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 */ |
cad2e5aa |
28 | # if defined(PERL_EXT_RE_DEBUG) && !defined(DEBUGGING) |
b9d5759e |
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 |
cad2e5aa |
38 | # define Perl_re_intuit_start my_re_intuit_start |
d06ea78c |
39 | /* *These* symbols are masked to allow static link. */ |
40 | # define Perl_pregexec my_pregexec |
d88dccdf |
41 | # define Perl_reginitcolors my_reginitcolors |
c5be433b |
42 | |
43 | # define PERL_NO_GET_CONTEXT |
56953603 |
44 | #endif |
45 | |
f0fcb552 |
46 | /*SUPPRESS 112*/ |
a687059c |
47 | /* |
e50aee73 |
48 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c |
49 | * |
50 | * Copyright (c) 1986 by University of Toronto. |
51 | * Written by Henry Spencer. Not derived from licensed software. |
52 | * |
53 | * Permission is granted to anyone to use this software for any |
54 | * purpose on any computer system, and to redistribute it freely, |
55 | * subject to the following restrictions: |
56 | * |
57 | * 1. The author is not responsible for the consequences of use of |
58 | * this software, no matter how awful, even if they arise |
59 | * from defects in it. |
60 | * |
61 | * 2. The origin of this software must not be misrepresented, either |
62 | * by explicit claim or by omission. |
63 | * |
64 | * 3. Altered versions must be plainly marked as such, and must not |
65 | * be misrepresented as being the original software. |
66 | * |
67 | **** Alterations to Henry's code are... |
68 | **** |
4eb8286e |
69 | **** Copyright (c) 1991-1999, Larry Wall |
a687059c |
70 | **** |
9ef589d8 |
71 | **** You may distribute under the terms of either the GNU General Public |
72 | **** License or the Artistic License, as specified in the README file. |
a687059c |
73 | * |
74 | * Beware that some of this code is subtly aware of the way operator |
75 | * precedence is structured in regular expressions. Serious changes in |
76 | * regular-expression syntax might require a total rethink. |
77 | */ |
78 | #include "EXTERN.h" |
864dbfa3 |
79 | #define PERL_IN_REGEXEC_C |
a687059c |
80 | #include "perl.h" |
0f5d15d6 |
81 | |
c5be433b |
82 | #ifdef PERL_IN_XSUB_RE |
83 | # if defined(PERL_CAPI) || defined(PERL_OBJECT) |
84 | # include "XSUB.h" |
85 | # endif |
86 | #endif |
87 | |
a687059c |
88 | #include "regcomp.h" |
89 | |
c277df42 |
90 | #define RF_tainted 1 /* tainted information used? */ |
91 | #define RF_warned 2 /* warned about big count? */ |
ce862d02 |
92 | #define RF_evaled 4 /* Did an EVAL with setting? */ |
a0ed51b3 |
93 | #define RF_utf8 8 /* String contains multibyte chars? */ |
94 | |
95 | #define UTF (PL_reg_flags & RF_utf8) |
ce862d02 |
96 | |
97 | #define RS_init 1 /* eval environment created */ |
98 | #define RS_set 2 /* replsv value is set */ |
c277df42 |
99 | |
a687059c |
100 | #ifndef STATIC |
101 | #define STATIC static |
102 | #endif |
103 | |
c277df42 |
104 | /* |
105 | * Forwards. |
106 | */ |
107 | |
b8c5462f |
108 | #define REGINCLASS(p,c) (ANYOF_FLAGS(p) ? reginclass(p,c) : ANYOF_BITMAP_TEST(p,c)) |
a0ed51b3 |
109 | #define REGINCLASSUTF8(f,p) (ARG1(f) ? reginclassutf8(f,p) : swash_fetch((SV*)PL_regdata->data[ARG2(f)],p)) |
110 | |
111 | #define CHR_SVLEN(sv) (UTF ? sv_len_utf8(sv) : SvCUR(sv)) |
112 | #define CHR_DIST(a,b) (UTF ? utf8_distance(a,b) : a - b) |
113 | |
dfe13c55 |
114 | #define reghop_c(pos,off) ((char*)reghop((U8*)pos, off)) |
115 | #define reghopmaybe_c(pos,off) ((char*)reghopmaybe((U8*)pos, off)) |
116 | #define HOP(pos,off) (UTF ? reghop((U8*)pos, off) : (U8*)(pos + off)) |
117 | #define HOPMAYBE(pos,off) (UTF ? reghopmaybe((U8*)pos, off) : (U8*)(pos + off)) |
118 | #define HOPc(pos,off) ((char*)HOP(pos,off)) |
119 | #define HOPMAYBEc(pos,off) ((char*)HOPMAYBE(pos,off)) |
a0d0e21e |
120 | |
51371543 |
121 | static void restore_pos(pTHXo_ void *arg); |
122 | |
123 | |
76e3520e |
124 | STATIC CHECKPOINT |
cea2e8a9 |
125 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e |
126 | { |
11343788 |
127 | dTHR; |
3280af22 |
128 | int retval = PL_savestack_ix; |
129 | int i = (PL_regsize - parenfloor) * 4; |
a0d0e21e |
130 | int p; |
131 | |
132 | SSCHECK(i + 5); |
3280af22 |
133 | for (p = PL_regsize; p > parenfloor; p--) { |
cf93c79d |
134 | SSPUSHINT(PL_regendp[p]); |
135 | SSPUSHINT(PL_regstartp[p]); |
3280af22 |
136 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e |
137 | SSPUSHINT(p); |
138 | } |
3280af22 |
139 | SSPUSHINT(PL_regsize); |
140 | SSPUSHINT(*PL_reglastparen); |
141 | SSPUSHPTR(PL_reginput); |
a0d0e21e |
142 | SSPUSHINT(i + 3); |
143 | SSPUSHINT(SAVEt_REGCONTEXT); |
144 | return retval; |
145 | } |
146 | |
c277df42 |
147 | /* These are needed since we do not localize EVAL nodes: */ |
c3464db5 |
148 | # define REGCP_SET DEBUG_r(PerlIO_printf(Perl_debug_log, \ |
149 | " Setting an EVAL scope, savestack=%i\n", \ |
3280af22 |
150 | PL_savestack_ix)); lastcp = PL_savestack_ix |
c3464db5 |
151 | |
3280af22 |
152 | # define REGCP_UNWIND DEBUG_r(lastcp != PL_savestack_ix ? \ |
c3464db5 |
153 | PerlIO_printf(Perl_debug_log, \ |
154 | " Clearing an EVAL scope, savestack=%i..%i\n", \ |
3280af22 |
155 | lastcp, PL_savestack_ix) : 0); regcpblow(lastcp) |
c277df42 |
156 | |
76e3520e |
157 | STATIC char * |
cea2e8a9 |
158 | S_regcppop(pTHX) |
a0d0e21e |
159 | { |
11343788 |
160 | dTHR; |
a0d0e21e |
161 | I32 i = SSPOPINT; |
162 | U32 paren = 0; |
163 | char *input; |
cf93c79d |
164 | I32 tmps; |
a0d0e21e |
165 | assert(i == SAVEt_REGCONTEXT); |
166 | i = SSPOPINT; |
167 | input = (char *) SSPOPPTR; |
3280af22 |
168 | *PL_reglastparen = SSPOPINT; |
169 | PL_regsize = SSPOPINT; |
c277df42 |
170 | for (i -= 3; i > 0; i -= 4) { |
a0d0e21e |
171 | paren = (U32)SSPOPINT; |
3280af22 |
172 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d |
173 | PL_regstartp[paren] = SSPOPINT; |
174 | tmps = SSPOPINT; |
3280af22 |
175 | if (paren <= *PL_reglastparen) |
176 | PL_regendp[paren] = tmps; |
c277df42 |
177 | DEBUG_r( |
c3464db5 |
178 | PerlIO_printf(Perl_debug_log, |
179 | " restoring \\%d to %d(%d)..%d%s\n", |
cf93c79d |
180 | paren, PL_regstartp[paren], |
181 | PL_reg_start_tmp[paren] - PL_bostr, |
182 | PL_regendp[paren], |
3280af22 |
183 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 |
184 | ); |
a0d0e21e |
185 | } |
c277df42 |
186 | DEBUG_r( |
3280af22 |
187 | if (*PL_reglastparen + 1 <= PL_regnpar) { |
c3464db5 |
188 | PerlIO_printf(Perl_debug_log, |
189 | " restoring \\%d..\\%d to undef\n", |
3280af22 |
190 | *PL_reglastparen + 1, PL_regnpar); |
c277df42 |
191 | } |
192 | ); |
3280af22 |
193 | for (paren = *PL_reglastparen + 1; paren <= PL_regnpar; paren++) { |
194 | if (paren > PL_regsize) |
cf93c79d |
195 | PL_regstartp[paren] = -1; |
196 | PL_regendp[paren] = -1; |
a0d0e21e |
197 | } |
198 | return input; |
199 | } |
200 | |
0f5d15d6 |
201 | STATIC char * |
cea2e8a9 |
202 | S_regcp_set_to(pTHX_ I32 ss) |
0f5d15d6 |
203 | { |
46124e9e |
204 | dTHR; |
0f5d15d6 |
205 | I32 tmp = PL_savestack_ix; |
206 | |
207 | PL_savestack_ix = ss; |
208 | regcppop(); |
209 | PL_savestack_ix = tmp; |
942e002e |
210 | return Nullch; |
0f5d15d6 |
211 | } |
212 | |
213 | typedef struct re_cc_state |
214 | { |
215 | I32 ss; |
216 | regnode *node; |
217 | struct re_cc_state *prev; |
218 | CURCUR *cc; |
219 | regexp *re; |
220 | } re_cc_state; |
221 | |
c277df42 |
222 | #define regcpblow(cp) LEAVE_SCOPE(cp) |
a0d0e21e |
223 | |
a687059c |
224 | /* |
e50aee73 |
225 | * pregexec and friends |
a687059c |
226 | */ |
227 | |
228 | /* |
c277df42 |
229 | - pregexec - match a regexp against a string |
a687059c |
230 | */ |
c277df42 |
231 | I32 |
864dbfa3 |
232 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 |
233 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 |
234 | /* strend: pointer to null at end of string */ |
235 | /* strbeg: real beginning of string */ |
236 | /* minend: end of match must be >=minend after stringarg. */ |
237 | /* nosave: For optimizations. */ |
238 | { |
239 | return |
240 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
241 | nosave ? 0 : REXEC_COPY_STR); |
242 | } |
0f5d15d6 |
243 | |
244 | STATIC void |
cea2e8a9 |
245 | S_cache_re(pTHX_ regexp *prog) |
0f5d15d6 |
246 | { |
46124e9e |
247 | dTHR; |
0f5d15d6 |
248 | PL_regprecomp = prog->precomp; /* Needed for FAIL. */ |
249 | #ifdef DEBUGGING |
250 | PL_regprogram = prog->program; |
251 | #endif |
252 | PL_regnpar = prog->nparens; |
253 | PL_regdata = prog->data; |
254 | PL_reg_re = prog; |
255 | } |
22e551b9 |
256 | |
cad2e5aa |
257 | /* |
258 | * Need to implement the following flags for reg_anch: |
259 | * |
260 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first |
261 | * USE_INTUIT_ML |
262 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer |
263 | * INTUIT_AUTORITATIVE_ML |
264 | * INTUIT_ONCE_NOML - Intuit can match in one location only. |
265 | * INTUIT_ONCE_ML |
266 | * |
267 | * Another flag for this function: SECOND_TIME (so that float substrs |
268 | * with giant delta may be not rechecked). |
269 | */ |
270 | |
271 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ |
272 | |
2c2d71f5 |
273 | /* If SCREAM, then SvPVX(sv) should be compatible with strpos and strend. |
cad2e5aa |
274 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
275 | |
276 | /* XXXX We assume that strpos is strbeg unless sv. */ |
277 | |
2c2d71f5 |
278 | /* A failure to find a constant substring means that there is no need to make |
279 | an expensive call to REx engine, thus we celebrate a failure. Similarly, |
280 | finding a substring too deep into the string means that less calls to |
281 | regtry() should be needed. */ |
282 | |
cad2e5aa |
283 | char * |
284 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, |
285 | char *strend, U32 flags, re_scream_pos_data *data) |
286 | { |
2c2d71f5 |
287 | register I32 start_shift; |
cad2e5aa |
288 | /* Should be nonnegative! */ |
2c2d71f5 |
289 | register I32 end_shift; |
290 | register char *s; |
291 | register SV *check; |
cad2e5aa |
292 | char *t; |
293 | I32 ml_anch; |
2c2d71f5 |
294 | char *tmp; |
295 | register char *other_last = Nullch; |
cad2e5aa |
296 | |
297 | DEBUG_r( if (!PL_colorset) reginitcolors() ); |
298 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 |
299 | "%sGuessing start of match, REx%s `%s%.60s%s%s' against `%s%.*s%s%s'...\n", |
cad2e5aa |
300 | PL_colors[4],PL_colors[5],PL_colors[0], |
301 | prog->precomp, |
302 | PL_colors[1], |
303 | (strlen(prog->precomp) > 60 ? "..." : ""), |
304 | PL_colors[0], |
305 | (strend - strpos > 60 ? 60 : strend - strpos), |
306 | strpos, PL_colors[1], |
307 | (strend - strpos > 60 ? "..." : "")) |
308 | ); |
309 | |
2c2d71f5 |
310 | if (prog->minlen > strend - strpos) { |
311 | DEBUG_r(PerlIO_printf(Perl_debug_log, "String too short...\n")); |
cad2e5aa |
312 | goto fail; |
2c2d71f5 |
313 | } |
314 | if (prog->reganch & ROPT_ANCH) { /* Match at beg-of-str or after \n */ |
cad2e5aa |
315 | ml_anch = !( (prog->reganch & ROPT_ANCH_SINGLE) |
316 | || ( (prog->reganch & ROPT_ANCH_BOL) |
2c2d71f5 |
317 | && !PL_multiline ) ); /* Check after \n? */ |
cad2e5aa |
318 | |
319 | if ((prog->check_offset_min == prog->check_offset_max) && !ml_anch) { |
2c2d71f5 |
320 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa |
321 | I32 slen; |
322 | |
323 | if ( !(prog->reganch & ROPT_ANCH_GPOS) /* Checked by the caller */ |
2c2d71f5 |
324 | && (sv && (strpos + SvCUR(sv) != strend)) ) { |
325 | DEBUG_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
cad2e5aa |
326 | goto fail; |
2c2d71f5 |
327 | } |
adac82c7 |
328 | PL_regeol = strend; /* Used in HOP() */ |
2c2d71f5 |
329 | s = HOPc(strpos, prog->check_offset_min); |
cad2e5aa |
330 | if (SvTAIL(prog->check_substr)) { |
331 | slen = SvCUR(prog->check_substr); /* >= 1 */ |
332 | |
2c2d71f5 |
333 | if ( strend - s > slen || strend - s < slen - 1 |
334 | || (strend - s == slen && strend[-1] != '\n')) { |
335 | DEBUG_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
336 | goto fail_finish; |
cad2e5aa |
337 | } |
338 | /* Now should match s[0..slen-2] */ |
339 | slen--; |
340 | if (slen && (*SvPVX(prog->check_substr) != *s |
341 | || (slen > 1 |
2c2d71f5 |
342 | && memNE(SvPVX(prog->check_substr), s, slen)))) { |
343 | report_neq: |
344 | DEBUG_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
345 | goto fail_finish; |
346 | } |
cad2e5aa |
347 | } |
348 | else if (*SvPVX(prog->check_substr) != *s |
349 | || ((slen = SvCUR(prog->check_substr)) > 1 |
350 | && memNE(SvPVX(prog->check_substr), s, slen))) |
2c2d71f5 |
351 | goto report_neq; |
352 | goto success_at_start; |
cad2e5aa |
353 | } |
2c2d71f5 |
354 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa |
355 | s = strpos; |
2c2d71f5 |
356 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
357 | /* Should be nonnegative! */ |
358 | end_shift = prog->minlen - start_shift - |
359 | CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0); |
360 | if (!ml_anch) { |
361 | I32 end = prog->check_offset_max + CHR_SVLEN(prog->check_substr) |
362 | - (SvTAIL(prog->check_substr) != 0); |
363 | I32 eshift = strend - s - end; |
364 | |
365 | if (end_shift < eshift) |
366 | end_shift = eshift; |
367 | } |
cad2e5aa |
368 | } |
2c2d71f5 |
369 | else { /* Can match at random position */ |
cad2e5aa |
370 | ml_anch = 0; |
371 | s = strpos; |
2c2d71f5 |
372 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
373 | /* Should be nonnegative! */ |
374 | end_shift = prog->minlen - start_shift - |
375 | CHR_SVLEN(prog->check_substr) + (SvTAIL(prog->check_substr) != 0); |
cad2e5aa |
376 | } |
377 | |
2c2d71f5 |
378 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d |
379 | if (end_shift < 0) |
6bbae5e6 |
380 | Perl_croak(aTHX_ "panic: end_shift"); |
2c2d71f5 |
381 | #endif |
382 | |
383 | check = prog->check_substr; |
384 | restart: |
385 | /* Find a possible match in the region s..strend by looking for |
386 | the "check" substring in the region corrected by start/end_shift. */ |
cad2e5aa |
387 | if (flags & REXEC_SCREAM) { |
cad2e5aa |
388 | char *strbeg = SvPVX(sv); /* XXXX Assume PV_force() on SCREAM! */ |
389 | I32 p = -1; /* Internal iterator of scream. */ |
390 | I32 *pp = data ? data->scream_pos : &p; |
391 | |
2c2d71f5 |
392 | if (PL_screamfirst[BmRARE(check)] >= 0 |
393 | || ( BmRARE(check) == '\n' |
394 | && (BmPREVIOUS(check) == SvCUR(check) - 1) |
395 | && SvTAIL(check) )) |
396 | s = screaminstr(sv, check, |
