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 */ |
3e901dc0 YO |
125 | /* it would be nice to rework regcomp.sym to generate this stuff. sigh */ |
126 | #define JUMPABLE(rn) ( \ | |
127 | OP(rn) == OPEN || \ | |
128 | (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \ | |
129 | OP(rn) == EVAL || \ | |
cca55fe3 JP |
130 | OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ |
131 | OP(rn) == PLUS || OP(rn) == MINMOD || \ | |
ee9b8eae | 132 | OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \ |
3dab1dad | 133 | (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \ |
e2d8ce26 | 134 | ) |
ee9b8eae | 135 | #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT) |
e2d8ce26 | 136 | |
ee9b8eae YO |
137 | #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF ) |
138 | ||
139 | #if 0 | |
140 | /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so | |
141 | we don't need this definition. */ | |
142 | #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF ) | |
143 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF ) | |
144 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL ) | |
145 | ||
146 | #else | |
147 | /* ... so we use this as its faster. */ | |
148 | #define IS_TEXT(rn) ( OP(rn)==EXACT ) | |
149 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF ) | |
150 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL ) | |
151 | ||
152 | #endif | |
e2d8ce26 | 153 | |
a84d97b6 HS |
154 | /* |
155 | Search for mandatory following text node; for lookahead, the text must | |
156 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
157 | */ | |
cca55fe3 | 158 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
3dab1dad YO |
159 | while (JUMPABLE(rn)) { \ |
160 | const OPCODE type = OP(rn); \ | |
161 | if (type == SUSPEND || PL_regkind[type] == CURLY) \ | |
e2d8ce26 | 162 | rn = NEXTOPER(NEXTOPER(rn)); \ |
3dab1dad | 163 | else if (type == PLUS) \ |
cca55fe3 | 164 | rn = NEXTOPER(rn); \ |
3dab1dad | 165 | else if (type == IFMATCH) \ |
a84d97b6 | 166 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ |
e2d8ce26 | 167 | else rn += NEXT_OFF(rn); \ |
3dab1dad | 168 | } \ |
5f80c4cf | 169 | } STMT_END |
74750237 | 170 | |
c476f425 | 171 | |
acfe0abc | 172 | static void restore_pos(pTHX_ void *arg); |
51371543 | 173 | |
76e3520e | 174 | STATIC CHECKPOINT |
cea2e8a9 | 175 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 176 | { |
97aff369 | 177 | dVAR; |
a3b680e6 | 178 | const int retval = PL_savestack_ix; |
b1ce53c5 | 179 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 180 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 181 | int p; |
40a82448 | 182 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 183 | |
e49a9654 IH |
184 | if (paren_elems_to_push < 0) |
185 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
186 | ||
f0ab9afb | 187 | #define REGCP_OTHER_ELEMS 7 |
4b3c1a47 | 188 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
7f69552c | 189 | |
3280af22 | 190 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 191 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
f0ab9afb NC |
192 | SSPUSHINT(PL_regoffs[p].end); |
193 | SSPUSHINT(PL_regoffs[p].start); | |
3280af22 | 194 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 195 | SSPUSHINT(p); |
e7707071 | 196 | DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log, |
40a82448 | 197 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", |
f0ab9afb | 198 | (UV)p, (IV)PL_regoffs[p].start, |
40a82448 | 199 | (IV)(PL_reg_start_tmp[p] - PL_bostr), |
f0ab9afb | 200 | (IV)PL_regoffs[p].end |
40a82448 | 201 | )); |
a0d0e21e | 202 | } |
b1ce53c5 | 203 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
f0ab9afb | 204 | SSPUSHPTR(PL_regoffs); |
3280af22 NIS |
205 | SSPUSHINT(PL_regsize); |
206 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 207 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 208 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
209 | #define REGCP_FRAME_ELEMS 2 |
210 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
211 | * are needed for the regexp context stack bookkeeping. */ | |
212 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 213 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 214 | |
a0d0e21e LW |
215 | return retval; |
216 | } | |
217 | ||
c277df42 | 218 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
219 | #define REGCP_SET(cp) \ |
220 | DEBUG_STATE_r( \ | |
ab3bbdeb | 221 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 222 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
223 | (IV)PL_savestack_ix)); \ |
224 | cp = PL_savestack_ix | |
c3464db5 | 225 | |
ab3bbdeb | 226 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 227 | DEBUG_STATE_r( \ |
ab3bbdeb | 228 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
229 | PerlIO_printf(Perl_debug_log, \ |
230 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
231 | (IV)(cp), (IV)PL_savestack_ix)); \ |
232 | regcpblow(cp) | |
c277df42 | 233 | |
76e3520e | 234 | STATIC char * |
097eb12c | 235 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 236 | { |
97aff369 | 237 | dVAR; |
3b6647e0 | 238 | U32 i; |
a0d0e21e | 239 | char *input; |
b1ce53c5 | 240 | |
a3621e74 YO |
241 | GET_RE_DEBUG_FLAGS_DECL; |
242 | ||
b1ce53c5 | 243 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 244 | i = SSPOPINT; |
b1ce53c5 JH |
245 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
246 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 247 | input = (char *) SSPOPPTR; |
a01268b5 | 248 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
249 | *PL_reglastparen = SSPOPINT; |
250 | PL_regsize = SSPOPINT; | |
f0ab9afb | 251 | PL_regoffs=(regexp_paren_pair *) SSPOPPTR; |
b1ce53c5 | 252 | |
6bda09f9 | 253 | |
b1ce53c5 | 254 | /* Now restore the parentheses context. */ |
41123dfd JH |
255 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
256 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 257 | I32 tmps; |
097eb12c | 258 | U32 paren = (U32)SSPOPINT; |
3280af22 | 259 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
f0ab9afb | 260 | PL_regoffs[paren].start = SSPOPINT; |
cf93c79d | 261 | tmps = SSPOPINT; |
3280af22 | 262 | if (paren <= *PL_reglastparen) |
f0ab9afb | 263 | PL_regoffs[paren].end = tmps; |
e7707071 | 264 | DEBUG_BUFFERS_r( |
c3464db5 | 265 | PerlIO_printf(Perl_debug_log, |
b900a521 | 266 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
f0ab9afb | 267 | (UV)paren, (IV)PL_regoffs[paren].start, |
b900a521 | 268 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
f0ab9afb | 269 | (IV)PL_regoffs[paren].end, |
3280af22 | 270 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 271 | ); |
a0d0e21e | 272 | } |
e7707071 | 273 | DEBUG_BUFFERS_r( |
bb7a0f54 | 274 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 275 | PerlIO_printf(Perl_debug_log, |
faccc32b | 276 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 277 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
278 | } |
279 | ); | |
daf18116 | 280 | #if 1 |
dafc8851 JH |
281 | /* It would seem that the similar code in regtry() |
282 | * already takes care of this, and in fact it is in | |
283 | * a better location to since this code can #if 0-ed out | |
284 | * but the code in regtry() is needed or otherwise tests | |
285 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
286 | * (as of patchlevel 7877) will fail. Then again, |
287 | * this code seems to be necessary or otherwise | |
225593e1 DM |
288 | * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ |
289 | * --jhi updated by dapm */ | |
3b6647e0 | 290 | for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { |
097eb12c | 291 | if (i > PL_regsize) |
f0ab9afb NC |
292 | PL_regoffs[i].start = -1; |
293 | PL_regoffs[i].end = -1; | |
a0d0e21e | 294 | } |
dafc8851 | 295 | #endif |
a0d0e21e LW |
296 | return input; |
297 | } | |
298 | ||
02db2b7b | 299 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 300 | |
a687059c | 301 | /* |
e50aee73 | 302 | * pregexec and friends |
a687059c LW |
303 | */ |
304 | ||
76234dfb | 305 | #ifndef PERL_IN_XSUB_RE |
a687059c | 306 | /* |
c277df42 | 307 | - pregexec - match a regexp against a string |
a687059c | 308 | */ |
c277df42 | 309 | I32 |
49d7dfbc | 310 | Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend, |
c3464db5 | 311 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
312 | /* strend: pointer to null at end of string */ |
313 | /* strbeg: real beginning of string */ | |
314 | /* minend: end of match must be >=minend after stringarg. */ | |
315 | /* nosave: For optimizations. */ | |
316 | { | |
317 | return | |
9041c2e3 | 318 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
319 | nosave ? 0 : REXEC_COPY_STR); |
320 | } | |
76234dfb | 321 | #endif |
22e551b9 | 322 | |
9041c2e3 | 323 | /* |
cad2e5aa JH |
