Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regexec.c |
2 | */ | |
3 | ||
4 | /* | |
5 | * "One Ring to rule them all, One Ring to find them..." | |
6 | */ | |
7 | ||
61296642 DM |
8 | /* This file contains functions for executing a regular expression. See |
9 | * also regcomp.c which funnily enough, contains functions for compiling | |
166f8a29 | 10 | * a regular expression. |
e4a054ea DM |
11 | * |
12 | * This file is also copied at build time to ext/re/re_exec.c, where | |
13 | * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT. | |
14 | * This causes the main functions to be compiled under new names and with | |
15 | * debugging support added, which makes "use re 'debug'" work. | |
166f8a29 DM |
16 | */ |
17 | ||
a687059c LW |
18 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
19 | * confused with the original package (see point 3 below). Thanks, Henry! | |
20 | */ | |
21 | ||
22 | /* Additional note: this code is very heavily munged from Henry's version | |
23 | * in places. In some spots I've traded clarity for efficiency, so don't | |
24 | * blame Henry for some of the lack of readability. | |
25 | */ | |
26 | ||
e50aee73 AD |
27 | /* The names of the functions have been changed from regcomp and |
28 | * regexec to pregcomp and pregexec in order to avoid conflicts | |
29 | * with the POSIX routines of the same names. | |
30 | */ | |
31 | ||
b9d5759e | 32 | #ifdef PERL_EXT_RE_BUILD |
54df2634 | 33 | #include "re_top.h" |
9041c2e3 | 34 | #endif |
56953603 | 35 | |
a687059c | 36 | /* |
e50aee73 | 37 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c LW |
38 | * |
39 | * Copyright (c) 1986 by University of Toronto. | |
40 | * Written by Henry Spencer. Not derived from licensed software. | |
41 | * | |
42 | * Permission is granted to anyone to use this software for any | |
43 | * purpose on any computer system, and to redistribute it freely, | |
44 | * subject to the following restrictions: | |
45 | * | |
46 | * 1. The author is not responsible for the consequences of use of | |
47 | * this software, no matter how awful, even if they arise | |
48 | * from defects in it. | |
49 | * | |
50 | * 2. The origin of this software must not be misrepresented, either | |
51 | * by explicit claim or by omission. | |
52 | * | |
53 | * 3. Altered versions must be plainly marked as such, and must not | |
54 | * be misrepresented as being the original software. | |
55 | * | |
56 | **** Alterations to Henry's code are... | |
57 | **** | |
4bb101f2 | 58 | **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
fdf8c088 | 59 | **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007 by Larry Wall and others |
a687059c | 60 | **** |
9ef589d8 LW |
61 | **** You may distribute under the terms of either the GNU General Public |
62 | **** License or the Artistic License, as specified in the README file. | |
a687059c LW |
63 | * |
64 | * Beware that some of this code is subtly aware of the way operator | |
65 | * precedence is structured in regular expressions. Serious changes in | |
66 | * regular-expression syntax might require a total rethink. | |
67 | */ | |
68 | #include "EXTERN.h" | |
864dbfa3 | 69 | #define PERL_IN_REGEXEC_C |
a687059c | 70 | #include "perl.h" |
0f5d15d6 | 71 | |
54df2634 NC |
72 | #ifdef PERL_IN_XSUB_RE |
73 | # include "re_comp.h" | |
74 | #else | |
75 | # include "regcomp.h" | |
76 | #endif | |
a687059c | 77 | |
c277df42 IZ |
78 | #define RF_tainted 1 /* tainted information used? */ |
79 | #define RF_warned 2 /* warned about big count? */ | |
faec1544 | 80 | |
ab3bbdeb | 81 | #define RF_utf8 8 /* Pattern contains multibyte chars? */ |
a0ed51b3 | 82 | |
eb160463 | 83 | #define UTF ((PL_reg_flags & RF_utf8) != 0) |
ce862d02 IZ |
84 | |
85 | #define RS_init 1 /* eval environment created */ | |
86 | #define RS_set 2 /* replsv value is set */ | |
c277df42 | 87 | |
a687059c LW |
88 | #ifndef STATIC |
89 | #define STATIC static | |
90 | #endif | |
91 | ||
32fc9b6a | 92 | #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c))) |
7d3e948e | 93 | |
c277df42 IZ |
94 | /* |
95 | * Forwards. | |
96 | */ | |
97 | ||
33b8afdf | 98 | #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv)) |
53c4c00c | 99 | #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b) |
a0ed51b3 | 100 | |
3dab1dad YO |
101 | #define HOPc(pos,off) \ |
102 | (char *)(PL_reg_match_utf8 \ | |
52657f30 | 103 | ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \ |
3dab1dad YO |
104 | : (U8*)(pos + off)) |
105 | #define HOPBACKc(pos, off) \ | |
07be1b83 YO |
106 | (char*)(PL_reg_match_utf8\ |
107 | ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \ | |
108 | : (pos - off >= PL_bostr) \ | |
8e11feef | 109 | ? (U8*)pos - off \ |
3dab1dad | 110 | : NULL) |
efb30f32 | 111 | |
e7409c1b | 112 | #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off)) |
1aa99e6b | 113 | #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) |
1aa99e6b | 114 | |
1a4fad37 AL |
115 | #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \ |
116 | if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END | |
117 | #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a") | |
118 | #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0") | |
119 | #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ") | |
120 | #define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86") | |
51371543 | 121 | |
3dab1dad YO |
122 | /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */ |
123 | ||
5f80c4cf | 124 | /* for use after a quantifier and before an EXACT-like node -- japhy */ |
e2d8ce26 JP |
125 | #define JUMPABLE(rn) ( \ |
126 | OP(rn) == OPEN || OP(rn) == CLOSE || OP(rn) == EVAL || \ | |
cca55fe3 JP |
127 | OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ |
128 | OP(rn) == PLUS || OP(rn) == MINMOD || \ | |
3dab1dad | 129 | (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \ |
e2d8ce26 JP |
130 | ) |
131 | ||
cca55fe3 | 132 | #define HAS_TEXT(rn) ( \ |
3dab1dad | 133 | PL_regkind[OP(rn)] == EXACT || PL_regkind[OP(rn)] == REF \ |
cca55fe3 | 134 | ) |
e2d8ce26 | 135 | |
a84d97b6 HS |
136 | /* |
137 | Search for mandatory following text node; for lookahead, the text must | |
138 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
139 | */ | |
cca55fe3 | 140 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
3dab1dad YO |
141 | while (JUMPABLE(rn)) { \ |
142 | const OPCODE type = OP(rn); \ | |
143 | if (type == SUSPEND || PL_regkind[type] == CURLY) \ | |
e2d8ce26 | 144 | rn = NEXTOPER(NEXTOPER(rn)); \ |
3dab1dad | 145 | else if (type == PLUS) \ |
cca55fe3 | 146 | rn = NEXTOPER(rn); \ |
3dab1dad | 147 | else if (type == IFMATCH) \ |
a84d97b6 | 148 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ |
e2d8ce26 | 149 | else rn += NEXT_OFF(rn); \ |
3dab1dad | 150 | } \ |
5f80c4cf | 151 | } STMT_END |
74750237 | 152 | |
c476f425 | 153 | |
acfe0abc | 154 | static void restore_pos(pTHX_ void *arg); |
51371543 | 155 | |
76e3520e | 156 | STATIC CHECKPOINT |
cea2e8a9 | 157 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 158 | { |
97aff369 | 159 | dVAR; |
a3b680e6 | 160 | const int retval = PL_savestack_ix; |
b1ce53c5 | 161 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 162 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 163 | int p; |
40a82448 | 164 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 165 | |
e49a9654 IH |
166 | if (paren_elems_to_push < 0) |
167 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
168 | ||
6bda09f9 | 169 | #define REGCP_OTHER_ELEMS 8 |
4b3c1a47 | 170 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
7f69552c | 171 | |
3280af22 | 172 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 173 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
cf93c79d IZ |
174 | SSPUSHINT(PL_regendp[p]); |
175 | SSPUSHINT(PL_regstartp[p]); | |
3280af22 | 176 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 177 | SSPUSHINT(p); |
40a82448 DM |
178 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
179 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", | |
180 | (UV)p, (IV)PL_regstartp[p], | |
181 | (IV)(PL_reg_start_tmp[p] - PL_bostr), | |
182 | (IV)PL_regendp[p] | |
183 | )); | |
a0d0e21e | 184 | } |
b1ce53c5 | 185 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
6bda09f9 YO |
186 | SSPUSHPTR(PL_regstartp); |
187 | SSPUSHPTR(PL_regendp); | |
3280af22 NIS |
188 | SSPUSHINT(PL_regsize); |
189 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 190 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 191 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
192 | #define REGCP_FRAME_ELEMS 2 |
193 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
194 | * are needed for the regexp context stack bookkeeping. */ | |
195 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 196 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 197 | |
a0d0e21e LW |
198 | return retval; |
199 | } | |
200 | ||
c277df42 | 201 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
202 | #define REGCP_SET(cp) \ |
203 | DEBUG_STATE_r( \ | |
ab3bbdeb | 204 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 205 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
206 | (IV)PL_savestack_ix)); \ |
207 | cp = PL_savestack_ix | |
c3464db5 | 208 | |
ab3bbdeb | 209 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 210 | DEBUG_STATE_r( \ |
ab3bbdeb | 211 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
212 | PerlIO_printf(Perl_debug_log, \ |
213 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
214 | (IV)(cp), (IV)PL_savestack_ix)); \ |
215 | regcpblow(cp) | |
c277df42 | 216 | |
76e3520e | 217 | STATIC char * |
097eb12c | 218 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 219 | { |
97aff369 | 220 | dVAR; |
3b6647e0 | 221 | U32 i; |
a0d0e21e | 222 | char *input; |
b1ce53c5 | 223 | |
a3621e74 YO |
224 | GET_RE_DEBUG_FLAGS_DECL; |
225 | ||
b1ce53c5 | 226 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 227 | i = SSPOPINT; |
b1ce53c5 JH |
228 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
229 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 230 | input = (char *) SSPOPPTR; |
a01268b5 | 231 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
232 | *PL_reglastparen = SSPOPINT; |
233 | PL_regsize = SSPOPINT; | |
6bda09f9 YO |
234 | PL_regendp=(I32 *) SSPOPPTR; |
235 | PL_regstartp=(I32 *) SSPOPPTR; | |
b1ce53c5 | 236 | |
6bda09f9 | 237 | |
b1ce53c5 | 238 | /* Now restore the parentheses context. */ |
41123dfd JH |
239 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
240 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 241 | I32 tmps; |
097eb12c | 242 | U32 paren = (U32)SSPOPINT; |
3280af22 | 243 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
cf93c79d IZ |
244 | PL_regstartp[paren] = SSPOPINT; |
245 | tmps = SSPOPINT; | |
3280af22 NIS |
246 | if (paren <= *PL_reglastparen) |
247 | PL_regendp[paren] = tmps; | |
a3621e74 | 248 | DEBUG_EXECUTE_r( |
c3464db5 | 249 | PerlIO_printf(Perl_debug_log, |
b900a521 | 250 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
9041c2e3 | 251 | (UV)paren, (IV)PL_regstartp[paren], |
b900a521 | 252 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
9041c2e3 | 253 | (IV)PL_regendp[paren], |
3280af22 | 254 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 255 | ); |
a0d0e21e | 256 | } |
a3621e74 | 257 | DEBUG_EXECUTE_r( |
bb7a0f54 | 258 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 259 | PerlIO_printf(Perl_debug_log, |
faccc32b | 260 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 261 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
262 | } |
263 | ); | |
daf18116 | 264 | #if 1 |
dafc8851 JH |
265 | /* It would seem that the similar code in regtry() |
266 | * already takes care of this, and in fact it is in | |
267 | * a better location to since this code can #if 0-ed out | |
268 | * but the code in regtry() is needed or otherwise tests | |
269 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
270 | * (as of patchlevel 7877) will fail. Then again, |
271 | * this code seems to be necessary or otherwise | |
272 | * building DynaLoader will fail: | |
273 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
274 | * --jhi */ | |
3b6647e0 | 275 | for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { |
097eb12c AL |
276 | if (i > PL_regsize) |
277 | PL_regstartp[i] = -1; | |
278 | PL_regendp[i] = -1; | |
a0d0e21e | 279 | } |
dafc8851 | 280 | #endif |
a0d0e21e LW |
281 | return input; |
282 | } | |
283 | ||
02db2b7b | 284 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 285 | |
a687059c | 286 | /* |
e50aee73 | 287 | * pregexec and friends |
a687059c LW |
288 | */ |
289 | ||
76234dfb | 290 | #ifndef PERL_IN_XSUB_RE |
a687059c | 291 | /* |
c277df42 | 292 | - pregexec - match a regexp against a string |
a687059c | 293 | */ |
c277df42 | 294 | I32 |
864dbfa3 | 295 | Perl_pregexec(pTHX_ register regexp *prog, char *stringarg, register char *strend, |
c3464db5 | 296 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
297 | /* strend: pointer to null at end of string */ |
298 | /* strbeg: real beginning of string */ | |
299 | /* minend: end of match must be >=minend after stringarg. */ | |
300 | /* nosave: For optimizations. */ | |
301 | { | |
302 | return | |
9041c2e3 | 303 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
304 | nosave ? 0 : REXEC_COPY_STR); |
305 | } | |
76234dfb | 306 | #endif |
22e551b9 | 307 | |
9041c2e3 | 308 | /* |
cad2e5aa JH |
309 | * Need to implement the following flags for reg_anch: |
310 | * | |
311 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
312 | * USE_INTUIT_ML | |
313 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
314 | * INTUIT_AUTORITATIVE_ML | |
315 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
316 | * INTUIT_ONCE_ML | |
317 | * | |
318 | * Another flag for this function: SECOND_TIME (so that float substrs | |
319 | * with giant delta may be not rechecked). | |
320 | */ | |
321 | ||
322 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
323 | ||
3f7c398e | 324 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
325 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
326 | ||
327 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
328 | ||
6eb5f6b9 JH |
329 | /* XXXX Some places assume that there is a fixed substring. |
330 | An update may be needed if optimizer marks as "INTUITable" | |
331 | RExen without fixed substrings. Similarly, it is assumed that | |
332 | lengths of all the strings are no more than minlen, thus they | |
333 | cannot come from lookahead. | |
40d049e4 YO |
334 | (Or minlen should take into account lookahead.) |
335 | NOTE: Some of this comment is not correct. minlen does now take account | |
336 | of lookahead/behind. Further research is required. -- demerphq | |
337 | ||
338 | */ | |
6eb5f6b9 | 339 | |
2c2d71f5 JH |
340 | /* A failure to find a constant substring means that there is no need to make |
341 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
342 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
343 | regtry() should be needed. |
344 | ||
345 | REx compiler's optimizer found 4 possible hints: | |
346 | a) Anchored substring; | |
347 | b) Fixed substring; | |
348 | c) Whether we are anchored (beginning-of-line or \G); | |
349 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 350 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
351 | string which does not contradict any of them. |
352 | */ | |
2c2d71f5 | 353 | |
6eb5f6b9 JH |
354 | /* Most of decisions we do here should have been done at compile time. |
355 | The nodes of the REx which we used for the search should have been | |
356 | deleted from the finite automaton. */ | |
357 | ||
cad2e5aa JH |
358 | char * |
359 | Perl_re_intuit_start(pTHX_ regexp *prog, SV *sv, char *strpos, | |
360 | char *strend, U32 flags, re_scream_pos_data *data) | |
361 | { | |
97aff369 | 362 | dVAR; |
b7953727 | 363 | register I32 start_shift = 0; |
cad2e5aa | 364 | /* Should be nonnegative! */ |
b7953727 | 365 | register I32 end_shift = 0; |
2c2d71f5 JH |
366 | register char *s; |
367 | register SV *check; | |
a1933d95 | 368 | char *strbeg; |
cad2e5aa | 369 | char *t; |
1de06328 | 370 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 371 | I32 ml_anch; |
bd61b366 SS |
372 | register char *other_last = NULL; /* other substr checked before this */ |
373 | char *check_at = NULL; /* check substr found at this pos */ | |
bbe252da | 374 | const I32 multiline = prog->extflags & RXf_PMf_MULTILINE; |
f8fc2ecf | 375 | RXi_GET_DECL(prog,progi); |
30944b6d | 376 | #ifdef DEBUGGING |
890ce7af | 377 | const char * const i_strpos = strpos; |
30944b6d | 378 | #endif |
a3621e74 YO |
379 | |
380 | GET_RE_DEBUG_FLAGS_DECL; | |
381 | ||
a30b2f1f | 382 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 383 | |
bbe252da | 384 | if (prog->extflags & RXf_UTF8) { |
b8d68ded JH |
385 | PL_reg_flags |= RF_utf8; |
386 | } | |
ab3bbdeb YO |
387 | DEBUG_EXECUTE_r( |
388 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
389 | sv ? "Guessing start of match in sv for" |
390 | : "Guessing start of match in string for"); | |
2a782b5b | 391 | ); |
cad2e5aa | 392 | |
c344f387 JH |
393 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
394 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 395 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 396 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 397 | goto fail; |
2c2d71f5 | 398 | } |
1de06328 | 399 | |
a1933d95 | 400 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 401 | PL_regeol = strend; |
33b8afdf JH |
402 | if (do_utf8) { |
403 | if (!prog->check_utf8 && prog->check_substr) | |
404 | to_utf8_substr(prog); | |
405 | check = prog->check_utf8; | |
406 | } else { | |
407 | if (!prog->check_substr && prog->check_utf8) | |
408 | to_byte_substr(prog); | |
409 | check = prog->check_substr; | |
410 | } | |
1de06328 | 411 | if (check == &PL_sv_undef) { |
a3621e74 | 412 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 413 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
414 | goto fail; |
415 | } | |
bbe252da YO |
416 | if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */ |
417 | ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE) | |
418 | || ( (prog->extflags & RXf_ANCH_BOL) | |
7fba1cd6 | 419 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 420 | |
7e25d62c | 421 | if (!ml_anch) { |
bbe252da YO |
422 | if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */ |
423 | && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */ | |
3f7c398e | 424 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
425 | && sv && !SvROK(sv) |
426 | && (strpos != strbeg)) { | |
a3621e74 | 427 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
428 | goto fail; |
429 | } | |
430 | if (prog->check_offset_min == prog->check_offset_max && | |
bbe252da | 431 | !(prog->extflags & RXf_CANY_SEEN)) { |
2c2d71f5 | 432 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
433 | I32 slen; |
434 | ||
1aa99e6b | 435 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 436 | |
653099ff GS |
437 | if (SvTAIL(check)) { |
438 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 439 | |
9041c2e3 | 440 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 441 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 442 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 443 | goto fail_finish; |
cad2e5aa JH |
444 | } |
445 | /* Now should match s[0..slen-2] */ | |
446 | slen--; | |
3f7c398e | 447 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 448 | || (slen > 1 |
3f7c398e | 449 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 450 | report_neq: |
a3621e74 | 451 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
452 | goto fail_finish; |
453 | } | |
cad2e5aa | 454 | } |
3f7c398e | 455 | else if (*SvPVX_const(check) != *s |
653099ff | 456 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 457 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 458 | goto report_neq; |
c315bfe8 | 459 | check_at = s; |
2c2d71f5 | 460 | goto success_at_start; |
7e25d62c | 461 | } |
cad2e5aa | 462 | } |
2c2d71f5 | 463 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 464 | s = strpos; |
2c2d71f5 | 465 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
466 | end_shift = prog->check_end_shift; |
467 | ||
2c2d71f5 | 468 | if (!ml_anch) { |
a3b680e6 | 469 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 470 | - (SvTAIL(check) != 0); |
a3b680e6 | 471 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
472 | |
473 | if (end_shift < eshift) | |
474 | end_shift = eshift; | |
475 | } | |
cad2e5aa | 476 | } |
2c2d71f5 | 477 | else { /* Can match at random position */ |
cad2e5aa JH |
478 | ml_anch = 0; |
479 | s = strpos; | |
1de06328 YO |
480 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
481 | end_shift = prog->check_end_shift; | |
482 | ||
483 | /* end shift should be non negative here */ | |
cad2e5aa JH |
484 | } |
485 | ||
2c2d71f5 | 486 | #ifdef DEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 487 | if (end_shift < 0) |
1de06328 | 488 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 489 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
490 | #endif |
491 | ||
2c2d71f5 JH |
492 | restart: |
493 | /* Find a possible match in the region s..strend by looking for | |
494 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
495 | |
496 | { | |
497 | I32 srch_start_shift = start_shift; | |
498 | I32 srch_end_shift = end_shift; | |
499 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
500 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
501 | srch_start_shift = strbeg - s; | |
502 | } | |
6bda09f9 | 503 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
504 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
505 | (IV)prog->check_offset_min, | |
506 | (IV)srch_start_shift, | |
507 | (IV)srch_end_shift, | |
508 | (IV)prog->check_end_shift); | |
509 | }); | |
510 | ||
cad2e5aa | 511 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 512 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 513 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 514 | |
2c2d71f5 JH |
515 | if (PL_screamfirst[BmRARE(check)] >= 0 |
516 | || ( BmRARE(check) == '\n' | |
85c508c3 | 517 | && (BmPREVIOUS(check) == SvCUR(check) - 1) |
2c2d71f5 | 518 | && SvTAIL(check) )) |
9041c2e3 | 519 | s = screaminstr(sv, check, |
1de06328 | 520 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 521 | else |
2c2d71f5 | 522 | goto fail_finish; |
4addbd3b HS |
523 | /* we may be pointing at the wrong string */ |
524 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 525 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
526 | if (data) |
527 | *data->scream_olds = s; | |
528 | } | |
1de06328 YO |
529 | else { |
530 | U8* start_point; | |
531 | U8* end_point; | |
bbe252da | 532 | if (prog->extflags & RXf_CANY_SEEN) { |
1de06328 YO |
533 | start_point= (U8*)(s + srch_start_shift); |
534 | end_point= (U8*)(strend - srch_end_shift); | |
535 | } else { | |
536 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
537 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
538 | } | |
6bda09f9 | 539 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 540 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 541 | (int)(end_point - start_point), |
fc8cd66c | 542 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
543 | start_point); |
544 | }); | |
545 | ||
546 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 547 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
548 | } |
549 | } | |
cad2e5aa JH |
550 | /* Update the count-of-usability, remove useless subpatterns, |
551 | unshift s. */ | |
2c2d71f5 | 552 | |
ab3bbdeb YO |
553 | DEBUG_EXECUTE_r({ |
554 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
555 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
556 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 557 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
558 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
559 | ? "anchored" : "floating"), | |
560 | quoted, | |
561 | RE_SV_TAIL(check), | |
562 | (s ? " at offset " : "...\n") ); | |
563 | }); | |
2c2d71f5 JH |
564 | |
565 | if (!s) | |
566 | goto fail_finish; | |
2c2d71f5 | 567 | /* Finish the diagnostic message */ |
a3621e74 | 568 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 569 | |
1de06328 YO |
570 | /* XXX dmq: first branch is for positive lookbehind... |
571 | Our check string is offset from the beginning of the pattern. | |
572 | So we need to do any stclass tests offset forward from that | |
573 | point. I think. :-( | |
574 | */ | |
575 | ||
576 | ||
577 | ||
578 | check_at=s; | |
579 | ||
580 | ||
2c2d71f5 JH |
581 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
582 | Start with the other substr. | |
583 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 584 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
585 | *always* match. Probably should be marked during compile... |
586 | Probably it is right to do no SCREAM here... | |
587 | */ | |
588 | ||
1de06328 YO |
589 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
590 | : (prog->float_substr && prog->anchored_substr)) | |
591 | { | |
30944b6d | 592 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
593 | /* XXXX May be hopelessly wrong for UTF... */ |
594 | if (!other_last) | |
6eb5f6b9 | 595 | other_last = strpos; |
33b8afdf | 596 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
597 | do_other_anchored: |
598 | { | |
890ce7af AL |
599 | char * const last = HOP3c(s, -start_shift, strbeg); |
600 | char *last1, *last2; | |
be8e71aa | 601 | char * const saved_s = s; |
33b8afdf | 602 | SV* must; |
2c2d71f5 | 603 | |
2c2d71f5 JH |
604 | t = s - prog->check_offset_max; |
605 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 606 | && (!do_utf8 |
0ce71af7 | 607 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 608 | && t > strpos))) |
6f207bd3 | 609 | NOOP; |
2c2d71f5 JH |
610 | else |
611 | t = strpos; | |
1aa99e6b | 612 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
613 | if (t < other_last) /* These positions already checked */ |
614 | t = other_last; | |
1aa99e6b | 615 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
616 | if (last < last1) |
617 | last1 = last; | |
1de06328 YO |
618 | /* XXXX It is not documented what units *_offsets are in. |
619 | We assume bytes, but this is clearly wrong. | |
620 | Meaning this code needs to be carefully reviewed for errors. | |
621 | dmq. | |
622 | */ | |
623 | ||
2c2d71f5 | 624 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
625 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
626 | if (must == &PL_sv_undef) { | |
627 | s = (char*)NULL; | |
1de06328 | 628 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
629 | } |
630 | else | |
631 | s = fbm_instr( | |
632 | (unsigned char*)t, | |
633 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
634 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
635 | must, | |
7fba1cd6 | 636 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 637 | ); |
ab3bbdeb YO |
638 | DEBUG_EXECUTE_r({ |
639 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
640 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
641 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 642 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
643 | quoted, RE_SV_TAIL(must)); |
644 | }); | |
645 | ||
646 | ||
2c2d71f5 JH |
647 | if (!s) { |
648 | if (last1 >= last2) { | |
a3621e74 | 649 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
650 | ", giving up...\n")); |
651 | goto fail_finish; | |
652 | } | |
a3621e74 | 653 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 654 | ", trying floating at offset %ld...\n", |
be8e71aa | 655 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
656 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