397 | start_shift + (s - strbeg), end_shift, pp, 0); |
cad2e5aa |
398 | else |
2c2d71f5 |
399 | goto fail_finish; |
cad2e5aa |
400 | if (data) |
401 | *data->scream_olds = s; |
402 | } |
403 | else |
404 | s = fbm_instr((unsigned char*)s + start_shift, |
405 | (unsigned char*)strend - end_shift, |
2c2d71f5 |
406 | check, PL_multiline ? FBMrf_MULTILINE : 0); |
cad2e5aa |
407 | |
408 | /* Update the count-of-usability, remove useless subpatterns, |
409 | unshift s. */ |
2c2d71f5 |
410 | |
411 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%s %s substr `%s%.*s%s'%s%s", |
412 | (s ? "Found" : "Did not find"), |
413 | ((check == prog->anchored_substr) ? "anchored" : "floating"), |
414 | PL_colors[0], |
415 | SvCUR(check) - (SvTAIL(check)!=0), SvPVX(check), |
416 | PL_colors[1], (SvTAIL(check) ? "$" : ""), |
417 | (s ? " at offset " : "...\n") ) ); |
418 | |
419 | if (!s) |
420 | goto fail_finish; |
421 | |
422 | /* Finish the diagnostic message */ |
423 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - strpos)) ); |
424 | |
425 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
426 | Start with the other substr. |
427 | XXXX no SCREAM optimization yet - and a very coarse implementation |
428 | XXXX /ttx+/ results in anchored=`ttx', floating=`x'. floating will |
429 | *always* match. Probably should be marked during compile... |
430 | Probably it is right to do no SCREAM here... |
431 | */ |
432 | |
433 | if (prog->float_substr && prog->anchored_substr) { |
434 | /* Take into account the anchored substring. */ |
435 | /* XXXX May be hopelessly wrong for UTF... */ |
436 | if (!other_last) |
437 | other_last = strpos - 1; |
438 | if (check == prog->float_substr) { |
439 | char *last = s - start_shift, *last1, *last2; |
440 | char *s1 = s; |
441 | |
442 | tmp = PL_bostr; |
443 | t = s - prog->check_offset_max; |
444 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
445 | && (!(prog->reganch & ROPT_UTF8) |
446 | || (PL_bostr = strpos, /* Used in regcopmaybe() */ |
447 | (t = reghopmaybe_c(s, -(prog->check_offset_max))) |
448 | && t > strpos))) |
449 | ; |
450 | else |
451 | t = strpos; |
452 | t += prog->anchored_offset; |
453 | if (t <= other_last) |
454 | t = other_last + 1; |
455 | PL_bostr = tmp; |
456 | last2 = last1 = strend - prog->minlen; |
457 | if (last < last1) |
458 | last1 = last; |
459 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
460 | /* On end-of-str: see comment below. */ |
461 | s = fbm_instr((unsigned char*)t, |
462 | (unsigned char*)last1 + prog->anchored_offset |
463 | + SvCUR(prog->anchored_substr) |
464 | - (SvTAIL(prog->anchored_substr)!=0), |
465 | prog->anchored_substr, PL_multiline ? FBMrf_MULTILINE : 0); |
466 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%s anchored substr `%s%.*s%s'%s", |
467 | (s ? "Found" : "Contradicts"), |
468 | PL_colors[0], |
469 | SvCUR(prog->anchored_substr) |
470 | - (SvTAIL(prog->anchored_substr)!=0), |
471 | SvPVX(prog->anchored_substr), |
472 | PL_colors[1], (SvTAIL(prog->anchored_substr) ? "$" : ""))); |
473 | if (!s) { |
474 | if (last1 >= last2) { |
475 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
476 | ", giving up...\n")); |
477 | goto fail_finish; |
478 | } |
479 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
480 | ", trying floating at offset %ld...\n", |
481 | (long)(s1 + 1 - strpos))); |
482 | PL_regeol = strend; /* Used in HOP() */ |
483 | other_last = last1 + prog->anchored_offset; |
484 | s = HOPc(last, 1); |
485 | goto restart; |
486 | } |
487 | else { |
488 | DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
489 | (long)(s - strpos))); |
490 | t = s - prog->anchored_offset; |
491 | other_last = s - 1; |
492 | if (t == strpos) |
493 | goto try_at_start; |
494 | s = s1; |
495 | goto try_at_offset; |
496 | } |
497 | } |
498 | else { /* Take into account the floating substring. */ |
499 | char *last, *last1; |
500 | char *s1 = s; |
501 | |
502 | t = s - start_shift; |
503 | last1 = last = strend - prog->minlen + prog->float_min_offset; |
504 | if (last - t > prog->float_max_offset) |
505 | last = t + prog->float_max_offset; |
506 | s = t + prog->float_min_offset; |
507 | if (s <= other_last) |
508 | s = other_last + 1; |
509 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
510 | /* fbm_instr() takes into account exact value of end-of-str |
511 | if the check is SvTAIL(ed). Since false positives are OK, |
512 | and end-of-str is not later than strend we are OK. */ |
513 | s = fbm_instr((unsigned char*)s, |
514 | (unsigned char*)last + SvCUR(prog->float_substr) |
515 | - (SvTAIL(prog->float_substr)!=0), |
516 | prog->float_substr, PL_multiline ? FBMrf_MULTILINE : 0); |
517 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%s floating substr `%s%.*s%s'%s", |
518 | (s ? "Found" : "Contradicts"), |
519 | PL_colors[0], |
520 | SvCUR(prog->float_substr) |
521 | - (SvTAIL(prog->float_substr)!=0), |
522 | SvPVX(prog->float_substr), |
523 | PL_colors[1], (SvTAIL(prog->float_substr) ? "$" : ""))); |
524 | if (!s) { |
525 | if (last1 == last) { |
526 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
527 | ", giving up...\n")); |
528 | goto fail_finish; |
529 | } |
530 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
531 | ", trying anchored starting at offset %ld...\n", |
532 | (long)(s1 + 1 - strpos))); |
533 | other_last = last; |
534 | PL_regeol = strend; /* Used in HOP() */ |
535 | s = HOPc(t, 1); |
536 | goto restart; |
537 | } |
538 | else { |
539 | DEBUG_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
540 | (long)(s - strpos))); |
541 | other_last = s - 1; |
542 | if (t == strpos) |
543 | goto try_at_start; |
544 | s = s1; |
545 | goto try_at_offset; |
546 | } |
547 | } |
cad2e5aa |
548 | } |
2c2d71f5 |
549 | |
550 | t = s - prog->check_offset_max; |
551 | tmp = PL_bostr; |
552 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
553 | && (!(prog->reganch & ROPT_UTF8) |
554 | || (PL_bostr = strpos, /* Used in regcopmaybe() */ |
555 | ((t = reghopmaybe_c(s, -(prog->check_offset_max))) |
556 | && t > strpos)))) { |
557 | PL_bostr = tmp; |
558 | /* Fixed substring is found far enough so that the match |
559 | cannot start at strpos. */ |
560 | try_at_offset: |
cad2e5aa |
561 | if (ml_anch && t[-1] != '\n') { |
2c2d71f5 |
562 | find_anchor: /* Eventually fbm_*() should handle this */ |
563 | while (t < strend - prog->minlen) { |
cad2e5aa |
564 | if (*t == '\n') { |
565 | if (t < s - prog->check_offset_min) { |
566 | s = t + 1; |
2c2d71f5 |
567 | DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
568 | PL_colors[0],PL_colors[1], (long)(s - strpos))); |
cad2e5aa |
569 | goto set_useful; |
570 | } |
2c2d71f5 |
571 | DEBUG_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting at offset %ld...\n", |
572 | PL_colors[0],PL_colors[1], (long)(t + 1 - strpos))); |
cad2e5aa |
573 | s = t + 1; |
574 | goto restart; |
575 | } |
576 | t++; |
577 | } |
2c2d71f5 |
578 | DEBUG_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
579 | PL_colors[0],PL_colors[1])); |
580 | goto fail_finish; |
cad2e5aa |
581 | } |
582 | s = t; |
583 | set_useful: |
2c2d71f5 |
584 | ++BmUSEFUL(prog->check_substr); /* hooray/5 */ |
cad2e5aa |
585 | } |
586 | else { |
2c2d71f5 |
587 | PL_bostr = tmp; |
588 | /* The found string does not prohibit matching at beg-of-str |
589 | - no optimization of calling REx engine can be performed, |
590 | unless it was an MBOL and we are not after MBOL. */ |
591 | try_at_start: |
592 | /* Even in this situation we may use MBOL flag if strpos is offset |
593 | wrt the start of the string. */ |
594 | if (ml_anch && sv |
cad2e5aa |
595 | && (strpos + SvCUR(sv) != strend) && strpos[-1] != '\n') { |
596 | t = strpos; |
597 | goto find_anchor; |
598 | } |
2c2d71f5 |
599 | success_at_start: |
cad2e5aa |
600 | if (!(prog->reganch & ROPT_NAUGHTY) |
601 | && --BmUSEFUL(prog->check_substr) < 0 |
602 | && prog->check_substr == prog->float_substr) { /* boo */ |
603 | /* If flags & SOMETHING - do not do it many times on the same match */ |
604 | SvREFCNT_dec(prog->check_substr); |
605 | prog->check_substr = Nullsv; /* disable */ |
606 | prog->float_substr = Nullsv; /* clear */ |
607 | s = strpos; |
608 | prog->reganch &= ~RE_USE_INTUIT; |
609 | } |
610 | else |
611 | s = strpos; |
612 | } |
613 | |
2c2d71f5 |
614 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sGuessed:%s match at offset %ld\n", |
615 | PL_colors[4], PL_colors[5], (long)(s - strpos)) ); |
cad2e5aa |
616 | return s; |
2c2d71f5 |
617 | |
618 | fail_finish: /* Substring not found */ |
619 | BmUSEFUL(prog->check_substr) += 5; /* hooray */ |
cad2e5aa |
620 | fail: |
2c2d71f5 |
621 | DEBUG_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
cad2e5aa |
622 | PL_colors[4],PL_colors[5])); |
623 | return Nullch; |
624 | } |
9661b544 |
625 | |
a687059c |
626 | /* |
c277df42 |
627 | - regexec_flags - match a regexp against a string |
a687059c |
628 | */ |
79072805 |
629 | I32 |
864dbfa3 |
630 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
22e551b9 |
631 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) |
c277df42 |
632 | /* strend: pointer to null at end of string */ |
633 | /* strbeg: real beginning of string */ |
634 | /* minend: end of match must be >=minend after stringarg. */ |
635 | /* data: May be used for some additional optimizations. */ |
636 | /* nosave: For optimizations. */ |
a687059c |
637 | { |
5c0ca799 |
638 | dTHR; |
a0d0e21e |
639 | register char *s; |
c277df42 |
640 | register regnode *c; |
a0d0e21e |
641 | register char *startpos = stringarg; |
642 | register I32 tmp; |
c277df42 |
643 | I32 minlen; /* must match at least this many chars */ |
a0d0e21e |
644 | I32 dontbother = 0; /* how many characters not to try at end */ |
645 | CURCUR cc; |
c277df42 |
646 | I32 start_shift = 0; /* Offset of the start to find |
a0ed51b3 |
647 | constant substr. */ /* CC */ |
648 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
c277df42 |
649 | I32 scream_pos = -1; /* Internal iterator of scream. */ |
650 | char *scream_olds; |
3280af22 |
651 | SV* oreplsv = GvSV(PL_replgv); |
a687059c |
652 | |
a0d0e21e |
653 | cc.cur = 0; |
4633a7c4 |
654 | cc.oldcc = 0; |
3280af22 |
655 | PL_regcc = &cc; |
a0d0e21e |
656 | |
0f5d15d6 |
657 | cache_re(prog); |
a0d0e21e |
658 | #ifdef DEBUGGING |
3280af22 |
659 | PL_regnarrate = PL_debug & 512; |
a0d0e21e |
660 | #endif |
661 | |
662 | /* Be paranoid... */ |
663 | if (prog == NULL || startpos == NULL) { |
cea2e8a9 |
664 | Perl_croak(aTHX_ "NULL regexp parameter"); |
a0d0e21e |
665 | return 0; |
666 | } |
667 | |
c277df42 |
668 | minlen = prog->minlen; |
669 | if (strend - startpos < minlen) goto phooey; |
670 | |
a0d0e21e |
671 | if (startpos == strbeg) /* is ^ valid at stringarg? */ |
3280af22 |
672 | PL_regprev = '\n'; |
a0d0e21e |
673 | else { |
a0ed51b3 |
674 | PL_regprev = (U32)stringarg[-1]; |
3280af22 |
675 | if (!PL_multiline && PL_regprev == '\n') |
676 | PL_regprev = '\0'; /* force ^ to NOT match */ |
a0d0e21e |
677 | } |
bbce6d69 |
678 | |
a0d0e21e |
679 | /* Check validity of program. */ |
22c35a8c |
680 | if (UCHARAT(prog->program) != REG_MAGIC) { |
cea2e8a9 |
681 | Perl_croak(aTHX_ "corrupted regexp program"); |
a0d0e21e |
682 | } |
683 | |
3280af22 |
684 | PL_reg_flags = 0; |
685 | PL_reg_eval_set = 0; |
2c2d71f5 |
686 | PL_reg_maxiter = 0; |
a0d0e21e |
687 | |
a0ed51b3 |
688 | if (prog->reganch & ROPT_UTF8) |
689 | PL_reg_flags |= RF_utf8; |
690 | |
691 | /* Mark beginning of line for ^ and lookbehind. */ |
692 | PL_regbol = startpos; |
693 | PL_bostr = strbeg; |
9661b544 |
694 | PL_reg_sv = sv; |
a0ed51b3 |
695 | |
696 | /* Mark end of line for $ (and such) */ |
697 | PL_regeol = strend; |
698 | |
699 | /* see how far we have to get to not match where we matched before */ |
700 | PL_regtill = startpos+minend; |
701 | |
0f5d15d6 |
702 | /* We start without call_cc context. */ |
703 | PL_reg_call_cc = 0; |
704 | |
a0d0e21e |
705 | /* If there is a "must appear" string, look for it. */ |
706 | s = startpos; |
cad2e5aa |
707 | |
708 | if (prog->reganch & ROPT_GPOS_SEEN) { |
709 | MAGIC *mg; |
710 | |
711 | if (!(flags & REXEC_IGNOREPOS) && sv && SvTYPE(sv) >= SVt_PVMG |
712 | && SvMAGIC(sv) && (mg = mg_find(sv, 'g')) && mg->mg_len >= 0) |
713 | PL_reg_ganch = strbeg + mg->mg_len; |
a0d0e21e |
714 | else |
cad2e5aa |
715 | PL_reg_ganch = startpos; |
716 | if (prog->reganch & ROPT_ANCH_GPOS) { |
717 | if (s > PL_reg_ganch) |
718 | goto phooey; |
719 | s = PL_reg_ganch; |
a0ed51b3 |
720 | } |
a0d0e21e |
721 | } |
a687059c |
722 | |
cad2e5aa |
723 | if (!(flags & REXEC_CHECKED) && prog->check_substr != Nullsv) { |
724 | re_scream_pos_data d; |
725 | |
726 | d.scream_olds = &scream_olds; |
727 | d.scream_pos = &scream_pos; |
728 | s = re_intuit_start(prog, sv, s, strend, flags, &d); |
729 | if (!s) |
730 | goto phooey; /* not present */ |
731 | } |
732 | |
733 | DEBUG_r( if (!PL_colorset) reginitcolors() ); |
734 | DEBUG_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 |
735 | "%sMatching REx%s `%s%.60s%s%s' against `%s%.*s%s%s'\n", |
8d300b32 |
736 | PL_colors[4],PL_colors[5],PL_colors[0], |
737 | prog->precomp, |
738 | PL_colors[1], |
c277df42 |
739 | (strlen(prog->precomp) > 60 ? "..." : ""), |
cad2e5aa |
740 | PL_colors[0], |
c277df42 |
741 | (strend - startpos > 60 ? 60 : strend - startpos), |
8d300b32 |
742 | startpos, PL_colors[1], |
c277df42 |
743 | (strend - startpos > 60 ? "..." : "")) |
744 | ); |
745 | |
a0d0e21e |
746 | /* Simplest case: anchored match need be tried only once. */ |
774d564b |
747 | /* [unless only anchor is BOL and multiline is set] */ |
22e551b9 |
748 | if (prog->reganch & (ROPT_ANCH & ~ROPT_ANCH_GPOS)) { |
cad2e5aa |
749 | if (s == startpos && regtry(prog, startpos)) |
a0d0e21e |
750 | goto got_it; |
22e551b9 |
751 | else if (PL_multiline || (prog->reganch & ROPT_IMPLICIT) |
752 | || (prog->reganch & ROPT_ANCH_MBOL)) /* XXXX SBOL? */ |
774d564b |
753 | { |
cad2e5aa |
754 | char *end; |
755 | |
a0d0e21e |
756 | if (minlen) |
757 | dontbother = minlen - 1; |
cad2e5aa |
758 | end = HOPc(strend, -dontbother) - 1; |
a0d0e21e |
759 | /* for multiline we only have to try after newlines */ |
cad2e5aa |
760 | if (prog->check_substr) { |
761 | while (1) { |
762 | if (regtry(prog, s)) |
a0d0e21e |
763 | goto got_it; |
cad2e5aa |
764 | if (s >= end) |
765 | goto phooey; |
766 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); |
767 | if (!s) |
768 | goto phooey; |