324 | * Need to implement the following flags for reg_anch: |
325 | * | |
326 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
327 | * USE_INTUIT_ML | |
328 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
329 | * INTUIT_AUTORITATIVE_ML | |
330 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
331 | * INTUIT_ONCE_ML | |
332 | * | |
333 | * Another flag for this function: SECOND_TIME (so that float substrs | |
334 | * with giant delta may be not rechecked). | |
335 | */ | |
336 | ||
337 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
338 | ||
3f7c398e | 339 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
340 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
341 | ||
342 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
343 | ||
6eb5f6b9 JH |
344 | /* XXXX Some places assume that there is a fixed substring. |
345 | An update may be needed if optimizer marks as "INTUITable" | |
346 | RExen without fixed substrings. Similarly, it is assumed that | |
347 | lengths of all the strings are no more than minlen, thus they | |
348 | cannot come from lookahead. | |
40d049e4 YO |
349 | (Or minlen should take into account lookahead.) |
350 | NOTE: Some of this comment is not correct. minlen does now take account | |
351 | of lookahead/behind. Further research is required. -- demerphq | |
352 | ||
353 | */ | |
6eb5f6b9 | 354 | |
2c2d71f5 JH |
355 | /* A failure to find a constant substring means that there is no need to make |
356 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
357 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
358 | regtry() should be needed. |
359 | ||
360 | REx compiler's optimizer found 4 possible hints: | |
361 | a) Anchored substring; | |
362 | b) Fixed substring; | |
363 | c) Whether we are anchored (beginning-of-line or \G); | |
364 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 365 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
366 | string which does not contradict any of them. |
367 | */ | |
2c2d71f5 | 368 | |
6eb5f6b9 JH |
369 | /* Most of decisions we do here should have been done at compile time. |
370 | The nodes of the REx which we used for the search should have been | |
371 | deleted from the finite automaton. */ | |
372 | ||
cad2e5aa | 373 | char * |
49d7dfbc | 374 | Perl_re_intuit_start(pTHX_ REGEXP * const prog, SV *sv, char *strpos, |
9f61653a | 375 | char *strend, const U32 flags, re_scream_pos_data *data) |
cad2e5aa | 376 | { |
97aff369 | 377 | dVAR; |
b7953727 | 378 | register I32 start_shift = 0; |
cad2e5aa | 379 | /* Should be nonnegative! */ |
b7953727 | 380 | register I32 end_shift = 0; |
2c2d71f5 JH |
381 | register char *s; |
382 | register SV *check; | |
a1933d95 | 383 | char *strbeg; |
cad2e5aa | 384 | char *t; |
1de06328 | 385 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 386 | I32 ml_anch; |
bd61b366 SS |
387 | register char *other_last = NULL; /* other substr checked before this */ |
388 | char *check_at = NULL; /* check substr found at this pos */ | |
bbe252da | 389 | const I32 multiline = prog->extflags & RXf_PMf_MULTILINE; |
f8fc2ecf | 390 | RXi_GET_DECL(prog,progi); |
30944b6d | 391 | #ifdef DEBUGGING |
890ce7af | 392 | const char * const i_strpos = strpos; |
30944b6d | 393 | #endif |
a3621e74 YO |
394 | |
395 | GET_RE_DEBUG_FLAGS_DECL; | |
396 | ||
a30b2f1f | 397 | RX_MATCH_UTF8_set(prog,do_utf8); |
cad2e5aa | 398 | |
bbe252da | 399 | if (prog->extflags & RXf_UTF8) { |
b8d68ded JH |
400 | PL_reg_flags |= RF_utf8; |
401 | } | |
ab3bbdeb YO |
402 | DEBUG_EXECUTE_r( |
403 | debug_start_match(prog, do_utf8, strpos, strend, | |
1de06328 YO |
404 | sv ? "Guessing start of match in sv for" |
405 | : "Guessing start of match in string for"); | |
2a782b5b | 406 | ); |
cad2e5aa | 407 | |
c344f387 JH |
408 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
409 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 410 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 411 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 412 | goto fail; |
2c2d71f5 | 413 | } |
1de06328 | 414 | |
a1933d95 | 415 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 416 | PL_regeol = strend; |
33b8afdf JH |
417 | if (do_utf8) { |
418 | if (!prog->check_utf8 && prog->check_substr) | |
419 | to_utf8_substr(prog); | |
420 | check = prog->check_utf8; | |
421 | } else { | |
422 | if (!prog->check_substr && prog->check_utf8) | |
423 | to_byte_substr(prog); | |
424 | check = prog->check_substr; | |
425 | } | |
1de06328 | 426 | if (check == &PL_sv_undef) { |
a3621e74 | 427 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 428 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
429 | goto fail; |
430 | } | |
bbe252da YO |
431 | if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */ |
432 | ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE) | |
433 | || ( (prog->extflags & RXf_ANCH_BOL) | |
7fba1cd6 | 434 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 435 | |
7e25d62c | 436 | if (!ml_anch) { |
bbe252da YO |
437 | if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */ |
438 | && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */ | |
3f7c398e | 439 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
440 | && sv && !SvROK(sv) |
441 | && (strpos != strbeg)) { | |
a3621e74 | 442 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
443 | goto fail; |
444 | } | |
445 | if (prog->check_offset_min == prog->check_offset_max && | |
bbe252da | 446 | !(prog->extflags & RXf_CANY_SEEN)) { |
2c2d71f5 | 447 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
448 | I32 slen; |
449 | ||
1aa99e6b | 450 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 451 | |
653099ff GS |
452 | if (SvTAIL(check)) { |
453 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 454 | |
9041c2e3 | 455 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 456 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 457 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 458 | goto fail_finish; |
cad2e5aa JH |
459 | } |
460 | /* Now should match s[0..slen-2] */ | |
461 | slen--; | |
3f7c398e | 462 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 463 | || (slen > 1 |
3f7c398e | 464 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 465 | report_neq: |
a3621e74 | 466 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
467 | goto fail_finish; |
468 | } | |
cad2e5aa | 469 | } |
3f7c398e | 470 | else if (*SvPVX_const(check) != *s |
653099ff | 471 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 472 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 473 | goto report_neq; |
c315bfe8 | 474 | check_at = s; |
2c2d71f5 | 475 | goto success_at_start; |
7e25d62c | 476 | } |
cad2e5aa | 477 | } |
2c2d71f5 | 478 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 479 | s = strpos; |
2c2d71f5 | 480 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
481 | end_shift = prog->check_end_shift; |
482 | ||
2c2d71f5 | 483 | if (!ml_anch) { |
a3b680e6 | 484 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 485 | - (SvTAIL(check) != 0); |
a3b680e6 | 486 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
487 | |
488 | if (end_shift < eshift) | |
489 | end_shift = eshift; | |
490 | } | |
cad2e5aa | 491 | } |
2c2d71f5 | 492 | else { /* Can match at random position */ |
cad2e5aa JH |
493 | ml_anch = 0; |
494 | s = strpos; | |
1de06328 YO |
495 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
496 | end_shift = prog->check_end_shift; | |
497 | ||
498 | /* end shift should be non negative here */ | |
cad2e5aa JH |
499 | } |
500 | ||
bcdf7404 | 501 | #ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 502 | if (end_shift < 0) |
1de06328 | 503 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
b57a0404 | 504 | (IV)end_shift, prog->precomp); |
2c2d71f5 JH |
505 | #endif |
506 | ||
2c2d71f5 JH |
507 | restart: |
508 | /* Find a possible match in the region s..strend by looking for | |
509 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
510 | |
511 | { | |
512 | I32 srch_start_shift = start_shift; | |
513 | I32 srch_end_shift = end_shift; | |
514 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
515 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
516 | srch_start_shift = strbeg - s; | |
517 | } | |
6bda09f9 | 518 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
519 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
520 | (IV)prog->check_offset_min, | |
521 | (IV)srch_start_shift, | |
522 | (IV)srch_end_shift, | |
523 | (IV)prog->check_end_shift); | |
524 | }); | |
525 | ||
cad2e5aa | 526 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 527 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 528 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 529 | |
2c2d71f5 JH |
530 | if (PL_screamfirst[BmRARE(check)] >= 0 |
