); |
76e3520e |
1368 | if (nextchr < 0) |
1369 | nextchr = UCHARAT(locinput); |
873ef191 |
1370 | if (!REGINCLASS(s, nextchr)) |
4633a7c4 |
1371 | sayNO; |
3280af22 |
1372 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1373 | sayNO; |
76e3520e |
1374 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1375 | break; |
bbce6d69 |
1376 | case ALNUML: |
3280af22 |
1377 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1378 | /* FALL THROUGH */ |
a0d0e21e |
1379 | case ALNUM: |
76e3520e |
1380 | if (!nextchr) |
4633a7c4 |
1381 | sayNO; |
bbce6d69 |
1382 | if (!(OP(scan) == ALNUM |
76e3520e |
1383 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
4633a7c4 |
1384 | sayNO; |
76e3520e |
1385 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1386 | break; |
a0ed51b3 |
1387 | case ALNUMLUTF8: |
1388 | PL_reg_flags |= RF_tainted; |
1389 | /* FALL THROUGH */ |
1390 | case ALNUMUTF8: |
1391 | if (!nextchr) |
1392 | sayNO; |
1393 | if (nextchr & 0x80) { |
1394 | if (!(OP(scan) == ALNUMUTF8 |
dfe13c55 |
1395 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1396 | : isALNUM_LC_utf8((U8*)locinput))) |
1397 | { |
a0ed51b3 |
1398 | sayNO; |
dfe13c55 |
1399 | } |
6f06b55f |
1400 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1401 | nextchr = UCHARAT(locinput); |
1402 | break; |
1403 | } |
1404 | if (!(OP(scan) == ALNUMUTF8 |
1405 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
1406 | sayNO; |
1407 | nextchr = UCHARAT(++locinput); |
1408 | break; |
bbce6d69 |
1409 | case NALNUML: |
3280af22 |
1410 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1411 | /* FALL THROUGH */ |
a0d0e21e |
1412 | case NALNUM: |
3280af22 |
1413 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1414 | sayNO; |
bbce6d69 |
1415 | if (OP(scan) == NALNUM |
76e3520e |
1416 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
4633a7c4 |
1417 | sayNO; |
76e3520e |
1418 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1419 | break; |
a0ed51b3 |
1420 | case NALNUMLUTF8: |
1421 | PL_reg_flags |= RF_tainted; |
1422 | /* FALL THROUGH */ |
1423 | case NALNUMUTF8: |
1424 | if (!nextchr && locinput >= PL_regeol) |
1425 | sayNO; |
1426 | if (nextchr & 0x80) { |
1427 | if (OP(scan) == NALNUMUTF8 |
dfe13c55 |
1428 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1429 | : isALNUM_LC_utf8((U8*)locinput)) |
1430 | { |
a0ed51b3 |
1431 | sayNO; |
dfe13c55 |
1432 | } |
6f06b55f |
1433 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1434 | nextchr = UCHARAT(locinput); |
1435 | break; |
1436 | } |
1437 | if (OP(scan) == NALNUMUTF8 |
1438 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
1439 | sayNO; |
1440 | nextchr = UCHARAT(++locinput); |
1441 | break; |
bbce6d69 |
1442 | case BOUNDL: |
1443 | case NBOUNDL: |
3280af22 |
1444 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1445 | /* FALL THROUGH */ |
a0d0e21e |
1446 | case BOUND: |
bbce6d69 |
1447 | case NBOUND: |
1448 | /* was last char in word? */ |
3280af22 |
1449 | ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev; |
bbce6d69 |
1450 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
1451 | ln = isALNUM(ln); |
76e3520e |
1452 | n = isALNUM(nextchr); |
bbce6d69 |
1453 | } |
1454 | else { |
1455 | ln = isALNUM_LC(ln); |
76e3520e |
1456 | n = isALNUM_LC(nextchr); |
bbce6d69 |
1457 | } |
95bac841 |
1458 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 |
1459 | sayNO; |
a0d0e21e |
1460 | break; |
a0ed51b3 |
1461 | case BOUNDLUTF8: |
1462 | case NBOUNDLUTF8: |
1463 | PL_reg_flags |= RF_tainted; |
1464 | /* FALL THROUGH */ |
1465 | case BOUNDUTF8: |
1466 | case NBOUNDUTF8: |
1467 | /* was last char in word? */ |
dfe13c55 |
1468 | ln = (locinput != PL_regbol) |
1469 | ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev; |
a0ed51b3 |
1470 | if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) { |
1471 | ln = isALNUM_uni(ln); |
dfe13c55 |
1472 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput); |
a0ed51b3 |
1473 | } |
1474 | else { |
1475 | ln = isALNUM_LC_uni(ln); |
dfe13c55 |
1476 | n = isALNUM_LC_utf8((U8*)locinput); |
a0ed51b3 |
1477 | } |
1478 | if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8)) |
1479 | sayNO; |
1480 | break; |
bbce6d69 |
1481 | case SPACEL: |
3280af22 |
1482 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1483 | /* FALL THROUGH */ |
a0d0e21e |
1484 | case SPACE: |
3280af22 |
1485 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1486 | sayNO; |
bbce6d69 |
1487 | if (!(OP(scan) == SPACE |
76e3520e |
1488 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
4633a7c4 |
1489 | sayNO; |
76e3520e |
1490 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1491 | break; |
a0ed51b3 |
1492 | case SPACELUTF8: |
1493 | PL_reg_flags |= RF_tainted; |
1494 | /* FALL THROUGH */ |
1495 | case SPACEUTF8: |
1496 | if (!nextchr && locinput >= PL_regeol) |
1497 | sayNO; |
1498 | if (nextchr & 0x80) { |
1499 | if (!(OP(scan) == SPACEUTF8 |
dfe13c55 |
1500 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1501 | : isSPACE_LC_utf8((U8*)locinput))) |
1502 | { |
a0ed51b3 |
1503 | sayNO; |
dfe13c55 |
1504 | } |
6f06b55f |
1505 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1506 | nextchr = UCHARAT(locinput); |
1507 | break; |
1508 | } |
1509 | if (!(OP(scan) == SPACEUTF8 |
1510 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
1511 | sayNO; |
1512 | nextchr = UCHARAT(++locinput); |
1513 | break; |
bbce6d69 |
1514 | case NSPACEL: |
3280af22 |
1515 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1516 | /* FALL THROUGH */ |
a0d0e21e |
1517 | case NSPACE: |
76e3520e |
1518 | if (!nextchr) |
4633a7c4 |
1519 | sayNO; |
bbce6d69 |
1520 | if (OP(scan) == SPACE |
76e3520e |
1521 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 |
1522 | sayNO; |
76e3520e |
1523 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1524 | break; |
a0ed51b3 |
1525 | case NSPACELUTF8: |
1526 | PL_reg_flags |= RF_tainted; |
1527 | /* FALL THROUGH */ |
1528 | case NSPACEUTF8: |
1529 | if (!nextchr) |
1530 | sayNO; |
1531 | if (nextchr & 0x80) { |
1532 | if (OP(scan) == NSPACEUTF8 |
dfe13c55 |
1533 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1534 | : isSPACE_LC_utf8((U8*)locinput)) |
1535 | { |
a0ed51b3 |
1536 | sayNO; |
dfe13c55 |
1537 | } |
6f06b55f |
1538 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1539 | nextchr = UCHARAT(locinput); |