657 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
658 | goto restart; |
659 | } | |
660 | else { | |
a3621e74 | 661 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 662 | (long)(s - i_strpos))); |
1aa99e6b IH |
663 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
664 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 665 | s = saved_s; |
2c2d71f5 JH |
666 | if (t == strpos) |
667 | goto try_at_start; | |
2c2d71f5 JH |
668 | goto try_at_offset; |
669 | } | |
30944b6d | 670 | } |
2c2d71f5 JH |
671 | } |
672 | else { /* Take into account the floating substring. */ | |
33b8afdf | 673 | char *last, *last1; |
be8e71aa | 674 | char * const saved_s = s; |
33b8afdf JH |
675 | SV* must; |
676 | ||
677 | t = HOP3c(s, -start_shift, strbeg); | |
678 | last1 = last = | |
679 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
680 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
681 | last = HOP3c(t, prog->float_max_offset, strend); | |
682 | s = HOP3c(t, prog->float_min_offset, strend); | |
683 | if (s < other_last) | |
684 | s = other_last; | |
2c2d71f5 | 685 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
686 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
687 | /* fbm_instr() takes into account exact value of end-of-str | |
688 | if the check is SvTAIL(ed). Since false positives are OK, | |
689 | and end-of-str is not later than strend we are OK. */ | |
690 | if (must == &PL_sv_undef) { | |
691 | s = (char*)NULL; | |
1de06328 | 692 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
693 | } |
694 | else | |
2c2d71f5 | 695 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
696 | (unsigned char*)last + SvCUR(must) |
697 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 698 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
699 | DEBUG_EXECUTE_r({ |
700 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
701 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
702 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 703 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
704 | quoted, RE_SV_TAIL(must)); |
705 | }); | |
33b8afdf JH |
706 | if (!s) { |
707 | if (last1 == last) { | |
a3621e74 | 708 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
709 | ", giving up...\n")); |
710 | goto fail_finish; | |
2c2d71f5 | 711 | } |
a3621e74 | 712 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 713 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 714 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
715 | other_last = last; |
716 | s = HOP3c(t, 1, strend); | |
717 | goto restart; | |
718 | } | |
719 | else { | |
a3621e74 | 720 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
721 | (long)(s - i_strpos))); |
722 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 723 | s = saved_s; |
33b8afdf JH |
724 | if (t == strpos) |
725 | goto try_at_start; | |
726 | goto try_at_offset; | |
727 | } | |
2c2d71f5 | 728 | } |
cad2e5aa | 729 | } |
2c2d71f5 | 730 | |
1de06328 | 731 | |
9ef43ace | 732 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 733 | |
6bda09f9 | 734 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
735 | PerlIO_printf(Perl_debug_log, |
736 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
737 | (IV)prog->check_offset_min, | |
738 | (IV)prog->check_offset_max, | |
739 | (IV)(s-strpos), | |
740 | (IV)(t-strpos), | |
741 | (IV)(t-s), | |
742 | (IV)(strend-strpos) | |
743 | ) | |
744 | ); | |
745 | ||
2c2d71f5 | 746 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 747 | && (!do_utf8 |
9ef43ace | 748 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
749 | && t > strpos))) |
750 | { | |
2c2d71f5 JH |
751 | /* Fixed substring is found far enough so that the match |
752 | cannot start at strpos. */ | |
753 | try_at_offset: | |
cad2e5aa | 754 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
755 | /* Eventually fbm_*() should handle this, but often |
756 | anchored_offset is not 0, so this check will not be wasted. */ | |
757 | /* XXXX In the code below we prefer to look for "^" even in | |
758 | presence of anchored substrings. And we search even | |
759 | beyond the found float position. These pessimizations | |
760 | are historical artefacts only. */ | |
761 | find_anchor: | |
2c2d71f5 | 762 | while (t < strend - prog->minlen) { |
cad2e5aa | 763 | if (*t == '\n') { |
4ee3650e | 764 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 765 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
766 | /* Since we moved from the found position, |
767 | we definitely contradict the found anchored | |
30944b6d IZ |
768 | substr. Due to the above check we do not |
769 | contradict "check" substr. | |
770 | Thus we can arrive here only if check substr | |
771 | is float. Redo checking for "other"=="fixed". | |
772 | */ | |
9041c2e3 | 773 | strpos = t + 1; |
a3621e74 | 774 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 775 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
776 | goto do_other_anchored; |
777 | } | |
4ee3650e GS |
778 | /* We don't contradict the found floating substring. */ |
779 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 780 | s = t + 1; |
a3621e74 | 781 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 782 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
783 | goto set_useful; |
784 | } | |
4ee3650e GS |
785 | /* Position contradicts check-string */ |
786 | /* XXXX probably better to look for check-string | |
787 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 788 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 789 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 790 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
791 | goto restart; |
792 | } | |
793 | t++; | |
794 | } | |
a3621e74 | 795 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 796 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 797 | goto fail_finish; |
cad2e5aa | 798 | } |
f5952150 | 799 | else { |
a3621e74 | 800 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 801 | PL_colors[0], PL_colors[1])); |
f5952150 | 802 | } |
cad2e5aa JH |
803 | s = t; |
804 | set_useful: | |
33b8afdf | 805 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
806 | } |
807 | else { | |
f5952150 | 808 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 809 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
810 | unless it was an MBOL and we are not after MBOL, |
811 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
812 | try_at_start: |
813 | /* Even in this situation we may use MBOL flag if strpos is offset | |
814 | wrt the start of the string. */ | |
05b4157f | 815 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 816 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d | 817 | /* May be due to an implicit anchor of m{.*foo} */ |
bbe252da | 818 | && !(prog->intflags & PREGf_IMPLICIT)) |
d506a20d | 819 | { |
cad2e5aa JH |
820 | t = strpos; |
821 | goto find_anchor; | |
822 | } | |
a3621e74 | 823 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 824 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
70685ca0 | 825 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 826 | ); |
2c2d71f5 | 827 | success_at_start: |
bbe252da | 828 | if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
829 | && (do_utf8 ? ( |
830 | prog->check_utf8 /* Could be deleted already */ | |
831 | && --BmUSEFUL(prog->check_utf8) < 0 | |
832 | && (prog->check_utf8 == prog->float_utf8) | |
833 | ) : ( | |
834 | prog->check_substr /* Could be deleted already */ | |
835 | && --BmUSEFUL(prog->check_substr) < 0 | |
836 | && (prog->check_substr == prog->float_substr) | |
837 | ))) | |
66e933ab | 838 | { |
cad2e5aa | 839 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 840 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
841 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
842 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
843 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
844 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
845 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
846 | check = NULL; /* abort */ | |
cad2e5aa | 847 | s = strpos; |
3cf5c195 IZ |
848 | /* XXXX This is a remnant of the old implementation. It |
849 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 850 | other heuristics. */ |
bbe252da | 851 | prog->extflags &= ~RXf_USE_INTUIT; |
cad2e5aa JH |
852 | } |
853 | else | |
854 | s = strpos; | |
855 | } | |
856 | ||
6eb5f6b9 JH |
857 | /* Last resort... */ |
858 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
859 | /* trie stclasses are too expensive to use here, we are better off to |
860 | leave it to regmatch itself */ | |
f8fc2ecf | 861 | if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
862 | /* minlen == 0 is possible if regstclass is \b or \B, |
863 | and the fixed substr is ''$. | |
864 | Since minlen is already taken into account, s+1 is before strend; | |
865 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
866 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
867 | regstclass does not come from lookahead... */ | |
868 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
869 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
f8fc2ecf YO |
870 | const U8* const str = (U8*)STRING(progi->regstclass); |
871 | const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT | |
872 | ? CHR_DIST(str+STR_LEN(progi->regstclass), str) | |
66e933ab | 873 | : 1); |
1de06328 YO |
874 | char * endpos; |
875 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
876 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
877 | else if (prog->float_substr || prog->float_utf8) | |
878 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
879 | else | |
880 | endpos= strend; | |
881 | ||
70685ca0 JH |
882 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n", |
883 | (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg))); | |
1de06328 | 884 | |
6eb5f6b9 | 885 | t = s; |
f8fc2ecf | 886 | s = find_byclass(prog, progi->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
887 | if (!s) { |
888 | #ifdef DEBUGGING | |
cbbf8932 | 889 | const char *what = NULL; |
6eb5f6b9 JH |
890 | #endif |
891 | if (endpos == strend) { | |
a3621e74 | 892 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
893 | "Could not match STCLASS...\n") ); |
894 | goto fail; | |
895 | } | |
a3621e74 | 896 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 897 | "This position contradicts STCLASS...\n") ); |
bbe252da | 898 | if ((prog->extflags & RXf_ANCH) && !ml_anch) |
653099ff | 899 | goto fail; |
6eb5f6b9 | 900 | /* Contradict one of substrings */ |
33b8afdf JH |
901 | if (prog->anchored_substr || prog->anchored_utf8) { |
902 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 903 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 904 | hop_and_restart: |
1aa99e6b | 905 | s = HOP3c(t, 1, strend); |
66e933ab GS |
906 | if (s + start_shift + end_shift > strend) { |
907 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 908 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
909 | "Could not match STCLASS...\n") ); |
910 | goto fail; | |
911 | } | |
5e39e1e5 HS |
912 | if (!check) |
913 | goto giveup; | |
a3621e74 | 914 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 915 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
916 | what, (long)(s + start_shift - i_strpos)) ); |
917 | goto restart; | |
918 | } | |
66e933ab | 919 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
920 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
921 | goto retry_floating_check; | |
922 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 923 | s = check_at; |
5e39e1e5 HS |
924 | if (!check) |
925 | goto giveup; | |
a3621e74 | 926 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 927 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
928 | (long)(other_last - i_strpos)) ); |
929 | goto do_other_anchored; | |
930 | } | |
60e71179 GS |
931 | /* Another way we could have checked stclass at the |
932 | current position only: */ | |
933 | if (ml_anch) { | |
934 | s = t = t + 1; | |
5e39e1e5 HS |
935 | if (!check) |
936 | goto giveup; | |
a3621e74 | 937 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 938 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 939 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 940 | goto try_at_offset; |
66e933ab | 941 | } |
33b8afdf | 942 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 943 | goto fail; |
6eb5f6b9 JH |
944 | /* Check is floating subtring. */ |
945 | retry_floating_check: | |
946 | t = check_at - start_shift; | |
a3621e74 | 947 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
948 | goto hop_and_restart; |
949 | } | |
b7953727 | 950 | if (t != s) { |
a3621e74 | 951 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 952 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
953 | (long)(t - i_strpos), (long)(s - i_strpos)) |
954 | ); | |
955 | } | |
956 | else { | |
a3621e74 | 957 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
958 | "Does not contradict STCLASS...\n"); |
959 | ); | |
960 | } | |
6eb5f6b9 | 961 | } |
5e39e1e5 | 962 | giveup: |
a3621e74 | 963 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
964 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
965 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 966 | return s; |
2c2d71f5 JH |
967 | |
968 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
969 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
970 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 971 | fail: |
a3621e74 | 972 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 973 | PL_colors[4], PL_colors[5])); |
bd61b366 | 974 | return NULL; |
cad2e5aa | 975 | } |
9661b544 | 976 | |
786e8c11 | 977 | |
3b0527fe | 978 | |
55eed653 NC |
979 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ |
980 | uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ | |
4cadc6a9 YO |
981 | switch (trie_type) { \ |
982 | case trie_utf8_fold: \ | |
983 | if ( foldlen>0 ) { \ | |
984 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
985 | foldlen -= len; \ | |
986 | uscan += len; \ | |
987 | len=0; \ | |
988 | } else { \ | |
989 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
990 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
991 | foldlen -= UNISKIP( uvc ); \ | |
992 | uscan = foldbuf + UNISKIP( uvc ); \ | |
993 | } \ | |
994 | break; \ | |
995 | case trie_utf8: \ | |
996 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
997 | break; \ | |
998 | case trie_plain: \ | |
999 | uvc = (UV)*uc; \ | |
1000 | len = 1; \ | |
1001 | } \ | |
1002 | \ | |
1003 | if (uvc < 256) { \ | |
1004 | charid = trie->charmap[ uvc ]; \ | |
1005 | } \ | |
1006 | else { \ | |
1007 | charid = 0; \ | |
55eed653 NC |
1008 | if (widecharmap) { \ |
1009 | SV** const svpp = hv_fetch(widecharmap, \ | |
4cadc6a9 YO |
1010 | (char*)&uvc, sizeof(UV), 0); \ |
1011 | if (svpp) \ | |
1012 | charid = (U16)SvIV(*svpp); \ | |
1013 | } \ | |
1014 | } \ | |
1015 | } STMT_END | |
1016 | ||
1017 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1018 | if ( (CoNd) \ | |
1019 | && (ln == len || \ | |
1020 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1021 | m, NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1022 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1023 | goto got_it; \ |
1024 | else { \ | |
1025 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1026 | uvchr_to_utf8(tmpbuf, c); \ | |
1027 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1028 | if ( f != c \ | |
1029 | && (f == c1 || f == c2) \ | |
1030 | && (ln == foldlen || \ | |
1031 | !ibcmp_utf8((char *) foldbuf, \ | |
1032 | NULL, foldlen, do_utf8, \ | |
1033 | m, \ | |
1034 | NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1035 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1036 | goto got_it; \ |
1037 | } \ | |
1038 | s += len | |
1039 | ||
1040 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1041 | STMT_START { \ | |
1042 | while (s <= e) { \ | |
1043 | if ( (CoNd) \ | |
1044 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1045 | ? ibcmp(s, m, ln) \ | |
1046 | : ibcmp_locale(s, m, ln))) \ | |
24b23f37 | 1047 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1048 | goto got_it; \ |
1049 | s++; \ | |
1050 | } \ | |
1051 | } STMT_END | |
1052 | ||
1053 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1054 | STMT_START { \ | |
1055 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1056 | CoDe \ | |
1057 | s += uskip; \ | |
1058 | } \ | |
1059 | } STMT_END | |
1060 | ||
1061 | #define REXEC_FBC_SCAN(CoDe) \ | |
1062 | STMT_START { \ | |
1063 | while (s < strend) { \ | |
1064 | CoDe \ | |
1065 | s++; \ | |
1066 | } \ | |
1067 | } STMT_END | |
1068 | ||
1069 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1070 | REXEC_FBC_UTF8_SCAN( \ | |
1071 | if (CoNd) { \ | |
24b23f37 | 1072 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1073 | goto got_it; \ |
1074 | else \ | |
1075 | tmp = doevery; \ | |
1076 | } \ | |
1077 | else \ | |
1078 | tmp = 1; \ | |
1079 | ) | |
1080 | ||
1081 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1082 | REXEC_FBC_SCAN( \ | |
1083 | if (CoNd) { \ | |
24b23f37 | 1084 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1085 | goto got_it; \ |
1086 | else \ | |
1087 | tmp = doevery; \ | |
1088 | } \ | |
1089 | else \ | |
1090 | tmp = 1; \ | |
1091 | ) | |
1092 | ||
1093 | #define REXEC_FBC_TRYIT \ | |
24b23f37 | 1094 | if ((!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1095 | goto got_it |
1096 | ||
1097 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ | |
1098 | if (do_utf8) { \ | |
1099 | UtFpReLoAd; \ | |
1100 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1101 | } \ | |
1102 | else { \ | |
1103 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1104 | } \ | |
1105 | break | |
1106 | ||
1107 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1108 | PL_reg_flags |= RF_tainted; \ | |
1109 | if (do_utf8) { \ | |
1110 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1111 | } \ | |
1112 | else { \ | |
1113 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1114 | } \ | |
1115 | break | |
1116 | ||
786e8c11 YO |
1117 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1118 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1119 | ||
1120 | /* We know what class REx starts with. Try to find this position... */ | |
1121 | /* if reginfo is NULL, its a dryrun */ | |
1122 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1123 | in regmatch. /grrr */ | |
1124 | ||
3c3eec57 | 1125 | STATIC char * |
07be1b83 | 1126 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
24b23f37 | 1127 | const char *strend, regmatch_info *reginfo) |
a687059c | 1128 | { |
27da23d5 | 1129 | dVAR; |
bbe252da | 1130 | const I32 doevery = (prog->intflags & PREGf_SKIP) == 0; |
6eb5f6b9 | 1131 | char *m; |
d8093b23 | 1132 | STRLEN ln; |
5dab1207 | 1133 | STRLEN lnc; |
078c425b | 1134 | register STRLEN uskip; |
d8093b23 G |
1135 | unsigned int c1; |
1136 | unsigned int c2; | |
6eb5f6b9 JH |
1137 | char *e; |
1138 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1139 | register const bool do_utf8 = PL_reg_match_utf8; |
f8fc2ecf YO |
1140 | RXi_GET_DECL(prog,progi); |
1141 | ||
6eb5f6b9 JH |
1142 | /* We know what class it must start with. */ |
1143 | switch (OP(c)) { | |
6eb5f6b9 | 1144 | case ANYOF: |
388cc4de | 1145 | if (do_utf8) { |
4cadc6a9 | 1146 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1147 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1148 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1149 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1150 | } |
1151 | else { | |
1152 | while (s < strend) { | |
1153 | STRLEN skip = 1; | |
1154 | ||
32fc9b6a | 1155 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1156 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1157 | /* The assignment of 2 is intentional: | |
1158 | * for the folded sharp s, the skip is 2. */ | |
1159 | (skip = SHARP_S_SKIP))) { | |
24b23f37 | 1160 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
388cc4de HS |
1161 | goto got_it; |
1162 | else | |
1163 | tmp = doevery; | |
1164 | } | |
1165 | else | |
1166 | tmp = 1; | |
1167 | s += skip; | |
1168 | } | |
a0d0e21e | 1169 | } |
6eb5f6b9 | 1170 | break; |
f33976b4 | 1171 | case CANY: |
4cadc6a9 | 1172 | REXEC_FBC_SCAN( |
24b23f37 | 1173 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
f33976b4 DB |
1174 | goto got_it; |
1175 | else | |
1176 | tmp = doevery; | |
4cadc6a9 | 1177 | ); |
f33976b4 | 1178 | break; |
6eb5f6b9 | 1179 | case EXACTF: |
5dab1207 NIS |
1180 | m = STRING(c); |
1181 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1182 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1183 | if (UTF) { |
a2a2844f | 1184 | STRLEN ulen1, ulen2; |
5dab1207 | 1185 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1186 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1187 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1188 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1189 | |
1190 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1191 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1192 | ||
89ebb4a3 | 1193 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1194 | 0, uniflags); |
89ebb4a3 | 1195 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1196 | 0, uniflags); |
5dab1207 NIS |
1197 | lnc = 0; |
1198 | while (sm < ((U8 *) m + ln)) { | |
1199 | lnc++; | |
1200 | sm += UTF8SKIP(sm); | |
1201 | } | |
1aa99e6b IH |
1202 | } |
1203 | else { | |
1204 | c1 = *(U8*)m; | |
1205 | c2 = PL_fold[c1]; | |
1206 | } | |
6eb5f6b9 JH |
1207 | goto do_exactf; |
1208 | case EXACTFL: | |
5dab1207 NIS |
1209 | m = STRING(c); |
1210 | ln = STR_LEN(c); | |
1211 | lnc = (I32) ln; | |
d8093b23 | 1212 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1213 | c2 = PL_fold_locale[c1]; |
1214 | do_exactf: | |
db12adc6 | 1215 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1216 | |
3b0527fe | 1217 | if (!reginfo && e < s) |
6eb5f6b9 | 1218 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1219 | |
60a8b682 JH |
1220 | /* The idea in the EXACTF* cases is to first find the |
1221 | * first character of the EXACTF* node and then, if | |
1222 | * necessary, case-insensitively compare the full | |
1223 | * text of the node. The c1 and c2 are the first | |
1224 | * characters (though in Unicode it gets a bit | |
1225 | * more complicated because there are more cases | |
7f16dd3d JH |
1226 | * than just upper and lower: one needs to use |
1227 | * the so-called folding case for case-insensitive | |
1228 | * matching (called "loose matching" in Unicode). | |
1229 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1230 | |
1aa99e6b | 1231 | if (do_utf8) { |
575cac57 | 1232 | UV c, f; |
89ebb4a3 | 1233 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1234 | STRLEN len, foldlen; |
4ad0818d | 1235 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1236 | if (c1 == c2) { |
5dab1207 NIS |
1237 | /* Upper and lower of 1st char are equal - |
1238 | * probably not a "letter". */ | |
1aa99e6b | 1239 | while (s <= e) { |
89ebb4a3 | 1240 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1241 | uniflags); |
4cadc6a9 | 1242 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1243 | } |
09091399 JH |
1244 | } |
1245 | else { | |
1aa99e6b | 1246 | while (s <= e) { |
89ebb4a3 | 1247 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1248 | uniflags); |
80aecb99 | 1249 | |
60a8b682 | 1250 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1251 | * Note that not all the possible combinations |
1252 | * are handled here: some of them are handled | |
1253 | * by the standard folding rules, and some of | |
1254 | * them (the character class or ANYOF cases) | |
1255 | * are handled during compiletime in | |
1256 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1257 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1258 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1259 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1260 | |
4cadc6a9 | 1261 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1262 | } |
09091399 | 1263 | } |
1aa99e6b IH |
1264 | } |
1265 | else { | |
1266 | if (c1 == c2) | |
4cadc6a9 | 1267 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1268 | else |
4cadc6a9 | 1269 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1270 | } |
1271 | break; | |
bbce6d69 | 1272 | case BOUNDL: |
3280af22 | 1273 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1274 | /* FALL THROUGH */ |
a0d0e21e | 1275 | case BOUND: |
ffc61ed2 | 1276 | if (do_utf8) { |
12d33761 | 1277 | if (s == PL_bostr) |
ffc61ed2 JH |
1278 | tmp = '\n'; |
1279 | else { | |
6136c704 | 1280 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1281 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1282 | } |
1283 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1284 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1285 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1286 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1287 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1288 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1289 | isALNUM_LC_utf8((U8*)s))) |
1290 | { | |
1291 | tmp = !tmp; | |
4cadc6a9 | 1292 | REXEC_FBC_TRYIT; |
a687059c | 1293 | } |
4cadc6a9 | 1294 | ); |
a0d0e21e | 1295 | } |
667bb95a | 1296 | else { |
12d33761 | 1297 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1298 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1299 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1300 | if (tmp == |
1301 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1302 | tmp = !tmp; | |
4cadc6a9 | 1303 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1304 | } |
4cadc6a9 | 1305 | ); |
a0ed51b3 | 1306 | } |
24b23f37 | 1307 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1308 | goto got_it; |
1309 | break; | |
bbce6d69 | 1310 | case NBOUNDL: |
3280af22 | 1311 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1312 | /* FALL THROUGH */ |
a0d0e21e | 1313 | case NBOUND: |
ffc61ed2 | 1314 | if (do_utf8) { |
12d33761 | 1315 | if (s == PL_bostr) |
ffc61ed2 JH |
1316 | tmp = '\n'; |
1317 | else { | |
6136c704 | 1318 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1319 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1320 | } |
1321 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1322 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1323 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1324 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1325 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1326 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1327 | isALNUM_LC_utf8((U8*)s))) |
1328 | tmp = !tmp; | |
4cadc6a9 YO |
1329 | else REXEC_FBC_TRYIT; |
1330 | ); | |
a0d0e21e | 1331 | } |
667bb95a | 1332 | else { |
12d33761 | 1333 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1334 | tmp = ((OP(c) == NBOUND ? |
1335 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1336 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1337 | if (tmp == |
1338 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1339 | tmp = !tmp; | |
4cadc6a9 YO |
1340 | else REXEC_FBC_TRYIT; |
1341 | ); | |
a0ed51b3 | 1342 | } |
24b23f37 | 1343 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1344 | goto got_it; |
1345 | break; | |
a0d0e21e | 1346 | case ALNUM: |
4cadc6a9 YO |
1347 | REXEC_FBC_CSCAN_PRELOAD( |
1348 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1349 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1350 | isALNUM(*s) | |
1351 | ); | |
bbce6d69 | 1352 | case ALNUML: |
4cadc6a9 YO |
1353 | REXEC_FBC_CSCAN_TAINT( |
1354 | isALNUM_LC_utf8((U8*)s), | |
1355 | isALNUM_LC(*s) | |
1356 | ); | |
a0d0e21e | 1357 | case NALNUM: |
4cadc6a9 YO |
1358 | REXEC_FBC_CSCAN_PRELOAD( |