769 | } |
770 | } else { |
771 | if (s > startpos) |
772 | s--; |
773 | while (s < end) { |
774 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ |
775 | if (regtry(prog, s)) |
776 | goto got_it; |
777 | } |
778 | } |
35c8bce7 |
779 | } |
35c8bce7 |
780 | } |
a0d0e21e |
781 | goto phooey; |
22e551b9 |
782 | } else if (prog->reganch & ROPT_ANCH_GPOS) { |
783 | if (regtry(prog, PL_reg_ganch)) |
784 | goto got_it; |
785 | goto phooey; |
a0d0e21e |
786 | } |
35c8bce7 |
787 | |
a0d0e21e |
788 | /* Messy cases: unanchored match. */ |
c277df42 |
789 | if (prog->anchored_substr && prog->reganch & ROPT_SKIP) { |
790 | /* we have /x+whatever/ */ |
cad2e5aa |
791 | /* it must be a one character string (XXXX Except UTF?) */ |
c277df42 |
792 | char ch = SvPVX(prog->anchored_substr)[0]; |
a0ed51b3 |
793 | if (UTF) { |
794 | while (s < strend) { |
795 | if (*s == ch) { |
796 | if (regtry(prog, s)) goto got_it; |
797 | s += UTF8SKIP(s); |
798 | while (s < strend && *s == ch) |
799 | s += UTF8SKIP(s); |
800 | } |
801 | s += UTF8SKIP(s); |
802 | } |
803 | } |
804 | else { |
805 | while (s < strend) { |
806 | if (*s == ch) { |
807 | if (regtry(prog, s)) goto got_it; |
c277df42 |
808 | s++; |
a0ed51b3 |
809 | while (s < strend && *s == ch) |
810 | s++; |
811 | } |
812 | s++; |
a0d0e21e |
813 | } |
a687059c |
814 | } |
c277df42 |
815 | } |
816 | /*SUPPRESS 560*/ |
817 | else if (prog->anchored_substr != Nullsv |
818 | || (prog->float_substr != Nullsv |
819 | && prog->float_max_offset < strend - s)) { |
820 | SV *must = prog->anchored_substr |
821 | ? prog->anchored_substr : prog->float_substr; |
822 | I32 back_max = |
823 | prog->anchored_substr ? prog->anchored_offset : prog->float_max_offset; |
824 | I32 back_min = |
825 | prog->anchored_substr ? prog->anchored_offset : prog->float_min_offset; |
826 | I32 delta = back_max - back_min; |
cf93c79d |
827 | char *last = HOPc(strend, /* Cannot start after this */ |
73c4f7a1 |
828 | -(I32)(CHR_SVLEN(must) |
829 | - (SvTAIL(must) != 0) + back_min)); |
a0ed51b3 |
830 | char *last1; /* Last position checked before */ |
831 | |
832 | if (s > PL_bostr) |
dfe13c55 |
833 | last1 = HOPc(s, -1); |
a0ed51b3 |
834 | else |
835 | last1 = s - 1; /* bogus */ |
c277df42 |
836 | |
837 | /* XXXX check_substr already used to find `s', can optimize if |
838 | check_substr==must. */ |
839 | scream_pos = -1; |
840 | dontbother = end_shift; |
dfe13c55 |
841 | strend = HOPc(strend, -dontbother); |
c277df42 |
842 | while ( (s <= last) && |
22e551b9 |
843 | ((flags & REXEC_SCREAM) |
844 | ? (s = screaminstr(sv, must, HOPc(s, back_min) - strbeg, |
c277df42 |
845 | end_shift, &scream_pos, 0)) |
a0ed51b3 |
846 | : (s = fbm_instr((unsigned char*)HOP(s, back_min), |
cf93c79d |
847 | (unsigned char*)strend, must, |
848 | PL_multiline ? FBMrf_MULTILINE : 0))) ) { |
dfe13c55 |
849 | if (HOPc(s, -back_max) > last1) { |
850 | last1 = HOPc(s, -back_min); |
851 | s = HOPc(s, -back_max); |
a0ed51b3 |
852 | } |
853 | else { |
dfe13c55 |
854 | char *t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
c277df42 |
855 | |
dfe13c55 |
856 | last1 = HOPc(s, -back_min); |
c277df42 |
857 | s = t; |
a0d0e21e |
858 | } |
a0ed51b3 |
859 | if (UTF) { |
860 | while (s <= last1) { |
861 | if (regtry(prog, s)) |
862 | goto got_it; |
863 | s += UTF8SKIP(s); |
864 | } |
865 | } |
866 | else { |
867 | while (s <= last1) { |
868 | if (regtry(prog, s)) |
869 | goto got_it; |
870 | s++; |
871 | } |
a0d0e21e |
872 | } |
873 | } |
874 | goto phooey; |
a0ed51b3 |
875 | } |
876 | else if (c = prog->regstclass) { |
a0d0e21e |
877 | I32 doevery = (prog->reganch & ROPT_SKIP) == 0; |
a0ed51b3 |
878 | char *cc; |
a687059c |
879 | |
a0d0e21e |
880 | if (minlen) |
881 | dontbother = minlen - 1; |
dfe13c55 |
882 | strend = HOPc(strend, -dontbother); /* don't bother with what can't match */ |
a0d0e21e |
883 | tmp = 1; |
884 | /* We know what class it must start with. */ |
885 | switch (OP(c)) { |
a0ed51b3 |
886 | case ANYOFUTF8: |
cd439c50 |
887 | cc = MASK(c); |
a0ed51b3 |
888 | while (s < strend) { |
889 | if (REGINCLASSUTF8(c, (U8*)s)) { |
890 | if (tmp && regtry(prog, s)) |
891 | goto got_it; |
892 | else |
893 | tmp = doevery; |
894 | } |
895 | else |
896 | tmp = 1; |
897 | s += UTF8SKIP(s); |
898 | } |
899 | break; |
a0d0e21e |
900 | case ANYOF: |
cd439c50 |
901 | cc = MASK(c); |
a0d0e21e |
902 | while (s < strend) { |
a0ed51b3 |
903 | if (REGINCLASS(cc, *s)) { |
a0d0e21e |
904 | if (tmp && regtry(prog, s)) |
905 | goto got_it; |
906 | else |
907 | tmp = doevery; |
a687059c |
908 | } |
a0d0e21e |
909 | else |
910 | tmp = 1; |
911 | s++; |
912 | } |
913 | break; |
bbce6d69 |
914 | case BOUNDL: |
3280af22 |
915 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
916 | /* FALL THROUGH */ |
a0d0e21e |
917 | case BOUND: |
a0ed51b3 |
918 | if (minlen) { |
919 | dontbother++; |
920 | strend -= 1; |
921 | } |
3280af22 |
922 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
923 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
924 | while (s < strend) { |
95bac841 |
925 | if (tmp == !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { |
a0d0e21e |
926 | tmp = !tmp; |
927 | if (regtry(prog, s)) |
928 | goto got_it; |
a687059c |
929 | } |
a0d0e21e |
930 | s++; |
931 | } |
932 | if ((minlen || tmp) && regtry(prog,s)) |
933 | goto got_it; |
934 | break; |
a0ed51b3 |
935 | case BOUNDLUTF8: |
936 | PL_reg_flags |= RF_tainted; |
937 | /* FALL THROUGH */ |
938 | case BOUNDUTF8: |
939 | if (minlen) { |
940 | dontbother++; |
dfe13c55 |
941 | strend = reghop_c(strend, -1); |
a0ed51b3 |
942 | } |
dfe13c55 |
943 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
944 | tmp = ((OP(c) == BOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
945 | while (s < strend) { |
dfe13c55 |
946 | if (tmp == !(OP(c) == BOUND ? |
947 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
948 | isALNUM_LC_utf8((U8*)s))) |
949 | { |
a0ed51b3 |
950 | tmp = !tmp; |
951 | if (regtry(prog, s)) |
952 | goto got_it; |
953 | } |
954 | s += UTF8SKIP(s); |
955 | } |
956 | if ((minlen || tmp) && regtry(prog,s)) |
957 | goto got_it; |
958 | break; |
bbce6d69 |
959 | case NBOUNDL: |
3280af22 |
960 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
961 | /* FALL THROUGH */ |
a0d0e21e |
962 | case NBOUND: |
a0ed51b3 |
963 | if (minlen) { |
964 | dontbother++; |
965 | strend -= 1; |
966 | } |
3280af22 |
967 | tmp = (s != startpos) ? UCHARAT(s - 1) : PL_regprev; |
95bac841 |
968 | tmp = ((OP(c) == NBOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
a0d0e21e |
969 | while (s < strend) { |
95bac841 |
970 | if (tmp == !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) |
a0d0e21e |
971 | tmp = !tmp; |
972 | else if (regtry(prog, s)) |
973 | goto got_it; |
974 | s++; |
975 | } |
976 | if ((minlen || !tmp) && regtry(prog,s)) |
977 | goto got_it; |
978 | break; |
a0ed51b3 |
979 | case NBOUNDLUTF8: |
980 | PL_reg_flags |= RF_tainted; |
981 | /* FALL THROUGH */ |
982 | case NBOUNDUTF8: |
983 | if (minlen) { |
984 | dontbother++; |
dfe13c55 |
985 | strend = reghop_c(strend, -1); |
a0ed51b3 |
986 | } |
dfe13c55 |
987 | tmp = (I32)(s != startpos) ? utf8_to_uv(reghop((U8*)s, -1), 0) : PL_regprev; |
a0ed51b3 |
988 | tmp = ((OP(c) == NBOUND ? isALNUM_uni(tmp) : isALNUM_LC_uni(tmp)) != 0); |
989 | while (s < strend) { |
dfe13c55 |
990 | if (tmp == !(OP(c) == NBOUND ? |
991 | swash_fetch(PL_utf8_alnum, (U8*)s) : |
992 | isALNUM_LC_utf8((U8*)s))) |
a0ed51b3 |
993 | tmp = !tmp; |
994 | else if (regtry(prog, s)) |
995 | goto got_it; |
996 | s += UTF8SKIP(s); |
997 | } |
998 | if ((minlen || !tmp) && regtry(prog,s)) |
999 | goto got_it; |
1000 | break; |
a0d0e21e |
1001 | case ALNUM: |
1002 | while (s < strend) { |
bbce6d69 |
1003 | if (isALNUM(*s)) { |
1004 | if (tmp && regtry(prog, s)) |
1005 | goto got_it; |
1006 | else |
1007 | tmp = doevery; |
1008 | } |
1009 | else |
1010 | tmp = 1; |
1011 | s++; |
1012 | } |
1013 | break; |
a0ed51b3 |
1014 | case ALNUMUTF8: |
1015 | while (s < strend) { |
dfe13c55 |
1016 | if (swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
1017 | if (tmp && regtry(prog, s)) |
1018 | goto got_it; |
1019 | else |
1020 | tmp = doevery; |
1021 | } |
1022 | else |
1023 | tmp = 1; |
1024 | s += UTF8SKIP(s); |
1025 | } |
1026 | break; |
bbce6d69 |
1027 | case ALNUML: |
3280af22 |
1028 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1029 | while (s < strend) { |
1030 | if (isALNUM_LC(*s)) { |
a0d0e21e |
1031 | if (tmp && regtry(prog, s)) |
1032 | goto got_it; |
a687059c |
1033 | else |
a0d0e21e |
1034 | tmp = doevery; |
1035 | } |
1036 | else |
1037 | tmp = 1; |
1038 | s++; |
1039 | } |
1040 | break; |
a0ed51b3 |
1041 | case ALNUMLUTF8: |
1042 | PL_reg_flags |= RF_tainted; |
1043 | while (s < strend) { |
dfe13c55 |
1044 | if (isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
1045 | if (tmp && regtry(prog, s)) |
1046 | goto got_it; |
1047 | else |
1048 | tmp = doevery; |
1049 | } |
1050 | else |
1051 | tmp = 1; |
1052 | s += UTF8SKIP(s); |
1053 | } |
1054 | break; |
a0d0e21e |
1055 | case NALNUM: |
1056 | while (s < strend) { |
bbce6d69 |
1057 | if (!isALNUM(*s)) { |
1058 | if (tmp && regtry(prog, s)) |
1059 | goto got_it; |
1060 | else |
1061 | tmp = doevery; |
1062 | } |
1063 | else |
1064 | tmp = 1; |
1065 | s++; |
1066 | } |
1067 | break; |
a0ed51b3 |
1068 | case NALNUMUTF8: |
1069 | while (s < strend) { |
dfe13c55 |
1070 | if (!swash_fetch(PL_utf8_alnum, (U8*)s)) { |
a0ed51b3 |
1071 | if (tmp && regtry(prog, s)) |
1072 | goto got_it; |
1073 | else |
1074 | tmp = doevery; |
1075 | } |
1076 | else |
1077 | tmp = 1; |
1078 | s += UTF8SKIP(s); |
1079 | } |
1080 | break; |
bbce6d69 |
1081 | case NALNUML: |
3280af22 |
1082 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1083 | while (s < strend) { |
1084 | if (!isALNUM_LC(*s)) { |
a0d0e21e |
1085 | if (tmp && regtry(prog, s)) |
1086 | goto got_it; |
a687059c |
1087 | else |
a0d0e21e |
1088 | tmp = doevery; |
a687059c |
1089 | } |
a0d0e21e |
1090 | else |
1091 | tmp = 1; |
1092 | s++; |
1093 | } |
1094 | break; |
a0ed51b3 |
1095 | case NALNUMLUTF8: |
1096 | PL_reg_flags |= RF_tainted; |
1097 | while (s < strend) { |
dfe13c55 |
1098 | if (!isALNUM_LC_utf8((U8*)s)) { |
a0ed51b3 |
1099 | if (tmp && regtry(prog, s)) |
1100 | goto got_it; |
1101 | else |
1102 | tmp = doevery; |
1103 | } |
1104 | else |
1105 | tmp = 1; |
1106 | s += UTF8SKIP(s); |
1107 | } |
1108 | break; |
a0d0e21e |
1109 | case SPACE: |
1110 | while (s < strend) { |
1111 | if (isSPACE(*s)) { |
1112 | if (tmp && regtry(prog, s)) |
1113 | goto got_it; |
1114 | else |
1115 | tmp = doevery; |
2304df62 |
1116 | } |
a0d0e21e |
1117 | else |
1118 | tmp = 1; |
1119 | s++; |
1120 | } |
1121 | break; |
a0ed51b3 |
1122 | case SPACEUTF8: |
1123 | while (s < strend) { |
dfe13c55 |
1124 | if (*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s)) { |
a0ed51b3 |
1125 | if (tmp && regtry(prog, s)) |
1126 | goto got_it; |
1127 | else |
1128 | tmp = doevery; |
1129 | } |
1130 | else |
1131 | tmp = 1; |
1132 | s += UTF8SKIP(s); |
1133 | } |
1134 | break; |
bbce6d69 |
1135 | case SPACEL: |
3280af22 |
1136 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1137 | while (s < strend) { |
1138 | if (isSPACE_LC(*s)) { |
1139 | if (tmp && regtry(prog, s)) |
1140 | goto got_it; |
1141 | else |
1142 | tmp = doevery; |
1143 | } |
1144 | else |
1145 | tmp = 1; |
1146 | s++; |
1147 | } |
1148 | break; |
a0ed51b3 |
1149 | case SPACELUTF8: |
1150 | PL_reg_flags |= RF_tainted; |
1151 | while (s < strend) { |
dfe13c55 |
1152 | if (*s == ' ' || isSPACE_LC_utf8((U8*)s)) { |
a0ed51b3 |
1153 | if (tmp && regtry(prog, s)) |
1154 | goto got_it; |
1155 | else |
1156 | tmp = doevery; |
1157 | } |
1158 | else |
1159 | tmp = 1; |
1160 | s += UTF8SKIP(s); |
1161 | } |
1162 | break; |
a0d0e21e |
1163 | case NSPACE: |
1164 | while (s < strend) { |
1165 | if (!isSPACE(*s)) { |
1166 | if (tmp && regtry(prog, s)) |
1167 | goto got_it; |
1168 | else |
1169 | tmp = doevery; |
a687059c |
1170 | } |
a0d0e21e |
1171 | else |
1172 | tmp = 1; |
1173 | s++; |
1174 | } |
1175 | break; |
a0ed51b3 |
1176 | case NSPACEUTF8: |
1177 | while (s < strend) { |
dfe13c55 |
1178 | if (!(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s))) { |
a0ed51b3 |
1179 | if (tmp && regtry(prog, s)) |
1180 | goto got_it; |
1181 | else |
1182 | tmp = doevery; |
1183 | } |
1184 | else |
1185 | tmp = 1; |
1186 | s += UTF8SKIP(s); |
1187 | } |
1188 | break; |
bbce6d69 |
1189 | case NSPACEL: |
3280af22 |
1190 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1191 | while (s < strend) { |
1192 | if (!isSPACE_LC(*s)) { |
1193 | if (tmp && regtry(prog, s)) |
1194 | goto got_it; |
1195 | else |
1196 | tmp = doevery; |
1197 | } |
1198 | else |
1199 | tmp = 1; |
1200 | s++; |
1201 | } |
1202 | break; |
a0ed51b3 |
1203 | case NSPACELUTF8: |
1204 | PL_reg_flags |= RF_tainted; |
1205 | while (s < strend) { |
dfe13c55 |
1206 | if (!(*s == ' ' || isSPACE_LC_utf8((U8*)s))) { |
a0ed51b3 |
1207 | if (tmp && regtry(prog, s)) |
1208 | goto got_it; |
1209 | else |
1210 | tmp = doevery; |
1211 | } |
1212 | else |
1213 | tmp = 1; |
1214 | s += UTF8SKIP(s); |
1215 | } |
1216 | break; |
a0d0e21e |
1217 | case DIGIT: |
1218 | while (s < strend) { |
1219 | if (isDIGIT(*s)) { |
1220 | if (tmp && regtry(prog, s)) |
1221 | goto got_it; |
1222 | else |
1223 | tmp = doevery; |
2b69d0c2 |
1224 | } |
a0d0e21e |
1225 | else |
1226 | tmp = 1; |
1227 | s++; |
1228 | } |
1229 | break; |
a0ed51b3 |
1230 | case DIGITUTF8: |
1231 | while (s < strend) { |
dfe13c55 |
1232 | if (swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
1233 | if (tmp && regtry(prog, s)) |
1234 | goto got_it; |
1235 | else |
1236 | tmp = doevery; |
1237 | } |
1238 | else |
1239 | tmp = 1; |
1240 | s += UTF8SKIP(s); |
1241 | } |
1242 | break; |
b8c5462f |
1243 | case DIGITL: |
1244 | PL_reg_flags |= RF_tainted; |
1245 | while (s < strend) { |
1246 | if (isDIGIT_LC(*s)) { |
1247 | if (tmp && regtry(prog, s)) |
1248 | goto got_it; |
1249 | else |
1250 | tmp = doevery; |
1251 | } |
1252 | else |
1253 | tmp = 1; |
1254 | s++; |
1255 | } |
1256 | break; |
1257 | case DIGITLUTF8: |
1258 | PL_reg_flags |= RF_tainted; |
1259 | while (s < strend) { |
1260 | if (isDIGIT_LC_utf8((U8*)s)) { |
1261 | if (tmp && regtry(prog, s)) |
1262 | goto got_it; |
1263 | else |
1264 | tmp = doevery; |
1265 | } |
1266 | else |
1267 | tmp = 1; |
1268 | s += UTF8SKIP(s); |
1269 | } |