1540 | break; |
1541 | } |
1542 | if (OP(scan) == NSPACEUTF8 |
1543 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
1544 | sayNO; |
1545 | nextchr = UCHARAT(++locinput); |
1546 | break; |
a0d0e21e |
1547 | case DIGIT: |
76e3520e |
1548 | if (!isDIGIT(nextchr)) |
4633a7c4 |
1549 | sayNO; |
76e3520e |
1550 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1551 | break; |
a0ed51b3 |
1552 | case DIGITUTF8: |
1553 | if (nextchr & 0x80) { |
dfe13c55 |
1554 | if (!(swash_fetch(PL_utf8_digit,(U8*)locinput))) |
a0ed51b3 |
1555 | sayNO; |
6f06b55f |
1556 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1557 | nextchr = UCHARAT(locinput); |
1558 | break; |
1559 | } |
1560 | if (!isDIGIT(nextchr)) |
1561 | sayNO; |
1562 | nextchr = UCHARAT(++locinput); |
1563 | break; |
a0d0e21e |
1564 | case NDIGIT: |
3280af22 |
1565 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1566 | sayNO; |
76e3520e |
1567 | if (isDIGIT(nextchr)) |
4633a7c4 |
1568 | sayNO; |
76e3520e |
1569 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1570 | break; |
a0ed51b3 |
1571 | case NDIGITUTF8: |
1572 | if (!nextchr && locinput >= PL_regeol) |
1573 | sayNO; |
1574 | if (nextchr & 0x80) { |
dfe13c55 |
1575 | if (swash_fetch(PL_utf8_digit,(U8*)locinput)) |
a0ed51b3 |
1576 | sayNO; |
6f06b55f |
1577 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1578 | nextchr = UCHARAT(locinput); |
1579 | break; |
1580 | } |
1581 | if (isDIGIT(nextchr)) |
1582 | sayNO; |
1583 | nextchr = UCHARAT(++locinput); |
1584 | break; |
1585 | case CLUMP: |
dfe13c55 |
1586 | if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
1587 | sayNO; |
6f06b55f |
1588 | locinput += PL_utf8skip[nextchr]; |
dfe13c55 |
1589 | while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
1590 | locinput += UTF8SKIP(locinput); |
1591 | if (locinput > PL_regeol) |
1592 | sayNO; |
1593 | nextchr = UCHARAT(locinput); |
1594 | break; |
c8756f30 |
1595 | case REFFL: |
3280af22 |
1596 | PL_reg_flags |= RF_tainted; |
c8756f30 |
1597 | /* FALL THROUGH */ |
c277df42 |
1598 | case REF: |
c8756f30 |
1599 | case REFF: |
c277df42 |
1600 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1601 | ln = PL_regstartp[n]; |
1602 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 |
1603 | sayNO; /* Do not match unless seen CLOSEn. */ |
cf93c79d |
1604 | if (ln == PL_regendp[n]) |
a0d0e21e |
1605 | break; |
a0ed51b3 |
1606 | |
cf93c79d |
1607 | s = PL_bostr + ln; |
a0ed51b3 |
1608 | if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ |
1609 | char *l = locinput; |
cf93c79d |
1610 | char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 |
1611 | /* |
1612 | * Note that we can't do the "other character" lookup trick as |
1613 | * in the 8-bit case (no pun intended) because in Unicode we |
1614 | * have to map both upper and title case to lower case. |
1615 | */ |
1616 | if (OP(scan) == REFF) { |
1617 | while (s < e) { |
1618 | if (l >= PL_regeol) |
1619 | sayNO; |
dfe13c55 |
1620 | if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l)) |
a0ed51b3 |
1621 | sayNO; |
1622 | s += UTF8SKIP(s); |
1623 | l += UTF8SKIP(l); |
1624 | } |
1625 | } |
1626 | else { |
1627 | while (s < e) { |
1628 | if (l >= PL_regeol) |
1629 | sayNO; |
dfe13c55 |
1630 | if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l)) |
a0ed51b3 |
1631 | sayNO; |
1632 | s += UTF8SKIP(s); |
1633 | l += UTF8SKIP(l); |
1634 | } |
1635 | } |
1636 | locinput = l; |
1637 | nextchr = UCHARAT(locinput); |
1638 | break; |
1639 | } |
1640 | |
a0d0e21e |
1641 | /* Inline the first character, for speed. */ |
76e3520e |
1642 | if (UCHARAT(s) != nextchr && |
c8756f30 |
1643 | (OP(scan) == REF || |
1644 | (UCHARAT(s) != ((OP(scan) == REFF |
22c35a8c |
1645 | ? PL_fold : PL_fold_locale)[nextchr])))) |
4633a7c4 |
1646 | sayNO; |
cf93c79d |
1647 | ln = PL_regendp[n] - ln; |
3280af22 |
1648 | if (locinput + ln > PL_regeol) |
4633a7c4 |
1649 | sayNO; |
c8756f30 |
1650 | if (ln > 1 && (OP(scan) == REF |
1651 | ? memNE(s, locinput, ln) |
1652 | : (OP(scan) == REFF |
1653 | ? ibcmp(s, locinput, ln) |
1654 | : ibcmp_locale(s, locinput, ln)))) |
4633a7c4 |
1655 | sayNO; |
a0d0e21e |
1656 | locinput += ln; |
76e3520e |
1657 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1658 | break; |
1659 | |
1660 | case NOTHING: |
c277df42 |
1661 | case TAIL: |
a0d0e21e |
1662 | break; |
1663 | case BACK: |
1664 | break; |
c277df42 |
1665 | case EVAL: |
1666 | { |
1667 | dSP; |
533c011a |
1668 | OP_4tree *oop = PL_op; |
3280af22 |
1669 | COP *ocurcop = PL_curcop; |
1670 | SV **ocurpad = PL_curpad; |
c277df42 |
1671 | SV *ret; |
1672 | |
1673 | n = ARG(scan); |
533c011a |
1674 | PL_op = (OP_4tree*)PL_regdata->data[n]; |
1675 | DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) ); |
dfad63ad |
1676 | PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]); |
cf93c79d |
1677 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
c277df42 |
1678 | |
cea2e8a9 |
1679 | CALLRUNOPS(aTHX); /* Scalar context. */ |
c277df42 |
1680 | SPAGAIN; |
1681 | ret = POPs; |
1682 | PUTBACK; |
1683 | |
0f5d15d6 |
1684 | PL_op = oop; |
1685 | PL_curpad = ocurpad; |
1686 | PL_curcop = ocurcop; |
c277df42 |
1687 | if (logical) { |
0f5d15d6 |
1688 | if (logical == 2) { /* Postponed subexpression. */ |
1689 | regexp *re; |
22c35a8c |
1690 | MAGIC *mg = Null(MAGIC*); |
0f5d15d6 |
1691 | re_cc_state state; |
1692 | CURCUR cctmp; |
1693 | CHECKPOINT cp, lastcp; |
1694 | |
1695 | if(SvROK(ret) || SvRMAGICAL(ret)) { |
1696 | SV *sv = SvROK(ret) ? SvRV(ret) : ret; |
1697 | |
1698 | if(SvMAGICAL(sv)) |
1699 | mg = mg_find(sv, 'r'); |
1700 | } |
1701 | if (mg) { |
1702 | re = (regexp *)mg->mg_obj; |
df0003d4 |
1703 | (void)ReREFCNT_inc(re); |
0f5d15d6 |
1704 | } |
1705 | else { |
1706 | STRLEN len; |
1707 | char *t = SvPV(ret, len); |
1708 | PMOP pm; |
1709 | char *oprecomp = PL_regprecomp; |
1710 | I32 osize = PL_regsize; |
1711 | I32 onpar = PL_regnpar; |
1712 | |
1713 | pm.op_pmflags = 0; |
cea2e8a9 |
1714 | re = CALLREGCOMP(aTHX_ t, t + len, &pm); |
0f5d15d6 |
1715 | if (!(SvFLAGS(ret) |
1716 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) |
1717 | sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); |
1718 | PL_regprecomp = oprecomp; |
1719 | PL_regsize = osize; |
1720 | PL_regnpar = onpar; |
1721 | } |
1722 | DEBUG_r( |
1723 | PerlIO_printf(Perl_debug_log, |