531 | || ( BmRARE(check) == '\n' | |
85c508c3 | 532 | && (BmPREVIOUS(check) == SvCUR(check) - 1) |
2c2d71f5 | 533 | && SvTAIL(check) )) |
9041c2e3 | 534 | s = screaminstr(sv, check, |
1de06328 | 535 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 536 | else |
2c2d71f5 | 537 | goto fail_finish; |
4addbd3b HS |
538 | /* we may be pointing at the wrong string */ |
539 | if (s && RX_MATCH_COPIED(prog)) | |
3f7c398e | 540 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
541 | if (data) |
542 | *data->scream_olds = s; | |
543 | } | |
1de06328 YO |
544 | else { |
545 | U8* start_point; | |
546 | U8* end_point; | |
bbe252da | 547 | if (prog->extflags & RXf_CANY_SEEN) { |
1de06328 YO |
548 | start_point= (U8*)(s + srch_start_shift); |
549 | end_point= (U8*)(strend - srch_end_shift); | |
550 | } else { | |
551 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
552 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
553 | } | |
6bda09f9 | 554 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 555 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 556 | (int)(end_point - start_point), |
fc8cd66c | 557 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
558 | start_point); |
559 | }); | |
560 | ||
561 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 562 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
563 | } |
564 | } | |
cad2e5aa JH |
565 | /* Update the count-of-usability, remove useless subpatterns, |
566 | unshift s. */ | |
2c2d71f5 | 567 | |
ab3bbdeb YO |
568 | DEBUG_EXECUTE_r({ |
569 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
570 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
571 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 572 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
573 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
574 | ? "anchored" : "floating"), | |
575 | quoted, | |
576 | RE_SV_TAIL(check), | |
577 | (s ? " at offset " : "...\n") ); | |
578 | }); | |
2c2d71f5 JH |
579 | |
580 | if (!s) | |
581 | goto fail_finish; | |
2c2d71f5 | 582 | /* Finish the diagnostic message */ |
a3621e74 | 583 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 584 | |
1de06328 YO |
585 | /* XXX dmq: first branch is for positive lookbehind... |
586 | Our check string is offset from the beginning of the pattern. | |
587 | So we need to do any stclass tests offset forward from that | |
588 | point. I think. :-( | |
589 | */ | |
590 | ||
591 | ||
592 | ||
593 | check_at=s; | |
594 | ||
595 | ||
2c2d71f5 JH |
596 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
597 | Start with the other substr. | |
598 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 599 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
600 | *always* match. Probably should be marked during compile... |
601 | Probably it is right to do no SCREAM here... | |
602 | */ | |
603 | ||
1de06328 YO |
604 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
605 | : (prog->float_substr && prog->anchored_substr)) | |
606 | { | |
30944b6d | 607 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
608 | /* XXXX May be hopelessly wrong for UTF... */ |
609 | if (!other_last) | |
6eb5f6b9 | 610 | other_last = strpos; |
33b8afdf | 611 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
612 | do_other_anchored: |
613 | { | |
890ce7af AL |
614 | char * const last = HOP3c(s, -start_shift, strbeg); |
615 | char *last1, *last2; | |
be8e71aa | 616 | char * const saved_s = s; |
33b8afdf | 617 | SV* must; |
2c2d71f5 | 618 | |
2c2d71f5 JH |
619 | t = s - prog->check_offset_max; |
620 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 621 | && (!do_utf8 |
0ce71af7 | 622 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 623 | && t > strpos))) |
6f207bd3 | 624 | NOOP; |
2c2d71f5 JH |
625 | else |
626 | t = strpos; | |
1aa99e6b | 627 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
628 | if (t < other_last) /* These positions already checked */ |
629 | t = other_last; | |
1aa99e6b | 630 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
631 | if (last < last1) |
632 | last1 = last; | |
1de06328 YO |
633 | /* XXXX It is not documented what units *_offsets are in. |
634 | We assume bytes, but this is clearly wrong. | |
635 | Meaning this code needs to be carefully reviewed for errors. | |
636 | dmq. | |
637 | */ | |
638 | ||
2c2d71f5 | 639 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
640 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
641 | if (must == &PL_sv_undef) { | |
642 | s = (char*)NULL; | |
1de06328 | 643 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
644 | } |
645 | else | |
646 | s = fbm_instr( | |
647 | (unsigned char*)t, | |
648 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
649 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
650 | must, | |
7fba1cd6 | 651 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 652 | ); |
ab3bbdeb YO |
653 | DEBUG_EXECUTE_r({ |
654 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
655 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
656 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 657 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
658 | quoted, RE_SV_TAIL(must)); |
659 | }); | |
660 | ||
661 | ||
2c2d71f5 JH |
662 | if (!s) { |
663 | if (last1 >= last2) { | |
a3621e74 | 664 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
665 | ", giving up...\n")); |
666 | goto fail_finish; | |
667 | } | |
a3621e74 | 668 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 669 | ", trying floating at offset %ld...\n", |
be8e71aa | 670 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
671 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
672 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
673 | goto restart; |
674 | } | |
675 | else { | |
a3621e74 | 676 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 677 | (long)(s - i_strpos))); |
1aa99e6b IH |
678 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
679 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 680 | s = saved_s; |
2c2d71f5 JH |
681 | if (t == strpos) |
682 | goto try_at_start; | |
2c2d71f5 JH |
683 | goto try_at_offset; |
684 | } | |
30944b6d | 685 | } |
2c2d71f5 JH |
686 | } |
687 | else { /* Take into account the floating substring. */ | |
33b8afdf | 688 | char *last, *last1; |
be8e71aa | 689 | char * const saved_s = s; |
33b8afdf JH |
690 | SV* must; |
691 | ||
692 | t = HOP3c(s, -start_shift, strbeg); | |
693 | last1 = last = | |
694 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
695 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
696 | last = HOP3c(t, prog->float_max_offset, strend); | |
697 | s = HOP3c(t, prog->float_min_offset, strend); | |
698 | if (s < other_last) | |
699 | s = other_last; | |
2c2d71f5 | 700 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
701 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
702 | /* fbm_instr() takes into account exact value of end-of-str | |
703 | if the check is SvTAIL(ed). Since false positives are OK, | |
704 | and end-of-str is not later than strend we are OK. */ | |
705 | if (must == &PL_sv_undef) { | |
706 | s = (char*)NULL; | |
1de06328 | 707 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
708 | } |
709 | else | |
2c2d71f5 | 710 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
711 | (unsigned char*)last + SvCUR(must) |
712 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 713 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
714 | DEBUG_EXECUTE_r({ |
715 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
716 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
717 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 718 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
719 | quoted, RE_SV_TAIL(must)); |
720 | }); | |
33b8afdf JH |
721 | if (!s) { |
722 | if (last1 == last) { | |
a3621e74 | 723 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
724 | ", giving up...\n")); |
725 | goto fail_finish; | |
2c2d71f5 | 726 | } |
a3621e74 | 727 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 728 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 729 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
730 | other_last = last; |
731 | s = HOP3c(t, 1, strend); | |
732 | goto restart; | |
733 | } | |
734 | else { | |
a3621e74 | 735 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
736 | (long)(s - i_strpos))); |
737 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 738 | s = saved_s; |
33b8afdf JH |
739 | if (t == strpos) |
740 | goto try_at_start; | |
741 | goto try_at_offset; | |
742 | } | |
2c2d71f5 | 743 | } |
cad2e5aa | 744 | } |
2c2d71f5 | 745 | |
1de06328 | 746 | |
9ef43ace | 747 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 748 | |
6bda09f9 | 749 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
750 | PerlIO_printf(Perl_debug_log, |