1270 | break; |
a0d0e21e |
1271 | case NDIGIT: |
1272 | while (s < strend) { |
1273 | if (!isDIGIT(*s)) { |
1274 | if (tmp && regtry(prog, s)) |
1275 | goto got_it; |
1276 | else |
1277 | tmp = doevery; |
a687059c |
1278 | } |
a0d0e21e |
1279 | else |
1280 | tmp = 1; |
1281 | s++; |
1282 | } |
1283 | break; |
a0ed51b3 |
1284 | case NDIGITUTF8: |
1285 | while (s < strend) { |
dfe13c55 |
1286 | if (!swash_fetch(PL_utf8_digit,(U8*)s)) { |
a0ed51b3 |
1287 | if (tmp && regtry(prog, s)) |
1288 | goto got_it; |
1289 | else |
1290 | tmp = doevery; |
1291 | } |
1292 | else |
1293 | tmp = 1; |
1294 | s += UTF8SKIP(s); |
1295 | } |
1296 | break; |
b8c5462f |
1297 | case NDIGITL: |
1298 | PL_reg_flags |= RF_tainted; |
1299 | while (s < strend) { |
1300 | if (!isDIGIT_LC(*s)) { |
1301 | if (tmp && regtry(prog, s)) |
1302 | goto got_it; |
1303 | else |
1304 | tmp = doevery; |
1305 | } |
1306 | else |
1307 | tmp = 1; |
1308 | s++; |
a0ed51b3 |
1309 | } |
b8c5462f |
1310 | break; |
1311 | case NDIGITLUTF8: |
1312 | PL_reg_flags |= RF_tainted; |
1313 | while (s < strend) { |
1314 | if (!isDIGIT_LC_utf8((U8*)s)) { |
1315 | if (tmp && regtry(prog, s)) |
1316 | goto got_it; |
cf93c79d |
1317 | else |
b8c5462f |
1318 | tmp = doevery; |
1319 | } |
1320 | else |
1321 | tmp = 1; |
1322 | s += UTF8SKIP(s); |
1323 | } |
1324 | break; |
d6a28714 |
1325 | } |
1326 | } |
1327 | else { |
1328 | dontbother = 0; |
1329 | if (prog->float_substr != Nullsv) { /* Trim the end. */ |
1330 | char *last; |
1331 | I32 oldpos = scream_pos; |
1332 | |
1333 | if (flags & REXEC_SCREAM) { |
1334 | last = screaminstr(sv, prog->float_substr, s - strbeg, |
1335 | end_shift, &scream_pos, 1); /* last one */ |
1336 | if (!last) |
1337 | last = scream_olds; /* Only one occurence. */ |
b8c5462f |
1338 | } |
d6a28714 |
1339 | else { |
1340 | STRLEN len; |
1341 | char *little = SvPV(prog->float_substr, len); |
1342 | |
1343 | if (SvTAIL(prog->float_substr)) { |
1344 | if (memEQ(strend - len + 1, little, len - 1)) |
1345 | last = strend - len + 1; |
1346 | else if (!PL_multiline) |
1347 | last = memEQ(strend - len, little, len) |
1348 | ? strend - len : Nullch; |
b8c5462f |
1349 | else |
d6a28714 |
1350 | goto find_last; |
1351 | } else { |
1352 | find_last: |
1353 | if (len) |
1354 | last = rninstr(s, strend, little, little + len); |
b8c5462f |
1355 | else |
d6a28714 |
1356 | last = strend; /* matching `$' */ |
b8c5462f |
1357 | } |
b8c5462f |
1358 | } |
d6a28714 |
1359 | if (last == NULL) goto phooey; /* Should not happen! */ |
1360 | dontbother = strend - last + prog->float_min_offset; |
1361 | } |
1362 | if (minlen && (dontbother < minlen)) |
1363 | dontbother = minlen - 1; |
1364 | strend -= dontbother; /* this one's always in bytes! */ |
1365 | /* We don't know much -- general case. */ |
1366 | if (UTF) { |
1367 | for (;;) { |
1368 | if (regtry(prog, s)) |
1369 | goto got_it; |
1370 | if (s >= strend) |
1371 | break; |
b8c5462f |
1372 | s += UTF8SKIP(s); |
d6a28714 |
1373 | }; |
1374 | } |
1375 | else { |
1376 | do { |
1377 | if (regtry(prog, s)) |
1378 | goto got_it; |
1379 | } while (s++ < strend); |
1380 | } |
1381 | } |
1382 | |
1383 | /* Failure. */ |
1384 | goto phooey; |
1385 | |
1386 | got_it: |
1387 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); |
1388 | |
1389 | if (PL_reg_eval_set) { |
1390 | /* Preserve the current value of $^R */ |
1391 | if (oreplsv != GvSV(PL_replgv)) |
1392 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is |
1393 | restored, the value remains |
1394 | the same. */ |
1395 | restore_pos(aTHXo_ 0); |
1396 | } |
1397 | |
1398 | /* make sure $`, $&, $', and $digit will work later */ |
1399 | if ( !(flags & REXEC_NOT_FIRST) ) { |
1400 | if (RX_MATCH_COPIED(prog)) { |
1401 | Safefree(prog->subbeg); |
1402 | RX_MATCH_COPIED_off(prog); |
1403 | } |
1404 | if (flags & REXEC_COPY_STR) { |
1405 | I32 i = PL_regeol - startpos + (stringarg - strbeg); |
1406 | |
1407 | s = savepvn(strbeg, i); |
1408 | prog->subbeg = s; |
1409 | prog->sublen = i; |
1410 | RX_MATCH_COPIED_on(prog); |
1411 | } |
1412 | else { |
1413 | prog->subbeg = strbeg; |
1414 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ |
1415 | } |
1416 | } |
1417 | |
1418 | return 1; |
1419 | |
1420 | phooey: |
1421 | if (PL_reg_eval_set) |
1422 | restore_pos(aTHXo_ 0); |
1423 | return 0; |
1424 | } |
1425 | |
1426 | /* |
1427 | - regtry - try match at specific point |
1428 | */ |
1429 | STATIC I32 /* 0 failure, 1 success */ |
1430 | S_regtry(pTHX_ regexp *prog, char *startpos) |
1431 | { |
1432 | dTHR; |
1433 | register I32 i; |
1434 | register I32 *sp; |
1435 | register I32 *ep; |
1436 | CHECKPOINT lastcp; |
1437 | |
1438 | if ((prog->reganch & ROPT_EVAL_SEEN) && !PL_reg_eval_set) { |
1439 | MAGIC *mg; |
1440 | |
1441 | PL_reg_eval_set = RS_init; |
1442 | DEBUG_r(DEBUG_s( |
1443 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %i\n", |
1444 | PL_stack_sp - PL_stack_base); |
1445 | )); |
1446 | SAVEINT(cxstack[cxstack_ix].blk_oldsp); |
1447 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
1448 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ |
1449 | SAVETMPS; |
1450 | /* Apparently this is not needed, judging by wantarray. */ |
1451 | /* SAVEINT(cxstack[cxstack_ix].blk_gimme); |
1452 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
1453 | |
1454 | if (PL_reg_sv) { |
1455 | /* Make $_ available to executed code. */ |
1456 | if (PL_reg_sv != DEFSV) { |
1457 | /* SAVE_DEFSV does *not* suffice here for USE_THREADS */ |
1458 | SAVESPTR(DEFSV); |
1459 | DEFSV = PL_reg_sv; |
b8c5462f |
1460 | } |
d6a28714 |
1461 | |
1462 | if (!(SvTYPE(PL_reg_sv) >= SVt_PVMG && SvMAGIC(PL_reg_sv) |
1463 | && (mg = mg_find(PL_reg_sv, 'g')))) { |
1464 | /* prepare for quick setting of pos */ |
1465 | sv_magic(PL_reg_sv, (SV*)0, 'g', Nullch, 0); |
1466 | mg = mg_find(PL_reg_sv, 'g'); |
1467 | mg->mg_len = -1; |
b8c5462f |
1468 | } |
d6a28714 |
1469 | PL_reg_magic = mg; |
1470 | PL_reg_oldpos = mg->mg_len; |
1471 | SAVEDESTRUCTOR(restore_pos, 0); |
1472 | } |
1473 | if (!PL_reg_curpm) |
1474 | New(22,PL_reg_curpm, 1, PMOP); |
1475 | PL_reg_curpm->op_pmregexp = prog; |
1476 | PL_reg_oldcurpm = PL_curpm; |
1477 | PL_curpm = PL_reg_curpm; |
1478 | if (RX_MATCH_COPIED(prog)) { |
1479 | /* Here is a serious problem: we cannot rewrite subbeg, |
1480 | since it may be needed if this match fails. Thus |
1481 | $` inside (?{}) could fail... */ |
1482 | PL_reg_oldsaved = prog->subbeg; |
1483 | PL_reg_oldsavedlen = prog->sublen; |
1484 | RX_MATCH_COPIED_off(prog); |
1485 | } |
1486 | else |
1487 | PL_reg_oldsaved = Nullch; |
1488 | prog->subbeg = PL_bostr; |
1489 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ |
1490 | } |
1491 | prog->startp[0] = startpos - PL_bostr; |
1492 | PL_reginput = startpos; |
1493 | PL_regstartp = prog->startp; |
1494 | PL_regendp = prog->endp; |
1495 | PL_reglastparen = &prog->lastparen; |
1496 | prog->lastparen = 0; |
1497 | PL_regsize = 0; |
1498 | DEBUG_r(PL_reg_starttry = startpos); |
1499 | if (PL_reg_start_tmpl <= prog->nparens) { |
1500 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; |
1501 | if(PL_reg_start_tmp) |
1502 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
1503 | else |
1504 | New(22,PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
1505 | } |
1506 | |
1507 | /* XXXX What this code is doing here?!!! There should be no need |
1508 | to do this again and again, PL_reglastparen should take care of |
1509 | this! */ |
1510 | sp = prog->startp; |
1511 | ep = prog->endp; |
1512 | if (prog->nparens) { |
1513 | for (i = prog->nparens; i >= 1; i--) { |
1514 | *++sp = -1; |
1515 | *++ep = -1; |
1516 | } |
1517 | } |
1518 | REGCP_SET; |
1519 | if (regmatch(prog->program + 1)) { |
1520 | prog->endp[0] = PL_reginput - PL_bostr; |
1521 | return 1; |
1522 | } |
1523 | REGCP_UNWIND; |
1524 | return 0; |
1525 | } |
1526 | |
1527 | /* |
1528 | - regmatch - main matching routine |
1529 | * |
1530 | * Conceptually the strategy is simple: check to see whether the current |
1531 | * node matches, call self recursively to see whether the rest matches, |
1532 | * and then act accordingly. In practice we make some effort to avoid |
1533 | * recursion, in particular by going through "ordinary" nodes (that don't |
1534 | * need to know whether the rest of the match failed) by a loop instead of |
1535 | * by recursion. |
1536 | */ |
1537 | /* [lwall] I've hoisted the register declarations to the outer block in order to |
1538 | * maybe save a little bit of pushing and popping on the stack. It also takes |
1539 | * advantage of machines that use a register save mask on subroutine entry. |
1540 | */ |
1541 | STATIC I32 /* 0 failure, 1 success */ |
1542 | S_regmatch(pTHX_ regnode *prog) |
1543 | { |
1544 | dTHR; |
1545 | register regnode *scan; /* Current node. */ |
1546 | regnode *next; /* Next node. */ |
1547 | regnode *inner; /* Next node in internal branch. */ |
1548 | register I32 nextchr; /* renamed nextchr - nextchar colides with |
1549 | function of same name */ |
1550 | register I32 n; /* no or next */ |
1551 | register I32 ln; /* len or last */ |
1552 | register char *s; /* operand or save */ |
1553 | register char *locinput = PL_reginput; |
1554 | register I32 c1, c2, paren; /* case fold search, parenth */ |
1555 | int minmod = 0, sw = 0, logical = 0; |
1556 | #ifdef DEBUGGING |
1557 | PL_regindent++; |
1558 | #endif |
1559 | |
1560 | /* Note that nextchr is a byte even in UTF */ |
1561 | nextchr = UCHARAT(locinput); |
1562 | scan = prog; |
1563 | while (scan != NULL) { |
1564 | #define sayNO_L (logical ? (logical = 0, sw = 0, goto cont) : sayNO) |
1565 | #ifdef DEBUGGING |
1566 | # define sayYES goto yes |
1567 | # define sayNO goto no |
1568 | # define saySAME(x) if (x) goto yes; else goto no |
1569 | # define REPORT_CODE_OFF 24 |
1570 | #else |
1571 | # define sayYES return 1 |
1572 | # define sayNO return 0 |
1573 | # define saySAME(x) return x |
1574 | #endif |
1575 | DEBUG_r( { |
1576 | SV *prop = sv_newmortal(); |
1577 | int docolor = *PL_colors[0]; |
1578 | int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
1579 | int l = (PL_regeol - locinput > taill ? taill : PL_regeol - locinput); |
1580 | /* The part of the string before starttry has one color |
1581 | (pref0_len chars), between starttry and current |
1582 | position another one (pref_len - pref0_len chars), |
1583 | after the current position the third one. |
1584 | We assume that pref0_len <= pref_len, otherwise we |
1585 | decrease pref0_len. */ |
1586 | int pref_len = (locinput - PL_bostr > (5 + taill) - l |
1587 | ? (5 + taill) - l : locinput - PL_bostr); |
1588 | int pref0_len = pref_len - (locinput - PL_reg_starttry); |
1589 | |
1590 | if (l + pref_len < (5 + taill) && l < PL_regeol - locinput) |
1591 | l = ( PL_regeol - locinput > (5 + taill) - pref_len |
1592 | ? (5 + taill) - pref_len : PL_regeol - locinput); |
1593 | if (pref0_len < 0) |
1594 | pref0_len = 0; |
1595 | if (pref0_len > pref_len) |
1596 | pref0_len = pref_len; |
1597 | regprop(prop, scan); |
1598 | PerlIO_printf(Perl_debug_log, |
1599 | "%4i <%s%.*s%s%s%.*s%s%s%s%.*s%s>%*s|%3d:%*s%s\n", |
1600 | locinput - PL_bostr, |
1601 | PL_colors[4], pref0_len, |
1602 | locinput - pref_len, PL_colors[5], |
1603 | PL_colors[2], pref_len - pref0_len, |
1604 | locinput - pref_len + pref0_len, PL_colors[3], |
1605 | (docolor ? "" : "> <"), |
1606 | PL_colors[0], l, locinput, PL_colors[1], |
1607 | 15 - l - pref_len + 1, |
1608 | "", |
1609 | scan - PL_regprogram, PL_regindent*2, "", |
1610 | SvPVX(prop)); |
1611 | } ); |
1612 | |
1613 | next = scan + NEXT_OFF(scan); |
1614 | if (next == scan) |
1615 | next = NULL; |
1616 | |
1617 | switch (OP(scan)) { |
1618 | case BOL: |
1619 | if (locinput == PL_bostr |
1620 | ? PL_regprev == '\n' |
1621 | : (PL_multiline && |
1622 | (nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
1623 | { |
1624 | /* regtill = regbol; */ |
b8c5462f |
1625 | break; |
1626 | } |
d6a28714 |
1627 | sayNO; |
1628 | case MBOL: |
1629 | if (locinput == PL_bostr |
1630 | ? PL_regprev == '\n' |
1631 | : ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n') ) |
1632 | { |
b8c5462f |
1633 | break; |
1634 | } |
d6a28714 |
1635 | sayNO; |
1636 | case SBOL: |
1637 | if (locinput == PL_regbol && PL_regprev == '\n') |
b8c5462f |
1638 | break; |
d6a28714 |
1639 | sayNO; |
1640 | case GPOS: |
1641 | if (locinput == PL_reg_ganch) |
1642 | break; |
1643 | sayNO; |
1644 | case EOL: |
1645 | if (PL_multiline) |
1646 | goto meol; |
1647 | else |
1648 | goto seol; |
1649 | case MEOL: |
1650 | meol: |
1651 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f |
1652 | sayNO; |
b8c5462f |
1653 | break; |
d6a28714 |
1654 | case SEOL: |
1655 | seol: |
1656 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f |
1657 | sayNO; |
d6a28714 |
1658 | if (PL_regeol - locinput > 1) |
b8c5462f |
1659 | sayNO; |
b8c5462f |
1660 | break; |
d6a28714 |
1661 | case EOS: |
1662 | if (PL_regeol != locinput) |
b8c5462f |
1663 | sayNO; |
d6a28714 |
1664 | break; |
1665 | case SANYUTF8: |
b8c5462f |
1666 | if (nextchr & 0x80) { |
b8c5462f |
1667 | locinput += PL_utf8skip[nextchr]; |
d6a28714 |
1668 | if (locinput > PL_regeol) |
1669 | sayNO; |
b8c5462f |
1670 | nextchr = UCHARAT(locinput); |
1671 | break; |
1672 | } |
d6a28714 |
1673 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1674 | sayNO; |
b8c5462f |
1675 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1676 | break; |
d6a28714 |
1677 | case SANY: |
1678 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f |
1679 | sayNO; |
1680 | nextchr = UCHARAT(++locinput); |
b85d18e9 |
1681 | break; |
d6a28714 |
1682 | case ANYUTF8: |
a0ed51b3 |
1683 | if (nextchr & 0x80) { |
b8c5462f |
1684 | locinput += PL_utf8skip[nextchr]; |
d6a28714 |
1685 | if (locinput > PL_regeol) |
1686 | sayNO; |
a0ed51b3 |
1687 | nextchr = UCHARAT(locinput); |
1688 | break; |
1689 | } |
d6a28714 |
1690 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
a0ed51b3 |
1691 | sayNO; |
1692 | nextchr = UCHARAT(++locinput); |
1693 | break; |
d6a28714 |
1694 | case REG_ANY: |
1695 | if (!nextchr && locinput >= PL_regeol || nextchr == '\n') |
4633a7c4 |
1696 | sayNO; |
76e3520e |
1697 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1698 | break; |
d6a28714 |
1699 | case EXACT: |
cd439c50 |
1700 | s = STRING(scan); |
1701 | ln = STR_LEN(scan); |
d6a28714 |
1702 | /* Inline the first character, for speed. */ |