1724 | "Entering embedded `%s%.60s%s%s'\n", |
1725 | PL_colors[0], |
1726 | re->precomp, |
1727 | PL_colors[1], |
1728 | (strlen(re->precomp) > 60 ? "..." : "")) |
1729 | ); |
1730 | state.node = next; |
1731 | state.prev = PL_reg_call_cc; |
1732 | state.cc = PL_regcc; |
1733 | state.re = PL_reg_re; |
1734 | |
1735 | cctmp.cur = 0; |
1736 | cctmp.oldcc = 0; |
1737 | PL_regcc = &cctmp; |
1738 | |
1739 | cp = regcppush(0); /* Save *all* the positions. */ |
1740 | REGCP_SET; |
1741 | cache_re(re); |
1742 | state.ss = PL_savestack_ix; |
1743 | *PL_reglastparen = 0; |
1744 | PL_reg_call_cc = &state; |
1745 | PL_reginput = locinput; |
1746 | if (regmatch(re->program + 1)) { |
1747 | ReREFCNT_dec(re); |
1748 | regcpblow(cp); |
1749 | sayYES; |
1750 | } |
1751 | DEBUG_r( |
1752 | PerlIO_printf(Perl_debug_log, |
1753 | "%*s failed...\n", |
1754 | REPORT_CODE_OFF+PL_regindent*2, "") |
1755 | ); |
1756 | ReREFCNT_dec(re); |
1757 | REGCP_UNWIND; |
1758 | regcppop(); |
1759 | PL_reg_call_cc = state.prev; |
1760 | PL_regcc = state.cc; |
1761 | PL_reg_re = state.re; |
d3790889 |
1762 | cache_re(PL_reg_re); |
0f5d15d6 |
1763 | sayNO; |
1764 | } |
c277df42 |
1765 | sw = SvTRUE(ret); |
0f5d15d6 |
1766 | logical = 0; |
a0ed51b3 |
1767 | } |
1768 | else |
3280af22 |
1769 | sv_setsv(save_scalar(PL_replgv), ret); |
c277df42 |
1770 | break; |
1771 | } |
a0d0e21e |
1772 | case OPEN: |
c277df42 |
1773 | n = ARG(scan); /* which paren pair */ |
3280af22 |
1774 | PL_reg_start_tmp[n] = locinput; |
1775 | if (n > PL_regsize) |
1776 | PL_regsize = n; |
a0d0e21e |
1777 | break; |
1778 | case CLOSE: |
c277df42 |
1779 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1780 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
1781 | PL_regendp[n] = locinput - PL_bostr; |
3280af22 |
1782 | if (n > *PL_reglastparen) |
1783 | *PL_reglastparen = n; |
a0d0e21e |
1784 | break; |
c277df42 |
1785 | case GROUPP: |
1786 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
1787 | sw = (*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 |
1788 | break; |
1789 | case IFTHEN: |
1790 | if (sw) |
1791 | next = NEXTOPER(NEXTOPER(scan)); |
1792 | else { |
1793 | next = scan + ARG(scan); |
1794 | if (OP(next) == IFTHEN) /* Fake one. */ |
1795 | next = NEXTOPER(NEXTOPER(next)); |
1796 | } |
1797 | break; |
1798 | case LOGICAL: |
0f5d15d6 |
1799 | logical = scan->flags; |
c277df42 |
1800 | break; |
a0d0e21e |
1801 | case CURLYX: { |
1802 | CURCUR cc; |
3280af22 |
1803 | CHECKPOINT cp = PL_savestack_ix; |
c277df42 |
1804 | |
1805 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ |
1806 | next += ARG(next); |
3280af22 |
1807 | cc.oldcc = PL_regcc; |
1808 | PL_regcc = &cc; |
1809 | cc.parenfloor = *PL_reglastparen; |
a0d0e21e |
1810 | cc.cur = -1; |
1811 | cc.min = ARG1(scan); |
1812 | cc.max = ARG2(scan); |
c277df42 |
1813 | cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
a0d0e21e |
1814 | cc.next = next; |
1815 | cc.minmod = minmod; |
1816 | cc.lastloc = 0; |
3280af22 |
1817 | PL_reginput = locinput; |
a0d0e21e |
1818 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ |
1819 | regcpblow(cp); |
3280af22 |
1820 | PL_regcc = cc.oldcc; |
4633a7c4 |
1821 | saySAME(n); |
a0d0e21e |
1822 | } |
1823 | /* NOT REACHED */ |
1824 | case WHILEM: { |
1825 | /* |
1826 | * This is really hard to understand, because after we match |
1827 | * what we're trying to match, we must make sure the rest of |
1828 | * the RE is going to match for sure, and to do that we have |
1829 | * to go back UP the parse tree by recursing ever deeper. And |
1830 | * if it fails, we have to reset our parent's current state |
1831 | * that we can try again after backing off. |
1832 | */ |
1833 | |
c277df42 |
1834 | CHECKPOINT cp, lastcp; |
3280af22 |
1835 | CURCUR* cc = PL_regcc; |
c277df42 |
1836 | char *lastloc = cc->lastloc; /* Detection of 0-len. */ |
1837 | |
4633a7c4 |
1838 | n = cc->cur + 1; /* how many we know we matched */ |
3280af22 |
1839 | PL_reginput = locinput; |
a0d0e21e |
1840 | |
c277df42 |
1841 | DEBUG_r( |
1842 | PerlIO_printf(Perl_debug_log, |
1843 | "%*s %ld out of %ld..%ld cc=%lx\n", |
3280af22 |
1844 | REPORT_CODE_OFF+PL_regindent*2, "", |
c277df42 |
1845 | (long)n, (long)cc->min, |
1846 | (long)cc->max, (long)cc) |
1847 | ); |
4633a7c4 |
1848 | |
a0d0e21e |
1849 | /* If degenerate scan matches "", assume scan done. */ |
1850 | |
579cf2c3 |
1851 | if (locinput == cc->lastloc && n >= cc->min) { |
3280af22 |
1852 | PL_regcc = cc->oldcc; |
1853 | ln = PL_regcc->cur; |
c277df42 |
1854 | DEBUG_r( |
c3464db5 |
1855 | PerlIO_printf(Perl_debug_log, |
1856 | "%*s empty match detected, try continuation...\n", |
3280af22 |
1857 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1858 | ); |
a0d0e21e |
1859 | if (regmatch(cc->next)) |
4633a7c4 |
1860 | sayYES; |
c277df42 |
1861 | DEBUG_r( |
c3464db5 |
1862 | PerlIO_printf(Perl_debug_log, |
1863 | "%*s failed...\n", |
3280af22 |
1864 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1865 | ); |
3280af22 |
1866 | PL_regcc->cur = ln; |
1867 | PL_regcc = cc; |
4633a7c4 |
1868 | sayNO; |
a0d0e21e |
1869 | } |
1870 | |
1871 | /* First just match a string of min scans. */ |
1872 | |
1873 | if (n < cc->min) { |
1874 | cc->cur = n; |
1875 | cc->lastloc = locinput; |
4633a7c4 |
1876 | if (regmatch(cc->scan)) |
1877 | sayYES; |
1878 | cc->cur = n - 1; |
c277df42 |
1879 | cc->lastloc = lastloc; |
1880 | DEBUG_r( |
c3464db5 |
1881 | PerlIO_printf(Perl_debug_log, |
1882 | "%*s failed...\n", |
3280af22 |
1883 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1884 | ); |
4633a7c4 |
1885 | sayNO; |
a0d0e21e |
1886 | } |
1887 | |
1888 | /* Prefer next over scan for minimal matching. */ |
1889 | |
1890 | if (cc->minmod) { |
3280af22 |
1891 | PL_regcc = cc->oldcc; |
1892 | ln = PL_regcc->cur; |
5f05dabc |
1893 | cp = regcppush(cc->parenfloor); |
c277df42 |
1894 | REGCP_SET; |
5f05dabc |
1895 | if (regmatch(cc->next)) { |
c277df42 |
1896 | regcpblow(cp); |
4633a7c4 |
1897 | sayYES; /* All done. */ |
5f05dabc |
1898 | } |
c277df42 |
1899 | REGCP_UNWIND; |
5f05dabc |
1900 | regcppop(); |
3280af22 |
1901 | PL_regcc->cur = ln; |
1902 | PL_regcc = cc; |
a0d0e21e |
1903 | |
c277df42 |
1904 | if (n >= cc->max) { /* Maximum greed exceeded? */ |
599cee73 |
1905 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
3280af22 |