751 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
752 | (IV)prog->check_offset_min, | |
753 | (IV)prog->check_offset_max, | |
754 | (IV)(s-strpos), | |
755 | (IV)(t-strpos), | |
756 | (IV)(t-s), | |
757 | (IV)(strend-strpos) | |
758 | ) | |
759 | ); | |
760 | ||
2c2d71f5 | 761 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 762 | && (!do_utf8 |
9ef43ace | 763 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
764 | && t > strpos))) |
765 | { | |
2c2d71f5 JH |
766 | /* Fixed substring is found far enough so that the match |
767 | cannot start at strpos. */ | |
768 | try_at_offset: | |
cad2e5aa | 769 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
770 | /* Eventually fbm_*() should handle this, but often |
771 | anchored_offset is not 0, so this check will not be wasted. */ | |
772 | /* XXXX In the code below we prefer to look for "^" even in | |
773 | presence of anchored substrings. And we search even | |
774 | beyond the found float position. These pessimizations | |
775 | are historical artefacts only. */ | |
776 | find_anchor: | |
2c2d71f5 | 777 | while (t < strend - prog->minlen) { |
cad2e5aa | 778 | if (*t == '\n') { |
4ee3650e | 779 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 780 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
781 | /* Since we moved from the found position, |
782 | we definitely contradict the found anchored | |
30944b6d IZ |
783 | substr. Due to the above check we do not |
784 | contradict "check" substr. | |
785 | Thus we can arrive here only if check substr | |
786 | is float. Redo checking for "other"=="fixed". | |
787 | */ | |
9041c2e3 | 788 | strpos = t + 1; |
a3621e74 | 789 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 790 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
791 | goto do_other_anchored; |
792 | } | |
4ee3650e GS |
793 | /* We don't contradict the found floating substring. */ |
794 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 795 | s = t + 1; |
a3621e74 | 796 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 797 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
798 | goto set_useful; |
799 | } | |
4ee3650e GS |
800 | /* Position contradicts check-string */ |
801 | /* XXXX probably better to look for check-string | |
802 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 803 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 804 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 805 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
806 | goto restart; |
807 | } | |
808 | t++; | |
809 | } | |
a3621e74 | 810 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 811 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 812 | goto fail_finish; |
cad2e5aa | 813 | } |
f5952150 | 814 | else { |
a3621e74 | 815 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 816 | PL_colors[0], PL_colors[1])); |
f5952150 | 817 | } |
cad2e5aa JH |
818 | s = t; |
819 | set_useful: | |
33b8afdf | 820 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
821 | } |
822 | else { | |
f5952150 | 823 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 824 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
825 | unless it was an MBOL and we are not after MBOL, |
826 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
827 | try_at_start: |
828 | /* Even in this situation we may use MBOL flag if strpos is offset | |
829 | wrt the start of the string. */ | |
05b4157f | 830 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 831 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d | 832 | /* May be due to an implicit anchor of m{.*foo} */ |
bbe252da | 833 | && !(prog->intflags & PREGf_IMPLICIT)) |
d506a20d | 834 | { |
cad2e5aa JH |
835 | t = strpos; |
836 | goto find_anchor; | |
837 | } | |
a3621e74 | 838 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 839 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
70685ca0 | 840 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 841 | ); |
2c2d71f5 | 842 | success_at_start: |
bbe252da | 843 | if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
844 | && (do_utf8 ? ( |
845 | prog->check_utf8 /* Could be deleted already */ | |
846 | && --BmUSEFUL(prog->check_utf8) < 0 | |
847 | && (prog->check_utf8 == prog->float_utf8) | |
848 | ) : ( | |
849 | prog->check_substr /* Could be deleted already */ | |
850 | && --BmUSEFUL(prog->check_substr) < 0 | |
851 | && (prog->check_substr == prog->float_substr) | |
852 | ))) | |
66e933ab | 853 | { |
cad2e5aa | 854 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 855 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
856 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
857 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
858 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
859 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
860 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
861 | check = NULL; /* abort */ | |
cad2e5aa | 862 | s = strpos; |
3cf5c195 IZ |
863 | /* XXXX This is a remnant of the old implementation. It |
864 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 865 | other heuristics. */ |
bbe252da | 866 | prog->extflags &= ~RXf_USE_INTUIT; |
cad2e5aa JH |
867 | } |
868 | else | |
869 | s = strpos; | |
870 | } | |
871 | ||
6eb5f6b9 JH |
872 | /* Last resort... */ |
873 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
874 | /* trie stclasses are too expensive to use here, we are better off to |
875 | leave it to regmatch itself */ | |
f8fc2ecf | 876 | if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
877 | /* minlen == 0 is possible if regstclass is \b or \B, |
878 | and the fixed substr is ''$. | |
879 | Since minlen is already taken into account, s+1 is before strend; | |
880 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
881 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
882 | regstclass does not come from lookahead... */ | |
883 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
884 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
f8fc2ecf YO |
885 | const U8* const str = (U8*)STRING(progi->regstclass); |
886 | const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT | |
887 | ? CHR_DIST(str+STR_LEN(progi->regstclass), str) | |
66e933ab | 888 | : 1); |
1de06328 YO |
889 | char * endpos; |
890 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
891 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
892 | else if (prog->float_substr || prog->float_utf8) | |
893 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
894 | else | |
895 | endpos= strend; | |
896 | ||
70685ca0 JH |
897 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n", |
898 | (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg))); | |
1de06328 | 899 | |
6eb5f6b9 | 900 | t = s; |
f8fc2ecf | 901 | s = find_byclass(prog, progi->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
902 | if (!s) { |
903 | #ifdef DEBUGGING | |
cbbf8932 | 904 | const char *what = NULL; |
6eb5f6b9 JH |
905 | #endif |
906 | if (endpos == strend) { | |
a3621e74 | 907 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
908 | "Could not match STCLASS...\n") ); |
909 | goto fail; | |
910 | } | |
a3621e74 | 911 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 912 | "This position contradicts STCLASS...\n") ); |
bbe252da | 913 | if ((prog->extflags & RXf_ANCH) && !ml_anch) |
653099ff | 914 | goto fail; |
6eb5f6b9 | 915 | /* Contradict one of substrings */ |
33b8afdf JH |
916 | if (prog->anchored_substr || prog->anchored_utf8) { |
917 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 918 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 919 | hop_and_restart: |
1aa99e6b | 920 | s = HOP3c(t, 1, strend); |
66e933ab GS |
921 | if (s + start_shift + end_shift > strend) { |
922 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 923 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
924 | "Could not match STCLASS...\n") ); |
925 | goto fail; | |
926 | } | |
5e39e1e5 HS |
927 | if (!check) |
928 | goto giveup; | |
a3621e74 | 929 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 930 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
931 | what, (long)(s + start_shift - i_strpos)) ); |
932 | goto restart; | |
933 | } | |
66e933ab | 934 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
935 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
936 | goto retry_floating_check; | |
937 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 938 | s = check_at; |
5e39e1e5 HS |
939 | if (!check) |
940 | goto giveup; | |
a3621e74 | 941 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 942 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
943 | (long)(other_last - i_strpos)) ); |
944 | goto do_other_anchored; | |
945 | } | |
60e71179 GS |
946 | /* Another way we could have checked stclass at the |