1703 | if (UCHARAT(s) != nextchr) |
1704 | sayNO; |
1705 | if (PL_regeol - locinput < ln) |
1706 | sayNO; |
1707 | if (ln > 1 && memNE(s, locinput, ln)) |
1708 | sayNO; |
1709 | locinput += ln; |
1710 | nextchr = UCHARAT(locinput); |
1711 | break; |
1712 | case EXACTFL: |
b8c5462f |
1713 | PL_reg_flags |= RF_tainted; |
1714 | /* FALL THROUGH */ |
d6a28714 |
1715 | case EXACTF: |
cd439c50 |
1716 | s = STRING(scan); |
1717 | ln = STR_LEN(scan); |
d6a28714 |
1718 | |
1719 | if (UTF) { |
1720 | char *l = locinput; |
1721 | char *e = s + ln; |
1722 | c1 = OP(scan) == EXACTF; |
1723 | while (s < e) { |
1724 | if (l >= PL_regeol) |
1725 | sayNO; |
1726 | if (utf8_to_uv((U8*)s, 0) != (c1 ? |
1727 | toLOWER_utf8((U8*)l) : |
1728 | toLOWER_LC_utf8((U8*)l))) |
1729 | { |
1730 | sayNO; |
1731 | } |
1732 | s += UTF8SKIP(s); |
1733 | l += UTF8SKIP(l); |
b8c5462f |
1734 | } |
d6a28714 |
1735 | locinput = l; |
a0ed51b3 |
1736 | nextchr = UCHARAT(locinput); |
1737 | break; |
1738 | } |
d6a28714 |
1739 | |
1740 | /* Inline the first character, for speed. */ |
1741 | if (UCHARAT(s) != nextchr && |
1742 | UCHARAT(s) != ((OP(scan) == EXACTF) |
1743 | ? PL_fold : PL_fold_locale)[nextchr]) |
a0ed51b3 |
1744 | sayNO; |
d6a28714 |
1745 | if (PL_regeol - locinput < ln) |
b8c5462f |
1746 | sayNO; |
d6a28714 |
1747 | if (ln > 1 && (OP(scan) == EXACTF |
1748 | ? ibcmp(s, locinput, ln) |
1749 | : ibcmp_locale(s, locinput, ln))) |
4633a7c4 |
1750 | sayNO; |
d6a28714 |
1751 | locinput += ln; |
1752 | nextchr = UCHARAT(locinput); |
a0d0e21e |
1753 | break; |
d6a28714 |
1754 | case ANYOFUTF8: |
cd439c50 |
1755 | s = MASK(scan); |
d6a28714 |
1756 | if (!REGINCLASSUTF8(scan, (U8*)locinput)) |
4633a7c4 |
1757 | sayNO; |
d6a28714 |
1758 | if (locinput >= PL_regeol) |
1759 | sayNO; |
1760 | locinput += PL_utf8skip[nextchr]; |
1761 | nextchr = UCHARAT(locinput); |
1762 | break; |
1763 | case ANYOF: |
cd439c50 |
1764 | s = MASK(scan); |
d6a28714 |
1765 | if (nextchr < 0) |
b8c5462f |
1766 | nextchr = UCHARAT(locinput); |
d6a28714 |
1767 | if (!REGINCLASS(s, nextchr)) |
1768 | sayNO; |
1769 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1770 | sayNO; |
b8c5462f |
1771 | nextchr = UCHARAT(++locinput); |
1772 | break; |
d6a28714 |
1773 | case ALNUML: |
b8c5462f |
1774 | PL_reg_flags |= RF_tainted; |
1775 | /* FALL THROUGH */ |
d6a28714 |
1776 | case ALNUM: |
b8c5462f |
1777 | if (!nextchr) |
4633a7c4 |
1778 | sayNO; |
d6a28714 |
1779 | if (!(OP(scan) == ALNUM |
1780 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
b8c5462f |
1781 | sayNO; |
1782 | nextchr = UCHARAT(++locinput); |
bbce6d69 |
1783 | break; |
d6a28714 |
1784 | case ALNUMLUTF8: |
3280af22 |
1785 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1786 | /* FALL THROUGH */ |
d6a28714 |
1787 | case ALNUMUTF8: |
b8c5462f |
1788 | if (!nextchr) |
1789 | sayNO; |
1790 | if (nextchr & 0x80) { |
d6a28714 |
1791 | if (!(OP(scan) == ALNUMUTF8 |
1792 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1793 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f |
1794 | { |
1795 | sayNO; |
a0ed51b3 |
1796 | } |
b8c5462f |
1797 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1798 | nextchr = UCHARAT(locinput); |
1799 | break; |
1800 | } |
d6a28714 |
1801 | if (!(OP(scan) == ALNUMUTF8 |
1802 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 |
1803 | sayNO; |
b8c5462f |
1804 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1805 | break; |
d6a28714 |
1806 | case NALNUML: |
b8c5462f |
1807 | PL_reg_flags |= RF_tainted; |
1808 | /* FALL THROUGH */ |
d6a28714 |
1809 | case NALNUM: |
1810 | if (!nextchr && locinput >= PL_regeol) |
a0ed51b3 |
1811 | sayNO; |
d6a28714 |
1812 | if (OP(scan) == NALNUM |
1813 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
a0ed51b3 |
1814 | sayNO; |
b8c5462f |
1815 | nextchr = UCHARAT(++locinput); |
a0ed51b3 |
1816 | break; |
d6a28714 |
1817 | case NALNUMLUTF8: |
b8c5462f |
1818 | PL_reg_flags |= RF_tainted; |
1819 | /* FALL THROUGH */ |
d6a28714 |
1820 | case NALNUMUTF8: |
3280af22 |
1821 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1822 | sayNO; |
b8c5462f |
1823 | if (nextchr & 0x80) { |
d6a28714 |
1824 | if (OP(scan) == NALNUMUTF8 |
1825 | ? swash_fetch(PL_utf8_alnum, (U8*)locinput) |
1826 | : isALNUM_LC_utf8((U8*)locinput)) |
1827 | { |
b8c5462f |
1828 | sayNO; |
d6a28714 |
1829 | } |
b8c5462f |
1830 | locinput += PL_utf8skip[nextchr]; |
1831 | nextchr = UCHARAT(locinput); |
1832 | break; |
1833 | } |
d6a28714 |
1834 | if (OP(scan) == NALNUMUTF8 |
1835 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f |
1836 | sayNO; |
76e3520e |
1837 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1838 | break; |
d6a28714 |
1839 | case BOUNDL: |
1840 | case NBOUNDL: |
3280af22 |
1841 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1842 | /* FALL THROUGH */ |
d6a28714 |
1843 | case BOUND: |
1844 | case NBOUND: |
1845 | /* was last char in word? */ |
1846 | ln = (locinput != PL_regbol) ? UCHARAT(locinput - 1) : PL_regprev; |
1847 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
1848 | ln = isALNUM(ln); |
1849 | n = isALNUM(nextchr); |
1850 | } |
1851 | else { |
1852 | ln = isALNUM_LC(ln); |
1853 | n = isALNUM_LC(nextchr); |
1854 | } |
1855 | if (((!ln) == (!n)) == (OP(scan) == BOUND || OP(scan) == BOUNDL)) |
4633a7c4 |
1856 | sayNO; |
a0d0e21e |
1857 | break; |
d6a28714 |
1858 | case BOUNDLUTF8: |
1859 | case NBOUNDLUTF8: |
a0ed51b3 |
1860 | PL_reg_flags |= RF_tainted; |
1861 | /* FALL THROUGH */ |
d6a28714 |
1862 | case BOUNDUTF8: |
1863 | case NBOUNDUTF8: |
1864 | /* was last char in word? */ |
1865 | ln = (locinput != PL_regbol) |
1866 | ? utf8_to_uv(reghop((U8*)locinput, -1), 0) : PL_regprev; |
1867 | if (OP(scan) == BOUNDUTF8 || OP(scan) == NBOUNDUTF8) { |
1868 | ln = isALNUM_uni(ln); |
1869 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput); |
a0ed51b3 |
1870 | } |
d6a28714 |
1871 | else { |
1872 | ln = isALNUM_LC_uni(ln); |
1873 | n = isALNUM_LC_utf8((U8*)locinput); |
1874 | } |
1875 | if (((!ln) == (!n)) == (OP(scan) == BOUNDUTF8 || OP(scan) == BOUNDLUTF8)) |
a0ed51b3 |
1876 | sayNO; |
a0ed51b3 |
1877 | break; |
d6a28714 |
1878 | case SPACEL: |
3280af22 |
1879 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1880 | /* FALL THROUGH */ |
d6a28714 |
1881 | case SPACE: |
1882 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 |
1883 | sayNO; |
d6a28714 |
1884 | if (!(OP(scan) == SPACE |
1885 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
4633a7c4 |
1886 | sayNO; |
76e3520e |
1887 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1888 | break; |
d6a28714 |
1889 | case SPACELUTF8: |
a0ed51b3 |
1890 | PL_reg_flags |= RF_tainted; |
1891 | /* FALL THROUGH */ |
d6a28714 |
1892 | case SPACEUTF8: |
a0ed51b3 |
1893 | if (!nextchr && locinput >= PL_regeol) |
1894 | sayNO; |
1895 | if (nextchr & 0x80) { |
d6a28714 |
1896 | if (!(OP(scan) == SPACEUTF8 |
1897 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1898 | : isSPACE_LC_utf8((U8*)locinput))) |
1899 | { |
a0ed51b3 |
1900 | sayNO; |
d6a28714 |
1901 | } |
6f06b55f |
1902 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1903 | nextchr = UCHARAT(locinput); |
1904 | break; |
1905 | } |
d6a28714 |
1906 | if (!(OP(scan) == SPACEUTF8 |
1907 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) |
a0ed51b3 |
1908 | sayNO; |
1909 | nextchr = UCHARAT(++locinput); |
1910 | break; |
d6a28714 |
1911 | case NSPACEL: |
3280af22 |
1912 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
1913 | /* FALL THROUGH */ |
d6a28714 |
1914 | case NSPACE: |
b8c5462f |
1915 | if (!nextchr) |
1916 | sayNO; |
d6a28714 |
1917 | if (OP(scan) == SPACE |
1918 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 |
1919 | sayNO; |
b8c5462f |
1920 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1921 | break; |
d6a28714 |
1922 | case NSPACELUTF8: |
a0ed51b3 |
1923 | PL_reg_flags |= RF_tainted; |
1924 | /* FALL THROUGH */ |
d6a28714 |
1925 | case NSPACEUTF8: |
b8c5462f |
1926 | if (!nextchr) |
1927 | sayNO; |
1928 | if (nextchr & 0x80) { |
d6a28714 |
1929 | if (OP(scan) == NSPACEUTF8 |
1930 | ? swash_fetch(PL_utf8_space,(U8*)locinput) |
1931 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f |
1932 | { |
1933 | sayNO; |
1934 | } |
1935 | locinput += PL_utf8skip[nextchr]; |
1936 | nextchr = UCHARAT(locinput); |
1937 | break; |
a0ed51b3 |
1938 | } |
d6a28714 |
1939 | if (OP(scan) == NSPACEUTF8 |
1940 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 |
1941 | sayNO; |
76e3520e |
1942 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1943 | break; |
d6a28714 |
1944 | case DIGITL: |
a0ed51b3 |
1945 | PL_reg_flags |= RF_tainted; |
1946 | /* FALL THROUGH */ |
d6a28714 |
1947 | case DIGIT: |
a0ed51b3 |
1948 | if (!nextchr && locinput >= PL_regeol) |
1949 | sayNO; |
d6a28714 |
1950 | if (!(OP(scan) == DIGIT |
1951 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
4633a7c4 |
1952 | sayNO; |
76e3520e |
1953 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1954 | break; |
d6a28714 |
1955 | case DIGITLUTF8: |
a0ed51b3 |
1956 | PL_reg_flags |= RF_tainted; |
1957 | /* FALL THROUGH */ |
d6a28714 |
1958 | case DIGITUTF8: |
a0ed51b3 |
1959 | if (!nextchr) |
1960 | sayNO; |
1961 | if (nextchr & 0x80) { |
d6a28714 |
1962 | if (OP(scan) == NDIGITUTF8 |
1963 | ? swash_fetch(PL_utf8_digit,(U8*)locinput) |
1964 | : isDIGIT_LC_utf8((U8*)locinput)) |
dfe13c55 |
1965 | { |
a0ed51b3 |
1966 | sayNO; |
dfe13c55 |
1967 | } |
6f06b55f |
1968 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1969 | nextchr = UCHARAT(locinput); |
1970 | break; |
1971 | } |
d6a28714 |
1972 | if (!isDIGIT(nextchr)) |
a0ed51b3 |
1973 | sayNO; |
1974 | nextchr = UCHARAT(++locinput); |
1975 | break; |
d6a28714 |
1976 | case NDIGITL: |
b8c5462f |
1977 | PL_reg_flags |= RF_tainted; |
1978 | /* FALL THROUGH */ |
d6a28714 |
1979 | case NDIGIT: |
b8c5462f |
1980 | if (!nextchr) |
1981 | sayNO; |
d6a28714 |
1982 | if (OP(scan) == DIGIT |
1983 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
4633a7c4 |
1984 | sayNO; |
76e3520e |
1985 | nextchr = UCHARAT(++locinput); |
a0d0e21e |
1986 | break; |
d6a28714 |
1987 | case NDIGITLUTF8: |
b8c5462f |
1988 | PL_reg_flags |= RF_tainted; |
1989 | /* FALL THROUGH */ |
d6a28714 |
1990 | case NDIGITUTF8: |
b8c5462f |
1991 | if (!nextchr && locinput >= PL_regeol) |
1992 | sayNO; |
a0ed51b3 |
1993 | if (nextchr & 0x80) { |
d6a28714 |
1994 | if (swash_fetch(PL_utf8_digit,(U8*)locinput)) |
a0ed51b3 |
1995 | sayNO; |
6f06b55f |
1996 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 |
1997 | nextchr = UCHARAT(locinput); |
1998 | break; |
1999 | } |
d6a28714 |
2000 | if (isDIGIT(nextchr)) |
a0ed51b3 |
2001 | sayNO; |
2002 | nextchr = UCHARAT(++locinput); |
2003 | break; |
2004 | case CLUMP: |
dfe13c55 |
2005 | if (locinput >= PL_regeol || swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
2006 | sayNO; |
6f06b55f |
2007 | locinput += PL_utf8skip[nextchr]; |
dfe13c55 |
2008 | while (locinput < PL_regeol && swash_fetch(PL_utf8_mark,(U8*)locinput)) |
a0ed51b3 |
2009 | locinput += UTF8SKIP(locinput); |
2010 | if (locinput > PL_regeol) |
2011 | sayNO; |
2012 | nextchr = UCHARAT(locinput); |
2013 | break; |
c8756f30 |
2014 | case REFFL: |
3280af22 |
2015 | PL_reg_flags |= RF_tainted; |
c8756f30 |
2016 | /* FALL THROUGH */ |
c277df42 |
2017 | case REF: |
c8756f30 |
2018 | case REFF: |
c277df42 |
2019 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
2020 | ln = PL_regstartp[n]; |
2c2d71f5 |
2021 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
cf93c79d |
2022 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 |
2023 | sayNO; /* Do not match unless seen CLOSEn. */ |
cf93c79d |
2024 | if (ln == PL_regendp[n]) |
a0d0e21e |
2025 | break; |
a0ed51b3 |
2026 | |
cf93c79d |
2027 | s = PL_bostr + ln; |
a0ed51b3 |
2028 | if (UTF && OP(scan) != REF) { /* REF can do byte comparison */ |
2029 | char *l = locinput; |
cf93c79d |
2030 | char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 |
2031 | /* |
2032 | * Note that we can't do the "other character" lookup trick as |
2033 | * in the 8-bit case (no pun intended) because in Unicode we |
2034 | * have to map both upper and title case to lower case. |
2035 | */ |
2036 | if (OP(scan) == REFF) { |
2037 | while (s < e) { |
2038 | if (l >= PL_regeol) |
2039 | sayNO; |
dfe13c55 |
2040 | if (toLOWER_utf8((U8*)s) != toLOWER_utf8((U8*)l)) |
a0ed51b3 |
2041 | sayNO; |
2042 | s += UTF8SKIP(s); |
2043 | l += UTF8SKIP(l); |
2044 | } |
2045 | } |
2046 | else { |
2047 | while (s < e) { |
2048 | if (l >= PL_regeol) |
2049 | sayNO; |
dfe13c55 |
2050 | if (toLOWER_LC_utf8((U8*)s) != toLOWER_LC_utf8((U8*)l)) |
a0ed51b3 |
2051 | sayNO; |
2052 | s += UTF8SKIP(s); |
2053 | l += UTF8SKIP(l); |
2054 | } |
2055 | } |
2056 | locinput = l; |
2057 | nextchr = UCHARAT(locinput); |
2058 | break; |
2059 | } |
2060 | |
a0d0e21e |
2061 | /* Inline the first character, for speed. */ |
76e3520e |
2062 | if (UCHARAT(s) != nextchr && |
c8756f30 |
2063 | (OP(scan) == REF || |
2064 | (UCHARAT(s) != ((OP(scan) == REFF |
22c35a8c |
2065 | ? PL_fold : PL_fold_locale)[nextchr])))) |
4633a7c4 |
2066 | sayNO; |
cf93c79d |
2067 | ln = PL_regendp[n] - ln; |
3280af22 |
2068 | if (locinput + ln > PL_regeol) |
4633a7c4 |
2069 | sayNO; |
c8756f30 |
2070 | if (ln > 1 && (OP(scan) == REF |
2071 | ? memNE(s, locinput, ln) |
2072 | : (OP(scan) == REFF |
2073 | ? ibcmp(s, locinput, ln) |
2074 | : ibcmp_locale(s, locinput, ln)))) |
4633a7c4 |
2075 | sayNO; |
a0d0e21e |
2076 | locinput += ln; |
76e3520e |
2077 | nextchr = UCHARAT(locinput); |
a0d0e21e |
2078 | break; |
2079 | |
2080 | case NOTHING: |
c277df42 |
2081 | case TAIL: |
a0d0e21e |
2082 | break; |
2083 | case BACK: |
2084 | break; |
c277df42 |
2085 | case EVAL: |
2086 | { |
2087 | dSP; |
533c011a |
2088 | OP_4tree *oop = PL_op; |
3280af22 |
2089 | COP *ocurcop = PL_curcop; |
2090 | SV **ocurpad = PL_curpad; |
c277df42 |
2091 | SV *ret; |
2092 | |
2093 | n = ARG(scan); |
533c011a |
2094 | PL_op = (OP_4tree*)PL_regdata->data[n]; |
2095 | DEBUG_r( PerlIO_printf(Perl_debug_log, " re_eval 0x%x\n", PL_op) ); |
dfad63ad |
2096 | PL_curpad = AvARRAY((AV*)PL_regdata->data[n + 2]); |
cf93c79d |
2097 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
c277df42 |
2098 | |
cea2e8a9 |
2099 | CALLRUNOPS(aTHX); /* Scalar context. */ |
c277df42 |
2100 | SPAGAIN; |
2101 | ret = POPs; |
2102 | PUTBACK; |
2103 | |
0f5d15d6 |
2104 | PL_op = oop; |