1906 | && !(PL_reg_flags & RF_warned)) { |
1907 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
1908 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
2f3ca594 |
1909 | "Complex regular subexpression recursion", |
1910 | REG_INFTY - 1); |
c277df42 |
1911 | } |
4633a7c4 |
1912 | sayNO; |
c277df42 |
1913 | } |
a687059c |
1914 | |
c277df42 |
1915 | DEBUG_r( |
c3464db5 |
1916 | PerlIO_printf(Perl_debug_log, |
1917 | "%*s trying longer...\n", |
3280af22 |
1918 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1919 | ); |
a0d0e21e |
1920 | /* Try scanning more and see if it helps. */ |
3280af22 |
1921 | PL_reginput = locinput; |
a0d0e21e |
1922 | cc->cur = n; |
1923 | cc->lastloc = locinput; |
5f05dabc |
1924 | cp = regcppush(cc->parenfloor); |
c277df42 |
1925 | REGCP_SET; |
5f05dabc |
1926 | if (regmatch(cc->scan)) { |
c277df42 |
1927 | regcpblow(cp); |
4633a7c4 |
1928 | sayYES; |
5f05dabc |
1929 | } |
c277df42 |
1930 | DEBUG_r( |
c3464db5 |
1931 | PerlIO_printf(Perl_debug_log, |
1932 | "%*s failed...\n", |
3280af22 |
1933 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1934 | ); |
1935 | REGCP_UNWIND; |
5f05dabc |
1936 | regcppop(); |
4633a7c4 |
1937 | cc->cur = n - 1; |
c277df42 |
1938 | cc->lastloc = lastloc; |
4633a7c4 |
1939 | sayNO; |
a0d0e21e |
1940 | } |
1941 | |
1942 | /* Prefer scan over next for maximal matching. */ |
1943 | |
1944 | if (n < cc->max) { /* More greed allowed? */ |
5f05dabc |
1945 | cp = regcppush(cc->parenfloor); |
a0d0e21e |
1946 | cc->cur = n; |
1947 | cc->lastloc = locinput; |
c277df42 |
1948 | REGCP_SET; |
5f05dabc |
1949 | if (regmatch(cc->scan)) { |
c277df42 |
1950 | regcpblow(cp); |
4633a7c4 |
1951 | sayYES; |
5f05dabc |
1952 | } |
c277df42 |
1953 | REGCP_UNWIND; |
a0d0e21e |
1954 | regcppop(); /* Restore some previous $<digit>s? */ |
3280af22 |
1955 | PL_reginput = locinput; |
c277df42 |
1956 | DEBUG_r( |
c3464db5 |
1957 | PerlIO_printf(Perl_debug_log, |
1958 | "%*s failed, try continuation...\n", |
3280af22 |
1959 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1960 | ); |
1961 | } |
599cee73 |
1962 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
1963 | && !(PL_reg_flags & RF_warned)) { |
3280af22 |
1964 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
1965 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
cb5d145d |
1966 | "Complex regular subexpression recursion", |
1967 | REG_INFTY - 1); |
a0d0e21e |
1968 | } |
1969 | |
1970 | /* Failed deeper matches of scan, so see if this one works. */ |
3280af22 |
1971 | PL_regcc = cc->oldcc; |
1972 | ln = PL_regcc->cur; |
a0d0e21e |
1973 | if (regmatch(cc->next)) |
4633a7c4 |
1974 | sayYES; |
c277df42 |
1975 | DEBUG_r( |
c3464db5 |
1976 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", |
3280af22 |
1977 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
1978 | ); |
3280af22 |
1979 | PL_regcc->cur = ln; |
1980 | PL_regcc = cc; |
4633a7c4 |
1981 | cc->cur = n - 1; |
c277df42 |
1982 | cc->lastloc = lastloc; |
4633a7c4 |
1983 | sayNO; |
a0d0e21e |
1984 | } |
1985 | /* NOT REACHED */ |
c277df42 |
1986 | case BRANCHJ: |
1987 | next = scan + ARG(scan); |
1988 | if (next == scan) |
1989 | next = NULL; |
1990 | inner = NEXTOPER(NEXTOPER(scan)); |
1991 | goto do_branch; |
1992 | case BRANCH: |
1993 | inner = NEXTOPER(scan); |
1994 | do_branch: |
1995 | { |
1996 | CHECKPOINT lastcp; |
1997 | c1 = OP(scan); |
1998 | if (OP(next) != c1) /* No choice. */ |
1999 | next = inner; /* Avoid recursion. */ |
a0d0e21e |
2000 | else { |
3280af22 |
2001 | int lastparen = *PL_reglastparen; |
c277df42 |
2002 | |
2003 | REGCP_SET; |
a0d0e21e |
2004 | do { |
3280af22 |
2005 | PL_reginput = locinput; |
c277df42 |
2006 | if (regmatch(inner)) |
4633a7c4 |
2007 | sayYES; |
c277df42 |
2008 | REGCP_UNWIND; |
3280af22 |
2009 | for (n = *PL_reglastparen; n > lastparen; n--) |
cf93c79d |
2010 | PL_regendp[n] = -1; |
3280af22 |
2011 | *PL_reglastparen = n; |
c277df42 |
2012 | scan = next; |
a0d0e21e |
2013 | /*SUPPRESS 560*/ |
c277df42 |
2014 | if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) |
2015 | next += n; |
a0d0e21e |
2016 | else |
c277df42 |
2017 | next = NULL; |
c277df42 |
2018 | inner = NEXTOPER(scan); |
2019 | if (c1 == BRANCHJ) { |
2020 | inner = NEXTOPER(inner); |
2021 | } |
2022 | } while (scan != NULL && OP(scan) == c1); |
4633a7c4 |
2023 | sayNO; |
a0d0e21e |
2024 | /* NOTREACHED */ |
a687059c |
2025 | } |
a0d0e21e |
2026 | } |
2027 | break; |
2028 | case MINMOD: |
2029 | minmod = 1; |
2030 | break; |
c277df42 |
2031 | case CURLYM: |
2032 | { |
00db4c45 |
2033 | I32 l = 0; |
c277df42 |
2034 | CHECKPOINT lastcp; |
2035 | |
2036 | /* We suppose that the next guy does not need |
2037 | backtracking: in particular, it is of constant length, |
2038 | and has no parenths to influence future backrefs. */ |
2039 | ln = ARG1(scan); /* min to match */ |
2040 | n = ARG2(scan); /* max to match */ |
c277df42 |
2041 | paren = scan->flags; |
2042 | if (paren) { |
3280af22 |
2043 | if (paren > PL_regsize) |
2044 | PL_regsize = paren; |
2045 | if (paren > *PL_reglastparen) |
2046 | *PL_reglastparen = paren; |
c277df42 |
2047 | } |
dc45a647 |
2048 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
c277df42 |
2049 | if (paren) |
2050 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
3280af22 |
2051 | PL_reginput = locinput; |
c277df42 |
2052 | if (minmod) { |
2053 | minmod = 0; |
2054 | if (ln && regrepeat_hard(scan, ln, &l) < ln) |
2055 | sayNO; |
5f4b28b2 |
2056 | if (ln && l == 0 && n >= ln |
c277df42 |
2057 | /* In fact, this is tricky. If paren, then the |
2058 | fact that we did/didnot match may influence |
2059 | future execution. */ |
2060 | && !(paren && ln == 0)) |
2061 | ln = n; |
3280af22 |
2062 | locinput = PL_reginput; |
22c35a8c |
2063 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
c277df42 |
2064 | c1 = UCHARAT(OPERAND(next) + 1); |
2065 | if (OP(next) == EXACTF) |
22c35a8c |
2066 | c2 = PL_fold[c1]; |
c277df42 |
2067 | else if (OP(next) == EXACTFL) |
22c35a8c |
2068 | c2 = PL_fold_locale[c1]; |
c277df42 |
2069 | else |
2070 | c2 = c1; |
a0ed51b3 |
2071 | } |
2072 | else |
c277df42 |
2073 | c1 = c2 = -1000; |
2074 | REGCP_SET; |
5f4b28b2 |
2075 | /* This may be improved if l == 0. */ |
c277df42 |
2076 | while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ |
2077 | /* If it could work, try it. */ |
2078 | if (c1 == -1000 || |