1359 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1360 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1361 | !isALNUM(*s) | |
1362 | ); | |
bbce6d69 | 1363 | case NALNUML: |
4cadc6a9 YO |
1364 | REXEC_FBC_CSCAN_TAINT( |
1365 | !isALNUM_LC_utf8((U8*)s), | |
1366 | !isALNUM_LC(*s) | |
1367 | ); | |
a0d0e21e | 1368 | case SPACE: |
4cadc6a9 YO |
1369 | REXEC_FBC_CSCAN_PRELOAD( |
1370 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1371 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1372 | isSPACE(*s) | |
1373 | ); | |
bbce6d69 | 1374 | case SPACEL: |
4cadc6a9 YO |
1375 | REXEC_FBC_CSCAN_TAINT( |
1376 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1377 | isSPACE_LC(*s) | |
1378 | ); | |
a0d0e21e | 1379 | case NSPACE: |
4cadc6a9 YO |
1380 | REXEC_FBC_CSCAN_PRELOAD( |
1381 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1382 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1383 | !isSPACE(*s) | |
1384 | ); | |
bbce6d69 | 1385 | case NSPACEL: |
4cadc6a9 YO |
1386 | REXEC_FBC_CSCAN_TAINT( |
1387 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1388 | !isSPACE_LC(*s) | |
1389 | ); | |
a0d0e21e | 1390 | case DIGIT: |
4cadc6a9 YO |
1391 | REXEC_FBC_CSCAN_PRELOAD( |
1392 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1393 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1394 | isDIGIT(*s) | |
1395 | ); | |
b8c5462f | 1396 | case DIGITL: |
4cadc6a9 YO |
1397 | REXEC_FBC_CSCAN_TAINT( |
1398 | isDIGIT_LC_utf8((U8*)s), | |
1399 | isDIGIT_LC(*s) | |
1400 | ); | |
a0d0e21e | 1401 | case NDIGIT: |
4cadc6a9 YO |
1402 | REXEC_FBC_CSCAN_PRELOAD( |
1403 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1404 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1405 | !isDIGIT(*s) | |
1406 | ); | |
b8c5462f | 1407 | case NDIGITL: |
4cadc6a9 YO |
1408 | REXEC_FBC_CSCAN_TAINT( |
1409 | !isDIGIT_LC_utf8((U8*)s), | |
1410 | !isDIGIT_LC(*s) | |
1411 | ); | |
1de06328 YO |
1412 | case AHOCORASICKC: |
1413 | case AHOCORASICK: | |
07be1b83 YO |
1414 | { |
1415 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1416 | trie_type = do_utf8 ? | |
1417 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1418 | : trie_plain; | |
1419 | /* what trie are we using right now */ | |
1420 | reg_ac_data *aho | |
f8fc2ecf | 1421 | = (reg_ac_data*)progi->data->data[ ARG( c ) ]; |
3251b653 NC |
1422 | reg_trie_data *trie |
1423 | = (reg_trie_data*)progi->data->data[ aho->trie ]; | |
55eed653 | 1424 | HV *widecharmap = (HV*) progi->data->data[ aho->trie + 1 ]; |
07be1b83 YO |
1425 | |
1426 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1427 | #ifdef DEBUGGING |
07be1b83 | 1428 | const char *real_start = s; |
6148ee25 | 1429 | #endif |
07be1b83 | 1430 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1431 | SV *sv_points; |
1432 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1433 | when reading a given char. For ASCII this |
be8e71aa YO |
1434 | is unnecessary overhead as the relationship |
1435 | is always 1:1, but for unicode, especially | |
1436 | case folded unicode this is not true. */ | |
f9e705e8 | 1437 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1438 | U8 *bitmap=NULL; |
1439 | ||
07be1b83 YO |
1440 | |
1441 | GET_RE_DEBUG_FLAGS_DECL; | |
1442 | ||
be8e71aa YO |
1443 | /* We can't just allocate points here. We need to wrap it in |
1444 | * an SV so it gets freed properly if there is a croak while | |
1445 | * running the match */ | |
1446 | ENTER; | |
1447 | SAVETMPS; | |
1448 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1449 | SvCUR_set(sv_points, | |
1450 | maxlen * sizeof(U8 *)); | |
1451 | SvPOK_on(sv_points); | |
1452 | sv_2mortal(sv_points); | |
1453 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1454 | if ( trie_type != trie_utf8_fold |
1455 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1456 | { | |
786e8c11 YO |
1457 | if (trie->bitmap) |
1458 | bitmap=(U8*)trie->bitmap; | |
1459 | else | |
1460 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1461 | } |
786e8c11 YO |
1462 | /* this is the Aho-Corasick algorithm modified a touch |
1463 | to include special handling for long "unknown char" | |
1464 | sequences. The basic idea being that we use AC as long | |
1465 | as we are dealing with a possible matching char, when | |
1466 | we encounter an unknown char (and we have not encountered | |
1467 | an accepting state) we scan forward until we find a legal | |
1468 | starting char. | |
1469 | AC matching is basically that of trie matching, except | |
1470 | that when we encounter a failing transition, we fall back | |
1471 | to the current states "fail state", and try the current char | |
1472 | again, a process we repeat until we reach the root state, | |
1473 | state 1, or a legal transition. If we fail on the root state | |
1474 | then we can either terminate if we have reached an accepting | |
1475 | state previously, or restart the entire process from the beginning | |
1476 | if we have not. | |
1477 | ||
1478 | */ | |
07be1b83 YO |
1479 | while (s <= last_start) { |
1480 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1481 | U8 *uc = (U8*)s; | |
1482 | U16 charid = 0; | |
1483 | U32 base = 1; | |
1484 | U32 state = 1; | |
1485 | UV uvc = 0; | |
1486 | STRLEN len = 0; | |
1487 | STRLEN foldlen = 0; | |
1488 | U8 *uscan = (U8*)NULL; | |
1489 | U8 *leftmost = NULL; | |
786e8c11 YO |
1490 | #ifdef DEBUGGING |
1491 | U32 accepted_word= 0; | |
1492 | #endif | |
07be1b83 YO |
1493 | U32 pointpos = 0; |
1494 | ||
1495 | while ( state && uc <= (U8*)strend ) { | |
1496 | int failed=0; | |
786e8c11 YO |
1497 | U32 word = aho->states[ state ].wordnum; |
1498 | ||
1de06328 YO |
1499 | if( state==1 ) { |
1500 | if ( bitmap ) { | |
1501 | DEBUG_TRIE_EXECUTE_r( | |
1502 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1503 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1504 | (char *)uc, do_utf8 ); | |
1505 | PerlIO_printf( Perl_debug_log, | |
1506 | " Scanning for legal start char...\n"); | |
1507 | } | |
1508 | ); | |
1509 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1510 | uc++; | |
786e8c11 | 1511 | } |
1de06328 | 1512 | s= (char *)uc; |
786e8c11 | 1513 | } |
786e8c11 YO |
1514 | if (uc >(U8*)last_start) break; |
1515 | } | |
1516 | ||
1517 | if ( word ) { | |
1518 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1519 | if (!leftmost || lpos < leftmost) { | |
1520 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1521 | leftmost= lpos; |
786e8c11 | 1522 | } |
07be1b83 | 1523 | if (base==0) break; |
786e8c11 | 1524 | |
07be1b83 YO |
1525 | } |
1526 | points[pointpos++ % maxlen]= uc; | |
55eed653 NC |
1527 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
1528 | uscan, len, uvc, charid, foldlen, | |
1529 | foldbuf, uniflags); | |
786e8c11 YO |
1530 | DEBUG_TRIE_EXECUTE_r({ |
1531 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1532 | s, do_utf8 ); | |
07be1b83 | 1533 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1534 | " Charid:%3u CP:%4"UVxf" ", |
1535 | charid, uvc); | |
1536 | }); | |
07be1b83 YO |
1537 | |
1538 | do { | |
6148ee25 | 1539 | #ifdef DEBUGGING |
786e8c11 | 1540 | word = aho->states[ state ].wordnum; |
6148ee25 | 1541 | #endif |
07be1b83 YO |
1542 | base = aho->states[ state ].trans.base; |
1543 | ||
786e8c11 YO |
1544 | DEBUG_TRIE_EXECUTE_r({ |
1545 | if (failed) | |
1546 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1547 | s, do_utf8 ); | |
07be1b83 | 1548 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1549 | "%sState: %4"UVxf", word=%"UVxf, |
1550 | failed ? " Fail transition to " : "", | |
1551 | (UV)state, (UV)word); | |
1552 | }); | |
07be1b83 YO |
1553 | if ( base ) { |
1554 | U32 tmp; | |
1555 | if (charid && | |
1556 | (base + charid > trie->uniquecharcount ) | |
1557 | && (base + charid - 1 - trie->uniquecharcount | |
1558 | < trie->lasttrans) | |
1559 | && trie->trans[base + charid - 1 - | |
1560 | trie->uniquecharcount].check == state | |
1561 | && (tmp=trie->trans[base + charid - 1 - | |
1562 | trie->uniquecharcount ].next)) | |
1563 | { | |
786e8c11 YO |
1564 | DEBUG_TRIE_EXECUTE_r( |
1565 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1566 | state = tmp; |
1567 | break; | |
1568 | } | |
1569 | else { | |
786e8c11 YO |
1570 | DEBUG_TRIE_EXECUTE_r( |
1571 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1572 | failed = 1; | |
1573 | state = aho->fail[state]; | |
07be1b83 YO |
1574 | } |
1575 | } | |
1576 | else { | |
1577 | /* we must be accepting here */ | |
786e8c11 YO |
1578 | DEBUG_TRIE_EXECUTE_r( |
1579 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1580 | failed = 1; | |
07be1b83 YO |
1581 | break; |
1582 | } | |
1583 | } while(state); | |
786e8c11 | 1584 | uc += len; |
07be1b83 YO |
1585 | if (failed) { |
1586 | if (leftmost) | |
1587 | break; | |
786e8c11 | 1588 | if (!state) state = 1; |
07be1b83 YO |
1589 | } |
1590 | } | |
1591 | if ( aho->states[ state ].wordnum ) { | |
1592 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1593 | if (!leftmost || lpos < leftmost) { |
1594 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1595 | leftmost = lpos; |
786e8c11 | 1596 | } |
07be1b83 | 1597 | } |
07be1b83 YO |
1598 | if (leftmost) { |
1599 | s = (char*)leftmost; | |
786e8c11 YO |
1600 | DEBUG_TRIE_EXECUTE_r({ |
1601 | PerlIO_printf( | |
70685ca0 JH |
1602 | Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n", |
1603 | (UV)accepted_word, (IV)(s - real_start) | |
786e8c11 YO |
1604 | ); |
1605 | }); | |
24b23f37 | 1606 | if (!reginfo || regtry(reginfo, &s)) { |
be8e71aa YO |
1607 | FREETMPS; |
1608 | LEAVE; | |
07be1b83 | 1609 | goto got_it; |
be8e71aa | 1610 | } |
07be1b83 | 1611 | s = HOPc(s,1); |
786e8c11 YO |
1612 | DEBUG_TRIE_EXECUTE_r({ |
1613 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1614 | }); | |
07be1b83 | 1615 | } else { |
786e8c11 YO |
1616 | DEBUG_TRIE_EXECUTE_r( |
1617 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1618 | break; |
1619 | } | |
1620 | } | |
be8e71aa YO |
1621 | FREETMPS; |
1622 | LEAVE; | |
07be1b83 YO |
1623 | } |
1624 | break; | |
b3c9acc1 | 1625 | default: |
3c3eec57 GS |
1626 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1627 | break; | |
d6a28714 | 1628 | } |
6eb5f6b9 JH |
1629 | return 0; |
1630 | got_it: | |
1631 | return s; | |
1632 | } | |
1633 | ||
1634 | /* | |
1635 | - regexec_flags - match a regexp against a string | |
1636 | */ | |
1637 | I32 | |
1638 | Perl_regexec_flags(pTHX_ register regexp *prog, char *stringarg, register char *strend, | |
1639 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) | |
1640 | /* strend: pointer to null at end of string */ | |
1641 | /* strbeg: real beginning of string */ | |
1642 | /* minend: end of match must be >=minend after stringarg. */ | |
58e23c8d YO |
1643 | /* data: May be used for some additional optimizations. |
1644 | Currently its only used, with a U32 cast, for transmitting | |
1645 | the ganch offset when doing a /g match. This will change */ | |
6eb5f6b9 JH |
1646 | /* nosave: For optimizations. */ |
1647 | { | |
97aff369 | 1648 | dVAR; |
24b23f37 | 1649 | /*register*/ char *s; |
6eb5f6b9 | 1650 | register regnode *c; |
24b23f37 | 1651 | /*register*/ char *startpos = stringarg; |
6eb5f6b9 JH |
1652 | I32 minlen; /* must match at least this many chars */ |
1653 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1654 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1655 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1656 | char *scream_olds = NULL; |
3dab1dad | 1657 | SV* const oreplsv = GvSV(PL_replgv); |
f9f4320a | 1658 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1659 | I32 multiline; |
f8fc2ecf | 1660 | RXi_GET_DECL(prog,progi); |
3b0527fe | 1661 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
a3621e74 YO |
1662 | |
1663 | GET_RE_DEBUG_FLAGS_DECL; | |
1664 | ||
9d4ba2ae | 1665 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1666 | |
1667 | /* Be paranoid... */ | |
1668 | if (prog == NULL || startpos == NULL) { | |
1669 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1670 | return 0; | |
1671 | } | |
1672 | ||
bbe252da | 1673 | multiline = prog->extflags & RXf_PMf_MULTILINE; |
3b0527fe | 1674 | reginfo.prog = prog; |
2757e526 | 1675 | |
bac06658 | 1676 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1677 | DEBUG_EXECUTE_r( |
1678 | debug_start_match(prog, do_utf8, startpos, strend, | |
1679 | "Matching"); | |
1680 | ); | |
bac06658 | 1681 | |
6eb5f6b9 | 1682 | minlen = prog->minlen; |
1de06328 YO |
1683 | |
1684 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1685 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1686 | "String too short [regexec_flags]...\n")); |
1687 | goto phooey; | |
1aa99e6b | 1688 | } |
6eb5f6b9 | 1689 | |
1de06328 | 1690 | |
6eb5f6b9 | 1691 | /* Check validity of program. */ |
f8fc2ecf | 1692 | if (UCHARAT(progi->program) != REG_MAGIC) { |
6eb5f6b9 JH |
1693 | Perl_croak(aTHX_ "corrupted regexp program"); |
1694 | } | |
1695 | ||
1696 | PL_reg_flags = 0; | |
1697 | PL_reg_eval_set = 0; | |
1698 | PL_reg_maxiter = 0; | |
1699 | ||
bbe252da | 1700 | if (prog->extflags & RXf_UTF8) |
6eb5f6b9 JH |
1701 | PL_reg_flags |= RF_utf8; |
1702 | ||
1703 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1704 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1705 | PL_bostr = strbeg; |
3b0527fe | 1706 | reginfo.sv = sv; |
6eb5f6b9 JH |
1707 | |
1708 | /* Mark end of line for $ (and such) */ | |
1709 | PL_regeol = strend; | |
1710 | ||
1711 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1712 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1713 | |
6eb5f6b9 JH |
1714 | /* If there is a "must appear" string, look for it. */ |
1715 | s = startpos; | |
1716 | ||
bbe252da | 1717 | if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1718 | MAGIC *mg; |
1719 | ||
1720 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
58e23c8d | 1721 | reginfo.ganch = startpos + prog->gofs; |
6eb5f6b9 JH |
1722 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1723 | && SvMAGIC(sv) | |
14befaf4 DM |
1724 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1725 | && mg->mg_len >= 0) { | |
3b0527fe | 1726 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
bbe252da | 1727 | if (prog->extflags & RXf_ANCH_GPOS) { |
3b0527fe | 1728 | if (s > reginfo.ganch) |
6eb5f6b9 | 1729 | goto phooey; |
58e23c8d | 1730 | s = reginfo.ganch - prog->gofs; |
6eb5f6b9 JH |
1731 | } |
1732 | } | |
58e23c8d | 1733 | else if (data) { |
70685ca0 | 1734 | reginfo.ganch = strbeg + PTR2UV(data); |
58e23c8d | 1735 | } else /* pos() not defined */ |
3b0527fe | 1736 | reginfo.ganch = strbeg; |
6eb5f6b9 | 1737 | } |
c74340f9 YO |
1738 | if (PL_curpm && (PM_GETRE(PL_curpm) == prog)) { |
1739 | I32 *t; | |
f8fc2ecf | 1740 | if (!progi->swap) { |
c74340f9 YO |
1741 | /* We have to be careful. If the previous successful match |
1742 | was from this regex we don't want a subsequent paritally | |
1743 | successful match to clobber the old results. | |
1744 | So when we detect this possibility we add a swap buffer | |
1745 | to the re, and switch the buffer each match. If we fail | |
1746 | we switch it back, otherwise we leave it swapped. | |
1747 | */ | |
f8fc2ecf | 1748 | Newxz(progi->swap, 1, regexp_paren_ofs); |
c74340f9 | 1749 | /* no need to copy these */ |
f8fc2ecf YO |
1750 | Newxz(progi->swap->startp, prog->nparens + 1, I32); |
1751 | Newxz(progi->swap->endp, prog->nparens + 1, I32); | |
c74340f9 | 1752 | } |
f8fc2ecf YO |
1753 | t = progi->swap->startp; |
1754 | progi->swap->startp = prog->startp; | |
c74340f9 | 1755 | prog->startp = t; |
f8fc2ecf YO |
1756 | t = progi->swap->endp; |
1757 | progi->swap->endp = prog->endp; | |
c74340f9 YO |
1758 | prog->endp = t; |
1759 | } | |
a0714e2c | 1760 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1761 | re_scream_pos_data d; |
1762 | ||
1763 | d.scream_olds = &scream_olds; | |
1764 | d.scream_pos = &scream_pos; | |
1765 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1766 | if (!s) { |
a3621e74 | 1767 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1768 | goto phooey; /* not present */ |
3fa9c3d7 | 1769 | } |
6eb5f6b9 JH |
1770 | } |
1771 | ||
1de06328 | 1772 | |
6eb5f6b9 JH |
1773 | |
1774 | /* Simplest case: anchored match need be tried only once. */ | |
1775 | /* [unless only anchor is BOL and multiline is set] */ | |
bbe252da | 1776 | if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) { |
24b23f37 | 1777 | if (s == startpos && regtry(®info, &startpos)) |
6eb5f6b9 | 1778 | goto got_it; |
bbe252da YO |
1779 | else if (multiline || (prog->intflags & PREGf_IMPLICIT) |
1780 | || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */ | |
6eb5f6b9 JH |
1781 | { |
1782 | char *end; | |
1783 | ||
1784 | if (minlen) | |
1785 | dontbother = minlen - 1; | |
1aa99e6b | 1786 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1787 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1788 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1789 | if (s == startpos) |
1790 | goto after_try; | |
1791 | while (1) { | |
24b23f37 | 1792 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1793 | goto got_it; |
1794 | after_try: | |
1795 | if (s >= end) | |
1796 | goto phooey; | |
bbe252da | 1797 | if (prog->extflags & RXf_USE_INTUIT) { |
6eb5f6b9 JH |
1798 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); |
1799 | if (!s) | |
1800 | goto phooey; | |
1801 | } | |
1802 | else | |
1803 | s++; | |
1804 | } | |
1805 | } else { | |
1806 | if (s > startpos) | |
1807 | s--; | |
1808 | while (s < end) { | |
1809 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
24b23f37 | 1810 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1811 | goto got_it; |
1812 | } | |
1813 | } | |
1814 | } | |
1815 | } | |
1816 | goto phooey; | |
bbe252da | 1817 | } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK)) |
f9f4320a YO |
1818 | { |
1819 | /* the warning about reginfo.ganch being used without intialization | |
bbe252da | 1820 | is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN |
f9f4320a | 1821 | and we only enter this block when the same bit is set. */ |
58e23c8d YO |
1822 | char *tmp_s = reginfo.ganch - prog->gofs; |
1823 | if (regtry(®info, &tmp_s)) | |
6eb5f6b9 JH |
1824 | goto got_it; |
1825 | goto phooey; | |
1826 | } | |
1827 | ||
1828 | /* Messy cases: unanchored match. */ | |
bbe252da | 1829 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) { |
6eb5f6b9 JH |
1830 | /* we have /x+whatever/ */ |
1831 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1832 | char ch; |
bf93d4cc GS |
1833 | #ifdef DEBUGGING |
1834 | int did_match = 0; | |
1835 | #endif | |
33b8afdf JH |
1836 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1837 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1838 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1839 | |
1aa99e6b | 1840 | if (do_utf8) { |
4cadc6a9 | 1841 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1842 | if (*s == ch) { |
a3621e74 | 1843 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1844 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1845 | s += UTF8SKIP(s); |
1846 | while (s < strend && *s == ch) | |
1847 | s += UTF8SKIP(s); | |
1848 | } | |
4cadc6a9 | 1849 | ); |
6eb5f6b9 JH |
1850 | } |
1851 | else { | |
4cadc6a9 | 1852 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1853 | if (*s == ch) { |
a3621e74 | 1854 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1855 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1856 | s++; |
1857 | while (s < strend && *s == ch) | |
1858 | s++; | |
1859 | } | |
4cadc6a9 | 1860 | ); |
6eb5f6b9 | 1861 | } |
a3621e74 | 1862 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1863 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1864 | "Did not find anchored character...\n") |
1865 | ); | |
6eb5f6b9 | 1866 | } |
a0714e2c SS |
1867 | else if (prog->anchored_substr != NULL |
1868 | || prog->anchored_utf8 != NULL | |
1869 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1870 | && prog->float_max_offset < strend - s)) { |
1871 | SV *must; | |
1872 | I32 back_max; | |
1873 | I32 back_min; | |
1874 | char *last; | |
6eb5f6b9 | 1875 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1876 | #ifdef DEBUGGING |
1877 | int did_match = 0; | |
1878 | #endif | |
33b8afdf JH |
1879 | if (prog->anchored_substr || prog->anchored_utf8) { |
1880 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1881 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1882 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1883 | back_max = back_min = prog->anchored_offset; | |
1884 | } else { | |
1885 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1886 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1887 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1888 | back_max = prog->float_max_offset; | |
1889 | back_min = prog->float_min_offset; | |
1890 | } | |
1de06328 YO |
1891 | |
1892 | ||
33b8afdf JH |
1893 | if (must == &PL_sv_undef) |
1894 | /* could not downgrade utf8 check substring, so must fail */ | |
1895 | goto phooey; | |
1896 | ||
1de06328 YO |
1897 | if (back_min<0) { |
1898 | last = strend; | |
1899 | } else { | |
1900 | last = HOP3c(strend, /* Cannot start after this */ | |
1901 | -(I32)(CHR_SVLEN(must) | |
1902 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1903 | } | |
6eb5f6b9 JH |
1904 | if (s > PL_bostr) |
1905 | last1 = HOPc(s, -1); | |
1906 | else | |
1907 | last1 = s - 1; /* bogus */ | |
1908 | ||
a0288114 | 1909 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1910 | check_substr==must. */ |
1911 | scream_pos = -1; | |
1912 | dontbother = end_shift; | |
1913 | strend = HOPc(strend, -dontbother); | |
1914 | while ( (s <= last) && | |
9041c2e3 | 1915 | ((flags & REXEC_SCREAM) |
1de06328 | 1916 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1917 | end_shift, &scream_pos, 0)) |
1de06328 | 1918 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1919 | (unsigned char*)strend, must, |
7fba1cd6 | 1920 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1921 | /* we may be pointing at the wrong string */ |
1922 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1923 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1924 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1925 | if (HOPc(s, -back_max) > last1) { |
1926 | last1 = HOPc(s, -back_min); | |
1927 | s = HOPc(s, -back_max); | |
1928 | } | |
1929 | else { | |
52657f30 | 1930 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1931 | |
1932 | last1 = HOPc(s, -back_min); | |
52657f30 | 1933 | s = t; |
6eb5f6b9 | 1934 | } |
1aa99e6b | 1935 | if (do_utf8) { |
6eb5f6b9 | 1936 | while (s <= last1) { |
24b23f37 | 1937 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1938 | goto got_it; |
1939 | s += UTF8SKIP(s); | |
1940 | } | |
1941 | } | |
1942 | else { | |
1943 | while (s <= last1) { | |
24b23f37 | 1944 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1945 | goto got_it; |
1946 | s++; | |
1947 | } | |
1948 | } | |
1949 | } | |
ab3bbdeb YO |
1950 | DEBUG_EXECUTE_r(if (!did_match) { |
1951 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
1952 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
1953 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 1954 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 1955 | ? "anchored" : "floating"), |
ab3bbdeb YO |
1956 | quoted, RE_SV_TAIL(must)); |
1957 | }); | |
6eb5f6b9 JH |
1958 | goto phooey; |
1959 | } | |
f8fc2ecf | 1960 | else if ( (c = progi->regstclass) ) { |
f14c76ed | 1961 | if (minlen) { |
f8fc2ecf | 1962 | const OPCODE op = OP(progi->regstclass); |
66e933ab | 1963 | /* don't bother with what can't match */ |
786e8c11 | 1964 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
1965 | strend = HOPc(strend, -(minlen - 1)); |
1966 | } | |
a3621e74 | 1967 | DEBUG_EXECUTE_r({ |
be8e71aa | 1968 | SV * const prop = sv_newmortal(); |
32fc9b6a | 1969 | regprop(prog, prop, c); |
0df25f3d | 1970 | { |
ab3bbdeb YO |
1971 | RE_PV_QUOTED_DECL(quoted,UTF,PERL_DEBUG_PAD_ZERO(1), |
1972 | s,strend-s,60); | |
0df25f3d | 1973 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 1974 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 1975 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 1976 | quoted, (int)(strend - s)); |
0df25f3d | 1977 | } |
ffc61ed2 | 1978 | }); |
3b0527fe | 1979 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 1980 | goto got_it; |
07be1b83 | 1981 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
1982 | } |
1983 | else { | |
1984 | dontbother = 0; | |
a0714e2c | 1985 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 1986 | /* Trim the end. */ |
d6a28714 | 1987 | char *last; |
33b8afdf JH |
1988 | SV* float_real; |
1989 | ||
1990 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1991 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1992 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
1993 | |
1994 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 1995 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
1996 | end_shift, &scream_pos, 1); /* last one */ |
1997 | if (!last) | |
ffc61ed2 | 1998 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
1999 | /* we may be pointing at the wrong string */ |
2000 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 2001 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 2002 | } |
d6a28714 JH |
2003 | else { |
2004 | STRLEN len; | |
cfd0369c | 2005 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 2006 | |
33b8afdf | 2007 | if (SvTAIL(float_real)) { |
d6a28714 JH |
2008 | if (memEQ(strend - len + 1, little, len - 1)) |
2009 | last = strend - len + 1; | |
7fba1cd6 | 2010 | else if (!multiline) |
9041c2e3 | 2011 | last = memEQ(strend - len, little, len) |
bd61b366 | 2012 | ? strend - len : NULL; |
b8c5462f | 2013 | else |
d6a28714 JH |
2014 | goto find_last; |
2015 | } else { | |
2016 | find_last: | |
9041c2e3 | 2017 | if (len) |
d6a28714 | 2018 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 2019 | else |
a0288114 | 2020 | last = strend; /* matching "$" */ |
b8c5462f | 2021 | } |
b8c5462f | 2022 | } |
bf93d4cc | 2023 | if (last == NULL) { |
6bda09f9 YO |
2024 | DEBUG_EXECUTE_r( |
2025 | PerlIO_printf(Perl_debug_log, | |
2026 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
2027 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
2028 | goto phooey; /* Should not happen! */ |
2029 | } | |
d6a28714 JH |
2030 | dontbother = strend - last + prog->float_min_offset; |
2031 | } | |
2032 | if (minlen && (dontbother < minlen)) | |
2033 | dontbother = minlen - 1; | |
2034 | strend -= dontbother; /* this one's always in bytes! */ | |
2035 | /* We don't know much -- general case. */ | |
1aa99e6b | 2036 | if (do_utf8) { |
d6a28714 | 2037 | for (;;) { |
24b23f37 | 2038 | if (regtry(®info, &s)) |
d6a28714 JH |
2039 | goto got_it; |
2040 | if (s >= strend) | |
2041 | break; | |
b8c5462f | 2042 | s += UTF8SKIP(s); |
d6a28714 JH |
2043 | }; |
2044 | } | |
2045 | else { | |
2046 | do { | |
24b23f37 | 2047 | if (regtry(®info, &s)) |
d6a28714 JH |
2048 | goto got_it; |
2049 | } while (s++ < strend); | |
2050 | } | |
2051 | } | |
2052 | ||
2053 | /* Failure. */ | |
2054 | goto phooey; | |
2055 | ||
2056 | got_it: | |
2057 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2058 | ||
2059 | if (PL_reg_eval_set) { | |
2060 | /* Preserve the current value of $^R */ | |
2061 | if (oreplsv != GvSV(PL_replgv)) | |
2062 | sv_setsv(oreplsv, GvSV(PL_replgv));/* So that when GvSV(replgv) is | |
2063 | restored, the value remains | |
2064 | the same. */ | |
4f639d21 | 2065 | restore_pos(aTHX_ prog); |
d6a28714 | 2066 | } |
81714fb9 YO |
2067 | if (prog->paren_names) |
2068 | (void)hv_iterinit(prog->paren_names); | |
d6a28714 JH |
2069 | |
2070 | /* make sure $`, $&, $', and $digit will work later */ | |