947 | current position only: */ | |
948 | if (ml_anch) { | |
949 | s = t = t + 1; | |
5e39e1e5 HS |
950 | if (!check) |
951 | goto giveup; | |
a3621e74 | 952 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 953 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 954 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 955 | goto try_at_offset; |
66e933ab | 956 | } |
33b8afdf | 957 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 958 | goto fail; |
6eb5f6b9 JH |
959 | /* Check is floating subtring. */ |
960 | retry_floating_check: | |
961 | t = check_at - start_shift; | |
a3621e74 | 962 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
963 | goto hop_and_restart; |
964 | } | |
b7953727 | 965 | if (t != s) { |
a3621e74 | 966 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 967 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
968 | (long)(t - i_strpos), (long)(s - i_strpos)) |
969 | ); | |
970 | } | |
971 | else { | |
a3621e74 | 972 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
973 | "Does not contradict STCLASS...\n"); |
974 | ); | |
975 | } | |
6eb5f6b9 | 976 | } |
5e39e1e5 | 977 | giveup: |
a3621e74 | 978 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
979 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
980 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 981 | return s; |
2c2d71f5 JH |
982 | |
983 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
984 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
985 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 986 | fail: |
a3621e74 | 987 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 988 | PL_colors[4], PL_colors[5])); |
bd61b366 | 989 | return NULL; |
cad2e5aa | 990 | } |
9661b544 | 991 | |
786e8c11 | 992 | |
3b0527fe | 993 | |
55eed653 NC |
994 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ |
995 | uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ | |
4cadc6a9 YO |
996 | switch (trie_type) { \ |
997 | case trie_utf8_fold: \ | |
998 | if ( foldlen>0 ) { \ | |
999 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
1000 | foldlen -= len; \ | |
1001 | uscan += len; \ | |
1002 | len=0; \ | |
1003 | } else { \ | |
1004 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
1005 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ | |
1006 | foldlen -= UNISKIP( uvc ); \ | |
1007 | uscan = foldbuf + UNISKIP( uvc ); \ | |
1008 | } \ | |
1009 | break; \ | |
1010 | case trie_utf8: \ | |
1011 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
1012 | break; \ | |
1013 | case trie_plain: \ | |
1014 | uvc = (UV)*uc; \ | |
1015 | len = 1; \ | |
1016 | } \ | |
1017 | \ | |
1018 | if (uvc < 256) { \ | |
1019 | charid = trie->charmap[ uvc ]; \ | |
1020 | } \ | |
1021 | else { \ | |
1022 | charid = 0; \ | |
55eed653 NC |
1023 | if (widecharmap) { \ |
1024 | SV** const svpp = hv_fetch(widecharmap, \ | |
4cadc6a9 YO |
1025 | (char*)&uvc, sizeof(UV), 0); \ |
1026 | if (svpp) \ | |
1027 | charid = (U16)SvIV(*svpp); \ | |
1028 | } \ | |
1029 | } \ | |
1030 | } STMT_END | |
1031 | ||
1032 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ | |
1033 | if ( (CoNd) \ | |
1034 | && (ln == len || \ | |
1035 | ibcmp_utf8(s, NULL, 0, do_utf8, \ | |
1036 | m, NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1037 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1038 | goto got_it; \ |
1039 | else { \ | |
1040 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1041 | uvchr_to_utf8(tmpbuf, c); \ | |
1042 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1043 | if ( f != c \ | |
1044 | && (f == c1 || f == c2) \ | |
1045 | && (ln == foldlen || \ | |
1046 | !ibcmp_utf8((char *) foldbuf, \ | |
1047 | NULL, foldlen, do_utf8, \ | |
1048 | m, \ | |
1049 | NULL, ln, (bool)UTF)) \ | |
24b23f37 | 1050 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1051 | goto got_it; \ |
1052 | } \ | |
1053 | s += len | |
1054 | ||
1055 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1056 | STMT_START { \ | |
1057 | while (s <= e) { \ | |
1058 | if ( (CoNd) \ | |
1059 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1060 | ? ibcmp(s, m, ln) \ | |
1061 | : ibcmp_locale(s, m, ln))) \ | |
24b23f37 | 1062 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1063 | goto got_it; \ |
1064 | s++; \ | |
1065 | } \ | |
1066 | } STMT_END | |
1067 | ||
1068 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1069 | STMT_START { \ | |
1070 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1071 | CoDe \ | |
1072 | s += uskip; \ | |
1073 | } \ | |
1074 | } STMT_END | |
1075 | ||
1076 | #define REXEC_FBC_SCAN(CoDe) \ | |
1077 | STMT_START { \ | |
1078 | while (s < strend) { \ | |
1079 | CoDe \ | |
1080 | s++; \ | |
1081 | } \ | |
1082 | } STMT_END | |
1083 | ||
1084 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1085 | REXEC_FBC_UTF8_SCAN( \ | |
1086 | if (CoNd) { \ | |
24b23f37 | 1087 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1088 | goto got_it; \ |
1089 | else \ | |
1090 | tmp = doevery; \ | |
1091 | } \ | |
1092 | else \ | |
1093 | tmp = 1; \ | |
1094 | ) | |
1095 | ||
1096 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1097 | REXEC_FBC_SCAN( \ | |
1098 | if (CoNd) { \ | |
24b23f37 | 1099 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1100 | goto got_it; \ |
1101 | else \ | |
1102 | tmp = doevery; \ | |
1103 | } \ | |
1104 | else \ | |
1105 | tmp = 1; \ | |
1106 | ) | |
1107 | ||
1108 | #define REXEC_FBC_TRYIT \ | |
24b23f37 | 1109 | if ((!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1110 | goto got_it |
1111 | ||
e1d1eefb YO |
1112 | #define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \ |
1113 | if (do_utf8) { \ | |
1114 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1115 | } \ | |
1116 | else { \ | |
1117 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1118 | } \ | |
1119 | break | |
1120 | ||
4cadc6a9 YO |
1121 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ |
1122 | if (do_utf8) { \ | |
1123 | UtFpReLoAd; \ | |
1124 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1125 | } \ | |
1126 | else { \ | |
1127 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1128 | } \ | |
1129 | break | |
1130 | ||
1131 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1132 | PL_reg_flags |= RF_tainted; \ | |
1133 | if (do_utf8) { \ | |
1134 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1135 | } \ | |
1136 | else { \ | |
1137 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1138 | } \ | |
1139 | break | |
1140 | ||
786e8c11 YO |
1141 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1142 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1143 | ||
1144 | /* We know what class REx starts with. Try to find this position... */ | |
1145 | /* if reginfo is NULL, its a dryrun */ | |
1146 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1147 | in regmatch. /grrr */ | |
1148 | ||
3c3eec57 | 1149 | STATIC char * |
07be1b83 | 1150 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
24b23f37 | 1151 | const char *strend, regmatch_info *reginfo) |
a687059c | 1152 | { |
27da23d5 | 1153 | dVAR; |
bbe252da | 1154 | const I32 doevery = (prog->intflags & PREGf_SKIP) == 0; |
6eb5f6b9 | 1155 | char *m; |
d8093b23 | 1156 | STRLEN ln; |
5dab1207 | 1157 | STRLEN lnc; |
078c425b | 1158 | register STRLEN uskip; |
d8093b23 G |
1159 | unsigned int c1; |
1160 | unsigned int c2; | |
6eb5f6b9 JH |
1161 | char *e; |
1162 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1163 | register const bool do_utf8 = PL_reg_match_utf8; |
f8fc2ecf YO |
1164 | RXi_GET_DECL(prog,progi); |
1165 | ||
6eb5f6b9 JH |
1166 | /* We know what class it must start with. */ |
1167 | switch (OP(c)) { | |
6eb5f6b9 | 1168 | case ANYOF: |
388cc4de | 1169 | if (do_utf8) { |
4cadc6a9 | 1170 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1171 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1172 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1173 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1174 | } |
1175 | else { | |
1176 | while (s < strend) { | |
1177 | STRLEN skip = 1; | |
1178 | ||
32fc9b6a | 1179 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1180 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1181 | /* The assignment of 2 is intentional: | |
1182 | * for the folded sharp s, the skip is 2. */ | |
1183 | (skip = SHARP_S_SKIP))) { | |
24b23f37 | 1184 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
388cc4de HS |
1185 | goto got_it; |
1186 | else | |
1187 | tmp = doevery; | |
1188 | } | |
1189 | else | |
1190 | tmp = 1; | |
1191 | s += skip; | |
1192 | } | |
a0d0e21e | 1193 | } |
6eb5f6b9 | 1194 | break; |
f33976b4 | 1195 | case CANY: |
4cadc6a9 | 1196 | REXEC_FBC_SCAN( |
24b23f37 | 1197 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
f33976b4 DB |