3280af22 |
2079 | UCHARAT(PL_reginput) == c1 || |
2080 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2081 | { |
2082 | if (paren) { |
2083 | if (n) { |
cf93c79d |
2084 | PL_regstartp[paren] = |
2085 | HOPc(PL_reginput, -l) - PL_bostr; |
2086 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2087 | } |
2088 | else |
cf93c79d |
2089 | PL_regendp[paren] = -1; |
c277df42 |
2090 | } |
2091 | if (regmatch(next)) |
2092 | sayYES; |
2093 | REGCP_UNWIND; |
2094 | } |
2095 | /* Couldn't or didn't -- move forward. */ |
3280af22 |
2096 | PL_reginput = locinput; |
c277df42 |
2097 | if (regrepeat_hard(scan, 1, &l)) { |
2098 | ln++; |
3280af22 |
2099 | locinput = PL_reginput; |
c277df42 |
2100 | } |
2101 | else |
2102 | sayNO; |
2103 | } |
a0ed51b3 |
2104 | } |
2105 | else { |
c277df42 |
2106 | n = regrepeat_hard(scan, n, &l); |
2107 | if (n != 0 && l == 0 |
2108 | /* In fact, this is tricky. If paren, then the |
2109 | fact that we did/didnot match may influence |
2110 | future execution. */ |
2111 | && !(paren && ln == 0)) |
2112 | ln = n; |
3280af22 |
2113 | locinput = PL_reginput; |
c277df42 |
2114 | DEBUG_r( |
5c0ca799 |
2115 | PerlIO_printf(Perl_debug_log, |
2116 | "%*s matched %ld times, len=%ld...\n", |
3280af22 |
2117 | REPORT_CODE_OFF+PL_regindent*2, "", n, l) |
c277df42 |
2118 | ); |
2119 | if (n >= ln) { |
22c35a8c |
2120 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
c277df42 |
2121 | c1 = UCHARAT(OPERAND(next) + 1); |
2122 | if (OP(next) == EXACTF) |
22c35a8c |
2123 | c2 = PL_fold[c1]; |
c277df42 |
2124 | else if (OP(next) == EXACTFL) |
22c35a8c |
2125 | c2 = PL_fold_locale[c1]; |
c277df42 |
2126 | else |
2127 | c2 = c1; |
a0ed51b3 |
2128 | } |
2129 | else |
c277df42 |
2130 | c1 = c2 = -1000; |
2131 | } |
2132 | REGCP_SET; |
2133 | while (n >= ln) { |
2134 | /* If it could work, try it. */ |
2135 | if (c1 == -1000 || |
3280af22 |
2136 | UCHARAT(PL_reginput) == c1 || |
2137 | UCHARAT(PL_reginput) == c2) |
a0ed51b3 |
2138 | { |
2139 | DEBUG_r( |
c3464db5 |
2140 | PerlIO_printf(Perl_debug_log, |
2141 | "%*s trying tail with n=%ld...\n", |
3280af22 |
2142 | REPORT_CODE_OFF+PL_regindent*2, "", n) |
a0ed51b3 |
2143 | ); |
2144 | if (paren) { |
2145 | if (n) { |
cf93c79d |
2146 | PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr; |
2147 | PL_regendp[paren] = PL_reginput - PL_bostr; |
c277df42 |
2148 | } |
a0ed51b3 |
2149 | else |
cf93c79d |
2150 | PL_regendp[paren] = -1; |
c277df42 |
2151 | } |
a0ed51b3 |
2152 | if (regmatch(next)) |
2153 | sayYES; |
2154 | REGCP_UNWIND; |
2155 | } |
c277df42 |
2156 | /* Couldn't or didn't -- back up. */ |
2157 | n--; |
dfe13c55 |
2158 | locinput = HOPc(locinput, -l); |
3280af22 |
2159 | PL_reginput = locinput; |
c277df42 |
2160 | } |
2161 | } |
2162 | sayNO; |
2163 | break; |
2164 | } |
2165 | case CURLYN: |
2166 | paren = scan->flags; /* Which paren to set */ |
3280af22 |
2167 | if (paren > PL_regsize) |
2168 | PL_regsize = paren; |
2169 | if (paren > *PL_reglastparen) |
2170 | *PL_reglastparen = paren; |
c277df42 |
2171 | ln = ARG1(scan); /* min to match */ |
2172 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2173 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
c277df42 |
2174 | goto repeat; |
a0d0e21e |
2175 | case CURLY: |
c277df42 |
2176 | paren = 0; |
a0d0e21e |
2177 | ln = ARG1(scan); /* min to match */ |
2178 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2179 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
a0d0e21e |
2180 | goto repeat; |
2181 | case STAR: |
2182 | ln = 0; |
c277df42 |
2183 | n = REG_INFTY; |
a0d0e21e |
2184 | scan = NEXTOPER(scan); |
c277df42 |
2185 | paren = 0; |
a0d0e21e |
2186 | goto repeat; |
2187 | case PLUS: |
c277df42 |
2188 | ln = 1; |
2189 | n = REG_INFTY; |
2190 | scan = NEXTOPER(scan); |
2191 | paren = 0; |
2192 | repeat: |
a0d0e21e |
2193 | /* |
2194 | * Lookahead to avoid useless match attempts |
2195 | * when we know what character comes next. |
2196 | */ |
22c35a8c |
2197 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
bbce6d69 |
2198 | c1 = UCHARAT(OPERAND(next) + 1); |
2199 | if (OP(next) == EXACTF) |
22c35a8c |
2200 | c2 = PL_fold[c1]; |
bbce6d69 |
2201 | else if (OP(next) == EXACTFL) |
22c35a8c |
2202 | c2 = PL_fold_locale[c1]; |
bbce6d69 |
2203 | else |
2204 | c2 = c1; |
2205 | } |
a0d0e21e |
2206 | else |
bbce6d69 |
2207 | c1 = c2 = -1000; |
3280af22 |
2208 | PL_reginput = locinput; |
a0d0e21e |
2209 | if (minmod) { |
c277df42 |
2210 | CHECKPOINT lastcp; |
a0d0e21e |
2211 | minmod = 0; |
2212 | if (ln && regrepeat(scan, ln) < ln) |
4633a7c4 |
2213 | sayNO; |
a0ed51b3 |
2214 | locinput = PL_reginput; |
c277df42 |
2215 | REGCP_SET; |
0fe9bf95 |
2216 | if (c1 != -1000) { |
2217 | char *e = locinput + n - ln; /* Should not check after this */ |
2218 | char *old = locinput; |
2219 | |
2220 | if (e >= PL_regeol || (n == REG_INFTY)) |
2221 | e = PL_regeol - 1; |
2222 | while (1) { |
2223 | /* Find place 'next' could work */ |
2224 | if (c1 == c2) { |
2225 | while (locinput <= e && *locinput != c1) |
2226 | locinput++; |
2227 | } else { |
2228 | while (locinput <= e |
2229 | && *locinput != c1 |
2230 | && *locinput != c2) |
2231 | locinput++; |
2232 | } |
2233 | if (locinput > e) |
2234 | sayNO; |
2235 | /* PL_reginput == old now */ |
2236 | if (locinput != old) { |
2237 | ln = 1; /* Did some */ |
2238 | if (regrepeat(scan, locinput - old) < |
2239 | locinput - old) |
2240 | sayNO; |
2241 | } |
2242 | /* PL_reginput == locinput now */ |
2243 | if (paren) { |
2244 | if (ln) { |
cf93c79d |
2245 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; |
2246 | PL_regendp[paren] = locinput - PL_bostr; |
0fe9bf95 |
2247 | } |
2248 | else |
cf93c79d |
2249 | PL_regendp[paren] = -1; |
0fe9bf95 |
2250 | } |
2251 | if (regmatch(next)) |
2252 | sayYES; |
2253 | PL_reginput = locinput; /* Could be reset... */ |
2254 | REGCP_UNWIND; |
2255 | /* Couldn't or didn't -- move forward. */ |
2256 | old = locinput++; |
2257 | } |
2258 | } |
2259 | else |
c277df42 |
2260 | while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ |
a0d0e21e |
2261 | /* If it could work, try it. */ |
bbce6d69 |
2262 | if (c1 == -1000 || |
3280af22 |
2263 | UCHARAT(PL_reginput) == c1 || |
2264 | UCHARAT(PL_reginput) == c2) |
bbce6d69 |
2265 | { |
c277df42 |
2266 | if (paren) { |
2267 | if (n) { |
cf93c79d |