2071 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2072 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2073 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2074 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2075 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2076 | if ((SvIsCOW(sv) |
2077 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2078 | if (DEBUG_C_TEST) { | |
2079 | PerlIO_printf(Perl_debug_log, | |
2080 | "Copy on write: regexp capture, type %d\n", | |
2081 | (int) SvTYPE(sv)); | |
2082 | } | |
2083 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2084 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2085 | assert (SvPOKp(prog->saved_copy)); |
2086 | } else | |
2087 | #endif | |
2088 | { | |
2089 | RX_MATCH_COPIED_on(prog); | |
2090 | s = savepvn(strbeg, i); | |
2091 | prog->subbeg = s; | |
2092 | } | |
d6a28714 | 2093 | prog->sublen = i; |
d6a28714 JH |
2094 | } |
2095 | else { | |
2096 | prog->subbeg = strbeg; | |
2097 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2098 | } | |
2099 | } | |
9041c2e3 | 2100 | |
d6a28714 JH |
2101 | return 1; |
2102 | ||
2103 | phooey: | |
a3621e74 | 2104 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2105 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2106 | if (PL_reg_eval_set) |
4f639d21 | 2107 | restore_pos(aTHX_ prog); |
f8fc2ecf | 2108 | if (progi->swap) { |
c74340f9 YO |
2109 | /* we failed :-( roll it back */ |
2110 | I32 *t; | |
f8fc2ecf YO |
2111 | t = progi->swap->startp; |
2112 | progi->swap->startp = prog->startp; | |
c74340f9 | 2113 | prog->startp = t; |
f8fc2ecf YO |
2114 | t = progi->swap->endp; |
2115 | progi->swap->endp = prog->endp; | |
c74340f9 YO |
2116 | prog->endp = t; |
2117 | } | |
d6a28714 JH |
2118 | return 0; |
2119 | } | |
2120 | ||
6bda09f9 | 2121 | |
d6a28714 JH |
2122 | /* |
2123 | - regtry - try match at specific point | |
2124 | */ | |
2125 | STATIC I32 /* 0 failure, 1 success */ | |
24b23f37 | 2126 | S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) |
d6a28714 | 2127 | { |
97aff369 | 2128 | dVAR; |
d6a28714 JH |
2129 | register I32 *sp; |
2130 | register I32 *ep; | |
2131 | CHECKPOINT lastcp; | |
3b0527fe | 2132 | regexp *prog = reginfo->prog; |
f8fc2ecf | 2133 | RXi_GET_DECL(prog,progi); |
a3621e74 | 2134 | GET_RE_DEBUG_FLAGS_DECL; |
24b23f37 | 2135 | reginfo->cutpoint=NULL; |
d6a28714 | 2136 | |
bbe252da | 2137 | if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) { |
d6a28714 JH |
2138 | MAGIC *mg; |
2139 | ||
2140 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2141 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2142 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2143 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2144 | )); |
ea8d6ae1 | 2145 | SAVESTACK_CXPOS(); |
d6a28714 JH |
2146 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2147 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2148 | SAVETMPS; | |
2149 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2150 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2151 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2152 | ||
3b0527fe | 2153 | if (reginfo->sv) { |
d6a28714 | 2154 | /* Make $_ available to executed code. */ |
3b0527fe | 2155 | if (reginfo->sv != DEFSV) { |
59f00321 | 2156 | SAVE_DEFSV; |
3b0527fe | 2157 | DEFSV = reginfo->sv; |
b8c5462f | 2158 | } |
d6a28714 | 2159 | |
3b0527fe DM |
2160 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2161 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2162 | /* prepare for quick setting of pos */ |
d300d9fa | 2163 | #ifdef PERL_OLD_COPY_ON_WRITE |
51a9ea20 NC |
2164 | if (SvIsCOW(reginfo->sv)) |
2165 | sv_force_normal_flags(reginfo->sv, 0); | |
d300d9fa | 2166 | #endif |
3dab1dad | 2167 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2168 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2169 | mg->mg_len = -1; |
b8c5462f | 2170 | } |
d6a28714 JH |
2171 | PL_reg_magic = mg; |
2172 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2173 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2174 | } |
09687e5a | 2175 | if (!PL_reg_curpm) { |
a02a5408 | 2176 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2177 | #ifdef USE_ITHREADS |
2178 | { | |
be8e71aa | 2179 | SV* const repointer = newSViv(0); |
577e12cc | 2180 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2181 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2182 | av_push(PL_regex_padav,repointer); |
2183 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2184 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2185 | } | |
2186 | #endif | |
2187 | } | |
aaa362c4 | 2188 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2189 | PL_reg_oldcurpm = PL_curpm; |
2190 | PL_curpm = PL_reg_curpm; | |
2191 | if (RX_MATCH_COPIED(prog)) { | |
2192 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2193 | since it may be needed if this match fails. Thus | |
2194 | $` inside (?{}) could fail... */ | |
2195 | PL_reg_oldsaved = prog->subbeg; | |
2196 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2197 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2198 | PL_nrs = prog->saved_copy; |
2199 | #endif | |
d6a28714 JH |
2200 | RX_MATCH_COPIED_off(prog); |
2201 | } | |
2202 | else | |
bd61b366 | 2203 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2204 | prog->subbeg = PL_bostr; |
2205 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2206 | } | |
24b23f37 YO |
2207 | DEBUG_EXECUTE_r(PL_reg_starttry = *startpos); |
2208 | prog->startp[0] = *startpos - PL_bostr; | |
2209 | PL_reginput = *startpos; | |
d6a28714 | 2210 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2211 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2212 | prog->lastparen = 0; |
03994de8 | 2213 | prog->lastcloseparen = 0; |
d6a28714 | 2214 | PL_regsize = 0; |
6bda09f9 YO |
2215 | PL_regstartp = prog->startp; |
2216 | PL_regendp = prog->endp; | |
d6a28714 JH |
2217 | if (PL_reg_start_tmpl <= prog->nparens) { |
2218 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2219 | if(PL_reg_start_tmp) | |
2220 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2221 | else | |
a02a5408 | 2222 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2223 | } |
2224 | ||
2225 | /* XXXX What this code is doing here?!!! There should be no need | |
2226 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2227 | this! --ilya*/ |
dafc8851 JH |
2228 | |
2229 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2230 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 JH |
2231 | * PL_reglastparen), is not needed at all by the test suite |
2232 | * (op/regexp, op/pat, op/split), but that code is needed, oddly | |
2233 | * enough, for building DynaLoader, or otherwise this | |
2234 | * "Error: '*' not in typemap in DynaLoader.xs, line 164" | |
2235 | * will happen. Meanwhile, this code *is* needed for the | |
2236 | * above-mentioned test suite tests to succeed. The common theme | |
2237 | * on those tests seems to be returning null fields from matches. | |
2238 | * --jhi */ | |
dafc8851 | 2239 | #if 1 |
c74340f9 YO |
2240 | sp = PL_regstartp; |
2241 | ep = PL_regendp; | |
d6a28714 | 2242 | if (prog->nparens) { |
097eb12c | 2243 | register I32 i; |
eb160463 | 2244 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
d6a28714 JH |
2245 | *++sp = -1; |
2246 | *++ep = -1; | |
2247 | } | |
2248 | } | |
dafc8851 | 2249 | #endif |
02db2b7b | 2250 | REGCP_SET(lastcp); |
f8fc2ecf | 2251 | if (regmatch(reginfo, progi->program + 1)) { |
c74340f9 | 2252 | PL_regendp[0] = PL_reginput - PL_bostr; |
d6a28714 JH |
2253 | return 1; |
2254 | } | |
24b23f37 YO |
2255 | if (reginfo->cutpoint) |
2256 | *startpos= reginfo->cutpoint; | |
02db2b7b | 2257 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2258 | return 0; |
2259 | } | |
2260 | ||
02db2b7b | 2261 | |
8ba1375e MJD |
2262 | #define sayYES goto yes |
2263 | #define sayNO goto no | |
262b90c4 | 2264 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2265 | |
f9f4320a YO |
2266 | /* we dont use STMT_START/END here because it leads to |
2267 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2268 | #define CACHEsayNO \ | |
c476f425 DM |
2269 | if (ST.cache_mask) \ |
2270 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2271 | sayNO |
3298f257 | 2272 | |
a3621e74 | 2273 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2274 | 'failed...' are printed. It should be set such that messages |
2275 | are inline with the regop output that created them. | |
a3621e74 | 2276 | */ |
265c4333 | 2277 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2278 | |
2279 | ||
2280 | /* Make sure there is a test for this +1 options in re_tests */ | |
2281 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2282 | ||
40a82448 DM |
2283 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2284 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2285 | |
86545054 DM |
2286 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2287 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2288 | ||
5d9a96ca DM |
2289 | /* grab a new slab and return the first slot in it */ |
2290 | ||
2291 | STATIC regmatch_state * | |
2292 | S_push_slab(pTHX) | |
2293 | { | |
54df2634 NC |
2294 | #if PERL_VERSION < 9 |
2295 | dMY_CXT; | |
2296 | #endif | |
5d9a96ca DM |
2297 | regmatch_slab *s = PL_regmatch_slab->next; |
2298 | if (!s) { | |
2299 | Newx(s, 1, regmatch_slab); | |
2300 | s->prev = PL_regmatch_slab; | |
2301 | s->next = NULL; | |
2302 | PL_regmatch_slab->next = s; | |
2303 | } | |
2304 | PL_regmatch_slab = s; | |
86545054 | 2305 | return SLAB_FIRST(s); |
5d9a96ca | 2306 | } |
5b47454d | 2307 | |
95b24440 | 2308 | |
40a82448 DM |
2309 | /* push a new state then goto it */ |
2310 | ||
2311 | #define PUSH_STATE_GOTO(state, node) \ | |
2312 | scan = node; \ | |
2313 | st->resume_state = state; \ | |
2314 | goto push_state; | |
2315 | ||
2316 | /* push a new state with success backtracking, then goto it */ | |
2317 | ||
2318 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2319 | scan = node; \ | |
2320 | st->resume_state = state; \ | |
2321 | goto push_yes_state; | |
2322 | ||
aa283a38 | 2323 | |
aa283a38 | 2324 | |
d6a28714 | 2325 | /* |
95b24440 | 2326 | |
bf1f174e DM |
2327 | regmatch() - main matching routine |
2328 | ||
2329 | This is basically one big switch statement in a loop. We execute an op, | |
2330 | set 'next' to point the next op, and continue. If we come to a point which | |
2331 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2332 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2333 | state, and re-enter the loop at the state indicated. If there are no more | |
2334 | states to pop, we return failure. | |
2335 | ||
2336 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2337 | after successfully matching one A, we need to go back and try to | |
2338 | match another one; similarly for lookahead assertions: if the assertion | |
2339 | completes successfully, we backtrack to the state just before the assertion | |
2340 | and then carry on. In these cases, the pushed state is marked as | |
2341 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2342 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2343 | the nearest yes state, discarding any intermediate failure-only states. | |
2344 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2345 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2346 | it to free the inner regex. | |
2347 | ||
2348 | Note that failure backtracking rewinds the cursor position, while | |
2349 | success backtracking leaves it alone. | |
2350 | ||
2351 | A pattern is complete when the END op is executed, while a subpattern | |
2352 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2353 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2354 | behaviour. | |
2355 | ||
2356 | A common convention in this function is to use A and B to refer to the two | |
2357 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2358 | the subpattern to be matched possibly multiple times, while B is the entire | |
2359 | rest of the pattern. Variable and state names reflect this convention. | |
2360 | ||
2361 | The states in the main switch are the union of ops and failure/success of | |
2362 | substates associated with with that op. For example, IFMATCH is the op | |
2363 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2364 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2365 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2366 | just failed to match A. Resume states always come in pairs. The backtrack | |
2367 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2368 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2369 | on success or failure. | |
2370 | ||
2371 | The struct that holds a backtracking state is actually a big union, with | |
2372 | one variant for each major type of op. The variable st points to the | |
2373 | top-most backtrack struct. To make the code clearer, within each | |
2374 | block of code we #define ST to alias the relevant union. | |
2375 | ||
2376 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2377 | implementation: | |
2378 | ||
2379 | switch (state) { | |
2380 | .... | |
2381 | ||
2382 | #define ST st->u.ifmatch | |
2383 | ||
2384 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2385 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2386 | ... |
bf1f174e DM |
2387 | // push a yes backtrack state with a resume value of |
2388 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2389 | // first node of A: | |
2390 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2391 | // NOTREACHED | |
2392 | ||
2393 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2394 | next = B; | |
2395 | bar = ST.foo; // do something with the preserved value | |
2396 | break; | |
2397 | ||
2398 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2399 | ...; // do some housekeeping, then ... | |
2400 | sayNO; // propagate the failure | |
2401 | ||
2402 | #undef ST | |
95b24440 | 2403 | |
bf1f174e DM |
2404 | ... |
2405 | } | |
95b24440 | 2406 | |
bf1f174e DM |
2407 | For any old-timers reading this who are familiar with the old recursive |
2408 | approach, the code above is equivalent to: | |
95b24440 | 2409 | |
bf1f174e DM |
2410 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2411 | { | |
2412 | int foo = ... | |
95b24440 | 2413 | ... |
bf1f174e DM |
2414 | if (regmatch(A)) { |
2415 | next = B; | |
2416 | bar = foo; | |
2417 | break; | |
95b24440 | 2418 | } |
bf1f174e DM |
2419 | ...; // do some housekeeping, then ... |
2420 | sayNO; // propagate the failure | |
95b24440 | 2421 | } |
bf1f174e DM |
2422 | |
2423 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2424 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2425 | save, then do one of | |
2426 | ||
2427 | PUSH_STATE_GOTO(resume_state, node); | |
2428 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2429 | ||
2430 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2431 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2432 | On backtracking, the free slot is popped, and the saved state becomes the | |
2433 | new free state. An ST.foo field in this new top state can be temporarily | |
2434 | accessed to retrieve values, but once the main loop is re-entered, it | |
2435 | becomes available for reuse. | |
2436 | ||
2437 | Note that the depth of the backtrack stack constantly increases during the | |
2438 | left-to-right execution of the pattern, rather than going up and down with | |
2439 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2440 | end of the pattern, rather than at X in the following: | |
2441 | ||
2442 | /(((X)+)+)+....(Y)+....Z/ | |
2443 | ||
2444 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2445 | (?>A), which pop all the backtrack states associated with A before | |
2446 | continuing. | |
2447 | ||
2448 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2449 | PL_regmatch_state and st always point to the currently active state, | |
2450 | and PL_regmatch_slab points to the slab currently containing | |
2451 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2452 | allocated, and is never freed until interpreter destruction. When the slab | |
2453 | is full, a new one is allocated and chained to the end. At exit from | |
2454 | regmatch(), slabs allocated since entry are freed. | |
2455 | ||
2456 | */ | |
95b24440 | 2457 | |
40a82448 | 2458 | |
5bc10b2c | 2459 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2460 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2461 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2462 | PerlIO_printf(Perl_debug_log, \ | |
5d458dd8 | 2463 | " %*s"pp" %s%s%s%s%s\n", \ |
5bc10b2c | 2464 | depth*2, "", \ |
5d458dd8 YO |
2465 | reg_name[st->resume_state], \ |
2466 | ((st==yes_state||st==mark_state) ? "[" : ""), \ | |
2467 | ((st==yes_state) ? "Y" : ""), \ | |
2468 | ((st==mark_state) ? "M" : ""), \ | |
2469 | ((st==yes_state||st==mark_state) ? "]" : "") \ | |
2470 | ); \ | |
265c4333 | 2471 | }); |
5bc10b2c | 2472 | |
40a82448 | 2473 | |
3dab1dad | 2474 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2475 | |
3df15adc | 2476 | #ifdef DEBUGGING |
5bc10b2c | 2477 | |
ab3bbdeb YO |
2478 | STATIC void |
2479 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2480 | const char *start, const char *end, const char *blurb) | |
2481 | { | |
bbe252da | 2482 | const bool utf8_pat= prog->extflags & RXf_UTF8 ? 1 : 0; |
ab3bbdeb YO |
2483 | if (!PL_colorset) |
2484 | reginitcolors(); | |
2485 | { | |
2486 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2487 | prog->precomp, prog->prelen, 60); | |
2488 | ||
2489 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2490 | start, end - start, 60); | |
2491 | ||
2492 | PerlIO_printf(Perl_debug_log, | |
2493 | "%s%s REx%s %s against %s\n", | |
2494 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2495 | ||
2496 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2497 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2498 | utf8_pat ? "pattern" : "", | |
2499 | utf8_pat && do_utf8 ? " and " : "", | |
2500 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2501 | ); |
2502 | } | |
2503 | } | |
3df15adc YO |
2504 | |
2505 | STATIC void | |
786e8c11 YO |
2506 | S_dump_exec_pos(pTHX_ const char *locinput, |
2507 | const regnode *scan, | |
2508 | const char *loc_regeol, | |
2509 | const char *loc_bostr, | |
2510 | const char *loc_reg_starttry, | |
2511 | const bool do_utf8) | |
07be1b83 | 2512 | { |
786e8c11 | 2513 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2514 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2515 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2516 | /* The part of the string before starttry has one color |
2517 | (pref0_len chars), between starttry and current | |
2518 | position another one (pref_len - pref0_len chars), | |
2519 | after the current position the third one. | |
2520 | We assume that pref0_len <= pref_len, otherwise we | |
2521 | decrease pref0_len. */ | |
786e8c11 YO |
2522 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2523 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2524 | int pref0_len; |
2525 | ||
2526 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2527 | pref_len++; | |
786e8c11 YO |
2528 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2529 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2530 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2531 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2532 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2533 | l--; | |
2534 | if (pref0_len < 0) | |
2535 | pref0_len = 0; | |
2536 | if (pref0_len > pref_len) | |
2537 | pref0_len = pref_len; | |
2538 | { | |
3df15adc | 2539 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2540 | |
ab3bbdeb | 2541 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2542 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2543 | |
ab3bbdeb | 2544 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2545 | (locinput - pref_len + pref0_len), |
1de06328 | 2546 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2547 | |
ab3bbdeb | 2548 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2549 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2550 | |
1de06328 | 2551 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2552 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2553 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2554 | (IV)(locinput - loc_bostr), |
07be1b83 | 2555 | len0, s0, |
07be1b83 | 2556 | len1, s1, |
07be1b83 | 2557 | (docolor ? "" : "> <"), |
07be1b83 | 2558 | len2, s2, |
f9f4320a | 2559 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2560 | ""); |
2561 | } | |
2562 | } | |
3df15adc | 2563 | |
07be1b83 YO |
2564 | #endif |
2565 | ||
0a4db386 YO |
2566 | /* reg_check_named_buff_matched() |
2567 | * Checks to see if a named buffer has matched. The data array of | |
2568 | * buffer numbers corresponding to the buffer is expected to reside | |
2569 | * in the regexp->data->data array in the slot stored in the ARG() of | |
2570 | * node involved. Note that this routine doesn't actually care about the | |
2571 | * name, that information is not preserved from compilation to execution. | |
2572 | * Returns the index of the leftmost defined buffer with the given name | |
2573 | * or 0 if non of the buffers matched. | |
2574 | */ | |
2575 | STATIC I32 | |
2576 | S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) { | |
2577 | I32 n; | |
f8fc2ecf YO |
2578 | RXi_GET_DECL(rex,rexi); |
2579 | SV *sv_dat=(SV*)rexi->data->data[ ARG( scan ) ]; | |
0a4db386 YO |
2580 | I32 *nums=(I32*)SvPVX(sv_dat); |
2581 | for ( n=0; n<SvIVX(sv_dat); n++ ) { | |
2582 | if ((I32)*PL_reglastparen >= nums[n] && | |
2583 | PL_regendp[nums[n]] != -1) | |
2584 | { | |
2585 | return nums[n]; | |
2586 | } | |
2587 | } | |
2588 | return 0; | |
2589 | } | |
2590 | ||
d6a28714 | 2591 | STATIC I32 /* 0 failure, 1 success */ |
24b23f37 | 2592 | S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2593 | { |
54df2634 NC |
2594 | #if PERL_VERSION < 9 |
2595 | dMY_CXT; | |
2596 | #endif | |
27da23d5 | 2597 | dVAR; |
95b24440 | 2598 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2599 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2600 | |
3b0527fe | 2601 | regexp *rex = reginfo->prog; |
f8fc2ecf YO |
2602 | RXi_GET_DECL(rex,rexi); |
2603 | ||
5d9a96ca DM |
2604 | regmatch_slab *orig_slab; |
2605 | regmatch_state *orig_state; | |
a3621e74 | 2606 | |
5d9a96ca DM |
2607 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2608 | register regmatch_state *st; | |
95b24440 | 2609 | |
5d9a96ca DM |
2610 | /* cache heavy used fields of st in registers */ |
2611 | register regnode *scan; | |
2612 | register regnode *next; | |
438e9bae | 2613 | register U32 n = 0; /* general value; init to avoid compiler warning */ |
24d3c4a9 | 2614 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ |
5d9a96ca | 2615 | register char *locinput = PL_reginput; |
5d9a96ca | 2616 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2617 | |
b69b0499 | 2618 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2619 | int depth = 0; /* depth of backtrack stack */ |
1a147d38 | 2620 | int nochange_depth = 0; /* depth of GOSUB recursion with nochange*/ |
77cb431f DM |
2621 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2622 | subpattern */ | |
e2e6a0f1 YO |
2623 | /* mark_state piggy backs on the yes_state logic so that when we unwind |
2624 | the stack on success we can update the mark_state as we go */ | |
2625 | regmatch_state *mark_state = NULL; /* last mark state we have seen */ | |
5d458dd8 | 2626 | |
faec1544 | 2627 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2628 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2629 | U32 state_num; |
5d458dd8 YO |
2630 | bool no_final = 0; /* prevent failure from backtracking? */ |
2631 | bool do_cutgroup = 0; /* no_final only until next branch/trie entry */ | |
e2e6a0f1 | 2632 | char *startpoint = PL_reginput; |
5d458dd8 YO |
2633 | SV *popmark = NULL; /* are we looking for a mark? */ |
2634 | SV *sv_commit = NULL; /* last mark name seen in failure */ | |
2635 | SV *sv_yes_mark = NULL; /* last mark name we have seen | |
2636 | during a successfull match */ | |
2637 | U32 lastopen = 0; /* last open we saw */ | |
2638 | bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0; | |
58e23c8d | 2639 | |
5d458dd8 | 2640 | |
24d3c4a9 DM |
2641 | /* these three flags are set by various ops to signal information to |
2642 | * the very next op. They have a useful lifetime of exactly one loop | |
2643 | * iteration, and are not preserved or restored by state pushes/pops | |
2644 | */ | |
2645 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2646 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2647 | int logical = 0; /* the following EVAL is: | |
2648 | 0: (?{...}) | |
2649 | 1: (?(?{...})X|Y) | |
2650 | 2: (??{...}) | |
2651 | or the following IFMATCH/UNLESSM is: | |
2652 | false: plain (?=foo) | |
2653 | true: used as a condition: (?(?=foo)) | |
2654 | */ | |
2655 | ||
95b24440 | 2656 | #ifdef DEBUGGING |
e68ec53f | 2657 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2658 | #endif |
2659 | ||
58e23c8d | 2660 | DEBUG_OPTIMISE_r( { |
24b23f37 YO |
2661 | PerlIO_printf(Perl_debug_log,"regmatch start\n"); |
2662 | }); | |
5d9a96ca DM |
2663 | /* on first ever call to regmatch, allocate first slab */ |
2664 | if (!PL_regmatch_slab) { | |
2665 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2666 | PL_regmatch_slab->prev = NULL; | |
2667 | PL_regmatch_slab->next = NULL; | |
86545054 | 2668 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2669 | } |
2670 | ||
2671 | /* remember current high-water mark for exit */ | |
2672 | /* XXX this should be done with SAVE* instead */ | |
2673 | orig_slab = PL_regmatch_slab; | |
2674 | orig_state = PL_regmatch_state; | |
2675 | ||
2676 | /* grab next free state slot */ | |
2677 | st = ++PL_regmatch_state; | |
86545054 | 2678 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2679 | st = PL_regmatch_state = S_push_slab(aTHX); |
2680 | ||
d6a28714 JH |
2681 | /* Note that nextchr is a byte even in UTF */ |
2682 | nextchr = UCHARAT(locinput); | |
2683 | scan = prog; | |
2684 | while (scan != NULL) { | |
8ba1375e | 2685 | |
a3621e74 | 2686 | DEBUG_EXECUTE_r( { |
6136c704 | 2687 | SV * const prop = sv_newmortal(); |
1de06328 | 2688 | regnode *rnext=regnext(scan); |
786e8c11 | 2689 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2690 | regprop(rex, prop, scan); |
07be1b83 YO |
2691 | |
2692 | PerlIO_printf(Perl_debug_log, | |
2693 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
f8fc2ecf | 2694 | (IV)(scan - rexi->program), depth*2, "", |
07be1b83 | 2695 | SvPVX_const(prop), |
1de06328 | 2696 | (PL_regkind[OP(scan)] == END || !rnext) ? |
f8fc2ecf | 2697 | 0 : (IV)(rnext - rexi->program)); |
2a782b5b | 2698 | }); |
d6a28714 JH |
2699 | |
2700 | next = scan + NEXT_OFF(scan); | |
2701 | if (next == scan) | |
2702 | next = NULL; | |
40a82448 | 2703 | state_num = OP(scan); |
d6a28714 | 2704 | |
40a82448 DM |
2705 | reenter_switch: |
2706 | switch (state_num) { | |
d6a28714 | 2707 | case BOL: |
7fba1cd6 | 2708 | if (locinput == PL_bostr) |
d6a28714 | 2709 | { |
3b0527fe | 2710 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2711 | break; |
2712 | } | |
d6a28714 JH |
2713 | sayNO; |
2714 | case MBOL: | |
12d33761 HS |
2715 | if (locinput == PL_bostr || |
2716 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2717 | { |
b8c5462f JH |
2718 | break; |
2719 | } | |
d6a28714 JH |
2720 | sayNO; |
2721 | case SBOL: | |
c2a73568 | 2722 | if (locinput == PL_bostr) |
b8c5462f | 2723 | break; |
d6a28714 JH |
2724 | sayNO; |
2725 | case GPOS: | |
3b0527fe | 2726 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2727 | break; |
2728 | sayNO; | |