2105 | PL_curpad = ocurpad; |
2106 | PL_curcop = ocurcop; |
c277df42 |
2107 | if (logical) { |
0f5d15d6 |
2108 | if (logical == 2) { /* Postponed subexpression. */ |
2109 | regexp *re; |
22c35a8c |
2110 | MAGIC *mg = Null(MAGIC*); |
0f5d15d6 |
2111 | re_cc_state state; |
2112 | CURCUR cctmp; |
2113 | CHECKPOINT cp, lastcp; |
2114 | |
2115 | if(SvROK(ret) || SvRMAGICAL(ret)) { |
2116 | SV *sv = SvROK(ret) ? SvRV(ret) : ret; |
2117 | |
2118 | if(SvMAGICAL(sv)) |
2119 | mg = mg_find(sv, 'r'); |
2120 | } |
2121 | if (mg) { |
2122 | re = (regexp *)mg->mg_obj; |
df0003d4 |
2123 | (void)ReREFCNT_inc(re); |
0f5d15d6 |
2124 | } |
2125 | else { |
2126 | STRLEN len; |
2127 | char *t = SvPV(ret, len); |
2128 | PMOP pm; |
2129 | char *oprecomp = PL_regprecomp; |
2130 | I32 osize = PL_regsize; |
2131 | I32 onpar = PL_regnpar; |
2132 | |
2133 | pm.op_pmflags = 0; |
cea2e8a9 |
2134 | re = CALLREGCOMP(aTHX_ t, t + len, &pm); |
0f5d15d6 |
2135 | if (!(SvFLAGS(ret) |
2136 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY))) |
2137 | sv_magic(ret,(SV*)ReREFCNT_inc(re),'r',0,0); |
2138 | PL_regprecomp = oprecomp; |
2139 | PL_regsize = osize; |
2140 | PL_regnpar = onpar; |
2141 | } |
2142 | DEBUG_r( |
2143 | PerlIO_printf(Perl_debug_log, |
2144 | "Entering embedded `%s%.60s%s%s'\n", |
2145 | PL_colors[0], |
2146 | re->precomp, |
2147 | PL_colors[1], |
2148 | (strlen(re->precomp) > 60 ? "..." : "")) |
2149 | ); |
2150 | state.node = next; |
2151 | state.prev = PL_reg_call_cc; |
2152 | state.cc = PL_regcc; |
2153 | state.re = PL_reg_re; |
2154 | |
2155 | cctmp.cur = 0; |
2156 | cctmp.oldcc = 0; |
2157 | PL_regcc = &cctmp; |
2158 | |
2159 | cp = regcppush(0); /* Save *all* the positions. */ |
2160 | REGCP_SET; |
2161 | cache_re(re); |
2162 | state.ss = PL_savestack_ix; |
2163 | *PL_reglastparen = 0; |
2164 | PL_reg_call_cc = &state; |
2165 | PL_reginput = locinput; |
2c2d71f5 |
2166 | |
2167 | /* XXXX This is too dramatic a measure... */ |
2168 | PL_reg_maxiter = 0; |
2169 | |
0f5d15d6 |
2170 | if (regmatch(re->program + 1)) { |
2171 | ReREFCNT_dec(re); |
2172 | regcpblow(cp); |
2173 | sayYES; |
2174 | } |
2175 | DEBUG_r( |
2176 | PerlIO_printf(Perl_debug_log, |
2177 | "%*s failed...\n", |
2178 | REPORT_CODE_OFF+PL_regindent*2, "") |
2179 | ); |
2180 | ReREFCNT_dec(re); |
2181 | REGCP_UNWIND; |
2182 | regcppop(); |
2183 | PL_reg_call_cc = state.prev; |
2184 | PL_regcc = state.cc; |
2185 | PL_reg_re = state.re; |
d3790889 |
2186 | cache_re(PL_reg_re); |
2c2d71f5 |
2187 | |
2188 | /* XXXX This is too dramatic a measure... */ |
2189 | PL_reg_maxiter = 0; |
2190 | |
0f5d15d6 |
2191 | sayNO; |
2192 | } |
c277df42 |
2193 | sw = SvTRUE(ret); |
0f5d15d6 |
2194 | logical = 0; |
a0ed51b3 |
2195 | } |
2196 | else |
3280af22 |
2197 | sv_setsv(save_scalar(PL_replgv), ret); |
c277df42 |
2198 | break; |
2199 | } |
a0d0e21e |
2200 | case OPEN: |
c277df42 |
2201 | n = ARG(scan); /* which paren pair */ |
3280af22 |
2202 | PL_reg_start_tmp[n] = locinput; |
2203 | if (n > PL_regsize) |
2204 | PL_regsize = n; |
a0d0e21e |
2205 | break; |
2206 | case CLOSE: |
c277df42 |
2207 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
2208 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
2209 | PL_regendp[n] = locinput - PL_bostr; |
3280af22 |
2210 | if (n > *PL_reglastparen) |
2211 | *PL_reglastparen = n; |
a0d0e21e |
2212 | break; |
c277df42 |
2213 | case GROUPP: |
2214 | n = ARG(scan); /* which paren pair */ |
cf93c79d |
2215 | sw = (*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 |
2216 | break; |
2217 | case IFTHEN: |
2c2d71f5 |
2218 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
c277df42 |
2219 | if (sw) |
2220 | next = NEXTOPER(NEXTOPER(scan)); |
2221 | else { |
2222 | next = scan + ARG(scan); |
2223 | if (OP(next) == IFTHEN) /* Fake one. */ |
2224 | next = NEXTOPER(NEXTOPER(next)); |
2225 | } |
2226 | break; |
2227 | case LOGICAL: |
0f5d15d6 |
2228 | logical = scan->flags; |
c277df42 |
2229 | break; |
a0d0e21e |
2230 | case CURLYX: { |
2231 | CURCUR cc; |
3280af22 |
2232 | CHECKPOINT cp = PL_savestack_ix; |
c277df42 |
2233 | |
2234 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ |
2235 | next += ARG(next); |
3280af22 |
2236 | cc.oldcc = PL_regcc; |
2237 | PL_regcc = &cc; |
2238 | cc.parenfloor = *PL_reglastparen; |
a0d0e21e |
2239 | cc.cur = -1; |
2240 | cc.min = ARG1(scan); |
2241 | cc.max = ARG2(scan); |
c277df42 |
2242 | cc.scan = NEXTOPER(scan) + EXTRA_STEP_2ARGS; |
a0d0e21e |
2243 | cc.next = next; |
2244 | cc.minmod = minmod; |
2245 | cc.lastloc = 0; |
3280af22 |
2246 | PL_reginput = locinput; |
a0d0e21e |
2247 | n = regmatch(PREVOPER(next)); /* start on the WHILEM */ |
2248 | regcpblow(cp); |
3280af22 |
2249 | PL_regcc = cc.oldcc; |
4633a7c4 |
2250 | saySAME(n); |
a0d0e21e |
2251 | } |
2252 | /* NOT REACHED */ |
2253 | case WHILEM: { |
2254 | /* |
2255 | * This is really hard to understand, because after we match |
2256 | * what we're trying to match, we must make sure the rest of |
2c2d71f5 |
2257 | * the REx is going to match for sure, and to do that we have |
a0d0e21e |
2258 | * to go back UP the parse tree by recursing ever deeper. And |
2259 | * if it fails, we have to reset our parent's current state |
2260 | * that we can try again after backing off. |
2261 | */ |
2262 | |
c277df42 |
2263 | CHECKPOINT cp, lastcp; |
3280af22 |
2264 | CURCUR* cc = PL_regcc; |
c277df42 |
2265 | char *lastloc = cc->lastloc; /* Detection of 0-len. */ |
2266 | |
4633a7c4 |
2267 | n = cc->cur + 1; /* how many we know we matched */ |
3280af22 |
2268 | PL_reginput = locinput; |
a0d0e21e |
2269 | |
c277df42 |
2270 | DEBUG_r( |
2271 | PerlIO_printf(Perl_debug_log, |
2272 | "%*s %ld out of %ld..%ld cc=%lx\n", |
3280af22 |
2273 | REPORT_CODE_OFF+PL_regindent*2, "", |
c277df42 |
2274 | (long)n, (long)cc->min, |
2275 | (long)cc->max, (long)cc) |
2276 | ); |
4633a7c4 |
2277 | |
a0d0e21e |
2278 | /* If degenerate scan matches "", assume scan done. */ |
2279 | |
579cf2c3 |
2280 | if (locinput == cc->lastloc && n >= cc->min) { |
3280af22 |
2281 | PL_regcc = cc->oldcc; |
2282 | ln = PL_regcc->cur; |
c277df42 |
2283 | DEBUG_r( |
c3464db5 |
2284 | PerlIO_printf(Perl_debug_log, |
2285 | "%*s empty match detected, try continuation...\n", |
3280af22 |
2286 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2287 | ); |
a0d0e21e |
2288 | if (regmatch(cc->next)) |
4633a7c4 |
2289 | sayYES; |
c277df42 |
2290 | DEBUG_r( |
c3464db5 |
2291 | PerlIO_printf(Perl_debug_log, |
2292 | "%*s failed...\n", |
3280af22 |
2293 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2294 | ); |
3280af22 |
2295 | PL_regcc->cur = ln; |
2296 | PL_regcc = cc; |
4633a7c4 |
2297 | sayNO; |
a0d0e21e |
2298 | } |
2299 | |
2300 | /* First just match a string of min scans. */ |
2301 | |
2302 | if (n < cc->min) { |
2303 | cc->cur = n; |
2304 | cc->lastloc = locinput; |
4633a7c4 |
2305 | if (regmatch(cc->scan)) |
2306 | sayYES; |
2307 | cc->cur = n - 1; |
c277df42 |
2308 | cc->lastloc = lastloc; |
2309 | DEBUG_r( |
c3464db5 |
2310 | PerlIO_printf(Perl_debug_log, |
2311 | "%*s failed...\n", |
3280af22 |
2312 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2313 | ); |
4633a7c4 |
2314 | sayNO; |
a0d0e21e |
2315 | } |
2316 | |
2c2d71f5 |
2317 | if (scan->flags) { |
2318 | /* Check whether we already were at this position. |
2319 | Postpone detection until we know the match is not |
2320 | *that* much linear. */ |
2321 | if (!PL_reg_maxiter) { |
2322 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
2323 | PL_reg_leftiter = PL_reg_maxiter; |
2324 | } |
2325 | if (PL_reg_leftiter-- == 0) { |
2326 | I32 size = (PL_reg_maxiter + 7)/8; |
2327 | if (PL_reg_poscache) { |
2328 | if (PL_reg_poscache_size < size) { |
2329 | Renew(PL_reg_poscache, size, char); |
2330 | PL_reg_poscache_size = size; |
2331 | } |
2332 | Zero(PL_reg_poscache, size, char); |
2333 | } |
2334 | else { |
2335 | PL_reg_poscache_size = size; |
2336 | Newz(29, PL_reg_poscache, size, char); |
2337 | } |
2338 | DEBUG_r( |
2339 | PerlIO_printf(Perl_debug_log, |
2340 | "%sDetected a super-linear match, switching on caching%s...\n", |
2341 | PL_colors[4], PL_colors[5]) |
2342 | ); |
2343 | } |
2344 | if (PL_reg_leftiter < 0) { |
2345 | I32 o = locinput - PL_bostr, b; |
2346 | |
2347 | o = (scan->flags & 0xf) - 1 + o * (scan->flags>>4); |
2348 | b = o % 8; |
2349 | o /= 8; |
2350 | if (PL_reg_poscache[o] & (1<<b)) { |
2351 | DEBUG_r( |
2352 | PerlIO_printf(Perl_debug_log, |
2353 | "%*s already tried at this position...\n", |
2354 | REPORT_CODE_OFF+PL_regindent*2, "") |
2355 | ); |
2356 | sayNO; |
2357 | } |
2358 | PL_reg_poscache[o] |= (1<<b); |
2359 | } |
2360 | } |
2361 | |
a0d0e21e |
2362 | /* Prefer next over scan for minimal matching. */ |
2363 | |
2364 | if (cc->minmod) { |
3280af22 |
2365 | PL_regcc = cc->oldcc; |
2366 | ln = PL_regcc->cur; |
5f05dabc |
2367 | cp = regcppush(cc->parenfloor); |
c277df42 |
2368 | REGCP_SET; |
5f05dabc |
2369 | if (regmatch(cc->next)) { |
c277df42 |
2370 | regcpblow(cp); |
4633a7c4 |
2371 | sayYES; /* All done. */ |
5f05dabc |
2372 | } |
c277df42 |
2373 | REGCP_UNWIND; |
5f05dabc |
2374 | regcppop(); |
3280af22 |
2375 | PL_regcc->cur = ln; |
2376 | PL_regcc = cc; |
a0d0e21e |
2377 | |
c277df42 |
2378 | if (n >= cc->max) { /* Maximum greed exceeded? */ |
599cee73 |
2379 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
3280af22 |
2380 | && !(PL_reg_flags & RF_warned)) { |
2381 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
2382 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
2f3ca594 |
2383 | "Complex regular subexpression recursion", |
2384 | REG_INFTY - 1); |
c277df42 |
2385 | } |
4633a7c4 |
2386 | sayNO; |
c277df42 |
2387 | } |
a687059c |
2388 | |
c277df42 |
2389 | DEBUG_r( |
c3464db5 |
2390 | PerlIO_printf(Perl_debug_log, |
2391 | "%*s trying longer...\n", |
3280af22 |
2392 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2393 | ); |
a0d0e21e |
2394 | /* Try scanning more and see if it helps. */ |
3280af22 |
2395 | PL_reginput = locinput; |
a0d0e21e |
2396 | cc->cur = n; |
2397 | cc->lastloc = locinput; |
5f05dabc |
2398 | cp = regcppush(cc->parenfloor); |
c277df42 |
2399 | REGCP_SET; |
5f05dabc |
2400 | if (regmatch(cc->scan)) { |
c277df42 |
2401 | regcpblow(cp); |
4633a7c4 |
2402 | sayYES; |
5f05dabc |
2403 | } |
c277df42 |
2404 | DEBUG_r( |
c3464db5 |
2405 | PerlIO_printf(Perl_debug_log, |
2406 | "%*s failed...\n", |
3280af22 |
2407 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2408 | ); |
2409 | REGCP_UNWIND; |
5f05dabc |
2410 | regcppop(); |
4633a7c4 |
2411 | cc->cur = n - 1; |
c277df42 |
2412 | cc->lastloc = lastloc; |
4633a7c4 |
2413 | sayNO; |
a0d0e21e |
2414 | } |
2415 | |
2416 | /* Prefer scan over next for maximal matching. */ |
2417 | |
2418 | if (n < cc->max) { /* More greed allowed? */ |
5f05dabc |
2419 | cp = regcppush(cc->parenfloor); |
a0d0e21e |
2420 | cc->cur = n; |
2421 | cc->lastloc = locinput; |
c277df42 |
2422 | REGCP_SET; |
5f05dabc |
2423 | if (regmatch(cc->scan)) { |
c277df42 |
2424 | regcpblow(cp); |
4633a7c4 |
2425 | sayYES; |
5f05dabc |
2426 | } |
c277df42 |
2427 | REGCP_UNWIND; |
a0d0e21e |
2428 | regcppop(); /* Restore some previous $<digit>s? */ |
3280af22 |
2429 | PL_reginput = locinput; |
c277df42 |
2430 | DEBUG_r( |
c3464db5 |
2431 | PerlIO_printf(Perl_debug_log, |
2432 | "%*s failed, try continuation...\n", |
3280af22 |
2433 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2434 | ); |
2435 | } |
599cee73 |
2436 | if (ckWARN(WARN_UNSAFE) && n >= REG_INFTY |
2437 | && !(PL_reg_flags & RF_warned)) { |
3280af22 |
2438 | PL_reg_flags |= RF_warned; |
cea2e8a9 |
2439 | Perl_warner(aTHX_ WARN_UNSAFE, "%s limit (%d) exceeded", |
cb5d145d |
2440 | "Complex regular subexpression recursion", |
2441 | REG_INFTY - 1); |
a0d0e21e |
2442 | } |
2443 | |
2444 | /* Failed deeper matches of scan, so see if this one works. */ |
3280af22 |
2445 | PL_regcc = cc->oldcc; |
2446 | ln = PL_regcc->cur; |
a0d0e21e |
2447 | if (regmatch(cc->next)) |
4633a7c4 |
2448 | sayYES; |
c277df42 |
2449 | DEBUG_r( |
c3464db5 |
2450 | PerlIO_printf(Perl_debug_log, "%*s failed...\n", |
3280af22 |
2451 | REPORT_CODE_OFF+PL_regindent*2, "") |
c277df42 |
2452 | ); |
3280af22 |
2453 | PL_regcc->cur = ln; |
2454 | PL_regcc = cc; |
4633a7c4 |
2455 | cc->cur = n - 1; |
c277df42 |
2456 | cc->lastloc = lastloc; |
4633a7c4 |
2457 | sayNO; |
a0d0e21e |
2458 | } |
2459 | /* NOT REACHED */ |
c277df42 |
2460 | case BRANCHJ: |
2461 | next = scan + ARG(scan); |
2462 | if (next == scan) |
2463 | next = NULL; |
2464 | inner = NEXTOPER(NEXTOPER(scan)); |
2465 | goto do_branch; |
2466 | case BRANCH: |
2467 | inner = NEXTOPER(scan); |
2468 | do_branch: |
2469 | { |
2470 | CHECKPOINT lastcp; |
2471 | c1 = OP(scan); |
2472 | if (OP(next) != c1) /* No choice. */ |
2473 | next = inner; /* Avoid recursion. */ |
a0d0e21e |
2474 | else { |
3280af22 |
2475 | int lastparen = *PL_reglastparen; |
c277df42 |
2476 | |
2477 | REGCP_SET; |
a0d0e21e |
2478 | do { |
3280af22 |
2479 | PL_reginput = locinput; |
c277df42 |
2480 | if (regmatch(inner)) |
4633a7c4 |
2481 | sayYES; |
c277df42 |
2482 | REGCP_UNWIND; |
3280af22 |
2483 | for (n = *PL_reglastparen; n > lastparen; n--) |
cf93c79d |
2484 | PL_regendp[n] = -1; |
3280af22 |
2485 | *PL_reglastparen = n; |
c277df42 |
2486 | scan = next; |
a0d0e21e |
2487 | /*SUPPRESS 560*/ |
c277df42 |
2488 | if (n = (c1 == BRANCH ? NEXT_OFF(next) : ARG(next))) |
2489 | next += n; |
a0d0e21e |
2490 | else |
c277df42 |
2491 | next = NULL; |
c277df42 |
2492 | inner = NEXTOPER(scan); |
2493 | if (c1 == BRANCHJ) { |
2494 | inner = NEXTOPER(inner); |
2495 | } |
2496 | } while (scan != NULL && OP(scan) == c1); |
4633a7c4 |
2497 | sayNO; |
a0d0e21e |
2498 | /* NOTREACHED */ |
a687059c |
2499 | } |
a0d0e21e |
2500 | } |
2501 | break; |
2502 | case MINMOD: |
2503 | minmod = 1; |
2504 | break; |
c277df42 |
2505 | case CURLYM: |
2506 | { |
00db4c45 |
2507 | I32 l = 0; |
c277df42 |
2508 | CHECKPOINT lastcp; |
2509 | |
2510 | /* We suppose that the next guy does not need |
2511 | backtracking: in particular, it is of constant length, |
2512 | and has no parenths to influence future backrefs. */ |
2513 | ln = ARG1(scan); /* min to match */ |
2514 | n = ARG2(scan); /* max to match */ |
c277df42 |
2515 | paren = scan->flags; |
2516 | if (paren) { |
3280af22 |
2517 | if (paren > PL_regsize) |
2518 | PL_regsize = paren; |
2519 | if (paren > *PL_reglastparen) |