2268 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2269 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2270 | } |
2271 | else |
cf93c79d |
2272 | PL_regendp[paren] = -1; |
c277df42 |
2273 | } |
a0d0e21e |
2274 | if (regmatch(next)) |
4633a7c4 |
2275 | sayYES; |
c277df42 |
2276 | REGCP_UNWIND; |
bbce6d69 |
2277 | } |
c277df42 |
2278 | /* Couldn't or didn't -- move forward. */ |
a0ed51b3 |
2279 | PL_reginput = locinput; |
a0d0e21e |
2280 | if (regrepeat(scan, 1)) { |
2281 | ln++; |
a0ed51b3 |
2282 | locinput = PL_reginput; |
2283 | } |
2284 | else |
4633a7c4 |
2285 | sayNO; |
a0d0e21e |
2286 | } |
2287 | } |
2288 | else { |
c277df42 |
2289 | CHECKPOINT lastcp; |
a0d0e21e |
2290 | n = regrepeat(scan, n); |
a0ed51b3 |
2291 | locinput = PL_reginput; |
22c35a8c |
2292 | if (ln < n && PL_regkind[(U8)OP(next)] == EOL && |
3280af22 |
2293 | (!PL_multiline || OP(next) == SEOL)) |
a0d0e21e |
2294 | ln = n; /* why back off? */ |
c277df42 |
2295 | REGCP_SET; |
2296 | if (paren) { |
2297 | while (n >= ln) { |
2298 | /* If it could work, try it. */ |
2299 | if (c1 == -1000 || |
3280af22 |
2300 | UCHARAT(PL_reginput) == c1 || |
2301 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2302 | { |
2303 | if (paren && n) { |
2304 | if (n) { |
cf93c79d |
2305 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2306 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2307 | } |
2308 | else |
cf93c79d |
2309 | PL_regendp[paren] = -1; |
c277df42 |
2310 | } |
2311 | if (regmatch(next)) |
2312 | sayYES; |
2313 | REGCP_UNWIND; |
2314 | } |
2315 | /* Couldn't or didn't -- back up. */ |
2316 | n--; |
dfe13c55 |
2317 | PL_reginput = locinput = HOPc(locinput, -1); |
c277df42 |
2318 | } |
a0ed51b3 |
2319 | } |
2320 | else { |
c277df42 |
2321 | while (n >= ln) { |
2322 | /* If it could work, try it. */ |
2323 | if (c1 == -1000 || |
3280af22 |
2324 | UCHARAT(PL_reginput) == c1 || |
2325 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2326 | { |
2327 | if (regmatch(next)) |
2328 | sayYES; |
2329 | REGCP_UNWIND; |
2330 | } |
2331 | /* Couldn't or didn't -- back up. */ |
2332 | n--; |
dfe13c55 |
2333 | PL_reginput = locinput = HOPc(locinput, -1); |
bbce6d69 |
2334 | } |
a0d0e21e |
2335 | } |
2336 | } |
4633a7c4 |
2337 | sayNO; |
c277df42 |
2338 | break; |
a0d0e21e |
2339 | case END: |
0f5d15d6 |
2340 | if (PL_reg_call_cc) { |
2341 | re_cc_state *cur_call_cc = PL_reg_call_cc; |
2342 | CURCUR *cctmp = PL_regcc; |
2343 | regexp *re = PL_reg_re; |
2344 | CHECKPOINT cp, lastcp; |
2345 | |
2346 | cp = regcppush(0); /* Save *all* the positions. */ |
2347 | REGCP_SET; |
2348 | regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of |
2349 | the caller. */ |
2350 | PL_reginput = locinput; /* Make position available to |
2351 | the callcc. */ |
2352 | cache_re(PL_reg_call_cc->re); |
2353 | PL_regcc = PL_reg_call_cc->cc; |
2354 | PL_reg_call_cc = PL_reg_call_cc->prev; |
2355 | if (regmatch(cur_call_cc->node)) { |
2356 | PL_reg_call_cc = cur_call_cc; |
2357 | regcpblow(cp); |
2358 | sayYES; |
2359 | } |
2360 | REGCP_UNWIND; |
2361 | regcppop(); |
2362 | PL_reg_call_cc = cur_call_cc; |
2363 | PL_regcc = cctmp; |
2364 | PL_reg_re = re; |
2365 | cache_re(re); |
2366 | |
2367 | DEBUG_r( |
2368 | PerlIO_printf(Perl_debug_log, |
2369 | "%*s continuation failed...\n", |
2370 | REPORT_CODE_OFF+PL_regindent*2, "") |
2371 | ); |
2372 | sayNO; |
2373 | } |
3280af22 |
2374 | if (locinput < PL_regtill) |
7e5428c5 |
2375 | sayNO; /* Cannot match: too short. */ |
2376 | /* Fall through */ |
2377 | case SUCCEED: |
3280af22 |
2378 | PL_reginput = locinput; /* put where regtry can find it */ |
4633a7c4 |
2379 | sayYES; /* Success! */ |
c277df42 |
2380 | case SUSPEND: |
2381 | n = 1; |
9fe1d20c |
2382 | PL_reginput = locinput; |
c277df42 |
2383 | goto do_ifmatch; |
a0d0e21e |
2384 | case UNLESSM: |
c277df42 |
2385 | n = 0; |
a0ed51b3 |
2386 | if (scan->flags) { |
0fe9bf95 |
2387 | if (UTF) { /* XXXX This is absolutely |
2388 | broken, we read before |
2389 | start of string. */ |
2390 | s = HOPMAYBEc(locinput, -scan->flags); |
2391 | if (!s) |
2392 | goto say_yes; |
2393 | PL_reginput = s; |
2394 | } |
2395 | else { |
2396 | if (locinput < PL_bostr + scan->flags) |
2397 | goto say_yes; |
2398 | PL_reginput = locinput - scan->flags; |
2399 | goto do_ifmatch; |
2400 | } |
a0ed51b3 |
2401 | } |
2402 | else |
2403 | PL_reginput = locinput; |
c277df42 |
2404 | goto do_ifmatch; |
2405 | case IFMATCH: |
2406 | n = 1; |
a0ed51b3 |
2407 | if (scan->flags) { |
0fe9bf95 |
2408 | if (UTF) { /* XXXX This is absolutely |
2409 | broken, we read before |
2410 | start of string. */ |
2411 | s = HOPMAYBEc(locinput, -scan->flags); |
2412 | if (!s || s < PL_bostr) |
2413 | goto say_no; |
2414 | PL_reginput = s; |
2415 | } |
2416 | else { |
2417 | if (locinput < PL_bostr + scan->flags) |
2418 | goto say_no; |
2419 | PL_reginput = locinput - scan->flags; |
2420 | goto do_ifmatch; |
2421 | } |
a0ed51b3 |
2422 | } |
2423 | else |
2424 | PL_reginput = locinput; |
2425 | |
c277df42 |
2426 | do_ifmatch: |
c277df42 |
2427 | inner = NEXTOPER(NEXTOPER(scan)); |
2428 | if (regmatch(inner) != n) { |
2429 | say_no: |
2430 | if (logical) { |
2431 | logical = 0; |
2432 | sw = 0; |
2433 | goto do_longjump; |
a0ed51b3 |
2434 | } |
2435 | else |
c277df42 |
2436 | sayNO; |
2437 | } |
2438 | say_yes: |
2439 | if (logical) { |
2440 | logical = 0; |
2441 | sw = 1; |
2442 | } |
fe44a5e8 |
2443 | if (OP(scan) == SUSPEND) { |
3280af22 |
2444 | locinput = PL_reginput; |
565764a8 |
2445 | nextchr = UCHARAT(locinput); |
fe44a5e8 |
2446 | } |
c277df42 |
2447 | /* FALL THROUGH. */ |
2448 | case LONGJMP: |
2449 | do_longjump: |
2450 | next = scan + ARG(scan); |
2451 | if (next == scan) |
2452 | next = NULL; |
a0d0e21e |
2453 | break; |
2454 | default: |
c030ccd9 |
2455 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
c277df42 |
2456 | (unsigned long)scan, OP(scan)); |
cea2e8a9 |
2457 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c |
2458 | } |
a0d0e21e |
2459 | scan = next; |
2460 | } |
a687059c |
2461 | |
a0d0e21e |
2462 | /* |
2463 | * We get here only if there's trouble -- normally "case END" is |
2464 | * the terminating point. |
2465 | */ |
cea2e8a9 |
2466 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e |
2467 | /*NOTREACHED*/ |
4633a7c4 |
2468 | sayNO; |
2469 | |
2470 | yes: |
2471 | #ifdef DEBUGGING |