2729 | case EOL: | |
d6a28714 JH |
2730 | goto seol; |
2731 | case MEOL: | |
d6a28714 | 2732 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2733 | sayNO; |
b8c5462f | 2734 | break; |
d6a28714 JH |
2735 | case SEOL: |
2736 | seol: | |
2737 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2738 | sayNO; |
d6a28714 | 2739 | if (PL_regeol - locinput > 1) |
b8c5462f | 2740 | sayNO; |
b8c5462f | 2741 | break; |
d6a28714 JH |
2742 | case EOS: |
2743 | if (PL_regeol != locinput) | |
b8c5462f | 2744 | sayNO; |
d6a28714 | 2745 | break; |
ffc61ed2 | 2746 | case SANY: |
d6a28714 | 2747 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2748 | sayNO; |
f33976b4 DB |
2749 | if (do_utf8) { |
2750 | locinput += PL_utf8skip[nextchr]; | |
2751 | if (locinput > PL_regeol) | |
2752 | sayNO; | |
2753 | nextchr = UCHARAT(locinput); | |
2754 | } | |
2755 | else | |
2756 | nextchr = UCHARAT(++locinput); | |
2757 | break; | |
2758 | case CANY: | |
2759 | if (!nextchr && locinput >= PL_regeol) | |
2760 | sayNO; | |
b8c5462f | 2761 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2762 | break; |
ffc61ed2 | 2763 | case REG_ANY: |
1aa99e6b IH |
2764 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2765 | sayNO; | |
2766 | if (do_utf8) { | |
b8c5462f | 2767 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2768 | if (locinput > PL_regeol) |
2769 | sayNO; | |
a0ed51b3 | 2770 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2771 | } |
1aa99e6b IH |
2772 | else |
2773 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2774 | break; |
166ba7cd DM |
2775 | |
2776 | #undef ST | |
2777 | #define ST st->u.trie | |
786e8c11 YO |
2778 | case TRIEC: |
2779 | /* In this case the charclass data is available inline so | |
2780 | we can fail fast without a lot of extra overhead. | |
2781 | */ | |
2782 | if (scan->flags == EXACT || !do_utf8) { | |
2783 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2784 | DEBUG_EXECUTE_r( | |
2785 | PerlIO_printf(Perl_debug_log, | |
2786 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2787 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2788 | ); |
2789 | sayNO_SILENT; | |
2790 | /* NOTREACHED */ | |
2791 | } | |
2792 | } | |
2793 | /* FALL THROUGH */ | |
5b47454d | 2794 | case TRIE: |
3dab1dad | 2795 | { |
07be1b83 | 2796 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2797 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2798 | trie_type = do_utf8 ? | |
2799 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2800 | : trie_plain; | |
2801 | ||
2802 | /* what trie are we using right now */ | |
be8e71aa | 2803 | reg_trie_data * const trie |
f8fc2ecf | 2804 | = (reg_trie_data*)rexi->data->data[ ARG( scan ) ]; |
55eed653 | 2805 | HV * widecharmap = (HV *)rexi->data->data[ ARG( scan ) + 1 ]; |
3dab1dad | 2806 | U32 state = trie->startstate; |
166ba7cd | 2807 | |
3dab1dad YO |
2808 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2809 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2810 | ) { | |
2811 | if (trie->states[ state ].wordnum) { | |
2812 | DEBUG_EXECUTE_r( | |
2813 | PerlIO_printf(Perl_debug_log, | |
2814 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2815 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2816 | ); |
2817 | break; | |
2818 | } else { | |
2819 | DEBUG_EXECUTE_r( | |
2820 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2821 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2822 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2823 | ); |
2824 | sayNO_SILENT; | |
2825 | } | |
2826 | } | |
166ba7cd | 2827 | |
786e8c11 YO |
2828 | { |
2829 | U8 *uc = ( U8* )locinput; | |
2830 | ||
2831 | STRLEN len = 0; | |
2832 | STRLEN foldlen = 0; | |
2833 | U8 *uscan = (U8*)NULL; | |
2834 | STRLEN bufflen=0; | |
2835 | SV *sv_accept_buff = NULL; | |
2836 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2837 | ||
2838 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2839 | ST.B = next; | |
2840 | ST.jump = trie->jump; | |
786e8c11 | 2841 | ST.me = scan; |
786e8c11 | 2842 | |
07be1b83 YO |
2843 | /* |
2844 | traverse the TRIE keeping track of all accepting states | |
2845 | we transition through until we get to a failing node. | |
2846 | */ | |
2847 | ||
a3621e74 | 2848 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 2849 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 2850 | UV uvc = 0; |
786e8c11 YO |
2851 | U16 charid; |
2852 | /* We use charid to hold the wordnum as we don't use it | |
2853 | for charid until after we have done the wordnum logic. | |
2854 | We define an alias just so that the wordnum logic reads | |
2855 | more naturally. */ | |
2856 | ||
2857 | #define got_wordnum charid | |
2858 | got_wordnum = trie->states[ state ].wordnum; | |
2859 | ||
2860 | if ( got_wordnum ) { | |
2861 | if ( ! ST.accepted ) { | |
5b47454d DM |
2862 | ENTER; |
2863 | SAVETMPS; | |
2864 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; | |
2865 | sv_accept_buff=newSV(bufflen * | |
2866 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2867 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2868 | SvPOK_on(sv_accept_buff); |
2869 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2870 | SAVETMPS; |
2871 | ST.accept_buff = | |
5b47454d DM |
2872 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2873 | } | |
786e8c11 | 2874 | do { |
166ba7cd | 2875 | if (ST.accepted >= bufflen) { |
5b47454d | 2876 | bufflen *= 2; |
166ba7cd | 2877 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2878 | SvGROW(sv_accept_buff, |
2879 | bufflen * sizeof(reg_trie_accepted)); | |
2880 | } | |
2881 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2882 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2883 | |
786e8c11 YO |
2884 | |
2885 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2886 | ST.accept_buff[ST.accepted].endpos = uc; | |
2887 | ++ST.accepted; | |
2888 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2889 | } | |
2890 | #undef got_wordnum | |
a3621e74 | 2891 | |
07be1b83 | 2892 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2893 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2894 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2895 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2896 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2897 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2898 | }); |
a3621e74 YO |
2899 | |
2900 | if ( base ) { | |
55eed653 NC |
2901 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
2902 | uscan, len, uvc, charid, foldlen, | |
2903 | foldbuf, uniflags); | |
a3621e74 | 2904 | |
5b47454d DM |
2905 | if (charid && |
2906 | (base + charid > trie->uniquecharcount ) | |
2907 | && (base + charid - 1 - trie->uniquecharcount | |
2908 | < trie->lasttrans) | |
2909 | && trie->trans[base + charid - 1 - | |
2910 | trie->uniquecharcount].check == state) | |
2911 | { | |
2912 | state = trie->trans[base + charid - 1 - | |
2913 | trie->uniquecharcount ].next; | |
2914 | } | |
2915 | else { | |
2916 | state = 0; | |
2917 | } | |
2918 | uc += len; | |
2919 | ||
2920 | } | |
2921 | else { | |
a3621e74 YO |
2922 | state = 0; |
2923 | } | |
2924 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 2925 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2926 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 2927 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
2928 | ); |
2929 | } | |
166ba7cd | 2930 | if (!ST.accepted ) |
a3621e74 | 2931 | sayNO; |
a3621e74 | 2932 | |
166ba7cd DM |
2933 | DEBUG_EXECUTE_r( |
2934 | PerlIO_printf( Perl_debug_log, | |
2935 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 2936 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
2937 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
2938 | ); | |
786e8c11 | 2939 | }} |
166ba7cd DM |
2940 | |
2941 | /* FALL THROUGH */ | |
166ba7cd | 2942 | case TRIE_next_fail: /* we failed - try next alterative */ |
5d458dd8 YO |
2943 | if (do_cutgroup) { |
2944 | do_cutgroup = 0; | |
2945 | no_final = 0; | |
2946 | } | |
166ba7cd DM |
2947 | if ( ST.accepted == 1 ) { |
2948 | /* only one choice left - just continue */ | |
2949 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
2950 | AV *const trie_words |
2951 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
2952 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 2953 | ST.accept_buff[ 0 ].wordnum-1, 0 ); |
de734bd5 A |
2954 | SV *sv= tmp ? sv_newmortal() : NULL; |
2955 | ||
166ba7cd DM |
2956 | PerlIO_printf( Perl_debug_log, |
2957 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 2958 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 2959 | ST.accept_buff[ 0 ].wordnum, |
de734bd5 A |
2960 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
2961 | PL_colors[0], PL_colors[1], | |
2962 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
2963 | ) | |
2964 | : "not compiled under -Dr", | |
166ba7cd DM |
2965 | PL_colors[5] ); |
2966 | }); | |
2967 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
2968 | /* in this case we free tmps/leave before we call regmatch | |
2969 | as we wont be using accept_buff again. */ | |
5d458dd8 | 2970 | |
166ba7cd DM |
2971 | locinput = PL_reginput; |
2972 | nextchr = UCHARAT(locinput); | |
5d458dd8 YO |
2973 | if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum]) |
2974 | scan = ST.B; | |
2975 | else | |
2976 | scan = ST.me + ST.jump[ST.accept_buff[0].wordnum]; | |
2977 | if (!has_cutgroup) { | |
2978 | FREETMPS; | |
2979 | LEAVE; | |
2980 | } else { | |
2981 | ST.accepted--; | |
2982 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
2983 | } | |
786e8c11 | 2984 | |
166ba7cd DM |
2985 | continue; /* execute rest of RE */ |
2986 | } | |
2987 | ||
2988 | if (!ST.accepted-- ) { | |
5d458dd8 YO |
2989 | DEBUG_EXECUTE_r({ |
2990 | PerlIO_printf( Perl_debug_log, | |
2991 | "%*s %sTRIE failed...%s\n", | |
2992 | REPORT_CODE_OFF+depth*2, "", | |
2993 | PL_colors[4], | |
2994 | PL_colors[5] ); | |
2995 | }); | |
166ba7cd DM |
2996 | FREETMPS; |
2997 | LEAVE; | |
5d458dd8 | 2998 | sayNO_SILENT; |
166ba7cd DM |
2999 | } |
3000 | ||
a3621e74 | 3001 | /* |
166ba7cd DM |
3002 | There are at least two accepting states left. Presumably |
3003 | the number of accepting states is going to be low, | |
3004 | typically two. So we simply scan through to find the one | |
3005 | with lowest wordnum. Once we find it, we swap the last | |
3006 | state into its place and decrement the size. We then try to | |
3007 | match the rest of the pattern at the point where the word | |
3008 | ends. If we succeed, control just continues along the | |
3009 | regex; if we fail we return here to try the next accepting | |
3010 | state | |
3011 | */ | |
a3621e74 | 3012 | |
166ba7cd DM |
3013 | { |
3014 | U32 best = 0; | |
3015 | U32 cur; | |
3016 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
3017 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 3018 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 3019 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 3020 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
3021 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
3022 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
3023 | ); | |
3024 | ||
3025 | if (ST.accept_buff[cur].wordnum < | |
3026 | ST.accept_buff[best].wordnum) | |
3027 | best = cur; | |
a3621e74 | 3028 | } |
166ba7cd DM |
3029 | |
3030 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
3031 | AV *const trie_words |
3032 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
3033 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 3034 | ST.accept_buff[ best ].wordnum - 1, 0 ); |
7f69552c | 3035 | regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ? |
265c4333 | 3036 | ST.B : |
7f69552c | 3037 | ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
de734bd5 A |
3038 | SV *sv= tmp ? sv_newmortal() : NULL; |
3039 | ||
265c4333 YO |
3040 | PerlIO_printf( Perl_debug_log, |
3041 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 3042 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3043 | ST.accept_buff[best].wordnum, |
de734bd5 A |
3044 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3045 | PL_colors[0], PL_colors[1], | |
3046 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3047 | ) : "not compiled under -Dr", | |
265c4333 | 3048 | REG_NODE_NUM(nextop), |
166ba7cd DM |
3049 | PL_colors[5] ); |
3050 | }); | |
3051 | ||
3052 | if ( best<ST.accepted ) { | |
3053 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
3054 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
3055 | ST.accept_buff[ ST.accepted ] = tmp; | |
3056 | best = ST.accepted; | |
a3621e74 | 3057 | } |
166ba7cd | 3058 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
7f69552c | 3059 | if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) { |
5d458dd8 | 3060 | scan = ST.B; |
786e8c11 YO |
3061 | /* NOTREACHED */ |
3062 | } else { | |
5d458dd8 | 3063 | scan = ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
786e8c11 YO |
3064 | /* NOTREACHED */ |
3065 | } | |
5d458dd8 YO |
3066 | if (has_cutgroup) { |
3067 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
3068 | /* NOTREACHED */ | |
3069 | } else { | |
3070 | PUSH_STATE_GOTO(TRIE_next, scan); | |
3071 | /* NOTREACHED */ | |
3072 | } | |
786e8c11 | 3073 | /* NOTREACHED */ |
166ba7cd | 3074 | } |
166ba7cd | 3075 | /* NOTREACHED */ |
5d458dd8 YO |
3076 | case TRIE_next: |
3077 | FREETMPS; | |
3078 | LEAVE; | |
3079 | sayYES; | |
166ba7cd DM |
3080 | #undef ST |
3081 | ||
95b24440 DM |
3082 | case EXACT: { |
3083 | char *s = STRING(scan); | |
24d3c4a9 | 3084 | ln = STR_LEN(scan); |
eb160463 | 3085 | if (do_utf8 != UTF) { |
bc517b45 | 3086 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 3087 | char *l = locinput; |
24d3c4a9 | 3088 | const char * const e = s + ln; |
a72c7584 | 3089 | |
5ff6fc6d JH |
3090 | if (do_utf8) { |
3091 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 3092 | while (s < e) { |
a3b680e6 | 3093 | STRLEN ulen; |
1aa99e6b | 3094 | if (l >= PL_regeol) |
5ff6fc6d JH |
3095 | sayNO; |
3096 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 3097 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 3098 | uniflags)) |
5ff6fc6d | 3099 | sayNO; |
bc517b45 | 3100 | l += ulen; |
5ff6fc6d | 3101 | s ++; |
1aa99e6b | 3102 | } |
5ff6fc6d JH |
3103 | } |
3104 | else { | |
3105 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 3106 | while (s < e) { |
a3b680e6 | 3107 | STRLEN ulen; |
1aa99e6b IH |
3108 | if (l >= PL_regeol) |
3109 | sayNO; | |
5ff6fc6d | 3110 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 3111 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 3112 | uniflags)) |
1aa99e6b | 3113 | sayNO; |
bc517b45 | 3114 | s += ulen; |
a72c7584 | 3115 | l ++; |
1aa99e6b | 3116 | } |
5ff6fc6d | 3117 | } |
1aa99e6b IH |
3118 | locinput = l; |
3119 | nextchr = UCHARAT(locinput); | |
3120 | break; | |
3121 | } | |
bc517b45 | 3122 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
3123 | /* Inline the first character, for speed. */ |
3124 | if (UCHARAT(s) != nextchr) | |
3125 | sayNO; | |
24d3c4a9 | 3126 | if (PL_regeol - locinput < ln) |
d6a28714 | 3127 | sayNO; |
24d3c4a9 | 3128 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3129 | sayNO; |
24d3c4a9 | 3130 | locinput += ln; |
d6a28714 JH |
3131 | nextchr = UCHARAT(locinput); |
3132 | break; | |
95b24440 | 3133 | } |
d6a28714 | 3134 | case EXACTFL: |
b8c5462f JH |
3135 | PL_reg_flags |= RF_tainted; |
3136 | /* FALL THROUGH */ | |
95b24440 | 3137 | case EXACTF: { |
be8e71aa | 3138 | char * const s = STRING(scan); |
24d3c4a9 | 3139 | ln = STR_LEN(scan); |
d6a28714 | 3140 | |
d07ddd77 JH |
3141 | if (do_utf8 || UTF) { |
3142 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3143 | const char * const l = locinput; |
d07ddd77 | 3144 | char *e = PL_regeol; |
bc517b45 | 3145 | |
24d3c4a9 | 3146 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3147 | l, &e, 0, do_utf8)) { |
5486206c JH |
3148 | /* One more case for the sharp s: |
3149 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3150 | * the 0xC3 0x9F are the UTF-8 | |
3151 | * byte sequence for the U+00DF. */ | |
3152 | if (!(do_utf8 && | |
3153 | toLOWER(s[0]) == 's' && | |
24d3c4a9 | 3154 | ln >= 2 && |
5486206c JH |
3155 | toLOWER(s[1]) == 's' && |
3156 | (U8)l[0] == 0xC3 && | |
3157 | e - l >= 2 && | |
3158 | (U8)l[1] == 0x9F)) | |
3159 | sayNO; | |
3160 | } | |
d07ddd77 JH |
3161 | locinput = e; |
3162 | nextchr = UCHARAT(locinput); | |
3163 | break; | |
a0ed51b3 | 3164 | } |
d6a28714 | 3165 | |
bc517b45 JH |
3166 | /* Neither the target and the pattern are utf8. */ |
3167 | ||
d6a28714 JH |
3168 | /* Inline the first character, for speed. */ |
3169 | if (UCHARAT(s) != nextchr && | |
3170 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3171 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3172 | sayNO; |
24d3c4a9 | 3173 | if (PL_regeol - locinput < ln) |
b8c5462f | 3174 | sayNO; |
24d3c4a9 DM |
3175 | if (ln > 1 && (OP(scan) == EXACTF |
3176 | ? ibcmp(s, locinput, ln) | |
3177 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3178 | sayNO; |
24d3c4a9 | 3179 | locinput += ln; |
d6a28714 | 3180 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3181 | break; |
95b24440 | 3182 | } |
d6a28714 | 3183 | case ANYOF: |
ffc61ed2 | 3184 | if (do_utf8) { |
9e55ce06 JH |
3185 | STRLEN inclasslen = PL_regeol - locinput; |
3186 | ||
32fc9b6a | 3187 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3188 | goto anyof_fail; |
ffc61ed2 JH |
3189 | if (locinput >= PL_regeol) |
3190 | sayNO; | |
0f0076b4 | 3191 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3192 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3193 | break; |
ffc61ed2 JH |
3194 | } |
3195 | else { | |
3196 | if (nextchr < 0) | |
3197 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3198 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3199 | goto anyof_fail; |
ffc61ed2 JH |
3200 | if (!nextchr && locinput >= PL_regeol) |
3201 | sayNO; | |
3202 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3203 | break; |
3204 | } | |
262b90c4 | 3205 | anyof_fail: |
e0f9d4a8 JH |
3206 | /* If we might have the case of the German sharp s |
3207 | * in a casefolding Unicode character class. */ | |
3208 | ||
ebc501f0 JH |
3209 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3210 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3211 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3212 | } |
e0f9d4a8 JH |
3213 | else |
3214 | sayNO; | |
b8c5462f | 3215 | break; |
d6a28714 | 3216 | case ALNUML: |
b8c5462f JH |
3217 | PL_reg_flags |= RF_tainted; |
3218 | /* FALL THROUGH */ | |
d6a28714 | 3219 | case ALNUM: |
b8c5462f | 3220 | if (!nextchr) |
4633a7c4 | 3221 | sayNO; |
ffc61ed2 | 3222 | if (do_utf8) { |
1a4fad37 | 3223 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3224 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3225 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3226 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3227 | { |
3228 | sayNO; | |
a0ed51b3 | 3229 | } |
b8c5462f | 3230 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3231 | nextchr = UCHARAT(locinput); |
3232 | break; | |
3233 | } | |
ffc61ed2 | 3234 | if (!(OP(scan) == ALNUM |
d6a28714 | 3235 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3236 | sayNO; |
b8c5462f | 3237 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3238 | break; |
d6a28714 | 3239 | case NALNUML: |
b8c5462f JH |
3240 | PL_reg_flags |= RF_tainted; |
3241 | /* FALL THROUGH */ | |
d6a28714 JH |
3242 | case NALNUM: |
3243 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3244 | sayNO; |
ffc61ed2 | 3245 | if (do_utf8) { |
1a4fad37 | 3246 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3247 | if (OP(scan) == NALNUM |
bb7a0f54 | 3248 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3249 | : isALNUM_LC_utf8((U8*)locinput)) |
3250 | { | |
b8c5462f | 3251 | sayNO; |
d6a28714 | 3252 | } |
b8c5462f JH |
3253 | locinput += PL_utf8skip[nextchr]; |
3254 | nextchr = UCHARAT(locinput); | |
3255 | break; | |
3256 | } | |
ffc61ed2 | 3257 | if (OP(scan) == NALNUM |
d6a28714 | 3258 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3259 | sayNO; |
76e3520e | 3260 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3261 | break; |
d6a28714 JH |
3262 | case BOUNDL: |
3263 | case NBOUNDL: | |
3280af22 | 3264 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3265 | /* FALL THROUGH */ |
d6a28714 JH |
3266 | case BOUND: |
3267 | case NBOUND: | |
3268 | /* was last char in word? */ | |
ffc61ed2 | 3269 | if (do_utf8) { |
12d33761 | 3270 | if (locinput == PL_bostr) |
24d3c4a9 | 3271 | ln = '\n'; |
ffc61ed2 | 3272 | else { |
a3b680e6 | 3273 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3274 | |
24d3c4a9 | 3275 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3276 | } |
3277 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3278 | ln = isALNUM_uni(ln); |
1a4fad37 | 3279 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3280 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3281 | } |
3282 | else { | |
24d3c4a9 | 3283 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3284 | n = isALNUM_LC_utf8((U8*)locinput); |
3285 | } | |
a0ed51b3 | 3286 | } |
d6a28714 | 3287 | else { |
24d3c4a9 | 3288 | ln = (locinput != PL_bostr) ? |
12d33761 | 3289 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3290 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3291 | ln = isALNUM(ln); |
ffc61ed2 JH |
3292 | n = isALNUM(nextchr); |
3293 | } | |
3294 | else { | |
24d3c4a9 | 3295 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3296 | n = isALNUM_LC(nextchr); |
3297 | } | |
d6a28714 | 3298 | } |
24d3c4a9 | 3299 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3300 | OP(scan) == BOUNDL)) |
3301 | sayNO; | |
a0ed51b3 | 3302 | break; |
d6a28714 | 3303 | case SPACEL: |
3280af22 | 3304 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3305 | /* FALL THROUGH */ |
d6a28714 | 3306 | case SPACE: |
9442cb0e | 3307 | if (!nextchr) |
4633a7c4 | 3308 | sayNO; |
1aa99e6b | 3309 | if (do_utf8) { |
fd400ab9 | 3310 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3311 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3312 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3313 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3314 | : isSPACE_LC_utf8((U8*)locinput))) |
3315 | { | |
3316 | sayNO; | |
3317 | } | |
3318 | locinput += PL_utf8skip[nextchr]; | |
3319 | nextchr = UCHARAT(locinput); | |
3320 | break; | |
d6a28714 | 3321 | } |
ffc61ed2 JH |
3322 | if (!(OP(scan) == SPACE |
3323 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3324 | sayNO; | |
3325 | nextchr = UCHARAT(++locinput); | |
3326 | } | |
3327 | else { | |
3328 | if (!(OP(scan) == SPACE | |
3329 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3330 | sayNO; | |
3331 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3332 | } |
a0ed51b3 | 3333 | break; |
d6a28714 | 3334 | case NSPACEL: |
3280af22 | 3335 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3336 | /* FALL THROUGH */ |
d6a28714 | 3337 | case NSPACE: |
9442cb0e | 3338 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3339 | sayNO; |
1aa99e6b | 3340 | if (do_utf8) { |
1a4fad37 | 3341 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3342 | if (OP(scan) == NSPACE |
bb7a0f54 | 3343 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3344 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3345 | { |
3346 | sayNO; | |
3347 | } | |
3348 | locinput += PL_utf8skip[nextchr]; | |
3349 | nextchr = UCHARAT(locinput); | |
3350 | break; | |
a0ed51b3 | 3351 | } |
ffc61ed2 | 3352 | if (OP(scan) == NSPACE |
d6a28714 | 3353 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3354 | sayNO; |
76e3520e | 3355 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3356 | break; |
d6a28714 | 3357 | case DIGITL: |
a0ed51b3 LW |
3358 | PL_reg_flags |= RF_tainted; |
3359 | /* FALL THROUGH */ | |
d6a28714 | 3360 | case DIGIT: |
9442cb0e | 3361 | if (!nextchr) |
a0ed51b3 | 3362 | sayNO; |
1aa99e6b | 3363 | if (do_utf8) { |
1a4fad37 | 3364 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3365 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3366 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3367 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3368 | { |
a0ed51b3 | 3369 | sayNO; |
dfe13c55 | 3370 | } |
6f06b55f | 3371 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3372 | nextchr = UCHARAT(locinput); |
3373 | break; | |
3374 | } | |
ffc61ed2 | 3375 | if (!(OP(scan) == DIGIT |
9442cb0e | 3376 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3377 | sayNO; |
3378 | nextchr = UCHARAT(++locinput); | |
3379 | break; | |
d6a28714 | 3380 | case NDIGITL: |
b8c5462f JH |
3381 | PL_reg_flags |= RF_tainted; |
3382 | /* FALL THROUGH */ | |
d6a28714 | 3383 | case NDIGIT: |
9442cb0e | 3384 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3385 | sayNO; |
1aa99e6b | 3386 | if (do_utf8) { |
1a4fad37 | 3387 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3388 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3389 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3390 | : isDIGIT_LC_utf8((U8*)locinput)) |
3391 | { | |
a0ed51b3 | 3392 | sayNO; |
9442cb0e | 3393 | } |
6f06b55f | 3394 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3395 | nextchr = UCHARAT(locinput); |
3396 | break; | |
3397 | } | |
ffc61ed2 | 3398 | if (OP(scan) == NDIGIT |
9442cb0e | 3399 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3400 | sayNO; |
3401 | nextchr = UCHARAT(++locinput); | |
3402 | break; | |
3403 | case CLUMP: | |
b7c83a7e | 3404 | if (locinput >= PL_regeol) |
a0ed51b3 | 3405 | sayNO; |
b7c83a7e | 3406 | if (do_utf8) { |
1a4fad37 | 3407 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3408 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3409 | sayNO; | |
3410 | locinput += PL_utf8skip[nextchr]; | |
3411 | while (locinput < PL_regeol && | |
3412 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3413 | locinput += UTF8SKIP(locinput); | |
3414 | if (locinput > PL_regeol) | |
3415 | sayNO; | |
eb08e2da JH |
3416 | } |
3417 | else | |
3418 | locinput++; | |
a0ed51b3 LW |
3419 | nextchr = UCHARAT(locinput); |
3420 | break; | |
81714fb9 YO |
3421 | |
3422 | case NREFFL: | |
3423 | { | |
3424 | char *s; | |
ff1157ca | 3425 | char type; |
81714fb9 YO |
3426 | PL_reg_flags |= RF_tainted; |
3427 | /* FALL THROUGH */ | |
3428 | case NREF: | |
3429 | case NREFF: | |
ff1157ca | 3430 | type = OP(scan); |
0a4db386 YO |
3431 | n = reg_check_named_buff_matched(rex,scan); |
3432 | ||
3433 | if ( n ) { | |
3434 | type = REF + ( type - NREF ); | |
3435 | goto do_ref; | |
3436 | } else { | |
81714fb9 | 3437 | sayNO; |
0a4db386 YO |
3438 | } |
3439 | /* unreached */ | |
c8756f30 | 3440 | case REFFL: |
3280af22 | 3441 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3442 | /* FALL THROUGH */ |
c277df42 | 3443 | case REF: |
81714fb9 | 3444 | case REFF: |
c277df42 | 3445 | n = ARG(scan); /* which paren pair */ |
81714fb9 YO |
3446 | type = OP(scan); |
3447 | do_ref: | |
24d3c4a9 | 3448 | ln = PL_regstartp[n]; |
2c2d71f5 | 3449 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
3b6647e0 | 3450 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 | 3451 | sayNO; /* Do not match unless seen CLOSEn. */ |
24d3c4a9 | 3452 | if (ln == PL_regendp[n]) |
a0d0e21e | 3453 | break; |
a0ed51b3 | 3454 | |
24d3c4a9 | 3455 | s = PL_bostr + ln; |
81714fb9 | 3456 | if (do_utf8 && type != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3457 | char *l = locinput; |
a3b680e6 | 3458 | const char *e = PL_bostr + PL_regendp[n]; |
a0ed51b3 LW |
3459 | /* |
3460 | * Note that we can't do the "other character" lookup trick as | |
3461 | * in the 8-bit case (no pun intended) because in Unicode we | |