2520 | *PL_reglastparen = paren; |
c277df42 |
2521 | } |
dc45a647 |
2522 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
c277df42 |
2523 | if (paren) |
2524 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
3280af22 |
2525 | PL_reginput = locinput; |
c277df42 |
2526 | if (minmod) { |
2527 | minmod = 0; |
2528 | if (ln && regrepeat_hard(scan, ln, &l) < ln) |
2529 | sayNO; |
5f4b28b2 |
2530 | if (ln && l == 0 && n >= ln |
c277df42 |
2531 | /* In fact, this is tricky. If paren, then the |
2532 | fact that we did/didnot match may influence |
2533 | future execution. */ |
2534 | && !(paren && ln == 0)) |
2535 | ln = n; |
3280af22 |
2536 | locinput = PL_reginput; |
22c35a8c |
2537 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
cd439c50 |
2538 | c1 = (U8)*STRING(next); |
c277df42 |
2539 | if (OP(next) == EXACTF) |
22c35a8c |
2540 | c2 = PL_fold[c1]; |
c277df42 |
2541 | else if (OP(next) == EXACTFL) |
22c35a8c |
2542 | c2 = PL_fold_locale[c1]; |
c277df42 |
2543 | else |
2544 | c2 = c1; |
a0ed51b3 |
2545 | } |
2546 | else |
c277df42 |
2547 | c1 = c2 = -1000; |
2548 | REGCP_SET; |
5f4b28b2 |
2549 | /* This may be improved if l == 0. */ |
c277df42 |
2550 | while (n >= ln || (n == REG_INFTY && ln > 0 && l)) { /* ln overflow ? */ |
2551 | /* If it could work, try it. */ |
2552 | if (c1 == -1000 || |
3280af22 |
2553 | UCHARAT(PL_reginput) == c1 || |
2554 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2555 | { |
2556 | if (paren) { |
2557 | if (n) { |
cf93c79d |
2558 | PL_regstartp[paren] = |
2559 | HOPc(PL_reginput, -l) - PL_bostr; |
2560 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2561 | } |
2562 | else |
cf93c79d |
2563 | PL_regendp[paren] = -1; |
c277df42 |
2564 | } |
2565 | if (regmatch(next)) |
2566 | sayYES; |
2567 | REGCP_UNWIND; |
2568 | } |
2569 | /* Couldn't or didn't -- move forward. */ |
3280af22 |
2570 | PL_reginput = locinput; |
c277df42 |
2571 | if (regrepeat_hard(scan, 1, &l)) { |
2572 | ln++; |
3280af22 |
2573 | locinput = PL_reginput; |
c277df42 |
2574 | } |
2575 | else |
2576 | sayNO; |
2577 | } |
a0ed51b3 |
2578 | } |
2579 | else { |
c277df42 |
2580 | n = regrepeat_hard(scan, n, &l); |
2581 | if (n != 0 && l == 0 |
2582 | /* In fact, this is tricky. If paren, then the |
2583 | fact that we did/didnot match may influence |
2584 | future execution. */ |
2585 | && !(paren && ln == 0)) |
2586 | ln = n; |
3280af22 |
2587 | locinput = PL_reginput; |
c277df42 |
2588 | DEBUG_r( |
5c0ca799 |
2589 | PerlIO_printf(Perl_debug_log, |
2590 | "%*s matched %ld times, len=%ld...\n", |
3280af22 |
2591 | REPORT_CODE_OFF+PL_regindent*2, "", n, l) |
c277df42 |
2592 | ); |
2593 | if (n >= ln) { |
22c35a8c |
2594 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
cd439c50 |
2595 | c1 = (U8)*STRING(next); |
c277df42 |
2596 | if (OP(next) == EXACTF) |
22c35a8c |
2597 | c2 = PL_fold[c1]; |
c277df42 |
2598 | else if (OP(next) == EXACTFL) |
22c35a8c |
2599 | c2 = PL_fold_locale[c1]; |
c277df42 |
2600 | else |
2601 | c2 = c1; |
a0ed51b3 |
2602 | } |
2603 | else |
c277df42 |
2604 | c1 = c2 = -1000; |
2605 | } |
2606 | REGCP_SET; |
2607 | while (n >= ln) { |
2608 | /* If it could work, try it. */ |
2609 | if (c1 == -1000 || |
3280af22 |
2610 | UCHARAT(PL_reginput) == c1 || |
2611 | UCHARAT(PL_reginput) == c2) |
a0ed51b3 |
2612 | { |
2613 | DEBUG_r( |
c3464db5 |
2614 | PerlIO_printf(Perl_debug_log, |
2615 | "%*s trying tail with n=%ld...\n", |
3280af22 |
2616 | REPORT_CODE_OFF+PL_regindent*2, "", n) |
a0ed51b3 |
2617 | ); |
2618 | if (paren) { |
2619 | if (n) { |
cf93c79d |
2620 | PL_regstartp[paren] = HOPc(PL_reginput, -l) - PL_bostr; |
2621 | PL_regendp[paren] = PL_reginput - PL_bostr; |
c277df42 |
2622 | } |
a0ed51b3 |
2623 | else |
cf93c79d |
2624 | PL_regendp[paren] = -1; |
c277df42 |
2625 | } |
a0ed51b3 |
2626 | if (regmatch(next)) |
2627 | sayYES; |
2628 | REGCP_UNWIND; |
2629 | } |
c277df42 |
2630 | /* Couldn't or didn't -- back up. */ |
2631 | n--; |
dfe13c55 |
2632 | locinput = HOPc(locinput, -l); |
3280af22 |
2633 | PL_reginput = locinput; |
c277df42 |
2634 | } |
2635 | } |
2636 | sayNO; |
2637 | break; |
2638 | } |
2639 | case CURLYN: |
2640 | paren = scan->flags; /* Which paren to set */ |
3280af22 |
2641 | if (paren > PL_regsize) |
2642 | PL_regsize = paren; |
2643 | if (paren > *PL_reglastparen) |
2644 | *PL_reglastparen = paren; |
c277df42 |
2645 | ln = ARG1(scan); /* min to match */ |
2646 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2647 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
c277df42 |
2648 | goto repeat; |
a0d0e21e |
2649 | case CURLY: |
c277df42 |
2650 | paren = 0; |
a0d0e21e |
2651 | ln = ARG1(scan); /* min to match */ |
2652 | n = ARG2(scan); /* max to match */ |
dc45a647 |
2653 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
a0d0e21e |
2654 | goto repeat; |
2655 | case STAR: |
2656 | ln = 0; |
c277df42 |
2657 | n = REG_INFTY; |
a0d0e21e |
2658 | scan = NEXTOPER(scan); |
c277df42 |
2659 | paren = 0; |
a0d0e21e |
2660 | goto repeat; |
2661 | case PLUS: |
c277df42 |
2662 | ln = 1; |
2663 | n = REG_INFTY; |
2664 | scan = NEXTOPER(scan); |
2665 | paren = 0; |
2666 | repeat: |
a0d0e21e |
2667 | /* |
2668 | * Lookahead to avoid useless match attempts |
2669 | * when we know what character comes next. |
2670 | */ |
22c35a8c |
2671 | if (PL_regkind[(U8)OP(next)] == EXACT) { |
cd439c50 |
2672 | c1 = (U8)*STRING(next); |
bbce6d69 |
2673 | if (OP(next) == EXACTF) |
22c35a8c |
2674 | c2 = PL_fold[c1]; |
bbce6d69 |
2675 | else if (OP(next) == EXACTFL) |
22c35a8c |
2676 | c2 = PL_fold_locale[c1]; |
bbce6d69 |
2677 | else |
2678 | c2 = c1; |
2679 | } |
a0d0e21e |
2680 | else |
bbce6d69 |
2681 | c1 = c2 = -1000; |
3280af22 |
2682 | PL_reginput = locinput; |
a0d0e21e |
2683 | if (minmod) { |
c277df42 |
2684 | CHECKPOINT lastcp; |
a0d0e21e |
2685 | minmod = 0; |
2686 | if (ln && regrepeat(scan, ln) < ln) |
4633a7c4 |
2687 | sayNO; |
a0ed51b3 |
2688 | locinput = PL_reginput; |
c277df42 |
2689 | REGCP_SET; |
0fe9bf95 |
2690 | if (c1 != -1000) { |
2691 | char *e = locinput + n - ln; /* Should not check after this */ |
2692 | char *old = locinput; |
2693 | |
2694 | if (e >= PL_regeol || (n == REG_INFTY)) |
2695 | e = PL_regeol - 1; |
2696 | while (1) { |
2697 | /* Find place 'next' could work */ |
2698 | if (c1 == c2) { |
2699 | while (locinput <= e && *locinput != c1) |
2700 | locinput++; |
2701 | } else { |
2702 | while (locinput <= e |
2703 | && *locinput != c1 |
2704 | && *locinput != c2) |
2705 | locinput++; |
2706 | } |
2707 | if (locinput > e) |
2708 | sayNO; |
2709 | /* PL_reginput == old now */ |
2710 | if (locinput != old) { |
2711 | ln = 1; /* Did some */ |
2712 | if (regrepeat(scan, locinput - old) < |
2713 | locinput - old) |
2714 | sayNO; |
2715 | } |
2716 | /* PL_reginput == locinput now */ |
2717 | if (paren) { |
2718 | if (ln) { |
cf93c79d |
2719 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; |
2720 | PL_regendp[paren] = locinput - PL_bostr; |
0fe9bf95 |
2721 | } |
2722 | else |
cf93c79d |
2723 | PL_regendp[paren] = -1; |
0fe9bf95 |
2724 | } |
2725 | if (regmatch(next)) |
2726 | sayYES; |
2727 | PL_reginput = locinput; /* Could be reset... */ |
2728 | REGCP_UNWIND; |
2729 | /* Couldn't or didn't -- move forward. */ |
2730 | old = locinput++; |
2731 | } |
2732 | } |
2733 | else |
c277df42 |
2734 | while (n >= ln || (n == REG_INFTY && ln > 0)) { /* ln overflow ? */ |
a0d0e21e |
2735 | /* If it could work, try it. */ |
bbce6d69 |
2736 | if (c1 == -1000 || |
3280af22 |
2737 | UCHARAT(PL_reginput) == c1 || |
2738 | UCHARAT(PL_reginput) == c2) |
bbce6d69 |
2739 | { |
c277df42 |
2740 | if (paren) { |
2741 | if (n) { |
cf93c79d |
2742 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2743 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2744 | } |
2745 | else |
cf93c79d |
2746 | PL_regendp[paren] = -1; |
c277df42 |
2747 | } |
a0d0e21e |
2748 | if (regmatch(next)) |
4633a7c4 |
2749 | sayYES; |
c277df42 |
2750 | REGCP_UNWIND; |
bbce6d69 |
2751 | } |
c277df42 |
2752 | /* Couldn't or didn't -- move forward. */ |
a0ed51b3 |
2753 | PL_reginput = locinput; |
a0d0e21e |
2754 | if (regrepeat(scan, 1)) { |
2755 | ln++; |
a0ed51b3 |
2756 | locinput = PL_reginput; |
2757 | } |
2758 | else |
4633a7c4 |
2759 | sayNO; |
a0d0e21e |
2760 | } |
2761 | } |
2762 | else { |
c277df42 |
2763 | CHECKPOINT lastcp; |
a0d0e21e |
2764 | n = regrepeat(scan, n); |
a0ed51b3 |
2765 | locinput = PL_reginput; |
22c35a8c |
2766 | if (ln < n && PL_regkind[(U8)OP(next)] == EOL && |
3280af22 |
2767 | (!PL_multiline || OP(next) == SEOL)) |
a0d0e21e |
2768 | ln = n; /* why back off? */ |
c277df42 |
2769 | REGCP_SET; |
2770 | if (paren) { |
2771 | while (n >= ln) { |
2772 | /* If it could work, try it. */ |
2773 | if (c1 == -1000 || |
3280af22 |
2774 | UCHARAT(PL_reginput) == c1 || |
2775 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2776 | { |
2777 | if (paren && n) { |
2778 | if (n) { |
cf93c79d |
2779 | PL_regstartp[paren] = HOPc(PL_reginput, -1) - PL_bostr; |
2780 | PL_regendp[paren] = PL_reginput - PL_bostr; |
a0ed51b3 |
2781 | } |
2782 | else |
cf93c79d |
2783 | PL_regendp[paren] = -1; |
c277df42 |
2784 | } |
2785 | if (regmatch(next)) |
2786 | sayYES; |
2787 | REGCP_UNWIND; |
2788 | } |
2789 | /* Couldn't or didn't -- back up. */ |
2790 | n--; |
dfe13c55 |
2791 | PL_reginput = locinput = HOPc(locinput, -1); |
c277df42 |
2792 | } |
a0ed51b3 |
2793 | } |
2794 | else { |
c277df42 |
2795 | while (n >= ln) { |
2796 | /* If it could work, try it. */ |
2797 | if (c1 == -1000 || |
3280af22 |
2798 | UCHARAT(PL_reginput) == c1 || |
2799 | UCHARAT(PL_reginput) == c2) |
c277df42 |
2800 | { |
2801 | if (regmatch(next)) |
2802 | sayYES; |
2803 | REGCP_UNWIND; |
2804 | } |
2805 | /* Couldn't or didn't -- back up. */ |
2806 | n--; |
dfe13c55 |
2807 | PL_reginput = locinput = HOPc(locinput, -1); |
bbce6d69 |
2808 | } |
a0d0e21e |
2809 | } |
2810 | } |
4633a7c4 |
2811 | sayNO; |
c277df42 |
2812 | break; |
a0d0e21e |
2813 | case END: |
0f5d15d6 |
2814 | if (PL_reg_call_cc) { |
2815 | re_cc_state *cur_call_cc = PL_reg_call_cc; |
2816 | CURCUR *cctmp = PL_regcc; |
2817 | regexp *re = PL_reg_re; |
2818 | CHECKPOINT cp, lastcp; |
2819 | |
2820 | cp = regcppush(0); /* Save *all* the positions. */ |
2821 | REGCP_SET; |
2822 | regcp_set_to(PL_reg_call_cc->ss); /* Restore parens of |
2823 | the caller. */ |
2824 | PL_reginput = locinput; /* Make position available to |
2825 | the callcc. */ |
2826 | cache_re(PL_reg_call_cc->re); |
2827 | PL_regcc = PL_reg_call_cc->cc; |
2828 | PL_reg_call_cc = PL_reg_call_cc->prev; |
2829 | if (regmatch(cur_call_cc->node)) { |
2830 | PL_reg_call_cc = cur_call_cc; |
2831 | regcpblow(cp); |
2832 | sayYES; |
2833 | } |
2834 | REGCP_UNWIND; |
2835 | regcppop(); |
2836 | PL_reg_call_cc = cur_call_cc; |
2837 | PL_regcc = cctmp; |
2838 | PL_reg_re = re; |
2839 | cache_re(re); |
2840 | |
2841 | DEBUG_r( |
2842 | PerlIO_printf(Perl_debug_log, |
2843 | "%*s continuation failed...\n", |
2844 | REPORT_CODE_OFF+PL_regindent*2, "") |
2845 | ); |
2846 | sayNO; |
2847 | } |
3280af22 |
2848 | if (locinput < PL_regtill) |
7e5428c5 |
2849 | sayNO; /* Cannot match: too short. */ |
2850 | /* Fall through */ |
2851 | case SUCCEED: |
3280af22 |
2852 | PL_reginput = locinput; /* put where regtry can find it */ |
4633a7c4 |
2853 | sayYES; /* Success! */ |
c277df42 |
2854 | case SUSPEND: |
2855 | n = 1; |
9fe1d20c |
2856 | PL_reginput = locinput; |
c277df42 |
2857 | goto do_ifmatch; |
a0d0e21e |
2858 | case UNLESSM: |
c277df42 |
2859 | n = 0; |
a0ed51b3 |
2860 | if (scan->flags) { |
0fe9bf95 |
2861 | if (UTF) { /* XXXX This is absolutely |
2862 | broken, we read before |
2863 | start of string. */ |
2864 | s = HOPMAYBEc(locinput, -scan->flags); |
2865 | if (!s) |
2866 | goto say_yes; |
2867 | PL_reginput = s; |
2868 | } |
2869 | else { |
2870 | if (locinput < PL_bostr + scan->flags) |
2871 | goto say_yes; |
2872 | PL_reginput = locinput - scan->flags; |
2873 | goto do_ifmatch; |
2874 | } |
a0ed51b3 |
2875 | } |
2876 | else |
2877 | PL_reginput = locinput; |
c277df42 |
2878 | goto do_ifmatch; |
2879 | case IFMATCH: |
2880 | n = 1; |
a0ed51b3 |
2881 | if (scan->flags) { |
0fe9bf95 |
2882 | if (UTF) { /* XXXX This is absolutely |
2883 | broken, we read before |
2884 | start of string. */ |
2885 | s = HOPMAYBEc(locinput, -scan->flags); |
2886 | if (!s || s < PL_bostr) |
2887 | goto say_no; |
2888 | PL_reginput = s; |
2889 | } |
2890 | else { |
2891 | if (locinput < PL_bostr + scan->flags) |
2892 | goto say_no; |
2893 | PL_reginput = locinput - scan->flags; |
2894 | goto do_ifmatch; |
2895 | } |
a0ed51b3 |
2896 | } |
2897 | else |
2898 | PL_reginput = locinput; |
2899 | |
c277df42 |
2900 | do_ifmatch: |
c277df42 |
2901 | inner = NEXTOPER(NEXTOPER(scan)); |
2902 | if (regmatch(inner) != n) { |
2903 | say_no: |
2904 | if (logical) { |
2905 | logical = 0; |
2906 | sw = 0; |
2907 | goto do_longjump; |
a0ed51b3 |
2908 | } |
2909 | else |
c277df42 |
2910 | sayNO; |
2911 | } |
2912 | say_yes: |
2913 | if (logical) { |
2914 | logical = 0; |
2915 | sw = 1; |
2916 | } |
fe44a5e8 |
2917 | if (OP(scan) == SUSPEND) { |
3280af22 |
2918 | locinput = PL_reginput; |
565764a8 |
2919 | nextchr = UCHARAT(locinput); |
fe44a5e8 |
2920 | } |
c277df42 |
2921 | /* FALL THROUGH. */ |
2922 | case LONGJMP: |
2923 | do_longjump: |
2924 | next = scan + ARG(scan); |
2925 | if (next == scan) |
2926 | next = NULL; |
a0d0e21e |
2927 | break; |
2928 | default: |
c030ccd9 |
2929 | PerlIO_printf(PerlIO_stderr(), "%lx %d\n", |
c277df42 |
2930 | (unsigned long)scan, OP(scan)); |
cea2e8a9 |
2931 | Perl_croak(aTHX_ "regexp memory corruption"); |
a687059c |
2932 | } |
a0d0e21e |
2933 | scan = next; |
2934 | } |
a687059c |
2935 | |
a0d0e21e |
2936 | /* |
2937 | * We get here only if there's trouble -- normally "case END" is |
2938 | * the terminating point. |
2939 | */ |
cea2e8a9 |
2940 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e |
2941 | /*NOTREACHED*/ |
4633a7c4 |
2942 | sayNO; |
2943 | |
2944 | yes: |
2945 | #ifdef DEBUGGING |
3280af22 |
2946 | PL_regindent--; |
4633a7c4 |
2947 | #endif |
2948 | return 1; |
2949 | |
2950 | no: |
2951 | #ifdef DEBUGGING |
3280af22 |
2952 | PL_regindent--; |
4633a7c4 |
2953 | #endif |
a0d0e21e |
2954 | return 0; |
a687059c |
2955 | } |
2956 | |
2957 | /* |
2958 | - regrepeat - repeatedly match something simple, report how many |
2959 | */ |
2960 | /* |
2961 | * [This routine now assumes that it will only match on things of length 1. |