3280af22 |
2472 | PL_regindent--; |
4633a7c4 |
2473 | #endif |
2474 | return 1; |
2475 | |
2476 | no: |
2477 | #ifdef DEBUGGING |
3280af22 |
2478 | PL_regindent--; |
4633a7c4 |
2479 | #endif |
a0d0e21e |
2480 | return 0; |
a687059c |
2481 | } |
2482 | |
2483 | /* |
2484 | - regrepeat - repeatedly match something simple, report how many |
2485 | */ |
2486 | /* |
2487 | * [This routine now assumes that it will only match on things of length 1. |
2488 | * That was true before, but now we assume scan - reginput is the count, |
a0ed51b3 |
2489 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c |
2490 | */ |
76e3520e |
2491 | STATIC I32 |
cea2e8a9 |
2492 | S_regrepeat(pTHX_ regnode *p, I32 max) |
a687059c |
2493 | { |
5c0ca799 |
2494 | dTHR; |
a0d0e21e |
2495 | register char *scan; |
2496 | register char *opnd; |
2497 | register I32 c; |
3280af22 |
2498 | register char *loceol = PL_regeol; |
a0ed51b3 |
2499 | register I32 hardcount = 0; |
a0d0e21e |
2500 | |
3280af22 |
2501 | scan = PL_reginput; |
c277df42 |
2502 | if (max != REG_INFTY && max < loceol - scan) |
a0d0e21e |
2503 | loceol = scan + max; |
161b471a |
2504 | opnd = (char *) OPERAND(p); |
a0d0e21e |
2505 | switch (OP(p)) { |
22c35a8c |
2506 | case REG_ANY: |
a0d0e21e |
2507 | while (scan < loceol && *scan != '\n') |
2508 | scan++; |
2509 | break; |
2510 | case SANY: |
2511 | scan = loceol; |
2512 | break; |
a0ed51b3 |
2513 | case ANYUTF8: |
2514 | loceol = PL_regeol; |
2515 | while (scan < loceol && *scan != '\n') { |
2516 | scan += UTF8SKIP(scan); |
2517 | hardcount++; |
2518 | } |
2519 | break; |
2520 | case SANYUTF8: |
2521 | loceol = PL_regeol; |
2522 | while (scan < loceol) { |
2523 | scan += UTF8SKIP(scan); |
2524 | hardcount++; |
2525 | } |
2526 | break; |
bbce6d69 |
2527 | case EXACT: /* length of string is 1 */ |
2528 | c = UCHARAT(++opnd); |
2529 | while (scan < loceol && UCHARAT(scan) == c) |
2530 | scan++; |
2531 | break; |
2532 | case EXACTF: /* length of string is 1 */ |
2533 | c = UCHARAT(++opnd); |
2534 | while (scan < loceol && |
22c35a8c |
2535 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 |
2536 | scan++; |
2537 | break; |
2538 | case EXACTFL: /* length of string is 1 */ |
3280af22 |
2539 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2540 | c = UCHARAT(++opnd); |
2541 | while (scan < loceol && |
22c35a8c |
2542 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e |
2543 | scan++; |
2544 | break; |
a0ed51b3 |
2545 | case ANYOFUTF8: |
2546 | loceol = PL_regeol; |
2547 | while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) { |
2548 | scan += UTF8SKIP(scan); |
2549 | hardcount++; |
2550 | } |
2551 | break; |
a0d0e21e |
2552 | case ANYOF: |
ae5c130c |
2553 | while (scan < loceol && REGINCLASS(opnd, *scan)) |
a0d0e21e |
2554 | scan++; |
a0d0e21e |
2555 | break; |
2556 | case ALNUM: |
2557 | while (scan < loceol && isALNUM(*scan)) |
2558 | scan++; |
2559 | break; |
a0ed51b3 |
2560 | case ALNUMUTF8: |
2561 | loceol = PL_regeol; |
dfe13c55 |
2562 | while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
2563 | scan += UTF8SKIP(scan); |
2564 | hardcount++; |
2565 | } |
2566 | break; |
bbce6d69 |
2567 | case ALNUML: |
3280af22 |
2568 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2569 | while (scan < loceol && isALNUM_LC(*scan)) |
2570 | scan++; |
2571 | break; |
a0ed51b3 |
2572 | case ALNUMLUTF8: |
2573 | PL_reg_flags |= RF_tainted; |
2574 | loceol = PL_regeol; |
dfe13c55 |
2575 | while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
2576 | scan += UTF8SKIP(scan); |
2577 | hardcount++; |
2578 | } |
2579 | break; |
2580 | break; |
a0d0e21e |
2581 | case NALNUM: |
2582 | while (scan < loceol && !isALNUM(*scan)) |
2583 | scan++; |
2584 | break; |
a0ed51b3 |
2585 | case NALNUMUTF8: |
2586 | loceol = PL_regeol; |
dfe13c55 |
2587 | while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
2588 | scan += UTF8SKIP(scan); |
2589 | hardcount++; |
2590 | } |
2591 | break; |
bbce6d69 |
2592 | case NALNUML: |
3280af22 |
2593 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2594 | while (scan < loceol && !isALNUM_LC(*scan)) |
2595 | scan++; |
2596 | break; |
a0ed51b3 |
2597 | case NALNUMLUTF8: |
2598 | PL_reg_flags |= RF_tainted; |
2599 | loceol = PL_regeol; |
dfe13c55 |
2600 | while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
2601 | scan += UTF8SKIP(scan); |
2602 | hardcount++; |
2603 | } |
2604 | break; |
a0d0e21e |
2605 | case SPACE: |
2606 | while (scan < loceol && isSPACE(*scan)) |
2607 | scan++; |
2608 | break; |
a0ed51b3 |
2609 | case SPACEUTF8: |
2610 | loceol = PL_regeol; |
dfe13c55 |
2611 | while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
2612 | scan += UTF8SKIP(scan); |
2613 | hardcount++; |
2614 | } |
2615 | break; |
bbce6d69 |
2616 | case SPACEL: |
3280af22 |
2617 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2618 | while (scan < loceol && isSPACE_LC(*scan)) |
2619 | scan++; |
2620 | break; |
a0ed51b3 |
2621 | case SPACELUTF8: |
2622 | PL_reg_flags |= RF_tainted; |
2623 | loceol = PL_regeol; |
dfe13c55 |
2624 | while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
2625 | scan += UTF8SKIP(scan); |
2626 | hardcount++; |
2627 | } |
2628 | break; |
a0d0e21e |
2629 | case NSPACE: |
2630 | while (scan < loceol && !isSPACE(*scan)) |
2631 | scan++; |
2632 | break; |
a0ed51b3 |
2633 | case NSPACEUTF8: |
2634 | loceol = PL_regeol; |
dfe13c55 |
2635 | while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
2636 | scan += UTF8SKIP(scan); |
2637 | hardcount++; |
2638 | } |
2639 | break; |
bbce6d69 |
2640 | case NSPACEL: |
3280af22 |
2641 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2642 | while (scan < loceol && !isSPACE_LC(*scan)) |
2643 | scan++; |
2644 | break; |
a0ed51b3 |
2645 | case NSPACELUTF8: |
2646 | PL_reg_flags |= RF_tainted; |
2647 | loceol = PL_regeol; |
dfe13c55 |
2648 | while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
2649 | scan += UTF8SKIP(scan); |
2650 | hardcount++; |
2651 | } |
2652 | break; |
a0d0e21e |
2653 | case DIGIT: |
2654 | while (scan < loceol && isDIGIT(*scan)) |
2655 | scan++; |
2656 | break; |
a0ed51b3 |
2657 | case DIGITUTF8: |
2658 | loceol = PL_regeol; |
dfe13c55 |
2659 | while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
2660 | scan += UTF8SKIP(scan); |
2661 | hardcount++; |
2662 | } |
2663 | break; |
2664 | break; |