1198 | goto got_it; |
1199 | else | |
1200 | tmp = doevery; | |
4cadc6a9 | 1201 | ); |
f33976b4 | 1202 | break; |
6eb5f6b9 | 1203 | case EXACTF: |
5dab1207 NIS |
1204 | m = STRING(c); |
1205 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1206 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1207 | if (UTF) { |
a2a2844f | 1208 | STRLEN ulen1, ulen2; |
5dab1207 | 1209 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1210 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1211 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
4ad0818d | 1212 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
a2a2844f JH |
1213 | |
1214 | to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1215 | to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1216 | ||
89ebb4a3 | 1217 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1218 | 0, uniflags); |
89ebb4a3 | 1219 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1220 | 0, uniflags); |
5dab1207 NIS |
1221 | lnc = 0; |
1222 | while (sm < ((U8 *) m + ln)) { | |
1223 | lnc++; | |
1224 | sm += UTF8SKIP(sm); | |
1225 | } | |
1aa99e6b IH |
1226 | } |
1227 | else { | |
1228 | c1 = *(U8*)m; | |
1229 | c2 = PL_fold[c1]; | |
1230 | } | |
6eb5f6b9 JH |
1231 | goto do_exactf; |
1232 | case EXACTFL: | |
5dab1207 NIS |
1233 | m = STRING(c); |
1234 | ln = STR_LEN(c); | |
1235 | lnc = (I32) ln; | |
d8093b23 | 1236 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1237 | c2 = PL_fold_locale[c1]; |
1238 | do_exactf: | |
db12adc6 | 1239 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1240 | |
3b0527fe | 1241 | if (!reginfo && e < s) |
6eb5f6b9 | 1242 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1243 | |
60a8b682 JH |
1244 | /* The idea in the EXACTF* cases is to first find the |
1245 | * first character of the EXACTF* node and then, if | |
1246 | * necessary, case-insensitively compare the full | |
1247 | * text of the node. The c1 and c2 are the first | |
1248 | * characters (though in Unicode it gets a bit | |
1249 | * more complicated because there are more cases | |
7f16dd3d JH |
1250 | * than just upper and lower: one needs to use |
1251 | * the so-called folding case for case-insensitive | |
1252 | * matching (called "loose matching" in Unicode). | |
1253 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1254 | |
1aa99e6b | 1255 | if (do_utf8) { |
575cac57 | 1256 | UV c, f; |
89ebb4a3 | 1257 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
575cac57 | 1258 | STRLEN len, foldlen; |
4ad0818d | 1259 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1260 | if (c1 == c2) { |
5dab1207 NIS |
1261 | /* Upper and lower of 1st char are equal - |
1262 | * probably not a "letter". */ | |
1aa99e6b | 1263 | while (s <= e) { |
89ebb4a3 | 1264 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1265 | uniflags); |
4cadc6a9 | 1266 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1267 | } |
09091399 JH |
1268 | } |
1269 | else { | |
1aa99e6b | 1270 | while (s <= e) { |
89ebb4a3 | 1271 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, |
041457d9 | 1272 | uniflags); |
80aecb99 | 1273 | |
60a8b682 | 1274 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1275 | * Note that not all the possible combinations |
1276 | * are handled here: some of them are handled | |
1277 | * by the standard folding rules, and some of | |
1278 | * them (the character class or ANYOF cases) | |
1279 | * are handled during compiletime in | |
1280 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1281 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1282 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1283 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1284 | |
4cadc6a9 | 1285 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1286 | } |
09091399 | 1287 | } |
1aa99e6b IH |
1288 | } |
1289 | else { | |
1290 | if (c1 == c2) | |
4cadc6a9 | 1291 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1292 | else |
4cadc6a9 | 1293 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1294 | } |
1295 | break; | |
bbce6d69 | 1296 | case BOUNDL: |
3280af22 | 1297 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1298 | /* FALL THROUGH */ |
a0d0e21e | 1299 | case BOUND: |
ffc61ed2 | 1300 | if (do_utf8) { |
12d33761 | 1301 | if (s == PL_bostr) |
ffc61ed2 JH |
1302 | tmp = '\n'; |
1303 | else { | |
6136c704 | 1304 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1305 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1306 | } |
1307 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1308 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1309 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1310 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1311 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1312 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1313 | isALNUM_LC_utf8((U8*)s))) |
1314 | { | |
1315 | tmp = !tmp; | |
4cadc6a9 | 1316 | REXEC_FBC_TRYIT; |
a687059c | 1317 | } |
4cadc6a9 | 1318 | ); |
a0d0e21e | 1319 | } |
667bb95a | 1320 | else { |
12d33761 | 1321 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1322 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1323 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1324 | if (tmp == |
1325 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1326 | tmp = !tmp; | |
4cadc6a9 | 1327 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1328 | } |
4cadc6a9 | 1329 | ); |
a0ed51b3 | 1330 | } |
24b23f37 | 1331 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1332 | goto got_it; |
1333 | break; | |
bbce6d69 | 1334 | case NBOUNDL: |
3280af22 | 1335 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1336 | /* FALL THROUGH */ |
a0d0e21e | 1337 | case NBOUND: |
ffc61ed2 | 1338 | if (do_utf8) { |
12d33761 | 1339 | if (s == PL_bostr) |
ffc61ed2 JH |
1340 | tmp = '\n'; |
1341 | else { | |
6136c704 | 1342 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1343 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1344 | } |
1345 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1346 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1347 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1348 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1349 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1350 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1351 | isALNUM_LC_utf8((U8*)s))) |
1352 | tmp = !tmp; | |
4cadc6a9 YO |
1353 | else REXEC_FBC_TRYIT; |
1354 | ); | |
a0d0e21e | 1355 | } |
667bb95a | 1356 | else { |
12d33761 | 1357 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1358 | tmp = ((OP(c) == NBOUND ? |
1359 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1360 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1361 | if (tmp == |
1362 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1363 | tmp = !tmp; | |
4cadc6a9 YO |
1364 | else REXEC_FBC_TRYIT; |
1365 | ); | |
a0ed51b3 | 1366 | } |
24b23f37 | 1367 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1368 | goto got_it; |
1369 | break; | |
a0d0e21e | 1370 | case ALNUM: |
4cadc6a9 YO |
1371 | REXEC_FBC_CSCAN_PRELOAD( |
1372 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1373 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1374 | isALNUM(*s) | |
1375 | ); | |
bbce6d69 | 1376 | case ALNUML: |
4cadc6a9 YO |
1377 | REXEC_FBC_CSCAN_TAINT( |
1378 | isALNUM_LC_utf8((U8*)s), | |
1379 | isALNUM_LC(*s) | |
1380 | ); | |
a0d0e21e | 1381 | case NALNUM: |
4cadc6a9 YO |
1382 | REXEC_FBC_CSCAN_PRELOAD( |
1383 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1384 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1385 | !isALNUM(*s) | |
1386 | ); | |
bbce6d69 | 1387 | case NALNUML: |
4cadc6a9 YO |
1388 | REXEC_FBC_CSCAN_TAINT( |
1389 | !isALNUM_LC_utf8((U8*)s), | |
1390 | !isALNUM_LC(*s) | |
1391 | ); | |
a0d0e21e | 1392 | case SPACE: |
4cadc6a9 YO |
1393 | REXEC_FBC_CSCAN_PRELOAD( |
1394 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1395 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1396 | isSPACE(*s) | |
1397 | ); | |
bbce6d69 | 1398 | case SPACEL: |
4cadc6a9 YO |
1399 | REXEC_FBC_CSCAN_TAINT( |
1400 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1401 | isSPACE_LC(*s) | |
1402 | ); | |
a0d0e21e | 1403 | case NSPACE: |
4cadc6a9 YO |
1404 | REXEC_FBC_CSCAN_PRELOAD( |
1405 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1406 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1407 | !isSPACE(*s) | |
1408 | ); | |
bbce6d69 | 1409 | case NSPACEL: |
4cadc6a9 YO |
1410 | REXEC_FBC_CSCAN_TAINT( |
1411 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1412 | !isSPACE_LC(*s) | |
1413 | ); | |
a0d0e21e | 1414 | case DIGIT: |
4cadc6a9 YO |
1415 | REXEC_FBC_CSCAN_PRELOAD( |
1416 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1417 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1418 | isDIGIT(*s) | |