3462 | * have to map both upper and title case to lower case. | |
3463 | */ | |
81714fb9 | 3464 | if (type == REFF) { |
a0ed51b3 | 3465 | while (s < e) { |
a3b680e6 AL |
3466 | STRLEN ulen1, ulen2; |
3467 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3468 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3469 | ||
a0ed51b3 LW |
3470 | if (l >= PL_regeol) |
3471 | sayNO; | |
a2a2844f JH |
3472 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3473 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3474 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3475 | sayNO; |
a2a2844f JH |
3476 | s += ulen1; |
3477 | l += ulen2; | |
a0ed51b3 LW |
3478 | } |
3479 | } | |
3480 | locinput = l; | |
3481 | nextchr = UCHARAT(locinput); | |
3482 | break; | |
3483 | } | |
3484 | ||
a0d0e21e | 3485 | /* Inline the first character, for speed. */ |
76e3520e | 3486 | if (UCHARAT(s) != nextchr && |
81714fb9 YO |
3487 | (type == REF || |
3488 | (UCHARAT(s) != (type == REFF | |
3489 | ? PL_fold : PL_fold_locale)[nextchr]))) | |
4633a7c4 | 3490 | sayNO; |
24d3c4a9 DM |
3491 | ln = PL_regendp[n] - ln; |
3492 | if (locinput + ln > PL_regeol) | |
4633a7c4 | 3493 | sayNO; |
81714fb9 | 3494 | if (ln > 1 && (type == REF |
24d3c4a9 | 3495 | ? memNE(s, locinput, ln) |
81714fb9 | 3496 | : (type == REFF |
24d3c4a9 DM |
3497 | ? ibcmp(s, locinput, ln) |
3498 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 3499 | sayNO; |
24d3c4a9 | 3500 | locinput += ln; |
76e3520e | 3501 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3502 | break; |
81714fb9 | 3503 | } |
a0d0e21e | 3504 | case NOTHING: |
c277df42 | 3505 | case TAIL: |
a0d0e21e LW |
3506 | break; |
3507 | case BACK: | |
3508 | break; | |
40a82448 DM |
3509 | |
3510 | #undef ST | |
3511 | #define ST st->u.eval | |
c277df42 | 3512 | { |
c277df42 | 3513 | SV *ret; |
6bda09f9 | 3514 | regexp *re; |
f8fc2ecf | 3515 | regexp_internal *rei; |
1a147d38 YO |
3516 | regnode *startpoint; |
3517 | ||
3518 | case GOSTART: | |
3519 | case GOSUB: /* /(...(?1))/ */ | |
6bda09f9 | 3520 | if (cur_eval && cur_eval->locinput==locinput) { |
24b23f37 | 3521 | if (cur_eval->u.eval.close_paren == (U32)ARG(scan)) |
1a147d38 | 3522 | Perl_croak(aTHX_ "Infinite recursion in regex"); |
6bda09f9 | 3523 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) |
1a147d38 YO |
3524 | Perl_croak(aTHX_ |
3525 | "Pattern subroutine nesting without pos change" | |
3526 | " exceeded limit in regex"); | |
6bda09f9 YO |
3527 | } else { |
3528 | nochange_depth = 0; | |
1a147d38 | 3529 | } |
6bda09f9 | 3530 | re = rex; |
f8fc2ecf | 3531 | rei = rexi; |
6bda09f9 | 3532 | (void)ReREFCNT_inc(rex); |
1a147d38 | 3533 | if (OP(scan)==GOSUB) { |
6bda09f9 YO |
3534 | startpoint = scan + ARG2L(scan); |
3535 | ST.close_paren = ARG(scan); | |
3536 | } else { | |
f8fc2ecf | 3537 | startpoint = rei->program+1; |
6bda09f9 YO |
3538 | ST.close_paren = 0; |
3539 | } | |
3540 | goto eval_recurse_doit; | |
3541 | /* NOTREACHED */ | |
3542 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
3543 | if (cur_eval && cur_eval->locinput==locinput) { | |
3544 | if ( ++nochange_depth > MAX_RECURSE_EVAL_NOCHANGE_DEPTH ) | |
1a147d38 | 3545 | Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex"); |
6bda09f9 YO |
3546 | } else { |
3547 | nochange_depth = 0; | |
3548 | } | |
8e5e9ebe | 3549 | { |
4aabdb9b DM |
3550 | /* execute the code in the {...} */ |
3551 | dSP; | |
6136c704 | 3552 | SV ** const before = SP; |
4aabdb9b DM |
3553 | OP_4tree * const oop = PL_op; |
3554 | COP * const ocurcop = PL_curcop; | |
3555 | PAD *old_comppad; | |
4aabdb9b DM |
3556 | |
3557 | n = ARG(scan); | |
f8fc2ecf | 3558 | PL_op = (OP_4tree*)rexi->data->data[n]; |
24b23f37 YO |
3559 | DEBUG_STATE_r( PerlIO_printf(Perl_debug_log, |
3560 | " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); | |
f8fc2ecf | 3561 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]); |
4aabdb9b DM |
3562 | PL_regendp[0] = PL_reg_magic->mg_len = locinput - PL_bostr; |
3563 | ||
2bf803e2 YO |
3564 | if (sv_yes_mark) { |
3565 | SV *sv_mrk = get_sv("REGMARK", 1); | |
3566 | sv_setsv(sv_mrk, sv_yes_mark); | |
3567 | } | |
3568 | ||
8e5e9ebe RGS |
3569 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3570 | SPAGAIN; | |
3571 | if (SP == before) | |
075aa684 | 3572 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3573 | else { |
3574 | ret = POPs; | |
3575 | PUTBACK; | |
3576 | } | |
4aabdb9b DM |
3577 | |
3578 | PL_op = oop; | |
3579 | PAD_RESTORE_LOCAL(old_comppad); | |
3580 | PL_curcop = ocurcop; | |
24d3c4a9 | 3581 | if (!logical) { |
4aabdb9b DM |
3582 | /* /(?{...})/ */ |
3583 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3584 | break; |
3585 | } | |
8e5e9ebe | 3586 | } |
24d3c4a9 DM |
3587 | if (logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3588 | logical = 0; | |
4aabdb9b | 3589 | { |
4f639d21 DM |
3590 | /* extract RE object from returned value; compiling if |
3591 | * necessary */ | |
3592 | ||
6136c704 | 3593 | MAGIC *mg = NULL; |
be8e71aa | 3594 | const SV *sv; |
faf82a0b AE |
3595 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3596 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3597 | else if (SvSMAGICAL(ret)) { | |
3598 | if (SvGMAGICAL(ret)) | |
3599 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3600 | else | |
3601 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3602 | } |
faf82a0b | 3603 | |
0f5d15d6 IZ |
3604 | if (mg) { |
3605 | re = (regexp *)mg->mg_obj; | |
df0003d4 | 3606 | (void)ReREFCNT_inc(re); |
0f5d15d6 IZ |
3607 | } |
3608 | else { | |
3609 | STRLEN len; | |
6136c704 | 3610 | const char * const t = SvPV_const(ret, len); |
0f5d15d6 | 3611 | PMOP pm; |
a3b680e6 | 3612 | const I32 osize = PL_regsize; |
0f5d15d6 | 3613 | |
5fcd1c1b | 3614 | Zero(&pm, 1, PMOP); |
4aabdb9b | 3615 | if (DO_UTF8(ret)) pm.op_pmdynflags |= PMdf_DYN_UTF8; |
f9f4320a | 3616 | re = CALLREGCOMP((char*)t, (char*)t + len, &pm); |
9041c2e3 | 3617 | if (!(SvFLAGS(ret) |
faf82a0b AE |
3618 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3619 | | SVs_GMG))) | |
14befaf4 DM |
3620 | sv_magic(ret,(SV*)ReREFCNT_inc(re), |
3621 | PERL_MAGIC_qr,0,0); | |
0f5d15d6 | 3622 | PL_regsize = osize; |
0f5d15d6 | 3623 | } |
4aabdb9b | 3624 | } |
f8fc2ecf | 3625 | rei = RXi_GET(re); |
6bda09f9 YO |
3626 | DEBUG_EXECUTE_r( |
3627 | debug_start_match(re, do_utf8, locinput, PL_regeol, | |
3628 | "Matching embedded"); | |
3629 | ); | |
f8fc2ecf | 3630 | startpoint = rei->program + 1; |
1a147d38 | 3631 | ST.close_paren = 0; /* only used for GOSUB */ |
6bda09f9 YO |
3632 | /* borrowed from regtry */ |
3633 | if (PL_reg_start_tmpl <= re->nparens) { | |
3634 | PL_reg_start_tmpl = re->nparens*3/2 + 3; | |
3635 | if(PL_reg_start_tmp) | |
3636 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3637 | else | |
3638 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3639 | } | |
aa283a38 | 3640 | |
1a147d38 | 3641 | eval_recurse_doit: /* Share code with GOSUB below this line */ |
aa283a38 | 3642 | /* run the pattern returned from (??{...}) */ |
40a82448 DM |
3643 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3644 | REGCP_SET(ST.lastcp); | |
6bda09f9 | 3645 | |
1a147d38 YO |
3646 | PL_regstartp = re->startp; /* essentially NOOP on GOSUB */ |
3647 | PL_regendp = re->endp; /* essentially NOOP on GOSUB */ | |
6bda09f9 | 3648 | |
4aabdb9b DM |
3649 | *PL_reglastparen = 0; |
3650 | *PL_reglastcloseparen = 0; | |
4aabdb9b | 3651 | PL_reginput = locinput; |
ae0beba1 | 3652 | PL_regsize = 0; |
4aabdb9b DM |
3653 | |
3654 | /* XXXX This is too dramatic a measure... */ | |
3655 | PL_reg_maxiter = 0; | |
3656 | ||
faec1544 | 3657 | ST.toggle_reg_flags = PL_reg_flags; |
bbe252da | 3658 | if (re->extflags & RXf_UTF8) |
faec1544 DM |
3659 | PL_reg_flags |= RF_utf8; |
3660 | else | |
3661 | PL_reg_flags &= ~RF_utf8; | |
3662 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3663 | ||
40a82448 | 3664 | ST.prev_rex = rex; |
faec1544 | 3665 | ST.prev_curlyx = cur_curlyx; |
aa283a38 | 3666 | rex = re; |
f8fc2ecf | 3667 | rexi = rei; |
faec1544 | 3668 | cur_curlyx = NULL; |
40a82448 | 3669 | ST.B = next; |
faec1544 DM |
3670 | ST.prev_eval = cur_eval; |
3671 | cur_eval = st; | |
faec1544 | 3672 | /* now continue from first node in postoned RE */ |
6bda09f9 | 3673 | PUSH_YES_STATE_GOTO(EVAL_AB, startpoint); |
4aabdb9b | 3674 | /* NOTREACHED */ |
a0ed51b3 | 3675 | } |
24d3c4a9 DM |
3676 | /* logical is 1, /(?(?{...})X|Y)/ */ |
3677 | sw = (bool)SvTRUE(ret); | |
3678 | logical = 0; | |
c277df42 IZ |
3679 | break; |
3680 | } | |
40a82448 | 3681 | |
faec1544 DM |
3682 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3683 | /* note: this is called twice; first after popping B, then A */ | |
3684 | PL_reg_flags ^= ST.toggle_reg_flags; | |
40a82448 DM |
3685 | ReREFCNT_dec(rex); |
3686 | rex = ST.prev_rex; | |
f8fc2ecf | 3687 | rexi = RXi_GET(rex); |
faec1544 DM |
3688 | regcpblow(ST.cp); |
3689 | cur_eval = ST.prev_eval; | |
3690 | cur_curlyx = ST.prev_curlyx; | |
40a82448 DM |
3691 | /* XXXX This is too dramatic a measure... */ |
3692 | PL_reg_maxiter = 0; | |
262b90c4 | 3693 | sayYES; |
40a82448 | 3694 | |
40a82448 | 3695 | |
faec1544 DM |
3696 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3697 | /* note: this is called twice; first after popping B, then A */ | |
3698 | PL_reg_flags ^= ST.toggle_reg_flags; | |
3699 | ReREFCNT_dec(rex); | |
3700 | rex = ST.prev_rex; | |
f8fc2ecf | 3701 | rexi = RXi_GET(rex); |
40a82448 DM |
3702 | PL_reginput = locinput; |
3703 | REGCP_UNWIND(ST.lastcp); | |
3704 | regcppop(rex); | |
faec1544 DM |
3705 | cur_eval = ST.prev_eval; |
3706 | cur_curlyx = ST.prev_curlyx; | |
3707 | /* XXXX This is too dramatic a measure... */ | |
3708 | PL_reg_maxiter = 0; | |
40a82448 | 3709 | sayNO_SILENT; |
40a82448 DM |
3710 | #undef ST |
3711 | ||
a0d0e21e | 3712 | case OPEN: |
c277df42 | 3713 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3714 | PL_reg_start_tmp[n] = locinput; |
3715 | if (n > PL_regsize) | |
3716 | PL_regsize = n; | |
e2e6a0f1 | 3717 | lastopen = n; |
a0d0e21e LW |
3718 | break; |
3719 | case CLOSE: | |
c277df42 | 3720 | n = ARG(scan); /* which paren pair */ |
cf93c79d IZ |
3721 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; |
3722 | PL_regendp[n] = locinput - PL_bostr; | |
7f69552c YO |
3723 | /*if (n > PL_regsize) |
3724 | PL_regsize = n;*/ | |
3b6647e0 | 3725 | if (n > *PL_reglastparen) |
3280af22 | 3726 | *PL_reglastparen = n; |
a01268b5 | 3727 | *PL_reglastcloseparen = n; |
3b6647e0 | 3728 | if (cur_eval && cur_eval->u.eval.close_paren == n) { |
6bda09f9 YO |
3729 | goto fake_end; |
3730 | } | |
a0d0e21e | 3731 | break; |
e2e6a0f1 YO |
3732 | case ACCEPT: |
3733 | if (ARG(scan)){ | |
3734 | regnode *cursor; | |
3735 | for (cursor=scan; | |
3736 | cursor && OP(cursor)!=END; | |
3737 | cursor=regnext(cursor)) | |
3738 | { | |
3739 | if ( OP(cursor)==CLOSE ){ | |
3740 | n = ARG(cursor); | |
3741 | if ( n <= lastopen ) { | |
3742 | PL_regstartp[n] = PL_reg_start_tmp[n] - PL_bostr; | |
3743 | PL_regendp[n] = locinput - PL_bostr; | |
3744 | /*if (n > PL_regsize) | |
3745 | PL_regsize = n;*/ | |
3b6647e0 | 3746 | if (n > *PL_reglastparen) |
e2e6a0f1 YO |
3747 | *PL_reglastparen = n; |
3748 | *PL_reglastcloseparen = n; | |
3b6647e0 RB |
3749 | if ( n == ARG(scan) || (cur_eval && |
3750 | cur_eval->u.eval.close_paren == n)) | |
e2e6a0f1 YO |
3751 | break; |
3752 | } | |
3753 | } | |
3754 | } | |
3755 | } | |
3756 | goto fake_end; | |
3757 | /*NOTREACHED*/ | |
c277df42 IZ |
3758 | case GROUPP: |
3759 | n = ARG(scan); /* which paren pair */ | |
3b6647e0 | 3760 | sw = (bool)(*PL_reglastparen >= n && PL_regendp[n] != -1); |
c277df42 | 3761 | break; |
0a4db386 YO |
3762 | case NGROUPP: |
3763 | /* reg_check_named_buff_matched returns 0 for no match */ | |
3764 | sw = (bool)(0 < reg_check_named_buff_matched(rex,scan)); | |
3765 | break; | |
1a147d38 | 3766 | case INSUBP: |
0a4db386 | 3767 | n = ARG(scan); |
3b6647e0 | 3768 | sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n)); |
0a4db386 YO |
3769 | break; |
3770 | case DEFINEP: | |
3771 | sw = 0; | |
3772 | break; | |
c277df42 | 3773 | case IFTHEN: |
2c2d71f5 | 3774 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3775 | if (sw) |
c277df42 IZ |
3776 | next = NEXTOPER(NEXTOPER(scan)); |
3777 | else { | |
3778 | next = scan + ARG(scan); | |
3779 | if (OP(next) == IFTHEN) /* Fake one. */ | |
3780 | next = NEXTOPER(NEXTOPER(next)); | |
3781 | } | |
3782 | break; | |
3783 | case LOGICAL: | |
24d3c4a9 | 3784 | logical = scan->flags; |
c277df42 | 3785 | break; |
c476f425 | 3786 | |
2ab05381 | 3787 | /******************************************************************* |
2ab05381 | 3788 | |
c476f425 DM |
3789 | The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/ |
3790 | pattern, where A and B are subpatterns. (For simple A, CURLYM or | |
3791 | STAR/PLUS/CURLY/CURLYN are used instead.) | |
2ab05381 | 3792 | |
c476f425 | 3793 | A*B is compiled as <CURLYX><A><WHILEM><B> |
2ab05381 | 3794 | |
c476f425 DM |
3795 | On entry to the subpattern, CURLYX is called. This pushes a CURLYX |
3796 | state, which contains the current count, initialised to -1. It also sets | |
3797 | cur_curlyx to point to this state, with any previous value saved in the | |
3798 | state block. | |
2ab05381 | 3799 | |
c476f425 DM |
3800 | CURLYX then jumps straight to the WHILEM op, rather than executing A, |
3801 | since the pattern may possibly match zero times (i.e. it's a while {} loop | |
3802 | rather than a do {} while loop). | |
2ab05381 | 3803 | |
c476f425 DM |
3804 | Each entry to WHILEM represents a successful match of A. The count in the |
3805 | CURLYX block is incremented, another WHILEM state is pushed, and execution | |
3806 | passes to A or B depending on greediness and the current count. | |
2ab05381 | 3807 | |
c476f425 DM |
3808 | For example, if matching against the string a1a2a3b (where the aN are |
3809 | substrings that match /A/), then the match progresses as follows: (the | |
3810 | pushed states are interspersed with the bits of strings matched so far): | |
2ab05381 | 3811 | |
c476f425 DM |
3812 | <CURLYX cnt=-1> |
3813 | <CURLYX cnt=0><WHILEM> | |
3814 | <CURLYX cnt=1><WHILEM> a1 <WHILEM> | |
3815 | <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM> | |
3816 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> | |
3817 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b | |
2ab05381 | 3818 | |
c476f425 DM |
3819 | (Contrast this with something like CURLYM, which maintains only a single |
3820 | backtrack state: | |
2ab05381 | 3821 | |
c476f425 DM |
3822 | <CURLYM cnt=0> a1 |
3823 | a1 <CURLYM cnt=1> a2 | |
3824 | a1 a2 <CURLYM cnt=2> a3 | |
3825 | a1 a2 a3 <CURLYM cnt=3> b | |
3826 | ) | |
2ab05381 | 3827 | |
c476f425 DM |
3828 | Each WHILEM state block marks a point to backtrack to upon partial failure |
3829 | of A or B, and also contains some minor state data related to that | |
3830 | iteration. The CURLYX block, pointed to by cur_curlyx, contains the | |
3831 | overall state, such as the count, and pointers to the A and B ops. | |
2ab05381 | 3832 | |
c476f425 DM |
3833 | This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx |
3834 | must always point to the *current* CURLYX block, the rules are: | |
2ab05381 | 3835 | |
c476f425 DM |
3836 | When executing CURLYX, save the old cur_curlyx in the CURLYX state block, |
3837 | and set cur_curlyx to point the new block. | |
2ab05381 | 3838 | |
c476f425 DM |
3839 | When popping the CURLYX block after a successful or unsuccessful match, |
3840 | restore the previous cur_curlyx. | |
2ab05381 | 3841 | |
c476f425 DM |
3842 | When WHILEM is about to execute B, save the current cur_curlyx, and set it |
3843 | to the outer one saved in the CURLYX block. | |
2ab05381 | 3844 | |
c476f425 DM |
3845 | When popping the WHILEM block after a successful or unsuccessful B match, |
3846 | restore the previous cur_curlyx. | |
2ab05381 | 3847 | |
c476f425 DM |
3848 | Here's an example for the pattern (AI* BI)*BO |
3849 | I and O refer to inner and outer, C and W refer to CURLYX and WHILEM: | |
2ab05381 | 3850 | |
c476f425 DM |
3851 | cur_ |
3852 | curlyx backtrack stack | |
3853 | ------ --------------- | |
3854 | NULL | |
3855 | CO <CO prev=NULL> <WO> | |
3856 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3857 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi | |
3858 | NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo | |
2ab05381 | 3859 | |
c476f425 DM |
3860 | At this point the pattern succeeds, and we work back down the stack to |
3861 | clean up, restoring as we go: | |
95b24440 | 3862 | |
c476f425 DM |
3863 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi |
3864 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3865 | CO <CO prev=NULL> <WO> | |
3866 | NULL | |
a0374537 | 3867 | |
c476f425 DM |
3868 | *******************************************************************/ |
3869 | ||
3870 | #define ST st->u.curlyx | |
3871 | ||
3872 | case CURLYX: /* start of /A*B/ (for complex A) */ | |
3873 | { | |
3874 | /* No need to save/restore up to this paren */ | |
3875 | I32 parenfloor = scan->flags; | |
3876 | ||
3877 | assert(next); /* keep Coverity happy */ | |
3878 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
3879 | next += ARG(next); | |
3880 | ||
3881 | /* XXXX Probably it is better to teach regpush to support | |
3882 | parenfloor > PL_regsize... */ | |
3883 | if (parenfloor > (I32)*PL_reglastparen) | |
3884 | parenfloor = *PL_reglastparen; /* Pessimization... */ | |
3885 | ||
3886 | ST.prev_curlyx= cur_curlyx; | |
3887 | cur_curlyx = st; | |
3888 | ST.cp = PL_savestack_ix; | |
3889 | ||
3890 | /* these fields contain the state of the current curly. | |
3891 | * they are accessed by subsequent WHILEMs */ | |
3892 | ST.parenfloor = parenfloor; | |
3893 | ST.min = ARG1(scan); | |
3894 | ST.max = ARG2(scan); | |
3895 | ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
3896 | ST.B = next; | |
24d3c4a9 DM |
3897 | ST.minmod = minmod; |
3898 | minmod = 0; | |
c476f425 DM |
3899 | ST.count = -1; /* this will be updated by WHILEM */ |
3900 | ST.lastloc = NULL; /* this will be updated by WHILEM */ | |
3901 | ||
3902 | PL_reginput = locinput; | |
3903 | PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next)); | |
5f66b61c | 3904 | /* NOTREACHED */ |
c476f425 | 3905 | } |
a0d0e21e | 3906 | |
c476f425 | 3907 | case CURLYX_end: /* just finished matching all of A*B */ |
f0852a51 YO |
3908 | if (PL_reg_eval_set){ |
3909 | SV *pres= GvSV(PL_replgv); | |
3910 | SvREFCNT_inc(pres); | |
3911 | regcpblow(ST.cp); | |
3912 | sv_setsv(GvSV(PL_replgv), pres); | |
3913 | SvREFCNT_dec(pres); | |
3914 | } else { | |
3915 | regcpblow(ST.cp); | |
3916 | } | |
c476f425 DM |
3917 | cur_curlyx = ST.prev_curlyx; |
3918 | sayYES; | |
3919 | /* NOTREACHED */ | |
a0d0e21e | 3920 | |
c476f425 DM |
3921 | case CURLYX_end_fail: /* just failed to match all of A*B */ |
3922 | regcpblow(ST.cp); | |
3923 | cur_curlyx = ST.prev_curlyx; | |
3924 | sayNO; | |
3925 | /* NOTREACHED */ | |
4633a7c4 | 3926 | |
a0d0e21e | 3927 | |
c476f425 DM |
3928 | #undef ST |
3929 | #define ST st->u.whilem | |
3930 | ||
3931 | case WHILEM: /* just matched an A in /A*B/ (for complex A) */ | |
3932 | { | |
3933 | /* see the discussion above about CURLYX/WHILEM */ | |
c476f425 DM |
3934 | I32 n; |
3935 | assert(cur_curlyx); /* keep Coverity happy */ | |
3936 | n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */ | |
3937 | ST.save_lastloc = cur_curlyx->u.curlyx.lastloc; | |
3938 | ST.cache_offset = 0; | |
3939 | ST.cache_mask = 0; | |
3940 | ||
3941 | PL_reginput = locinput; | |
3942 | ||
3943 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3944 | "%*s whilem: matched %ld out of %ld..%ld\n", | |
3945 | REPORT_CODE_OFF+depth*2, "", (long)n, | |
3946 | (long)cur_curlyx->u.curlyx.min, | |
3947 | (long)cur_curlyx->u.curlyx.max) | |
3948 | ); | |
a0d0e21e | 3949 | |
c476f425 | 3950 | /* First just match a string of min A's. */ |
a0d0e21e | 3951 | |
c476f425 DM |
3952 | if (n < cur_curlyx->u.curlyx.min) { |
3953 | cur_curlyx->u.curlyx.lastloc = locinput; | |
3954 | PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A); | |
3955 | /* NOTREACHED */ | |
3956 | } | |
3957 | ||
3958 | /* If degenerate A matches "", assume A done. */ | |
3959 | ||
3960 | if (locinput == cur_curlyx->u.curlyx.lastloc) { | |
3961 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
3962 | "%*s whilem: empty match detected, trying continuation...\n", | |
3963 | REPORT_CODE_OFF+depth*2, "") | |
3964 | ); | |
3965 | goto do_whilem_B_max; | |
3966 | } | |
3967 | ||
3968 | /* super-linear cache processing */ | |
3969 | ||
3970 | if (scan->flags) { | |
a0d0e21e | 3971 | |
2c2d71f5 | 3972 | if (!PL_reg_maxiter) { |
c476f425 DM |
3973 | /* start the countdown: Postpone detection until we |
3974 | * know the match is not *that* much linear. */ | |
2c2d71f5 | 3975 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
66bf836d DM |
3976 | /* possible overflow for long strings and many CURLYX's */ |
3977 | if (PL_reg_maxiter < 0) | |
3978 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
3979 | PL_reg_leftiter = PL_reg_maxiter; |
3980 | } | |
c476f425 | 3981 | |
2c2d71f5 | 3982 | if (PL_reg_leftiter-- == 0) { |
c476f425 | 3983 | /* initialise cache */ |
3298f257 | 3984 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 3985 | if (PL_reg_poscache) { |
eb160463 | 3986 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
3987 | Renew(PL_reg_poscache, size, char); |
3988 | PL_reg_poscache_size = size; | |
3989 | } | |
3990 | Zero(PL_reg_poscache, size, char); | |
3991 | } | |
3992 | else { | |
3993 | PL_reg_poscache_size = size; | |
a02a5408 | 3994 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 3995 | } |
c476f425 DM |
3996 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
3997 | "%swhilem: Detected a super-linear match, switching on caching%s...\n", | |
3998 | PL_colors[4], PL_colors[5]) | |
3999 | ); | |
2c2d71f5 | 4000 | } |
c476f425 | 4001 | |
2c2d71f5 | 4002 | if (PL_reg_leftiter < 0) { |
c476f425 DM |
4003 | /* have we already failed at this position? */ |
4004 | I32 offset, mask; | |
4005 | offset = (scan->flags & 0xf) - 1 | |
4006 | + (locinput - PL_bostr) * (scan->flags>>4); | |
4007 | mask = 1 << (offset % 8); | |
4008 | offset /= 8; | |
4009 | if (PL_reg_poscache[offset] & mask) { | |
4010 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4011 | "%*s whilem: (cache) already tried at this position...\n", | |
4012 | REPORT_CODE_OFF+depth*2, "") | |
2c2d71f5 | 4013 | ); |
3298f257 | 4014 | sayNO; /* cache records failure */ |
2c2d71f5 | 4015 | } |
c476f425 DM |
4016 | ST.cache_offset = offset; |
4017 | ST.cache_mask = mask; | |
2c2d71f5 | 4018 | } |
c476f425 | 4019 | } |
2c2d71f5 | 4020 | |
c476f425 | 4021 | /* Prefer B over A for minimal matching. */ |
a687059c | 4022 | |
c476f425 DM |
4023 | if (cur_curlyx->u.curlyx.minmod) { |
4024 | ST.save_curlyx = cur_curlyx; | |
4025 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4026 | ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor); | |
4027 | REGCP_SET(ST.lastcp); | |
4028 | PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B); | |
4029 | /* NOTREACHED */ | |
4030 | } | |
a0d0e21e | 4031 | |
c476f425 DM |
4032 | /* Prefer A over B for maximal matching. */ |
4033 | ||
4034 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ | |
4035 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4036 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4037 | REGCP_SET(ST.lastcp); | |
4038 | PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A); | |
4039 | /* NOTREACHED */ | |
4040 | } | |
4041 | goto do_whilem_B_max; | |
4042 | } | |
4043 | /* NOTREACHED */ | |
4044 | ||
4045 | case WHILEM_B_min: /* just matched B in a minimal match */ | |
4046 | case WHILEM_B_max: /* just matched B in a maximal match */ | |
4047 | cur_curlyx = ST.save_curlyx; | |
4048 | sayYES; | |
4049 | /* NOTREACHED */ | |
4050 | ||
4051 | case WHILEM_B_max_fail: /* just failed to match B in a maximal match */ | |
4052 | cur_curlyx = ST.save_curlyx; | |
4053 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4054 | cur_curlyx->u.curlyx.count--; | |
4055 | CACHEsayNO; | |
4056 | /* NOTREACHED */ | |
4057 | ||
4058 | case WHILEM_A_min_fail: /* just failed to match A in a minimal match */ | |
4059 | REGCP_UNWIND(ST.lastcp); | |
4060 | regcppop(rex); | |
4061 | /* FALL THROUGH */ | |
4062 | case WHILEM_A_pre_fail: /* just failed to match even minimal A */ | |
4063 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4064 | cur_curlyx->u.curlyx.count--; | |
4065 | CACHEsayNO; | |
4066 | /* NOTREACHED */ | |
4067 | ||
4068 | case WHILEM_A_max_fail: /* just failed to match A in a maximal match */ | |
4069 | REGCP_UNWIND(ST.lastcp); | |
4070 | regcppop(rex); /* Restore some previous $<digit>s? */ | |
4071 | PL_reginput = locinput; | |
4072 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4073 | "%*s whilem: failed, trying continuation...\n", | |
4074 | REPORT_CODE_OFF+depth*2, "") | |
4075 | ); | |
4076 | do_whilem_B_max: | |
4077 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4078 | && ckWARN(WARN_REGEXP) | |
4079 | && !(PL_reg_flags & RF_warned)) | |
4080 | { | |
4081 | PL_reg_flags |= RF_warned; | |
4082 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", | |
4083 | "Complex regular subexpression recursion", | |
4084 | REG_INFTY - 1); | |
4085 | } | |
4086 | ||
4087 | /* now try B */ | |
4088 | ST.save_curlyx = cur_curlyx; | |
4089 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4090 | PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B); | |
4091 | /* NOTREACHED */ | |
4092 | ||
4093 | case WHILEM_B_min_fail: /* just failed to match B in a minimal match */ | |
4094 | cur_curlyx = ST.save_curlyx; | |
4095 | REGCP_UNWIND(ST.lastcp); | |
4096 | regcppop(rex); | |
4097 | ||
4098 | if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) { | |
4099 | /* Maximum greed exceeded */ | |
4100 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4101 | && ckWARN(WARN_REGEXP) | |
4102 | && !(PL_reg_flags & RF_warned)) | |
4103 | { | |
3280af22 | 4104 | PL_reg_flags |= RF_warned; |
c476f425 DM |
4105 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), |
4106 | "%s limit (%d) exceeded", | |
4107 | "Complex regular subexpression recursion", | |
4108 | REG_INFTY - 1); | |
a0d0e21e | 4109 | } |
c476f425 | 4110 | cur_curlyx->u.curlyx.count--; |
3ab3c9b4 | 4111 | CACHEsayNO; |
a0d0e21e | 4112 | } |
c476f425 DM |
4113 | |
4114 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4115 | "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "") | |
4116 | ); | |
4117 | /* Try grabbing another A and see if it helps. */ | |
4118 | PL_reginput = locinput; | |
4119 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4120 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4121 | REGCP_SET(ST.lastcp); | |
4122 | PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A); | |
5f66b61c | 4123 | /* NOTREACHED */ |
40a82448 DM |
4124 | |
4125 | #undef ST | |
4126 | #define ST st->u.branch | |
4127 | ||
4128 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
4129 | next = scan + ARG(scan); |
4130 | if (next == scan) | |
4131 | next = NULL; | |
40a82448 DM |
4132 | scan = NEXTOPER(scan); |
4133 | /* FALL THROUGH */ | |
c277df42 | 4134 | |
40a82448 DM |
4135 | case BRANCH: /* /(...|A|...)/ */ |
4136 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
5d458dd8 YO |
4137 | if ((!next || (OP(next) != BRANCH && OP(next) != BRANCHJ)) |
4138 | && !has_cutgroup) | |
4139 | { | |
40a82448 DM |
4140 | /* last branch; skip state push and jump direct to node */ |
4141 | continue; | |
5d458dd8 | 4142 | } |
40a82448 DM |
4143 | ST.lastparen = *PL_reglastparen; |
4144 | ST.next_branch = next; | |
4145 | REGCP_SET(ST.cp); | |
4146 | PL_reginput = locinput; | |