2962 | * That was true before, but now we assume scan - reginput is the count, |
a0ed51b3 |
2963 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c |
2964 | */ |
76e3520e |
2965 | STATIC I32 |
cea2e8a9 |
2966 | S_regrepeat(pTHX_ regnode *p, I32 max) |
a687059c |
2967 | { |
5c0ca799 |
2968 | dTHR; |
a0d0e21e |
2969 | register char *scan; |
2970 | register char *opnd; |
2971 | register I32 c; |
3280af22 |
2972 | register char *loceol = PL_regeol; |
a0ed51b3 |
2973 | register I32 hardcount = 0; |
a0d0e21e |
2974 | |
3280af22 |
2975 | scan = PL_reginput; |
c277df42 |
2976 | if (max != REG_INFTY && max < loceol - scan) |
a0d0e21e |
2977 | loceol = scan + max; |
a0d0e21e |
2978 | switch (OP(p)) { |
22c35a8c |
2979 | case REG_ANY: |
a0d0e21e |
2980 | while (scan < loceol && *scan != '\n') |
2981 | scan++; |
2982 | break; |
2983 | case SANY: |
2984 | scan = loceol; |
2985 | break; |
a0ed51b3 |
2986 | case ANYUTF8: |
2987 | loceol = PL_regeol; |
2988 | while (scan < loceol && *scan != '\n') { |
2989 | scan += UTF8SKIP(scan); |
2990 | hardcount++; |
2991 | } |
2992 | break; |
2993 | case SANYUTF8: |
2994 | loceol = PL_regeol; |
2995 | while (scan < loceol) { |
2996 | scan += UTF8SKIP(scan); |
2997 | hardcount++; |
2998 | } |
2999 | break; |
bbce6d69 |
3000 | case EXACT: /* length of string is 1 */ |
cd439c50 |
3001 | c = (U8)*STRING(p); |
bbce6d69 |
3002 | while (scan < loceol && UCHARAT(scan) == c) |
3003 | scan++; |
3004 | break; |
3005 | case EXACTF: /* length of string is 1 */ |
cd439c50 |
3006 | c = (U8)*STRING(p); |
bbce6d69 |
3007 | while (scan < loceol && |
22c35a8c |
3008 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 |
3009 | scan++; |
3010 | break; |
3011 | case EXACTFL: /* length of string is 1 */ |
3280af22 |
3012 | PL_reg_flags |= RF_tainted; |
cd439c50 |
3013 | c = (U8)*STRING(p); |
bbce6d69 |
3014 | while (scan < loceol && |
22c35a8c |
3015 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e |
3016 | scan++; |
3017 | break; |
a0ed51b3 |
3018 | case ANYOFUTF8: |
3019 | loceol = PL_regeol; |
3020 | while (scan < loceol && REGINCLASSUTF8(p, (U8*)scan)) { |
3021 | scan += UTF8SKIP(scan); |
3022 | hardcount++; |
3023 | } |
3024 | break; |
a0d0e21e |
3025 | case ANYOF: |
cd439c50 |
3026 | opnd = MASK(p); |
ae5c130c |
3027 | while (scan < loceol && REGINCLASS(opnd, *scan)) |
a0d0e21e |
3028 | scan++; |
a0d0e21e |
3029 | break; |
3030 | case ALNUM: |
3031 | while (scan < loceol && isALNUM(*scan)) |
3032 | scan++; |
3033 | break; |
a0ed51b3 |
3034 | case ALNUMUTF8: |
3035 | loceol = PL_regeol; |
dfe13c55 |
3036 | while (scan < loceol && swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
3037 | scan += UTF8SKIP(scan); |
3038 | hardcount++; |
3039 | } |
3040 | break; |
bbce6d69 |
3041 | case ALNUML: |
3280af22 |
3042 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
3043 | while (scan < loceol && isALNUM_LC(*scan)) |
3044 | scan++; |
3045 | break; |
a0ed51b3 |
3046 | case ALNUMLUTF8: |
3047 | PL_reg_flags |= RF_tainted; |
3048 | loceol = PL_regeol; |
dfe13c55 |
3049 | while (scan < loceol && isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
3050 | scan += UTF8SKIP(scan); |
3051 | hardcount++; |
3052 | } |
3053 | break; |
3054 | break; |
a0d0e21e |
3055 | case NALNUM: |
3056 | while (scan < loceol && !isALNUM(*scan)) |
3057 | scan++; |
3058 | break; |
a0ed51b3 |
3059 | case NALNUMUTF8: |
3060 | loceol = PL_regeol; |
dfe13c55 |
3061 | while (scan < loceol && !swash_fetch(PL_utf8_alnum, (U8*)scan)) { |
a0ed51b3 |
3062 | scan += UTF8SKIP(scan); |
3063 | hardcount++; |
3064 | } |
3065 | break; |
bbce6d69 |
3066 | case NALNUML: |
3280af22 |
3067 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
3068 | while (scan < loceol && !isALNUM_LC(*scan)) |
3069 | scan++; |
3070 | break; |
a0ed51b3 |
3071 | case NALNUMLUTF8: |
3072 | PL_reg_flags |= RF_tainted; |
3073 | loceol = PL_regeol; |
dfe13c55 |
3074 | while (scan < loceol && !isALNUM_LC_utf8((U8*)scan)) { |
a0ed51b3 |
3075 | scan += UTF8SKIP(scan); |
3076 | hardcount++; |
3077 | } |
3078 | break; |
a0d0e21e |
3079 | case SPACE: |
3080 | while (scan < loceol && isSPACE(*scan)) |
3081 | scan++; |
3082 | break; |
a0ed51b3 |
3083 | case SPACEUTF8: |
3084 | loceol = PL_regeol; |
dfe13c55 |
3085 | while (scan < loceol && (*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
3086 | scan += UTF8SKIP(scan); |
3087 | hardcount++; |
3088 | } |
3089 | break; |
bbce6d69 |
3090 | case SPACEL: |
3280af22 |
3091 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
3092 | while (scan < loceol && isSPACE_LC(*scan)) |
3093 | scan++; |
3094 | break; |
a0ed51b3 |
3095 | case SPACELUTF8: |
3096 | PL_reg_flags |= RF_tainted; |
3097 | loceol = PL_regeol; |
dfe13c55 |
3098 | while (scan < loceol && (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
3099 | scan += UTF8SKIP(scan); |
3100 | hardcount++; |
3101 | } |
3102 | break; |
a0d0e21e |
3103 | case NSPACE: |
3104 | while (scan < loceol && !isSPACE(*scan)) |
3105 | scan++; |
3106 | break; |
a0ed51b3 |
3107 | case NSPACEUTF8: |
3108 | loceol = PL_regeol; |
dfe13c55 |
3109 | while (scan < loceol && !(*scan == ' ' || swash_fetch(PL_utf8_space,(U8*)scan))) { |
a0ed51b3 |
3110 | scan += UTF8SKIP(scan); |
3111 | hardcount++; |
3112 | } |
3113 | break; |
bbce6d69 |
3114 | case NSPACEL: |
3280af22 |
3115 | PL_reg_flags |= RF_tainted; |
bbce6d69 |
3116 | while (scan < loceol && !isSPACE_LC(*scan)) |
3117 | scan++; |
3118 | break; |
a0ed51b3 |
3119 | case NSPACELUTF8: |
3120 | PL_reg_flags |= RF_tainted; |
3121 | loceol = PL_regeol; |
dfe13c55 |
3122 | while (scan < loceol && !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
a0ed51b3 |
3123 | scan += UTF8SKIP(scan); |
3124 | hardcount++; |
3125 | } |
3126 | break; |
a0d0e21e |
3127 | case DIGIT: |
3128 | while (scan < loceol && isDIGIT(*scan)) |
3129 | scan++; |
3130 | break; |
a0ed51b3 |
3131 | case DIGITUTF8: |
3132 | loceol = PL_regeol; |
dfe13c55 |
3133 | while (scan < loceol && swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
3134 | scan += UTF8SKIP(scan); |
3135 | hardcount++; |
3136 | } |
3137 | break; |
3138 | break; |
a0d0e21e |
3139 | case NDIGIT: |
3140 | while (scan < loceol && !isDIGIT(*scan)) |
3141 | scan++; |
3142 | break; |
a0ed51b3 |
3143 | case NDIGITUTF8: |
3144 | loceol = PL_regeol; |
dfe13c55 |
3145 | while (scan < loceol && !swash_fetch(PL_utf8_digit,(U8*)scan)) { |
a0ed51b3 |
3146 | scan += UTF8SKIP(scan); |
3147 | hardcount++; |
3148 | } |
3149 | break; |
a0d0e21e |
3150 | default: /* Called on something of 0 width. */ |
3151 | break; /* So match right here or not at all. */ |
3152 | } |
a687059c |
3153 | |
a0ed51b3 |
3154 | if (hardcount) |
3155 | c = hardcount; |
3156 | else |
3157 | c = scan - PL_reginput; |
3280af22 |
3158 | PL_reginput = scan; |
a687059c |
3159 | |
c277df42 |
3160 | DEBUG_r( |
3161 | { |
3162 | SV *prop = sv_newmortal(); |
3163 | |
3164 | regprop(prop, p); |
3165 | PerlIO_printf(Perl_debug_log, |
3166 | "%*s %s can match %ld times out of %ld...\n", |
3167 | REPORT_CODE_OFF+1, "", SvPVX(prop),c,max); |
3168 | }); |
3169 | |
a0d0e21e |
3170 | return(c); |
a687059c |
3171 | } |
3172 | |
3173 | /* |
c277df42 |
3174 | - regrepeat_hard - repeatedly match something, report total lenth and length |
3175 | * |
3176 | * The repeater is supposed to have constant length. |
3177 | */ |
3178 | |
76e3520e |
3179 | STATIC I32 |
cea2e8a9 |
3180 | S_regrepeat_hard(pTHX_ regnode *p, I32 max, I32 *lp) |
c277df42 |
3181 | { |
5c0ca799 |
3182 | dTHR; |
c277df42 |
3183 | register char *scan; |
3184 | register char *start; |
3280af22 |
3185 | register char *loceol = PL_regeol; |
a0ed51b3 |
3186 | I32 l = 0; |
708e3b05 |
3187 | I32 count = 0, res = 1; |
a0ed51b3 |
3188 | |
3189 | if (!max) |
3190 | return 0; |
c277df42 |
3191 | |
3280af22 |
3192 | start = PL_reginput; |
a0ed51b3 |
3193 | if (UTF) { |
708e3b05 |
3194 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
3195 | if (!count++) { |
3196 | l = 0; |
3197 | while (start < PL_reginput) { |
3198 | l++; |
3199 | start += UTF8SKIP(start); |
3200 | } |
3201 | *lp = l; |
3202 | if (l == 0) |
3203 | return max; |
3204 | } |
3205 | if (count == max) |
3206 | return count; |
3207 | } |
3208 | } |
3209 | else { |
708e3b05 |
3210 | while (PL_reginput < loceol && (scan = PL_reginput, res = regmatch(p))) { |
a0ed51b3 |
3211 | if (!count++) { |
3212 | *lp = l = PL_reginput - start; |
3213 | if (max != REG_INFTY && l*max < loceol - scan) |
3214 | loceol = scan + l*max; |
3215 | if (l == 0) |
3216 | return max; |
c277df42 |
3217 | } |
3218 | } |
3219 | } |
708e3b05 |
3220 | if (!res) |
3280af22 |
3221 | PL_reginput = scan; |
c277df42 |
3222 | |
a0ed51b3 |
3223 | return count; |
c277df42 |
3224 | } |
3225 | |
3226 | /* |
cb8d8820 |
3227 | - reginclass - determine if a character falls into a character class |
bbce6d69 |
3228 | */ |
3229 | |
76e3520e |
3230 | STATIC bool |
cea2e8a9 |
3231 | S_reginclass(pTHX_ register char *p, register I32 c) |
bbce6d69 |
3232 | { |
5c0ca799 |
3233 | dTHR; |
b8c5462f |
3234 | char flags = ANYOF_FLAGS(p); |
bbce6d69 |
3235 | bool match = FALSE; |
3236 | |
3237 | c &= 0xFF; |
b8c5462f |
3238 | if (ANYOF_BITMAP_TEST(p, c)) |
bbce6d69 |
3239 | match = TRUE; |
3240 | else if (flags & ANYOF_FOLD) { |
3241 | I32 cf; |
3242 | if (flags & ANYOF_LOCALE) { |
3280af22 |
3243 | PL_reg_flags |= RF_tainted; |
22c35a8c |
3244 | cf = PL_fold_locale[c]; |
bbce6d69 |
3245 | } |
3246 | else |
22c35a8c |
3247 | cf = PL_fold[c]; |
b8c5462f |
3248 | if (ANYOF_BITMAP_TEST(p, cf)) |
bbce6d69 |
3249 | match = TRUE; |
3250 | } |
3251 | |
b8c5462f |
3252 | if (!match && (flags & ANYOF_CLASS)) { |
3280af22 |
3253 | PL_reg_flags |= RF_tainted; |
b8c5462f |
3254 | if ( |
3255 | (ANYOF_CLASS_TEST(p, ANYOF_ALNUM) && isALNUM_LC(c)) || |
3256 | (ANYOF_CLASS_TEST(p, ANYOF_NALNUM) && !isALNUM_LC(c)) || |
3257 | (ANYOF_CLASS_TEST(p, ANYOF_SPACE) && isSPACE_LC(c)) || |
3258 | (ANYOF_CLASS_TEST(p, ANYOF_NSPACE) && !isSPACE_LC(c)) || |
3259 | (ANYOF_CLASS_TEST(p, ANYOF_DIGIT) && isDIGIT_LC(c)) || |
3260 | (ANYOF_CLASS_TEST(p, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || |
3261 | (ANYOF_CLASS_TEST(p, ANYOF_ALNUMC) && isALNUMC_LC(c)) || |
3262 | (ANYOF_CLASS_TEST(p, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || |
3263 | (ANYOF_CLASS_TEST(p, ANYOF_ALPHA) && isALPHA_LC(c)) || |
3264 | (ANYOF_CLASS_TEST(p, ANYOF_NALPHA) && !isALPHA_LC(c)) || |
3265 | (ANYOF_CLASS_TEST(p, ANYOF_ASCII) && isASCII(c)) || |
3266 | (ANYOF_CLASS_TEST(p, ANYOF_NASCII) && !isASCII(c)) || |
3267 | (ANYOF_CLASS_TEST(p, ANYOF_CNTRL) && isCNTRL_LC(c)) || |
3268 | (ANYOF_CLASS_TEST(p, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || |
3269 | (ANYOF_CLASS_TEST(p, ANYOF_GRAPH) && isGRAPH_LC(c)) || |
3270 | (ANYOF_CLASS_TEST(p, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || |
3271 | (ANYOF_CLASS_TEST(p, ANYOF_LOWER) && isLOWER_LC(c)) || |
3272 | (ANYOF_CLASS_TEST(p, ANYOF_NLOWER) && !isLOWER_LC(c)) || |
3273 | (ANYOF_CLASS_TEST(p, ANYOF_PRINT) && isPRINT_LC(c)) || |
3274 | (ANYOF_CLASS_TEST(p, ANYOF_NPRINT) && !isPRINT_LC(c)) || |
3275 | (ANYOF_CLASS_TEST(p, ANYOF_PUNCT) && isPUNCT_LC(c)) || |
3276 | (ANYOF_CLASS_TEST(p, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || |
3277 | (ANYOF_CLASS_TEST(p, ANYOF_UPPER) && isUPPER_LC(c)) || |
3278 | (ANYOF_CLASS_TEST(p, ANYOF_NUPPER) && !isUPPER_LC(c)) || |
3279 | (ANYOF_CLASS_TEST(p, ANYOF_XDIGIT) && isXDIGIT(c)) || |
3280 | (ANYOF_CLASS_TEST(p, ANYOF_NXDIGIT) && !isXDIGIT(c)) |
3281 | ) /* How's that for a conditional? */ |
bbce6d69 |
3282 | { |
3283 | match = TRUE; |
3284 | } |
3285 | } |
3286 | |
ae5c130c |
3287 | return (flags & ANYOF_INVERT) ? !match : match; |
bbce6d69 |
3288 | } |
3289 | |
a0ed51b3 |
3290 | STATIC bool |
cea2e8a9 |
3291 | S_reginclassutf8(pTHX_ regnode *f, U8 *p) |
c485e607 |
3292 | { |
3293 | dTHR; |
a0ed51b3 |
3294 | char flags = ARG1(f); |
3295 | bool match = FALSE; |
3296 | SV *sv = (SV*)PL_regdata->data[ARG2(f)]; |
3297 | |
3298 | if (swash_fetch(sv, p)) |
3299 | match = TRUE; |
3300 | else if (flags & ANYOF_FOLD) { |
3301 | I32 cf; |
dfe13c55 |
3302 | U8 tmpbuf[10]; |
a0ed51b3 |
3303 | if (flags & ANYOF_LOCALE) { |
3304 | PL_reg_flags |= RF_tainted; |
3305 | uv_to_utf8(tmpbuf, toLOWER_LC_utf8(p)); |
3306 | } |
3307 | else |
3308 | uv_to_utf8(tmpbuf, toLOWER_utf8(p)); |
3309 | if (swash_fetch(sv, tmpbuf)) |
3310 | match = TRUE; |
3311 | } |
3312 | |
b8c5462f |
3313 | /* UTF8 combined with ANYOF_CLASS is ill-defined. */ |
a0ed51b3 |
3314 | |
3315 | return (flags & ANYOF_INVERT) ? !match : match; |
3316 | } |
161b471a |
3317 | |
dfe13c55 |
3318 | STATIC U8 * |
cea2e8a9 |
3319 | S_reghop(pTHX_ U8 *s, I32 off) |
c485e607 |
3320 | { |
3321 | dTHR; |
a0ed51b3 |
3322 | if (off >= 0) { |
3323 | while (off-- && s < (U8*)PL_regeol) |
3324 | s += UTF8SKIP(s); |
3325 | } |
3326 | else { |
3327 | while (off++) { |
3328 | if (s > (U8*)PL_bostr) { |
3329 | s--; |
3330 | if (*s & 0x80) { |
3331 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
3332 | s--; |
3333 | } /* XXX could check well-formedness here */ |
3334 | } |
3335 | } |
3336 | } |
3337 | return s; |
3338 | } |
161b471a |
3339 | |
dfe13c55 |
3340 | STATIC U8 * |
cea2e8a9 |
3341 | S_reghopmaybe(pTHX_ U8* s, I32 off) |
a0ed51b3 |
3342 | { |
c485e607 |
3343 | dTHR; |
a0ed51b3 |
3344 | if (off >= 0) { |
3345 | while (off-- && s < (U8*)PL_regeol) |
3346 | s += UTF8SKIP(s); |
3347 | if (off >= 0) |
3348 | return 0; |
3349 | } |
3350 | else { |
3351 | while (off++) { |
3352 | if (s > (U8*)PL_bostr) { |
3353 | s--; |
3354 | if (*s & 0x80) { |
3355 | while (s > (U8*)PL_bostr && (*s & 0xc0) == 0x80) |
3356 | s--; |
3357 | } /* XXX could check well-formedness here */ |
3358 | } |
3359 | else |
3360 | break; |
3361 | } |
3362 | if (off <= 0) |
3363 | return 0; |
3364 | } |
3365 | return s; |
3366 | } |
51371543 |
3367 | |
3368 | #ifdef PERL_OBJECT |
3369 | #define NO_XSLOCKS |
3370 | #include "XSUB.h" |
3371 | #endif |
3372 | |
3373 | static void |
3374 | restore_pos(pTHXo_ void *arg) |
3375 | { |
3376 | dTHR; |
3377 | if (PL_reg_eval_set) { |
3378 | if (PL_reg_oldsaved) { |
3379 | PL_reg_re->subbeg = PL_reg_oldsaved; |
3380 | PL_reg_re->sublen = PL_reg_oldsavedlen; |
3381 | RX_MATCH_COPIED_on(PL_reg_re); |
3382 | } |
3383 | PL_reg_magic->mg_len = PL_reg_oldpos; |
3384 | PL_reg_eval_set = 0; |
3385 | PL_curpm = PL_reg_oldcurpm; |
3386 | } |
3387 | } |
3388 | |