1419 | ); | |
b8c5462f | 1420 | case DIGITL: |
4cadc6a9 YO |
1421 | REXEC_FBC_CSCAN_TAINT( |
1422 | isDIGIT_LC_utf8((U8*)s), | |
1423 | isDIGIT_LC(*s) | |
1424 | ); | |
a0d0e21e | 1425 | case NDIGIT: |
4cadc6a9 YO |
1426 | REXEC_FBC_CSCAN_PRELOAD( |
1427 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1428 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1429 | !isDIGIT(*s) | |
1430 | ); | |
b8c5462f | 1431 | case NDIGITL: |
4cadc6a9 YO |
1432 | REXEC_FBC_CSCAN_TAINT( |
1433 | !isDIGIT_LC_utf8((U8*)s), | |
1434 | !isDIGIT_LC(*s) | |
1435 | ); | |
e1d1eefb YO |
1436 | case LNBREAK: |
1437 | REXEC_FBC_CSCAN( | |
1438 | is_LNBREAK_utf8(s), | |
1439 | is_LNBREAK_latin1(s) | |
1440 | ); | |
1441 | case VERTWS: | |
1442 | REXEC_FBC_CSCAN( | |
1443 | is_VERTWS_utf8(s), | |
1444 | is_VERTWS_latin1(s) | |
1445 | ); | |
1446 | case NVERTWS: | |
1447 | REXEC_FBC_CSCAN( | |
1448 | !is_VERTWS_utf8(s), | |
1449 | !is_VERTWS_latin1(s) | |
1450 | ); | |
1451 | case HORIZWS: | |
1452 | REXEC_FBC_CSCAN( | |
1453 | is_HORIZWS_utf8(s), | |
1454 | is_HORIZWS_latin1(s) | |
1455 | ); | |
1456 | case NHORIZWS: | |
1457 | REXEC_FBC_CSCAN( | |
1458 | !is_HORIZWS_utf8(s), | |
1459 | !is_HORIZWS_latin1(s) | |
1460 | ); | |
1de06328 YO |
1461 | case AHOCORASICKC: |
1462 | case AHOCORASICK: | |
07be1b83 YO |
1463 | { |
1464 | const enum { trie_plain, trie_utf8, trie_utf8_fold } | |
1465 | trie_type = do_utf8 ? | |
1466 | (c->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
1467 | : trie_plain; | |
1468 | /* what trie are we using right now */ | |
1469 | reg_ac_data *aho | |
f8fc2ecf | 1470 | = (reg_ac_data*)progi->data->data[ ARG( c ) ]; |
3251b653 NC |
1471 | reg_trie_data *trie |
1472 | = (reg_trie_data*)progi->data->data[ aho->trie ]; | |
55eed653 | 1473 | HV *widecharmap = (HV*) progi->data->data[ aho->trie + 1 ]; |
07be1b83 YO |
1474 | |
1475 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1476 | #ifdef DEBUGGING |
07be1b83 | 1477 | const char *real_start = s; |
6148ee25 | 1478 | #endif |
07be1b83 | 1479 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1480 | SV *sv_points; |
1481 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1482 | when reading a given char. For ASCII this |
be8e71aa | 1483 | is unnecessary overhead as the relationship |
38a44b82 NC |
1484 | is always 1:1, but for Unicode, especially |
1485 | case folded Unicode this is not true. */ | |
f9e705e8 | 1486 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1487 | U8 *bitmap=NULL; |
1488 | ||
07be1b83 YO |
1489 | |
1490 | GET_RE_DEBUG_FLAGS_DECL; | |
1491 | ||
be8e71aa YO |
1492 | /* We can't just allocate points here. We need to wrap it in |
1493 | * an SV so it gets freed properly if there is a croak while | |
1494 | * running the match */ | |
1495 | ENTER; | |
1496 | SAVETMPS; | |
1497 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1498 | SvCUR_set(sv_points, | |
1499 | maxlen * sizeof(U8 *)); | |
1500 | SvPOK_on(sv_points); | |
1501 | sv_2mortal(sv_points); | |
1502 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1503 | if ( trie_type != trie_utf8_fold |
1504 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1505 | { | |
786e8c11 YO |
1506 | if (trie->bitmap) |
1507 | bitmap=(U8*)trie->bitmap; | |
1508 | else | |
1509 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1510 | } |
786e8c11 YO |
1511 | /* this is the Aho-Corasick algorithm modified a touch |
1512 | to include special handling for long "unknown char" | |
1513 | sequences. The basic idea being that we use AC as long | |
1514 | as we are dealing with a possible matching char, when | |
1515 | we encounter an unknown char (and we have not encountered | |
1516 | an accepting state) we scan forward until we find a legal | |
1517 | starting char. | |
1518 | AC matching is basically that of trie matching, except | |
1519 | that when we encounter a failing transition, we fall back | |
1520 | to the current states "fail state", and try the current char | |
1521 | again, a process we repeat until we reach the root state, | |
1522 | state 1, or a legal transition. If we fail on the root state | |
1523 | then we can either terminate if we have reached an accepting | |
1524 | state previously, or restart the entire process from the beginning | |
1525 | if we have not. | |
1526 | ||
1527 | */ | |
07be1b83 YO |
1528 | while (s <= last_start) { |
1529 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1530 | U8 *uc = (U8*)s; | |
1531 | U16 charid = 0; | |
1532 | U32 base = 1; | |
1533 | U32 state = 1; | |
1534 | UV uvc = 0; | |
1535 | STRLEN len = 0; | |
1536 | STRLEN foldlen = 0; | |
1537 | U8 *uscan = (U8*)NULL; | |
1538 | U8 *leftmost = NULL; | |
786e8c11 YO |
1539 | #ifdef DEBUGGING |
1540 | U32 accepted_word= 0; | |
1541 | #endif | |
07be1b83 YO |
1542 | U32 pointpos = 0; |
1543 | ||
1544 | while ( state && uc <= (U8*)strend ) { | |
1545 | int failed=0; | |
786e8c11 YO |
1546 | U32 word = aho->states[ state ].wordnum; |
1547 | ||
1de06328 YO |
1548 | if( state==1 ) { |
1549 | if ( bitmap ) { | |
1550 | DEBUG_TRIE_EXECUTE_r( | |
1551 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1552 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1553 | (char *)uc, do_utf8 ); | |
1554 | PerlIO_printf( Perl_debug_log, | |
1555 | " Scanning for legal start char...\n"); | |
1556 | } | |
1557 | ); | |
1558 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1559 | uc++; | |
786e8c11 | 1560 | } |
1de06328 | 1561 | s= (char *)uc; |
786e8c11 | 1562 | } |
786e8c11 YO |
1563 | if (uc >(U8*)last_start) break; |
1564 | } | |
1565 | ||
1566 | if ( word ) { | |
1567 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1568 | if (!leftmost || lpos < leftmost) { | |
1569 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1570 | leftmost= lpos; |
786e8c11 | 1571 | } |
07be1b83 | 1572 | if (base==0) break; |
786e8c11 | 1573 | |
07be1b83 YO |
1574 | } |
1575 | points[pointpos++ % maxlen]= uc; | |
55eed653 NC |
1576 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
1577 | uscan, len, uvc, charid, foldlen, | |
1578 | foldbuf, uniflags); | |
786e8c11 YO |
1579 | DEBUG_TRIE_EXECUTE_r({ |
1580 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1581 | s, do_utf8 ); | |
07be1b83 | 1582 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1583 | " Charid:%3u CP:%4"UVxf" ", |
1584 | charid, uvc); | |
1585 | }); | |
07be1b83 YO |
1586 | |
1587 | do { | |
6148ee25 | 1588 | #ifdef DEBUGGING |
786e8c11 | 1589 | word = aho->states[ state ].wordnum; |
6148ee25 | 1590 | #endif |
07be1b83 YO |
1591 | base = aho->states[ state ].trans.base; |
1592 | ||
786e8c11 YO |
1593 | DEBUG_TRIE_EXECUTE_r({ |
1594 | if (failed) | |
1595 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1596 | s, do_utf8 ); | |
07be1b83 | 1597 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1598 | "%sState: %4"UVxf", word=%"UVxf, |
1599 | failed ? " Fail transition to " : "", | |
1600 | (UV)state, (UV)word); | |
1601 | }); | |
07be1b83 YO |
1602 | if ( base ) { |
1603 | U32 tmp; | |
1604 | if (charid && | |
1605 | (base + charid > trie->uniquecharcount ) | |
1606 | && (base + charid - 1 - trie->uniquecharcount | |
1607 | < trie->lasttrans) | |
1608 | && trie->trans[base + charid - 1 - | |
1609 | trie->uniquecharcount].check == state | |
1610 | && (tmp=trie->trans[base + charid - 1 - | |
1611 | trie->uniquecharcount ].next)) | |
1612 | { | |
786e8c11 YO |
1613 | DEBUG_TRIE_EXECUTE_r( |
1614 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1615 | state = tmp; |
1616 | break; | |
1617 | } | |
1618 | else { | |
786e8c11 YO |
1619 | DEBUG_TRIE_EXECUTE_r( |
1620 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1621 | failed = 1; | |
1622 | state = aho->fail[state]; | |
07be1b83 YO |
1623 | } |
1624 | } | |
1625 | else { | |
1626 | /* we must be accepting here */ | |
786e8c11 YO |
1627 | DEBUG_TRIE_EXECUTE_r( |
1628 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1629 | failed = 1; | |
07be1b83 YO |
1630 | break; |
1631 | } | |
1632 | } while(state); | |
786e8c11 | 1633 | uc += len; |
07be1b83 YO |
1634 | if (failed) { |
1635 | if (leftmost) | |
1636 | break; | |
786e8c11 | 1637 | if (!state) state = 1; |
07be1b83 YO |
1638 | } |
1639 | } | |
1640 | if ( aho->states[ state ].wordnum ) { | |
1641 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1642 | if (!leftmost || lpos < leftmost) { |
1643 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1644 | leftmost = lpos; |
786e8c11 | 1645 | } |
07be1b83 | 1646 | } |
07be1b83 YO |
1647 | if (leftmost) { |
1648 | s = (char*)leftmost; | |
786e8c11 YO |
1649 | DEBUG_TRIE_EXECUTE_r({ |
1650 | PerlIO_printf( | |
70685ca0 JH |
1651 | Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n", |
1652 | (UV)accepted_word, (IV)(s - real_start) | |
786e8c11 YO |
1653 | ); |