02db2b7b | 4147 | |
40a82448 | 4148 | /* Now go into the branch */ |
5d458dd8 YO |
4149 | if (has_cutgroup) { |
4150 | PUSH_YES_STATE_GOTO(BRANCH_next, scan); | |
4151 | } else { | |
4152 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
4153 | } | |
40a82448 | 4154 | /* NOTREACHED */ |
5d458dd8 YO |
4155 | case CUTGROUP: |
4156 | PL_reginput = locinput; | |
4157 | sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL : | |
f8fc2ecf | 4158 | (SV*)rexi->data->data[ ARG( scan ) ]; |
5d458dd8 YO |
4159 | PUSH_STATE_GOTO(CUTGROUP_next,next); |
4160 | /* NOTREACHED */ | |
4161 | case CUTGROUP_next_fail: | |
4162 | do_cutgroup = 1; | |
4163 | no_final = 1; | |
4164 | if (st->u.mark.mark_name) | |
4165 | sv_commit = st->u.mark.mark_name; | |
4166 | sayNO; | |
4167 | /* NOTREACHED */ | |
4168 | case BRANCH_next: | |
4169 | sayYES; | |
4170 | /* NOTREACHED */ | |
40a82448 | 4171 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ |
5d458dd8 YO |
4172 | if (do_cutgroup) { |
4173 | do_cutgroup = 0; | |
4174 | no_final = 0; | |
4175 | } | |
40a82448 DM |
4176 | REGCP_UNWIND(ST.cp); |
4177 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
4178 | PL_regendp[n] = -1; | |
4179 | *PL_reglastparen = n; | |
0a4db386 | 4180 | /*dmq: *PL_reglastcloseparen = n; */ |
40a82448 DM |
4181 | scan = ST.next_branch; |
4182 | /* no more branches? */ | |
5d458dd8 YO |
4183 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) { |
4184 | DEBUG_EXECUTE_r({ | |
4185 | PerlIO_printf( Perl_debug_log, | |
4186 | "%*s %sBRANCH failed...%s\n", | |
4187 | REPORT_CODE_OFF+depth*2, "", | |
4188 | PL_colors[4], | |
4189 | PL_colors[5] ); | |
4190 | }); | |
4191 | sayNO_SILENT; | |
4192 | } | |
40a82448 DM |
4193 | continue; /* execute next BRANCH[J] op */ |
4194 | /* NOTREACHED */ | |
4195 | ||
a0d0e21e | 4196 | case MINMOD: |
24d3c4a9 | 4197 | minmod = 1; |
a0d0e21e | 4198 | break; |
40a82448 DM |
4199 | |
4200 | #undef ST | |
4201 | #define ST st->u.curlym | |
4202 | ||
4203 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
4204 | ||
4205 | /* This is an optimisation of CURLYX that enables us to push | |
4206 | * only a single backtracking state, no matter now many matches | |
4207 | * there are in {m,n}. It relies on the pattern being constant | |
4208 | * length, with no parens to influence future backrefs | |
4209 | */ | |
4210 | ||
4211 | ST.me = scan; | |
dc45a647 | 4212 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
4213 | |
4214 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
4215 | if (ST.me->flags) { | |
3b6647e0 | 4216 | U32 paren = ST.me->flags; |
40a82448 DM |
4217 | if (paren > PL_regsize) |
4218 | PL_regsize = paren; | |
3b6647e0 | 4219 | if (paren > *PL_reglastparen) |
40a82448 | 4220 | *PL_reglastparen = paren; |
c277df42 | 4221 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 4222 | } |
40a82448 DM |
4223 | ST.A = scan; |
4224 | ST.B = next; | |
4225 | ST.alen = 0; | |
4226 | ST.count = 0; | |
24d3c4a9 DM |
4227 | ST.minmod = minmod; |
4228 | minmod = 0; | |
40a82448 DM |
4229 | ST.c1 = CHRTEST_UNINIT; |
4230 | REGCP_SET(ST.cp); | |
6407bf3b | 4231 | |
40a82448 DM |
4232 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
4233 | goto curlym_do_B; | |
4234 | ||
4235 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 4236 | PL_reginput = locinput; |
40a82448 DM |
4237 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
4238 | /* NOTREACHED */ | |
5f80c4cf | 4239 | |
40a82448 DM |
4240 | case CURLYM_A: /* we've just matched an A */ |
4241 | locinput = st->locinput; | |
4242 | nextchr = UCHARAT(locinput); | |
4243 | ||
4244 | ST.count++; | |
4245 | /* after first match, determine A's length: u.curlym.alen */ | |
4246 | if (ST.count == 1) { | |
4247 | if (PL_reg_match_utf8) { | |
4248 | char *s = locinput; | |
4249 | while (s < PL_reginput) { | |
4250 | ST.alen++; | |
4251 | s += UTF8SKIP(s); | |
4252 | } | |
4253 | } | |
4254 | else { | |
4255 | ST.alen = PL_reginput - locinput; | |
4256 | } | |
4257 | if (ST.alen == 0) | |
4258 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
4259 | } | |
0cadcf80 DM |
4260 | DEBUG_EXECUTE_r( |
4261 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 4262 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 4263 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 4264 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
4265 | ); |
4266 | ||
40a82448 | 4267 | locinput = PL_reginput; |
0a4db386 YO |
4268 | |
4269 | if (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4270 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4271 | goto fake_end; |
4272 | ||
4273 | if ( ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)) ) | |
40a82448 DM |
4274 | goto curlym_do_A; /* try to match another A */ |
4275 | goto curlym_do_B; /* try to match B */ | |
5f80c4cf | 4276 | |
40a82448 DM |
4277 | case CURLYM_A_fail: /* just failed to match an A */ |
4278 | REGCP_UNWIND(ST.cp); | |
0a4db386 YO |
4279 | |
4280 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ | |
4281 | || (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4282 | cur_eval->u.eval.close_paren == (U32)ST.me->flags)) |
40a82448 | 4283 | sayNO; |
0cadcf80 | 4284 | |
40a82448 DM |
4285 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
4286 | PL_reginput = locinput; | |
4287 | if (ST.c1 == CHRTEST_UNINIT) { | |
4288 | /* calculate c1 and c2 for possible match of 1st char | |
4289 | * following curly */ | |
4290 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
4291 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
4292 | regnode *text_node = ST.B; | |
4293 | if (! HAS_TEXT(text_node)) | |
4294 | FIND_NEXT_IMPT(text_node); | |
4295 | if (HAS_TEXT(text_node) | |
4296 | && PL_regkind[OP(text_node)] != REF) | |
4297 | { | |
4298 | ST.c1 = (U8)*STRING(text_node); | |
4299 | ST.c2 = | |
4300 | (OP(text_node) == EXACTF || OP(text_node) == REFF) | |
4301 | ? PL_fold[ST.c1] | |
4302 | : (OP(text_node) == EXACTFL || OP(text_node) == REFFL) | |
4303 | ? PL_fold_locale[ST.c1] | |
4304 | : ST.c1; | |
c277df42 | 4305 | } |
c277df42 | 4306 | } |
40a82448 DM |
4307 | } |
4308 | ||
4309 | DEBUG_EXECUTE_r( | |
4310 | PerlIO_printf(Perl_debug_log, | |
4311 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 4312 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
4313 | "", (IV)ST.count) |
4314 | ); | |
4315 | if (ST.c1 != CHRTEST_VOID | |
4316 | && UCHARAT(PL_reginput) != ST.c1 | |
4317 | && UCHARAT(PL_reginput) != ST.c2) | |
4318 | { | |
4319 | /* simulate B failing */ | |
4320 | state_num = CURLYM_B_fail; | |
4321 | goto reenter_switch; | |
4322 | } | |
4323 | ||
4324 | if (ST.me->flags) { | |
4325 | /* mark current A as captured */ | |
4326 | I32 paren = ST.me->flags; | |
4327 | if (ST.count) { | |
4328 | PL_regstartp[paren] | |
4329 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; | |
4330 | PL_regendp[paren] = PL_reginput - PL_bostr; | |
0a4db386 | 4331 | /*dmq: *PL_reglastcloseparen = paren; */ |
c277df42 | 4332 | } |
40a82448 DM |
4333 | else |
4334 | PL_regendp[paren] = -1; | |
0a4db386 | 4335 | if (cur_eval && cur_eval->u.eval.close_paren && |
24b23f37 | 4336 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4337 | { |
4338 | if (ST.count) | |
4339 | goto fake_end; | |
4340 | else | |
4341 | sayNO; | |
4342 | } | |
c277df42 | 4343 | } |
0a4db386 | 4344 | |
40a82448 | 4345 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4346 | /* NOTREACHED */ |
40a82448 DM |
4347 | |
4348 | case CURLYM_B_fail: /* just failed to match a B */ | |
4349 | REGCP_UNWIND(ST.cp); | |
4350 | if (ST.minmod) { | |
4351 | if (ST.count == ARG2(ST.me) /* max */) | |
4352 | sayNO; | |
4353 | goto curlym_do_A; /* try to match a further A */ | |
4354 | } | |
4355 | /* backtrack one A */ | |
4356 | if (ST.count == ARG1(ST.me) /* min */) | |
4357 | sayNO; | |
4358 | ST.count--; | |
4359 | locinput = HOPc(locinput, -ST.alen); | |
4360 | goto curlym_do_B; /* try to match B */ | |
4361 | ||
c255a977 DM |
4362 | #undef ST |
4363 | #define ST st->u.curly | |
40a82448 | 4364 | |
c255a977 DM |
4365 | #define CURLY_SETPAREN(paren, success) \ |
4366 | if (paren) { \ | |
4367 | if (success) { \ | |
4368 | PL_regstartp[paren] = HOPc(locinput, -1) - PL_bostr; \ | |
4369 | PL_regendp[paren] = locinput - PL_bostr; \ | |
0a4db386 | 4370 | *PL_reglastcloseparen = paren; \ |
c255a977 DM |
4371 | } \ |
4372 | else \ | |
4373 | PL_regendp[paren] = -1; \ | |
4374 | } | |
4375 | ||
4376 | case STAR: /* /A*B/ where A is width 1 */ | |
4377 | ST.paren = 0; | |
4378 | ST.min = 0; | |
4379 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4380 | scan = NEXTOPER(scan); |
4381 | goto repeat; | |
c255a977 DM |
4382 | case PLUS: /* /A+B/ where A is width 1 */ |
4383 | ST.paren = 0; | |
4384 | ST.min = 1; | |
4385 | ST.max = REG_INFTY; | |
c277df42 | 4386 | scan = NEXTOPER(scan); |
c255a977 DM |
4387 | goto repeat; |
4388 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4389 | ST.paren = scan->flags; /* Which paren to set */ | |
4390 | if (ST.paren > PL_regsize) | |
4391 | PL_regsize = ST.paren; | |
3b6647e0 | 4392 | if (ST.paren > *PL_reglastparen) |
c255a977 DM |
4393 | *PL_reglastparen = ST.paren; |
4394 | ST.min = ARG1(scan); /* min to match */ | |
4395 | ST.max = ARG2(scan); /* max to match */ | |
0a4db386 | 4396 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4397 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4398 | ST.min=1; |
4399 | ST.max=1; | |
4400 | } | |
c255a977 DM |
4401 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
4402 | goto repeat; | |
4403 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4404 | ST.paren = 0; | |
4405 | ST.min = ARG1(scan); /* min to match */ | |
4406 | ST.max = ARG2(scan); /* max to match */ | |
4407 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4408 | repeat: |
a0d0e21e LW |
4409 | /* |
4410 | * Lookahead to avoid useless match attempts | |
4411 | * when we know what character comes next. | |
c255a977 | 4412 | * |
5f80c4cf JP |
4413 | * Used to only do .*x and .*?x, but now it allows |
4414 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4415 | * of the quantifier and the EXACT-like node. -- japhy | |
4416 | */ | |
4417 | ||
c255a977 DM |
4418 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4419 | sayNO; | |
cca55fe3 | 4420 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4421 | U8 *s; |
4422 | regnode *text_node = next; | |
4423 | ||
3dab1dad YO |
4424 | if (! HAS_TEXT(text_node)) |
4425 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4426 | |
9e137952 | 4427 | if (! HAS_TEXT(text_node)) |
c255a977 | 4428 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4429 | else { |
3dab1dad | 4430 | if (PL_regkind[OP(text_node)] == REF) { |
c255a977 | 4431 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4432 | goto assume_ok_easy; |
cca55fe3 | 4433 | } |
be8e71aa YO |
4434 | else |
4435 | s = (U8*)STRING(text_node); | |
5f80c4cf JP |
4436 | |
4437 | if (!UTF) { | |
c255a977 | 4438 | ST.c2 = ST.c1 = *s; |
f65d3ee7 | 4439 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) |
c255a977 | 4440 | ST.c2 = PL_fold[ST.c1]; |
f65d3ee7 | 4441 | else if (OP(text_node) == EXACTFL || OP(text_node) == REFFL) |
c255a977 | 4442 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4443 | } |
5f80c4cf | 4444 | else { /* UTF */ |
f65d3ee7 | 4445 | if (OP(text_node) == EXACTF || OP(text_node) == REFF) { |
a2a2844f | 4446 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4447 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4448 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4449 | |
4450 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4451 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4452 | #ifdef EBCDIC |
4453 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4454 | ckWARN(WARN_UTF8) ? | |
4455 | 0 : UTF8_ALLOW_ANY); | |
4456 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4457 | ckWARN(WARN_UTF8) ? | |
4458 | 0 : UTF8_ALLOW_ANY); | |
4459 | #else | |
c255a977 | 4460 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4461 | uniflags); |
c255a977 | 4462 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4463 | uniflags); |
4464 | #endif | |
5f80c4cf JP |
4465 | } |
4466 | else { | |
c255a977 | 4467 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4468 | uniflags); |
5f80c4cf | 4469 | } |
1aa99e6b IH |
4470 | } |
4471 | } | |
bbce6d69 | 4472 | } |
a0d0e21e | 4473 | else |
c255a977 | 4474 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4475 | assume_ok_easy: |
c255a977 DM |
4476 | |
4477 | ST.A = scan; | |
4478 | ST.B = next; | |
3280af22 | 4479 | PL_reginput = locinput; |
24d3c4a9 DM |
4480 | if (minmod) { |
4481 | minmod = 0; | |
e2e6a0f1 | 4482 | if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min) |
4633a7c4 | 4483 | sayNO; |
c255a977 | 4484 | ST.count = ST.min; |
a0ed51b3 | 4485 | locinput = PL_reginput; |
c255a977 DM |
4486 | REGCP_SET(ST.cp); |
4487 | if (ST.c1 == CHRTEST_VOID) | |
4488 | goto curly_try_B_min; | |
4489 | ||
4490 | ST.oldloc = locinput; | |
4491 | ||
4492 | /* set ST.maxpos to the furthest point along the | |
4493 | * string that could possibly match */ | |
4494 | if (ST.max == REG_INFTY) { | |
4495 | ST.maxpos = PL_regeol - 1; | |
4496 | if (do_utf8) | |
4497 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4498 | ST.maxpos--; | |
4499 | } | |
4500 | else if (do_utf8) { | |
4501 | int m = ST.max - ST.min; | |
4502 | for (ST.maxpos = locinput; | |
4503 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4504 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4505 | } | |
4506 | else { | |
4507 | ST.maxpos = locinput + ST.max - ST.min; | |
4508 | if (ST.maxpos >= PL_regeol) | |
4509 | ST.maxpos = PL_regeol - 1; | |
4510 | } | |
4511 | goto curly_try_B_min_known; | |
4512 | ||
4513 | } | |
4514 | else { | |
e2e6a0f1 | 4515 | ST.count = regrepeat(rex, ST.A, ST.max, depth); |
c255a977 DM |
4516 | locinput = PL_reginput; |
4517 | if (ST.count < ST.min) | |
4518 | sayNO; | |
4519 | if ((ST.count > ST.min) | |
4520 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4521 | { | |
4522 | /* A{m,n} must come at the end of the string, there's | |
4523 | * no point in backing off ... */ | |
4524 | ST.min = ST.count; | |
4525 | /* ...except that $ and \Z can match before *and* after | |
4526 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4527 | We may back off by one in this case. */ | |
4528 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4529 | ST.min--; | |
4530 | } | |
4531 | REGCP_SET(ST.cp); | |
4532 | goto curly_try_B_max; | |
4533 | } | |
4534 | /* NOTREACHED */ | |
4535 | ||
4536 | ||
4537 | case CURLY_B_min_known_fail: | |
4538 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4539 | if (ST.paren && ST.count) | |
4540 | PL_regendp[ST.paren] = -1; | |
4541 | ||
4542 | PL_reginput = locinput; /* Could be reset... */ | |
4543 | REGCP_UNWIND(ST.cp); | |
4544 | /* Couldn't or didn't -- move forward. */ | |
4545 | ST.oldloc = locinput; | |
4546 | if (do_utf8) | |
4547 | locinput += UTF8SKIP(locinput); | |
4548 | else | |
4549 | locinput++; | |
4550 | ST.count++; | |
4551 | curly_try_B_min_known: | |
4552 | /* find the next place where 'B' could work, then call B */ | |
4553 | { | |
4554 | int n; | |
4555 | if (do_utf8) { | |
4556 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4557 | if (ST.c1 == ST.c2) { | |
4558 | STRLEN len; | |
4559 | /* set n to utf8_distance(oldloc, locinput) */ | |
4560 | while (locinput <= ST.maxpos && | |
4561 | utf8n_to_uvchr((U8*)locinput, | |
4562 | UTF8_MAXBYTES, &len, | |
4563 | uniflags) != (UV)ST.c1) { | |
4564 | locinput += len; | |
4565 | n++; | |
4566 | } | |
1aa99e6b IH |
4567 | } |
4568 | else { | |
c255a977 DM |
4569 | /* set n to utf8_distance(oldloc, locinput) */ |
4570 | while (locinput <= ST.maxpos) { | |
4571 | STRLEN len; | |
4572 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4573 | UTF8_MAXBYTES, &len, | |
4574 | uniflags); | |
4575 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4576 | break; | |
4577 | locinput += len; | |
4578 | n++; | |
1aa99e6b | 4579 | } |
0fe9bf95 IZ |
4580 | } |
4581 | } | |
c255a977 DM |
4582 | else { |
4583 | if (ST.c1 == ST.c2) { | |
4584 | while (locinput <= ST.maxpos && | |
4585 | UCHARAT(locinput) != ST.c1) | |
4586 | locinput++; | |
bbce6d69 | 4587 | } |
c255a977 DM |
4588 | else { |
4589 | while (locinput <= ST.maxpos | |
4590 | && UCHARAT(locinput) != ST.c1 | |
4591 | && UCHARAT(locinput) != ST.c2) | |
4592 | locinput++; | |
a0ed51b3 | 4593 | } |
c255a977 DM |
4594 | n = locinput - ST.oldloc; |
4595 | } | |
4596 | if (locinput > ST.maxpos) | |
4597 | sayNO; | |
4598 | /* PL_reginput == oldloc now */ | |
4599 | if (n) { | |
4600 | ST.count += n; | |
e2e6a0f1 | 4601 | if (regrepeat(rex, ST.A, n, depth) < n) |
4633a7c4 | 4602 | sayNO; |
a0d0e21e | 4603 | } |
c255a977 DM |
4604 | PL_reginput = locinput; |
4605 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4606 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4607 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4608 | goto fake_end; |
4609 | } | |
c255a977 | 4610 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); |
a0d0e21e | 4611 | } |
c255a977 DM |
4612 | /* NOTREACHED */ |
4613 | ||
4614 | ||
4615 | case CURLY_B_min_fail: | |
4616 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4617 | if (ST.paren && ST.count) | |
4618 | PL_regendp[ST.paren] = -1; | |
4619 | ||
4620 | REGCP_UNWIND(ST.cp); | |
4621 | /* failed -- move forward one */ | |
4622 | PL_reginput = locinput; | |
e2e6a0f1 | 4623 | if (regrepeat(rex, ST.A, 1, depth)) { |
c255a977 | 4624 | ST.count++; |
a0ed51b3 | 4625 | locinput = PL_reginput; |
c255a977 DM |
4626 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4627 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4628 | { |
c255a977 DM |
4629 | curly_try_B_min: |
4630 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4631 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4632 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4633 | goto fake_end; |
4634 | } | |
c255a977 | 4635 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); |
a0d0e21e LW |
4636 | } |
4637 | } | |
4633a7c4 | 4638 | sayNO; |
c255a977 DM |
4639 | /* NOTREACHED */ |
4640 | ||
4641 | ||
4642 | curly_try_B_max: | |
4643 | /* a successful greedy match: now try to match B */ | |
40d049e4 | 4644 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4645 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
40d049e4 YO |
4646 | goto fake_end; |
4647 | } | |
c255a977 DM |
4648 | { |
4649 | UV c = 0; | |
4650 | if (ST.c1 != CHRTEST_VOID) | |
4651 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4652 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4653 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4654 | /* If it could work, try it. */ |
4655 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4656 | CURLY_SETPAREN(ST.paren, ST.count); | |
4657 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4658 | /* NOTREACHED */ | |
4659 | } | |
4660 | } | |
4661 | /* FALL THROUGH */ | |
4662 | case CURLY_B_max_fail: | |
4663 | /* failed to find B in a greedy match */ | |
4664 | if (ST.paren && ST.count) | |
4665 | PL_regendp[ST.paren] = -1; | |
4666 | ||
4667 | REGCP_UNWIND(ST.cp); | |
4668 | /* back up. */ | |
4669 | if (--ST.count < ST.min) | |
4670 | sayNO; | |
4671 | PL_reginput = locinput = HOPc(locinput, -1); | |
4672 | goto curly_try_B_max; | |
4673 | ||
4674 | #undef ST | |
4675 | ||
a0d0e21e | 4676 | case END: |
6bda09f9 | 4677 | fake_end: |
faec1544 DM |
4678 | if (cur_eval) { |
4679 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4680 | I32 tmpix; | |
4681 | ||
4682 | ||
4683 | st->u.eval.toggle_reg_flags | |
4684 | = cur_eval->u.eval.toggle_reg_flags; | |
4685 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4686 | ||
4687 | st->u.eval.prev_rex = rex; /* inner */ | |
f8fc2ecf YO |
4688 | rex = cur_eval->u.eval.prev_rex; /* outer */ |
4689 | rexi = RXi_GET(rex); | |
faec1544 DM |
4690 | cur_curlyx = cur_eval->u.eval.prev_curlyx; |
4691 | ReREFCNT_inc(rex); | |
4692 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ | |
4693 | REGCP_SET(st->u.eval.lastcp); | |
4694 | PL_reginput = locinput; | |
4695 | ||
4696 | /* Restore parens of the outer rex without popping the | |
4697 | * savestack */ | |
4698 | tmpix = PL_savestack_ix; | |
4699 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4700 | regcppop(rex); | |
4701 | PL_savestack_ix = tmpix; | |
4702 | ||
4703 | st->u.eval.prev_eval = cur_eval; | |
4704 | cur_eval = cur_eval->u.eval.prev_eval; | |
4705 | DEBUG_EXECUTE_r( | |
2a49f0f5 JH |
4706 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n", |
4707 | REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval));); | |
faec1544 DM |
4708 | PUSH_YES_STATE_GOTO(EVAL_AB, |
4709 | st->u.eval.prev_eval->u.eval.B); /* match B */ | |
4710 | } | |
4711 | ||
3b0527fe | 4712 | if (locinput < reginfo->till) { |
a3621e74 | 4713 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4714 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4715 | PL_colors[4], | |
4716 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4717 | (long)(reginfo->till - PL_reg_starttry), |
7821416a | 4718 | PL_colors[5])); |
58e23c8d | 4719 | |
262b90c4 | 4720 | sayNO_SILENT; /* Cannot match: too short. */ |
7821416a IZ |
4721 | } |
4722 | PL_reginput = locinput; /* put where regtry can find it */ | |
262b90c4 | 4723 | sayYES; /* Success! */ |
dad79028 DM |
4724 | |
4725 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4726 | DEBUG_EXECUTE_r( | |
4727 | PerlIO_printf(Perl_debug_log, | |
4728 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4729 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4730 | PL_reginput = locinput; /* put where regtry can find it */ |
262b90c4 | 4731 | sayYES; /* Success! */ |
dad79028 | 4732 | |
40a82448 DM |
4733 | #undef ST |
4734 | #define ST st->u.ifmatch | |
4735 | ||
4736 | case SUSPEND: /* (?>A) */ | |
4737 | ST.wanted = 1; | |
9fe1d20c | 4738 | PL_reginput = locinput; |
9041c2e3 | 4739 | goto do_ifmatch; |
dad79028 | 4740 | |
40a82448 DM |
4741 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4742 | ST.wanted = 0; | |
dad79028 DM |
4743 | goto ifmatch_trivial_fail_test; |
4744 | ||
40a82448 DM |
4745 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4746 | ST.wanted = 1; | |
dad79028 | 4747 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4748 | if (scan->flags) { |
52657f30 | 4749 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4750 | if (!s) { |
4751 | /* trivial fail */ | |
24d3c4a9 DM |
4752 | if (logical) { |
4753 | logical = 0; | |
4754 | sw = 1 - (bool)ST.wanted; | |
dad79028 | 4755 | } |
40a82448 | 4756 | else if (ST.wanted) |
dad79028 DM |
4757 | sayNO; |
4758 | next = scan + ARG(scan); | |
4759 | if (next == scan) | |
4760 | next = NULL; | |
4761 | break; | |
4762 | } | |
efb30f32 | 4763 | PL_reginput = s; |
a0ed51b3 LW |
4764 | } |
4765 | else | |
4766 | PL_reginput = locinput; | |
4767 | ||
c277df42 | 4768 | do_ifmatch: |
40a82448 | 4769 | ST.me = scan; |
24d3c4a9 | 4770 | ST.logical = logical; |
40a82448 DM |
4771 | /* execute body of (?...A) */ |
4772 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
4773 | /* NOTREACHED */ | |
4774 | ||
4775 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
4776 | ST.wanted = !ST.wanted; | |
4777 | /* FALL THROUGH */ | |
4778 | ||
4779 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
24d3c4a9 DM |
4780 | if (ST.logical) { |
4781 | sw = (bool)ST.wanted; | |
40a82448 DM |
4782 | } |
4783 | else if (!ST.wanted) | |
4784 | sayNO; | |
4785 | ||
4786 | if (OP(ST.me) == SUSPEND) | |
4787 | locinput = PL_reginput; | |
4788 | else { | |
4789 | locinput = PL_reginput = st->locinput; | |
4790 | nextchr = UCHARAT(locinput); | |
4791 | } | |
4792 | scan = ST.me + ARG(ST.me); | |
4793 | if (scan == ST.me) | |
4794 | scan = NULL; | |
4795 | continue; /* execute B */ | |
4796 | ||
4797 | #undef ST | |
dad79028 | 4798 | |
c277df42 | 4799 | case LONGJMP: |
c277df42 IZ |
4800 | next = scan + ARG(scan); |
4801 | if (next == scan) | |
4802 | next = NULL; | |
a0d0e21e | 4803 | break; |
54612592 | 4804 | case COMMIT: |
e2e6a0f1 YO |
4805 | reginfo->cutpoint = PL_regeol; |
4806 | /* FALLTHROUGH */ | |
5d458dd8 | 4807 | case PRUNE: |
24b23f37 | 4808 | PL_reginput = locinput; |
e2e6a0f1 | 4809 | if (!scan->flags) |
f8fc2ecf | 4810 | sv_yes_mark = sv_commit = (SV*)rexi->data->data[ ARG( scan ) ]; |
54612592 YO |
4811 | PUSH_STATE_GOTO(COMMIT_next,next); |
4812 | /* NOTREACHED */ | |
4813 | case COMMIT_next_fail: | |
4814 | no_final = 1; | |
4815 | /* FALLTHROUGH */ | |
7f69552c YO |
4816 | case OPFAIL: |
4817 | sayNO; | |
e2e6a0f1 YO |
4818 | /* NOTREACHED */ |
4819 | ||
4820 | #define ST st->u.mark | |
4821 | case MARKPOINT: | |
4822 | ST.prev_mark = mark_state; | |
5d458dd8 | 4823 | ST.mark_name = sv_commit = sv_yes_mark |
f8fc2ecf | 4824 | = (SV*)rexi->data->data[ ARG( scan ) ]; |
e2e6a0f1 YO |
4825 | mark_state = st; |
4826 | ST.mark_loc = PL_reginput = locinput; | |
4827 | PUSH_YES_STATE_GOTO(MARKPOINT_next,next); | |
4828 | /* NOTREACHED */ | |
4829 | case MARKPOINT_next: | |
4830 | mark_state = ST.prev_mark; | |
4831 | sayYES; | |
4832 | /* NOTREACHED */ | |
4833 | case MARKPOINT_next_fail: | |
5d458dd8 | 4834 | if (popmark && sv_eq(ST.mark_name,popmark)) |
e2e6a0f1 YO |
4835 | { |
4836 | if (ST.mark_loc > startpoint) | |
4837 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
4838 | popmark = NULL; /* we found our mark */ | |
4839 | sv_commit = ST.mark_name; | |
4840 | ||
4841 | DEBUG_EXECUTE_r({ | |
5d458dd8 | 4842 | PerlIO_printf(Perl_debug_log, |
e2e6a0f1 YO |
4843 | "%*s %ssetting cutpoint to mark:%"SVf"...%s\n", |
4844 | REPORT_CODE_OFF+depth*2, "", | |
be2597df | 4845 | PL_colors[4], SVfARG(sv_commit), PL_colors[5]); |
e2e6a0f1 YO |
4846 | }); |
4847 | } | |
4848 | mark_state = ST.prev_mark; | |
5d458dd8 YO |
4849 | sv_yes_mark = mark_state ? |
4850 | mark_state->u.mark.mark_name : NULL; | |
e2e6a0f1 YO |
4851 | sayNO; |
4852 | /* NOTREACHED */ | |
5d458dd8 YO |
4853 | case SKIP: |
4854 | PL_reginput = locinput; | |
4855 | if (scan->flags) { | |
2bf803e2 | 4856 | /* (*SKIP) : if we fail we cut here*/ |
5d458dd8 | 4857 | ST.mark_name = NULL; |
e2e6a0f1 | 4858 | ST.mark_loc = locinput; |
5d458dd8 YO |
4859 | PUSH_STATE_GOTO(SKIP_next,next); |
4860 | } else { | |
2bf803e2 | 4861 | /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was, |
5d458dd8 YO |
4862 | otherwise do nothing. Meaning we need to scan |
4863 | */ | |
4864 | regmatch_state *cur = mark_state; | |
f8fc2ecf | 4865 | SV *find = (SV*)rexi->data->data[ ARG( scan ) ]; |
5d458dd8 YO |
4866 | |
4867 | while (cur) { | |
4868 | if ( sv_eq( cur->u.mark.mark_name, | |
4869 | find ) ) | |
4870 | { | |
4871 | ST.mark_name = find; | |
4872 | PUSH_STATE_GOTO( SKIP_next, next ); | |
4873 | } | |
4874 | cur = cur->u.mark.prev_mark; | |
4875 | } | |
e2e6a0f1 | 4876 | } |
2bf803e2 | 4877 | /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */ |
5d458dd8 YO |
4878 | break; |
4879 | case SKIP_next_fail: | |
4880 | if (ST.mark_name) { | |
4881 | /* (*CUT:NAME) - Set up to search for the name as we | |
4882 | collapse the stack*/ | |
4883 | popmark = ST.mark_name; | |
4884 | } else { | |
4885 | /* (*CUT) - No name, we cut here.*/ | |
e2e6a0f1 YO |
4886 | if (ST.mark_loc > startpoint) |
4887 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
5d458dd8 YO |
4888 | /* but we set sv_commit to latest mark_name if there |
4889 | is one so they can test to see how things lead to this | |
4890 | cut */ | |
4891 | if (mark_state) | |
4892 | sv_commit=mark_state->u.mark.mark_name; | |
4893 | } | |
e2e6a0f1 YO |
4894 | no_final = 1; |
4895 | sayNO; | |
4896 | /* NOTREACHED */ | |
4897 | #undef ST | |
4898 | ||
a0d0e21e | 4899 | default: |