a0d0e21e |
2665 | case NDIGIT: |
2666 | while (scan < loceol && !isDIGIT(*scan)) |
2667 | scan++; |
2668 | break; |
a0ed51b3 |
2669 | case NDIGITUTF8: |
2670 | loceol = PL_regeol; |
dfe13c55 |
2671 | while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
2672 | scan += UTF8SKIP(scan); |
2673 | hardcount++; |
2674 | } |
2675 | break; |
a0d0e21e |
2676 | default: /* Called on something of 0 width. */ |
2677 | break; /* So match right here or not at all. */ |
2678 | } |
a687059c |
2679 | |
a0ed51b3 |
2680 | if (hardcount) |
2681 | c = hardcount; |
2682 | else |
2683 | c = scan - PL_reginput; |
3280af22 |
2684 | PL_reginput = scan; |
a687059c |
2685 | |
c277df42 |
2686 | DEBUG_r( |
2687 | { |
2688 | SV *prop = sv_newmortal(); |
2689 | |
2690 | regprop(prop, p); |
2691 | PerlIO_printf(Perl_debug_log, |
2692 | "%*s %s can match %ld times out of %ld...\n", |
2693 | REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); |
2694 | }); |
2695 | |
a0d0e21e |
2696 | return(c); |
a687059c |
2697 | } |
2698 | |
2699 | /* |
c277df42 |
2700 | - regrepeat_hard - repeatedly match something, report total lenth and length |
2701 | * |
2702 | * The repeater is supposed to have constant length. |
2703 | */ |
2704 | |
76e3520e |
2705 | STATIC I32 |
cea2e8a9 |
2706 | S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) |
c277df42 |
2707 | { |
5c0ca799 |
2708 | dTHR; |
c277df42 |
2709 | register char *scan; |
2710 | register char *start; |
3280af22 |
2711 | register char *loceol = PL_regeol; |
a0ed51b3 |
2712 | I32 l = 0; |
708e3b05 |
2713 | I32 count = 0, res = 1; |
a0ed51b3 |
2714 | |
2715 | if (!max) |
2716 | return 0; |
c277df42 |
2717 | |
3280af22 |
2718 | start = PL_reginput; |
a0ed51b3 |
2719 | if (UTF) { |
708e3b05 |
2720 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
2721 | if (!count++) { |
2722 | l = 0; |
2723 | while (start < PL_reginput) { |
2724 | l++; |
2725 | start += UTF8SKIP(start); |
2726 | } |
2727 | *lp = l; |
2728 | if (l == 0) |
2729 | return max; |
2730 | } |
2731 | if (count == max) |
2732 | return count; |
2733 | } |
2734 | } |
2735 | else { |
708e3b05 |
2736 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
2737 | if (!count++) { |
2738 | *lp = l = PL_reginput - start; |
2739 | if (max != REG_INFTY && l*max < loceol - scan) |
2740 | loceol = scan + l*max; |
2741 | if (l == 0) |
2742 | return max; |
c277df42 |
2743 | } |
2744 | } |
2745 | } |
708e3b05 |
2746 | if (!res) |
3280af22 |
2747 | PL_reginput = scan; |
c277df42 |
2748 | |
a0ed51b3 |
2749 | return count; |
c277df42 |
2750 | } |
2751 | |
2752 | /* |
cb8d8820 |
2753 | - reginclass - determine if a character falls into a character class |
bbce6d69 |
2754 | */ |
2755 | |
76e3520e |
2756 | STATIC bool |
cea2e8a9 |
2757 | S_reginclass(pTHX_ register char *p, register I32 c) |
bbce6d69 |
2758 | { |
5c0ca799 |
2759 | dTHR; |
bbce6d69 |
2760 | char flags = *p; |
2761 | bool match = FALSE; |
2762 | |
2763 | c &= 0xFF; |
ae5c130c |
2764 | if (ANYOF_TEST(p, c)) |
bbce6d69 |
2765 | match = TRUE; |
2766 | else if (flags & ANYOF_FOLD) { |
2767 | I32 cf; |
2768 | if (flags & ANYOF_LOCALE) { |
3280af22 |
2769 | PL_reg_flags |= RF_tainted; |
22c35a8c |
2770 | cf = PL_fold_locale[c]; |
bbce6d69 |
2771 | } |
2772 | else |
22c35a8c |
2773 | cf = PL_fold[c]; |
ae5c130c |
2774 | if (ANYOF_TEST(p, cf)) |
bbce6d69 |
2775 | match = TRUE; |
2776 | } |
2777 | |
2778 | if (!match && (flags & ANYOF_ISA)) { |
3280af22 |
2779 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
2780 | |
2781 | if (((flags & ANYOF_ALNUML) && isALNUM_LC(c)) || |
2782 | ((flags & ANYOF_NALNUML) && !isALNUM_LC(c)) || |
2783 | ((flags & ANYOF_SPACEL) && isSPACE_LC(c)) || |
2784 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC(c))) |
2785 | { |
2786 | match = TRUE; |
2787 | } |
2788 | } |
2789 | |
ae5c130c |
2790 | return (flags & ANYOF_INVERT) ? !match : match; |
bbce6d69 |
2791 | } |
2792 | |
a0ed51b3 |
2793 | STATIC bool |
cea2e8a9 |
2794 | S_reginclassutf8(pTHX_ regnode *f, U8 *p) |
c485e607 |
2795 | { |
2796 | dTHR; |
a0ed51b3 |
2797 | char flags = ARG1(f); |
2798 | bool match = FALSE; |
2799 | SV *sv = (SV*)PL_regdata->data[ARG2(f)]; |
2800 | |
2801 | if (swash_fetch(sv, p)) |
2802 | match = TRUE; |
2803 | else if (flags & ANYOF_FOLD) { |
2804 | I32 cf; |
dfe13c55 |
2805 | U8 tmpbuf[10]; |
a0ed51b3 |
2806 | if (flags & ANYOF_LOCALE) { |
2807 | PL_reg_flags |= RF_tainted; |
2808 | uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); |
2809 | } |
2810 | else |
2811 | uv_to_utf8(tmpbuf, toLOWER_utf8(p)); |
2812 | if (swash_fetch(sv, tmpbuf)) |
2813 | match = TRUE; |
2814 | } |
2815 | |
2816 | if (!match && (flags & ANYOF_ISA)) { |
2817 | PL_reg_flags |= RF_tainted; |
2818 | |
2819 | if (((flags & ANYOF_ALNUML) && isALNUM_LC_utf8(p)) || |
2820 | ((flags & ANYOF_NALNUML) && !isALNUM_LC_utf8(p)) || |
2821 | ((flags & ANYOF_SPACEL) && isSPACE_LC_utf8(p)) || |
2822 | ((flags & ANYOF_NSPACEL) && !isSPACE_LC_utf8(p))) |
2823 | { |
2824 | match = TRUE; |
2825 | } |
2826 | } |
2827 | |
2828 | return (flags & ANYOF_INVERT) ? !match : match; |
2829 | } |
161b471a |
2830 | |
dfe13c55 |
2831 | STATIC U8 * |
cea2e8a9 |
2832 | S_reghop(pTHX_ U8 *s, I32 off) |
c485e607 |
2833 | { |
2834 | dTHR; |
a0ed51b3 |
2835 | if (off >= 0) { |
2836 | while (off-- && s < (U8*)PL_regeol) |
2837 | s += UTF8SKIP(s); |
2838 | } |
2839 | else { |
2840 | while (off++) { |
2841 | if (s > (U8*)PL_bostr) { |
2842 | s--; |
2843 | if (*s & 0x80) { |
2844 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
2845 | s--; |
2846 | } /* XXX could check well-formedness here */ |
2847 | } |
2848 | } |
2849 | } |
2850 | return s; |
2851 | } |
161b471a |
2852 | |
dfe13c55 |
2853 | STATIC U8 * |
cea2e8a9 |
2854 | S_reghopmaybe(pTHX_ U8* s, I32 off) |
a0ed51b3 |
2855 | { |
c485e607 |
2856 | dTHR; |
a0ed51b3 |
2857 | if (off >= 0) { |
2858 | while (off-- && s < (U8*)PL_regeol) |
2859 | s += UTF8SKIP(s); |
2860 | if (off >= 0) |
2861 | return 0; |
2862 | } |
2863 | else { |
2864 | while (off++) { |
2865 | if (s > (U8*)PL_bostr) { |
2866 | s--; |
2867 | if (*s & 0x80) { |
2868 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
2869 | s--; |
2870 | } /* XXX could check well-formedness here */ |
2871 | } |
2872 | else |
2873 | break; |
2874 | } |
2875 | if (off <= 0) |
2876 | return 0; |
2877 | } |
2878 | return s; |
2879 | } |