b900a521 | 4900 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 4901 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 4902 | Perl_croak(aTHX_ "regexp memory corruption"); |
5d458dd8 YO |
4903 | |
4904 | } /* end switch */ | |
95b24440 | 4905 | |
5d458dd8 YO |
4906 | /* switch break jumps here */ |
4907 | scan = next; /* prepare to execute the next op and ... */ | |
4908 | continue; /* ... jump back to the top, reusing st */ | |
95b24440 DM |
4909 | /* NOTREACHED */ |
4910 | ||
40a82448 DM |
4911 | push_yes_state: |
4912 | /* push a state that backtracks on success */ | |
4913 | st->u.yes.prev_yes_state = yes_state; | |
4914 | yes_state = st; | |
4915 | /* FALL THROUGH */ | |
4916 | push_state: | |
4917 | /* push a new regex state, then continue at scan */ | |
4918 | { | |
4919 | regmatch_state *newst; | |
4920 | ||
24b23f37 YO |
4921 | DEBUG_STACK_r({ |
4922 | regmatch_state *cur = st; | |
4923 | regmatch_state *curyes = yes_state; | |
4924 | int curd = depth; | |
4925 | regmatch_slab *slab = PL_regmatch_slab; | |
4926 | for (;curd > -1;cur--,curd--) { | |
4927 | if (cur < SLAB_FIRST(slab)) { | |
4928 | slab = slab->prev; | |
4929 | cur = SLAB_LAST(slab); | |
4930 | } | |
4931 | PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n", | |
4932 | REPORT_CODE_OFF + 2 + depth * 2,"", | |
4933 | curd, reg_name[cur->resume_state], | |
4934 | (curyes == cur) ? "yes" : "" | |
4935 | ); | |
4936 | if (curyes == cur) | |
4937 | curyes = cur->u.yes.prev_yes_state; | |
4938 | } | |
4939 | } else | |
4940 | DEBUG_STATE_pp("push") | |
4941 | ); | |
40a82448 | 4942 | depth++; |
40a82448 DM |
4943 | st->locinput = locinput; |
4944 | newst = st+1; | |
4945 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
4946 | newst = S_push_slab(aTHX); | |
4947 | PL_regmatch_state = newst; | |
786e8c11 | 4948 | |
40a82448 DM |
4949 | locinput = PL_reginput; |
4950 | nextchr = UCHARAT(locinput); | |
4951 | st = newst; | |
4952 | continue; | |
4953 | /* NOTREACHED */ | |
4954 | } | |
a0d0e21e | 4955 | } |
a687059c | 4956 | |
a0d0e21e LW |
4957 | /* |
4958 | * We get here only if there's trouble -- normally "case END" is | |
4959 | * the terminating point. | |
4960 | */ | |
cea2e8a9 | 4961 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 4962 | /*NOTREACHED*/ |
4633a7c4 LW |
4963 | sayNO; |
4964 | ||
262b90c4 | 4965 | yes: |
77cb431f DM |
4966 | if (yes_state) { |
4967 | /* we have successfully completed a subexpression, but we must now | |
4968 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 4969 | assert(st != yes_state); |
5bc10b2c DM |
4970 | #ifdef DEBUGGING |
4971 | while (st != yes_state) { | |
4972 | st--; | |
4973 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
4974 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4975 | st = SLAB_LAST(PL_regmatch_slab); | |
4976 | } | |
e2e6a0f1 | 4977 | DEBUG_STATE_r({ |
54612592 YO |
4978 | if (no_final) { |
4979 | DEBUG_STATE_pp("pop (no final)"); | |
4980 | } else { | |
4981 | DEBUG_STATE_pp("pop (yes)"); | |
4982 | } | |
e2e6a0f1 | 4983 | }); |
5bc10b2c DM |
4984 | depth--; |
4985 | } | |
4986 | #else | |
77cb431f DM |
4987 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
4988 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
4989 | { | |
4990 | /* not in this slab, pop slab */ | |
4991 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
4992 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
4993 | st = SLAB_LAST(PL_regmatch_slab); | |
4994 | } | |
4995 | depth -= (st - yes_state); | |
5bc10b2c | 4996 | #endif |
77cb431f DM |
4997 | st = yes_state; |
4998 | yes_state = st->u.yes.prev_yes_state; | |
4999 | PL_regmatch_state = st; | |
24b23f37 | 5000 | |
5d458dd8 YO |
5001 | if (no_final) { |
5002 | locinput= st->locinput; | |
5003 | nextchr = UCHARAT(locinput); | |
5004 | } | |
54612592 | 5005 | state_num = st->resume_state + no_final; |
24d3c4a9 | 5006 | goto reenter_switch; |
77cb431f DM |
5007 | } |
5008 | ||
a3621e74 | 5009 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 5010 | PL_colors[4], PL_colors[5])); |
02db2b7b | 5011 | |
95b24440 | 5012 | result = 1; |
aa283a38 | 5013 | goto final_exit; |
4633a7c4 LW |
5014 | |
5015 | no: | |
a3621e74 | 5016 | DEBUG_EXECUTE_r( |
7821416a | 5017 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 5018 | "%*s %sfailed...%s\n", |
5bc10b2c | 5019 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 5020 | PL_colors[4], PL_colors[5]) |
7821416a | 5021 | ); |
aa283a38 | 5022 | |
262b90c4 | 5023 | no_silent: |
54612592 YO |
5024 | if (no_final) { |
5025 | if (yes_state) { | |
5026 | goto yes; | |
5027 | } else { | |
5028 | goto final_exit; | |
5029 | } | |
5030 | } | |
aa283a38 DM |
5031 | if (depth) { |
5032 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
5033 | st--; |
5034 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
5035 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5036 | st = SLAB_LAST(PL_regmatch_slab); | |
5037 | } | |
5038 | PL_regmatch_state = st; | |
40a82448 DM |
5039 | locinput= st->locinput; |
5040 | nextchr = UCHARAT(locinput); | |
5041 | ||
5bc10b2c DM |
5042 | DEBUG_STATE_pp("pop"); |
5043 | depth--; | |
262b90c4 DM |
5044 | if (yes_state == st) |
5045 | yes_state = st->u.yes.prev_yes_state; | |
5bc10b2c | 5046 | |
24d3c4a9 DM |
5047 | state_num = st->resume_state + 1; /* failure = success + 1 */ |
5048 | goto reenter_switch; | |
95b24440 | 5049 | } |
24d3c4a9 | 5050 | result = 0; |
aa283a38 | 5051 | |
262b90c4 | 5052 | final_exit: |
bbe252da | 5053 | if (rex->intflags & PREGf_VERBARG_SEEN) { |
5d458dd8 YO |
5054 | SV *sv_err = get_sv("REGERROR", 1); |
5055 | SV *sv_mrk = get_sv("REGMARK", 1); | |
5056 | if (result) { | |
e2e6a0f1 | 5057 | sv_commit = &PL_sv_no; |
5d458dd8 YO |
5058 | if (!sv_yes_mark) |
5059 | sv_yes_mark = &PL_sv_yes; | |
5060 | } else { | |
5061 | if (!sv_commit) | |
5062 | sv_commit = &PL_sv_yes; | |
5063 | sv_yes_mark = &PL_sv_no; | |
5064 | } | |
5065 | sv_setsv(sv_err, sv_commit); | |
5066 | sv_setsv(sv_mrk, sv_yes_mark); | |
e2e6a0f1 | 5067 | } |
5d9a96ca DM |
5068 | /* restore original high-water mark */ |
5069 | PL_regmatch_slab = orig_slab; | |
5070 | PL_regmatch_state = orig_state; | |
5071 | ||
5072 | /* free all slabs above current one */ | |
5073 | if (orig_slab->next) { | |
c4fd8992 | 5074 | regmatch_slab *sl = orig_slab->next; |
5d9a96ca DM |
5075 | orig_slab->next = NULL; |
5076 | while (sl) { | |
c4fd8992 | 5077 | regmatch_slab * const osl = sl; |
5d9a96ca | 5078 | sl = sl->next; |
ad65c075 | 5079 | Safefree(osl); |
5d9a96ca DM |
5080 | } |
5081 | } | |
5082 | ||
95b24440 | 5083 | return result; |
a687059c LW |
5084 | } |
5085 | ||
5086 | /* | |
5087 | - regrepeat - repeatedly match something simple, report how many | |
5088 | */ | |
5089 | /* | |
5090 | * [This routine now assumes that it will only match on things of length 1. | |
5091 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 5092 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 5093 | */ |
76e3520e | 5094 | STATIC I32 |
e2e6a0f1 | 5095 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth) |
a687059c | 5096 | { |
27da23d5 | 5097 | dVAR; |
a0d0e21e | 5098 | register char *scan; |
a0d0e21e | 5099 | register I32 c; |
3280af22 | 5100 | register char *loceol = PL_regeol; |
a0ed51b3 | 5101 | register I32 hardcount = 0; |
53c4c00c | 5102 | register bool do_utf8 = PL_reg_match_utf8; |
a0d0e21e | 5103 | |
3280af22 | 5104 | scan = PL_reginput; |
faf11cac HS |
5105 | if (max == REG_INFTY) |
5106 | max = I32_MAX; | |
5107 | else if (max < loceol - scan) | |
7f596f4c | 5108 | loceol = scan + max; |
a0d0e21e | 5109 | switch (OP(p)) { |
22c35a8c | 5110 | case REG_ANY: |
1aa99e6b | 5111 | if (do_utf8) { |
ffc61ed2 | 5112 | loceol = PL_regeol; |
1aa99e6b | 5113 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
5114 | scan += UTF8SKIP(scan); |
5115 | hardcount++; | |
5116 | } | |
5117 | } else { | |
5118 | while (scan < loceol && *scan != '\n') | |
5119 | scan++; | |
a0ed51b3 LW |
5120 | } |
5121 | break; | |
ffc61ed2 | 5122 | case SANY: |
def8e4ea JH |
5123 | if (do_utf8) { |
5124 | loceol = PL_regeol; | |
a0804c9e | 5125 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
5126 | scan += UTF8SKIP(scan); |
5127 | hardcount++; | |
5128 | } | |
5129 | } | |
5130 | else | |
5131 | scan = loceol; | |
a0ed51b3 | 5132 | break; |
f33976b4 DB |
5133 | case CANY: |
5134 | scan = loceol; | |
5135 | break; | |
090f7165 JH |
5136 | case EXACT: /* length of string is 1 */ |
5137 | c = (U8)*STRING(p); | |
5138 | while (scan < loceol && UCHARAT(scan) == c) | |
5139 | scan++; | |
bbce6d69 | 5140 | break; |
5141 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 5142 | c = (U8)*STRING(p); |
bbce6d69 | 5143 | while (scan < loceol && |
22c35a8c | 5144 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 5145 | scan++; |
5146 | break; | |
5147 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 5148 | PL_reg_flags |= RF_tainted; |
cd439c50 | 5149 | c = (U8)*STRING(p); |
bbce6d69 | 5150 | while (scan < loceol && |
22c35a8c | 5151 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
5152 | scan++; |
5153 | break; | |
5154 | case ANYOF: | |
ffc61ed2 JH |
5155 | if (do_utf8) { |
5156 | loceol = PL_regeol; | |
cfc92286 | 5157 | while (hardcount < max && scan < loceol && |
32fc9b6a | 5158 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
5159 | scan += UTF8SKIP(scan); |
5160 | hardcount++; | |
5161 | } | |
5162 | } else { | |
32fc9b6a | 5163 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
5164 | scan++; |
5165 | } | |
a0d0e21e LW |
5166 | break; |
5167 | case ALNUM: | |
1aa99e6b | 5168 | if (do_utf8) { |
ffc61ed2 | 5169 | loceol = PL_regeol; |
1a4fad37 | 5170 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5171 | while (hardcount < max && scan < loceol && |
3568d838 | 5172 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5173 | scan += UTF8SKIP(scan); |
5174 | hardcount++; | |
5175 | } | |
5176 | } else { | |
5177 | while (scan < loceol && isALNUM(*scan)) | |
5178 | scan++; | |
a0ed51b3 LW |
5179 | } |
5180 | break; | |
bbce6d69 | 5181 | case ALNUML: |
3280af22 | 5182 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5183 | if (do_utf8) { |
ffc61ed2 | 5184 | loceol = PL_regeol; |
1aa99e6b IH |
5185 | while (hardcount < max && scan < loceol && |
5186 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5187 | scan += UTF8SKIP(scan); |
5188 | hardcount++; | |
5189 | } | |
5190 | } else { | |
5191 | while (scan < loceol && isALNUM_LC(*scan)) | |
5192 | scan++; | |
a0ed51b3 LW |
5193 | } |
5194 | break; | |
a0d0e21e | 5195 | case NALNUM: |
1aa99e6b | 5196 | if (do_utf8) { |
ffc61ed2 | 5197 | loceol = PL_regeol; |
1a4fad37 | 5198 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5199 | while (hardcount < max && scan < loceol && |
3568d838 | 5200 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5201 | scan += UTF8SKIP(scan); |
5202 | hardcount++; | |
5203 | } | |
5204 | } else { | |
5205 | while (scan < loceol && !isALNUM(*scan)) | |
5206 | scan++; | |
a0ed51b3 LW |
5207 | } |
5208 | break; | |
bbce6d69 | 5209 | case NALNUML: |
3280af22 | 5210 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5211 | if (do_utf8) { |
ffc61ed2 | 5212 | loceol = PL_regeol; |
1aa99e6b IH |
5213 | while (hardcount < max && scan < loceol && |
5214 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5215 | scan += UTF8SKIP(scan); |
5216 | hardcount++; | |
5217 | } | |
5218 | } else { | |
5219 | while (scan < loceol && !isALNUM_LC(*scan)) | |
5220 | scan++; | |
a0ed51b3 LW |
5221 | } |
5222 | break; | |
a0d0e21e | 5223 | case SPACE: |
1aa99e6b | 5224 | if (do_utf8) { |
ffc61ed2 | 5225 | loceol = PL_regeol; |
1a4fad37 | 5226 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5227 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5228 | (*scan == ' ' || |
5229 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5230 | scan += UTF8SKIP(scan); |
5231 | hardcount++; | |
5232 | } | |
5233 | } else { | |
5234 | while (scan < loceol && isSPACE(*scan)) | |
5235 | scan++; | |
a0ed51b3 LW |
5236 | } |
5237 | break; | |
bbce6d69 | 5238 | case SPACEL: |
3280af22 | 5239 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5240 | if (do_utf8) { |
ffc61ed2 | 5241 | loceol = PL_regeol; |
1aa99e6b | 5242 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5243 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5244 | scan += UTF8SKIP(scan); | |
5245 | hardcount++; | |
5246 | } | |
5247 | } else { | |
5248 | while (scan < loceol && isSPACE_LC(*scan)) | |
5249 | scan++; | |
a0ed51b3 LW |
5250 | } |
5251 | break; | |
a0d0e21e | 5252 | case NSPACE: |
1aa99e6b | 5253 | if (do_utf8) { |
ffc61ed2 | 5254 | loceol = PL_regeol; |
1a4fad37 | 5255 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5256 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5257 | !(*scan == ' ' || |
5258 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5259 | scan += UTF8SKIP(scan); |
5260 | hardcount++; | |
5261 | } | |
5262 | } else { | |
5263 | while (scan < loceol && !isSPACE(*scan)) | |
5264 | scan++; | |
5265 | break; | |
a0ed51b3 | 5266 | } |
bbce6d69 | 5267 | case NSPACEL: |
3280af22 | 5268 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5269 | if (do_utf8) { |
ffc61ed2 | 5270 | loceol = PL_regeol; |
1aa99e6b | 5271 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5272 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5273 | scan += UTF8SKIP(scan); | |
5274 | hardcount++; | |
5275 | } | |
5276 | } else { | |
5277 | while (scan < loceol && !isSPACE_LC(*scan)) | |
5278 | scan++; | |
a0ed51b3 LW |
5279 | } |
5280 | break; | |
a0d0e21e | 5281 | case DIGIT: |
1aa99e6b | 5282 | if (do_utf8) { |
ffc61ed2 | 5283 | loceol = PL_regeol; |
1a4fad37 | 5284 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5285 | while (hardcount < max && scan < loceol && |
3568d838 | 5286 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5287 | scan += UTF8SKIP(scan); |
5288 | hardcount++; | |
5289 | } | |
5290 | } else { | |
5291 | while (scan < loceol && isDIGIT(*scan)) | |
5292 | scan++; | |
a0ed51b3 LW |
5293 | } |
5294 | break; | |
a0d0e21e | 5295 | case NDIGIT: |
1aa99e6b | 5296 | if (do_utf8) { |
ffc61ed2 | 5297 | loceol = PL_regeol; |
1a4fad37 | 5298 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5299 | while (hardcount < max && scan < loceol && |
3568d838 | 5300 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5301 | scan += UTF8SKIP(scan); |
5302 | hardcount++; | |
5303 | } | |
5304 | } else { | |
5305 | while (scan < loceol && !isDIGIT(*scan)) | |
5306 | scan++; | |
a0ed51b3 LW |
5307 | } |
5308 | break; | |
a0d0e21e LW |
5309 | default: /* Called on something of 0 width. */ |
5310 | break; /* So match right here or not at all. */ | |
5311 | } | |
a687059c | 5312 | |
a0ed51b3 LW |
5313 | if (hardcount) |
5314 | c = hardcount; | |
5315 | else | |
5316 | c = scan - PL_reginput; | |
3280af22 | 5317 | PL_reginput = scan; |
a687059c | 5318 | |
a3621e74 | 5319 | DEBUG_r({ |
e68ec53f | 5320 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 5321 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
5322 | SV * const prop = sv_newmortal(); |
5323 | regprop(prog, prop, p); | |
5324 | PerlIO_printf(Perl_debug_log, | |
be8e71aa | 5325 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
e2e6a0f1 | 5326 | REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max); |
a3621e74 | 5327 | }); |
be8e71aa | 5328 | }); |
9041c2e3 | 5329 | |
a0d0e21e | 5330 | return(c); |
a687059c LW |
5331 | } |
5332 | ||
c277df42 | 5333 | |
be8e71aa | 5334 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 5335 | /* |
ffc61ed2 JH |
5336 | - regclass_swash - prepare the utf8 swash |
5337 | */ | |
5338 | ||
5339 | SV * | |
32fc9b6a | 5340 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 5341 | { |
97aff369 | 5342 | dVAR; |
9e55ce06 JH |
5343 | SV *sw = NULL; |
5344 | SV *si = NULL; | |
5345 | SV *alt = NULL; | |
f8fc2ecf YO |
5346 | RXi_GET_DECL(prog,progi); |
5347 | const struct reg_data * const data = prog ? progi->data : NULL; | |
ffc61ed2 | 5348 | |
4f639d21 | 5349 | if (data && data->count) { |
a3b680e6 | 5350 | const U32 n = ARG(node); |
ffc61ed2 | 5351 | |
4f639d21 DM |
5352 | if (data->what[n] == 's') { |
5353 | SV * const rv = (SV*)data->data[n]; | |
890ce7af | 5354 | AV * const av = (AV*)SvRV((SV*)rv); |
2d03de9c | 5355 | SV **const ary = AvARRAY(av); |
9e55ce06 | 5356 | SV **a, **b; |
9041c2e3 | 5357 | |
711a919c | 5358 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
5359 | * documentation of these array elements. */ |
5360 | ||
b11f357e | 5361 | si = *ary; |
8f7f7219 | 5362 | a = SvROK(ary[1]) ? &ary[1] : 0; |
b11f357e JH |
5363 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : 0; |
5364 | ||
ffc61ed2 JH |
5365 | if (a) |
5366 | sw = *a; | |
5367 | else if (si && doinit) { | |
5368 | sw = swash_init("utf8", "", si, 1, 0); | |
5369 | (void)av_store(av, 1, sw); | |
5370 | } | |
9e55ce06 JH |
5371 | if (b) |
5372 | alt = *b; | |
ffc61ed2 JH |
5373 | } |
5374 | } | |
5375 | ||
9e55ce06 JH |
5376 | if (listsvp) |
5377 | *listsvp = si; | |
5378 | if (altsvp) | |
5379 | *altsvp = alt; | |
ffc61ed2 JH |
5380 | |
5381 | return sw; | |
5382 | } | |
76234dfb | 5383 | #endif |
ffc61ed2 JH |
5384 | |
5385 | /* | |
ba7b4546 | 5386 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
5387 | |
5388 | The n is the ANYOF regnode, the p is the target string, lenp | |
5389 | is pointer to the maximum length of how far to go in the p | |
5390 | (if the lenp is zero, UTF8SKIP(p) is used), | |
5391 | do_utf8 tells whether the target string is in UTF-8. | |
5392 | ||
bbce6d69 | 5393 | */ |
5394 | ||
76e3520e | 5395 | STATIC bool |
32fc9b6a | 5396 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 5397 | { |
27da23d5 | 5398 | dVAR; |
a3b680e6 | 5399 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 5400 | bool match = FALSE; |
cc07378b | 5401 | UV c = *p; |
ae9ddab8 | 5402 | STRLEN len = 0; |
9e55ce06 | 5403 | STRLEN plen; |
1aa99e6b | 5404 | |
19f67299 TS |
5405 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5406 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5407 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5408 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5409 | if (len == (STRLEN)-1) |
5410 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5411 | } |
bbce6d69 | 5412 | |
0f0076b4 | 5413 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5414 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5415 | if (lenp) |
5416 | *lenp = 0; | |
ffc61ed2 | 5417 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5418 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5419 | match = TRUE; | |
bbce6d69 | 5420 | } |
3568d838 | 5421 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5422 | match = TRUE; |
ffc61ed2 | 5423 | if (!match) { |
9e55ce06 | 5424 | AV *av; |
32fc9b6a | 5425 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5426 | |
5427 | if (sw) { | |
3568d838 | 5428 | if (swash_fetch(sw, p, do_utf8)) |
ffc61ed2 JH |
5429 | match = TRUE; |
5430 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5431 | if (!match && lenp && av) { |
5432 | I32 i; | |
9e55ce06 | 5433 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5434 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5435 | STRLEN len; |
890ce7af | 5436 | const char * const s = SvPV_const(sv, len); |
9e55ce06 | 5437 | |
061b10df | 5438 | if (len <= plen && memEQ(s, (char*)p, len)) { |
9e55ce06 JH |
5439 | *lenp = len; |
5440 | match = TRUE; | |
5441 | break; | |
5442 | } | |
5443 | } | |
5444 | } | |
5445 | if (!match) { | |
89ebb4a3 | 5446 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 JH |
5447 | STRLEN tmplen; |
5448 | ||
9e55ce06 JH |
5449 | to_utf8_fold(p, tmpbuf, &tmplen); |
5450 | if (swash_fetch(sw, tmpbuf, do_utf8)) | |
5451 | match = TRUE; | |
5452 | } | |
ffc61ed2 JH |
5453 | } |
5454 | } | |
bbce6d69 | 5455 | } |
9e55ce06 | 5456 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5457 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5458 | } |
1aa99e6b | 5459 | if (!match && c < 256) { |
ffc61ed2 JH |
5460 | if (ANYOF_BITMAP_TEST(n, c)) |
5461 | match = TRUE; | |
5462 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5463 | U8 f; |
a0ed51b3 | 5464 | |
ffc61ed2 JH |
5465 | if (flags & ANYOF_LOCALE) { |
5466 | PL_reg_flags |= RF_tainted; | |
5467 | f = PL_fold_locale[c]; | |
5468 | } | |
5469 | else | |
5470 | f = PL_fold[c]; | |
5471 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5472 | match = TRUE; | |
5473 | } | |
5474 | ||
5475 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5476 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5477 | if ( |
5478 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5479 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5480 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5481 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5482 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5483 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5484 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5485 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5486 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5487 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5488 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5489 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5490 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5491 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5492 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5493 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5494 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5495 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5496 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5497 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5498 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5499 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5500 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5501 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5502 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5503 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5504 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5505 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5506 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5507 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5508 | ) /* How's that for a conditional? */ | |
5509 | { | |
5510 | match = TRUE; | |
5511 | } | |
a0ed51b3 | 5512 | } |
a0ed51b3 LW |
5513 | } |
5514 | ||
a0ed51b3 LW |
5515 | return (flags & ANYOF_INVERT) ? !match : match; |
5516 | } | |
161b471a | 5517 | |
dfe13c55 | 5518 | STATIC U8 * |
0ce71af7 | 5519 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5520 | { |
97aff369 | 5521 | dVAR; |
a0ed51b3 | 5522 | if (off >= 0) { |
1aa99e6b | 5523 | while (off-- && s < lim) { |
ffc61ed2 | 5524 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5525 | s += UTF8SKIP(s); |
ffc61ed2 | 5526 | } |
a0ed51b3 LW |
5527 | } |
5528 | else { | |
1de06328 YO |
5529 | while (off++ && s > lim) { |
5530 | s--; | |
5531 | if (UTF8_IS_CONTINUED(*s)) { | |
5532 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5533 | s--; | |
a0ed51b3 | 5534 | } |
1de06328 | 5535 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5536 | } |
5537 | } | |
5538 | return s; | |
5539 | } | |
161b471a | 5540 | |
f9f4320a YO |
5541 | #ifdef XXX_dmq |
5542 | /* there are a bunch of places where we use two reghop3's that should | |
5543 | be replaced with this routine. but since thats not done yet | |
5544 | we ifdef it out - dmq | |
5545 | */ | |
dfe13c55 | 5546 | STATIC U8 * |
1de06328 YO |
5547 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5548 | { | |
5549 | dVAR; | |
5550 | if (off >= 0) { | |
5551 | while (off-- && s < rlim) { | |
5552 | /* XXX could check well-formedness here */ | |
5553 | s += UTF8SKIP(s); | |
5554 | } | |
5555 | } | |
5556 | else { | |
5557 | while (off++ && s > llim) { | |
5558 | s--; | |
5559 | if (UTF8_IS_CONTINUED(*s)) { | |
5560 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5561 | s--; | |
5562 | } | |
5563 | /* XXX could check well-formedness here */ | |
5564 | } | |
5565 | } | |
5566 | return s; | |
5567 | } | |
f9f4320a | 5568 | #endif |
1de06328 YO |
5569 | |
5570 | STATIC U8 * | |
0ce71af7 | 5571 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5572 | { |
97aff369 | 5573 | dVAR; |
a0ed51b3 | 5574 | if (off >= 0) { |
1aa99e6b | 5575 | while (off-- && s < lim) { |
ffc61ed2 | 5576 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5577 | s += UTF8SKIP(s); |
ffc61ed2 | 5578 | } |
a0ed51b3 | 5579 | if (off >= 0) |
3dab1dad | 5580 | return NULL; |
a0ed51b3 LW |
5581 | } |
5582 | else { | |
1de06328 YO |
5583 | while (off++ && s > lim) { |
5584 | s--; | |
5585 | if (UTF8_IS_CONTINUED(*s)) { | |
5586 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5587 | s--; | |
a0ed51b3 | 5588 | } |
1de06328 | 5589 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5590 | } |
5591 | if (off <= 0) | |
3dab1dad | 5592 | return NULL; |
a0ed51b3 LW |
5593 | } |
5594 | return s; | |
5595 | } | |
51371543 | 5596 | |
51371543 | 5597 | static void |
acfe0abc | 5598 | restore_pos(pTHX_ void *arg) |
51371543 | 5599 | { |
97aff369 | 5600 | dVAR; |
097eb12c | 5601 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5602 | if (PL_reg_eval_set) { |
5603 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5604 | rex->subbeg = PL_reg_oldsaved; |
5605 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5606 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5607 | rex->saved_copy = PL_nrs; |
ed252734 | 5608 | #endif |
4f639d21 | 5609 | RX_MATCH_COPIED_on(rex); |
51371543 GS |
5610 | } |
5611 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
5612 | PL_reg_eval_set = 0; | |
5613 | PL_curpm = PL_reg_oldcurpm; | |
5614 | } | |
5615 | } | |
33b8afdf JH |
5616 | |
5617 | STATIC void | |
5618 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
5619 | { | |
a1cac82e NC |
5620 | int i = 1; |
5621 | do { | |
5622 | if (prog->substrs->data[i].substr | |
5623 | && !prog->substrs->data[i].utf8_substr) { | |
5624 | SV* const sv = newSVsv(prog->substrs->data[i].substr); | |
5625 | prog->substrs->data[i].utf8_substr = sv; | |
5626 | sv_utf8_upgrade(sv); | |
5627 | if (SvVALID(prog->substrs->data[i].substr)) | |
5628 | fbm_compile(sv, 0); | |
5629 | if (SvTAIL(prog->substrs->data[i].substr)) | |
5630 | SvTAIL_on(sv); | |
5631 | if (prog->substrs->data[i].substr == prog->check_substr) | |
5632 | prog->check_utf8 = sv; | |
5633 | } | |
5634 | } while (i--); | |
33b8afdf JH |
5635 | } |
5636 | ||
5637 | STATIC void | |
5638 | S_to_byte_substr(pTHX_ register regexp *prog) | |
5639 | { | |
97aff369 | 5640 | dVAR; |
a1cac82e NC |
5641 | int i = 1; |
5642 | do { | |
5643 | if (prog->substrs->data[i].utf8_substr | |
5644 | && !prog->substrs->data[i].substr) { | |
5645 | SV* sv = newSVsv(prog->substrs->data[i].utf8_substr); | |
5646 | if (sv_utf8_downgrade(sv, TRUE)) { | |
5647 | if (SvVALID(prog->substrs->data[i].utf8_substr)) | |
5648 | fbm_compile(sv, 0); | |
5649 | if (SvTAIL(prog->substrs->data[i].utf8_substr)) | |
5650 | SvTAIL_on(sv); | |
5651 | } else { | |
5652 | SvREFCNT_dec(sv); | |
5653 | sv = &PL_sv_undef; | |
5654 | } | |
5655 | prog->substrs->data[i].substr = sv; | |
5656 | if (prog->substrs->data[i].utf8_substr == prog->check_utf8) | |
5657 | prog->check_substr = sv; | |
33b8afdf | 5658 | } |
a1cac82e | 5659 | } while (i--); |
33b8afdf | 5660 | } |
66610fdd RGS |
5661 | |
5662 | /* | |
5663 | * Local variables: | |
5664 | * c-indentation-style: bsd | |
5665 | * c-basic-offset: 4 | |
5666 | * indent-tabs-mode: t | |
5667 | * End: | |
5668 | * | |
37442d52 RGS |
5669 | * ex: set ts=8 sts=4 sw=4 noet: |
5670 | */ |