1654 | }); | |
24b23f37 | 1655 | if (!reginfo || regtry(reginfo, &s)) { |
be8e71aa YO |
1656 | FREETMPS; |
1657 | LEAVE; | |
07be1b83 | 1658 | goto got_it; |
be8e71aa | 1659 | } |
07be1b83 | 1660 | s = HOPc(s,1); |
786e8c11 YO |
1661 | DEBUG_TRIE_EXECUTE_r({ |
1662 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1663 | }); | |
07be1b83 | 1664 | } else { |
786e8c11 YO |
1665 | DEBUG_TRIE_EXECUTE_r( |
1666 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1667 | break; |
1668 | } | |
1669 | } | |
be8e71aa YO |
1670 | FREETMPS; |
1671 | LEAVE; | |
07be1b83 YO |
1672 | } |
1673 | break; | |
b3c9acc1 | 1674 | default: |
3c3eec57 GS |
1675 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1676 | break; | |
d6a28714 | 1677 | } |
6eb5f6b9 JH |
1678 | return 0; |
1679 | got_it: | |
1680 | return s; | |
1681 | } | |
1682 | ||
e6aad7ab | 1683 | static void |
fae667d5 | 1684 | S_swap_match_buff (pTHX_ regexp *prog) { |
f0ab9afb | 1685 | regexp_paren_pair *t; |
fae667d5 | 1686 | |
28d8d7f4 | 1687 | if (!prog->swap) { |
fae667d5 YO |
1688 | /* We have to be careful. If the previous successful match |
1689 | was from this regex we don't want a subsequent paritally | |
1690 | successful match to clobber the old results. | |
1691 | So when we detect this possibility we add a swap buffer | |
1692 | to the re, and switch the buffer each match. If we fail | |
1693 | we switch it back, otherwise we leave it swapped. | |
1694 | */ | |
f0ab9afb | 1695 | Newxz(prog->swap, (prog->nparens + 1), regexp_paren_pair); |
fae667d5 | 1696 | } |
f0ab9afb NC |
1697 | t = prog->swap; |
1698 | prog->swap = prog->offs; | |
1699 | prog->offs = t; | |
fae667d5 YO |
1700 | } |
1701 | ||
1702 | ||
6eb5f6b9 JH |
1703 | /* |
1704 | - regexec_flags - match a regexp against a string | |
1705 | */ | |
1706 | I32 | |
49d7dfbc | 1707 | Perl_regexec_flags(pTHX_ REGEXP * const prog, char *stringarg, register char *strend, |
6eb5f6b9 JH |
1708 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) |
1709 | /* strend: pointer to null at end of string */ | |
1710 | /* strbeg: real beginning of string */ | |
1711 | /* minend: end of match must be >=minend after stringarg. */ | |
58e23c8d YO |
1712 | /* data: May be used for some additional optimizations. |
1713 | Currently its only used, with a U32 cast, for transmitting | |
1714 | the ganch offset when doing a /g match. This will change */ | |
6eb5f6b9 JH |
1715 | /* nosave: For optimizations. */ |
1716 | { | |
97aff369 | 1717 | dVAR; |
24b23f37 | 1718 | /*register*/ char *s; |
6eb5f6b9 | 1719 | register regnode *c; |
24b23f37 | 1720 | /*register*/ char *startpos = stringarg; |
6eb5f6b9 JH |
1721 | I32 minlen; /* must match at least this many chars */ |
1722 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1723 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1724 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1725 | char *scream_olds = NULL; |
f9f4320a | 1726 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1727 | I32 multiline; |
f8fc2ecf | 1728 | RXi_GET_DECL(prog,progi); |
3b0527fe | 1729 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
fae667d5 | 1730 | bool swap_on_fail = 0; |
a3621e74 YO |
1731 | |
1732 | GET_RE_DEBUG_FLAGS_DECL; | |
1733 | ||
9d4ba2ae | 1734 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1735 | |
1736 | /* Be paranoid... */ | |
1737 | if (prog == NULL || startpos == NULL) { | |
1738 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1739 | return 0; | |
1740 | } | |
1741 | ||
bbe252da | 1742 | multiline = prog->extflags & RXf_PMf_MULTILINE; |
3b0527fe | 1743 | reginfo.prog = prog; |
2757e526 | 1744 | |
bac06658 | 1745 | RX_MATCH_UTF8_set(prog, do_utf8); |
1de06328 YO |
1746 | DEBUG_EXECUTE_r( |
1747 | debug_start_match(prog, do_utf8, startpos, strend, | |
1748 | "Matching"); | |
1749 | ); | |
bac06658 | 1750 | |
6eb5f6b9 | 1751 | minlen = prog->minlen; |
1de06328 YO |
1752 | |
1753 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1754 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1755 | "String too short [regexec_flags]...\n")); |
1756 | goto phooey; | |
1aa99e6b | 1757 | } |
6eb5f6b9 | 1758 | |
1de06328 | 1759 | |
6eb5f6b9 | 1760 | /* Check validity of program. */ |
f8fc2ecf | 1761 | if (UCHARAT(progi->program) != REG_MAGIC) { |
6eb5f6b9 JH |
1762 | Perl_croak(aTHX_ "corrupted regexp program"); |
1763 | } | |
1764 | ||
1765 | PL_reg_flags = 0; | |
1766 | PL_reg_eval_set = 0; | |
1767 | PL_reg_maxiter = 0; | |
1768 | ||
bbe252da | 1769 | if (prog->extflags & RXf_UTF8) |
6eb5f6b9 JH |
1770 | PL_reg_flags |= RF_utf8; |
1771 | ||
1772 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1773 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1774 | PL_bostr = strbeg; |
3b0527fe | 1775 | reginfo.sv = sv; |
6eb5f6b9 JH |
1776 | |
1777 | /* Mark end of line for $ (and such) */ | |
1778 | PL_regeol = strend; | |
1779 | ||
1780 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1781 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1782 | |
6eb5f6b9 JH |
1783 | /* If there is a "must appear" string, look for it. */ |
1784 | s = startpos; | |
1785 | ||
bbe252da | 1786 | if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 JH |
1787 | MAGIC *mg; |
1788 | ||
1789 | if (flags & REXEC_IGNOREPOS) /* Means: check only at start */ | |
58e23c8d | 1790 | reginfo.ganch = startpos + prog->gofs; |
6eb5f6b9 JH |
1791 | else if (sv && SvTYPE(sv) >= SVt_PVMG |
1792 | && SvMAGIC(sv) | |
14befaf4 DM |
1793 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1794 | && mg->mg_len >= 0) { | |
3b0527fe | 1795 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
bbe252da | 1796 | if (prog->extflags & RXf_ANCH_GPOS) { |
3b0527fe | 1797 | if (s > reginfo.ganch) |
6eb5f6b9 | 1798 | goto phooey; |
58e23c8d | 1799 | s = reginfo.ganch - prog->gofs; |
6eb5f6b9 JH |
1800 | } |
1801 | } | |
58e23c8d | 1802 | else if (data) { |
70685ca0 | 1803 | reginfo.ganch = strbeg + PTR2UV(data); |
58e23c8d | 1804 | } else /* pos() not defined */ |
3b0527fe | 1805 | reginfo.ganch = strbeg; |
6eb5f6b9 | 1806 | } |
c74340f9 | 1807 | if (PL_curpm && (PM_GETRE(PL_curpm) == prog)) { |
fae667d5 YO |
1808 | swap_on_fail = 1; |
1809 | swap_match_buff(prog); /* do we need a save destructor here for | |
1810 | eval dies? */ | |
c74340f9 | 1811 | } |
a0714e2c | 1812 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1813 | re_scream_pos_data d; |
1814 | ||
1815 | d.scream_olds = &scream_olds; | |
1816 | d.scream_pos = &scream_pos; | |
1817 | s = re_intuit_start(prog, sv, s, strend, flags, &d); | |
3fa9c3d7 | 1818 | if (!s) { |
a3621e74 | 1819 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1820 | goto phooey; /* not present */ |
3fa9c3d7 | 1821 | } |
6eb5f6b9 JH |
1822 | } |
1823 | ||
1de06328 | 1824 | |
6eb5f6b9 JH |
1825 | |
1826 | /* Simplest case: anchored match need be tried only once. */ | |
1827 | /* [unless only anchor is BOL and multiline is set] */ | |
bbe252da | 1828 | if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) { |
24b23f37 | 1829 | if (s == startpos && regtry(®info, &startpos)) |
6eb5f6b9 | 1830 | goto got_it; |
bbe252da YO |
1831 | else if (multiline || (prog->intflags & PREGf_IMPLICIT) |
1832 | || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */ | |
6eb5f6b9 JH |
1833 | { |
1834 | char *end; | |
1835 | ||
1836 | if (minlen) | |
1837 | dontbother = minlen - 1; | |
1aa99e6b | 1838 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1839 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1840 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1841 | if (s == startpos) |
1842 | goto after_try; | |
1843 | while (1) { | |
24b23f37 | 1844 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1845 | goto got_it; |
1846 | after_try: | |
5339e136 | 1847 | if (s > end) |
6eb5f6b9 | 1848 | goto phooey; |
bbe252da | 1849 | if (prog->extflags & RXf_USE_INTUIT) { |
6eb5f6b9 JH |
1850 | s = re_intuit_start(prog, sv, s + 1, strend, flags, NULL); |
1851 | if (!s) | |
1852 | goto phooey; | |
1853 | } | |
1854 | else | |
1855 | s++; | |
1856 | } | |
1857 | } else { | |
1858 | if (s > startpos) | |
1859 | s--; | |
1860 | while (s < end) { | |
1861 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
24b23f37 | 1862 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1863 | goto got_it; |
1864 | } | |
1865 | } | |
1866 | } | |
1867 | } | |
1868 | goto phooey; | |
bbe252da | 1869 | } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK)) |
f9f4320a YO |
1870 | { |
1871 | /* the warning about reginfo.ganch being used without intialization | |
bbe252da | 1872 | is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN |
f9f4320a | 1873 | and we only enter this block when the same bit is set. */ |
58e23c8d YO |
1874 | char *tmp_s = reginfo.ganch - prog->gofs; |
1875 | if (regtry(®info, &tmp_s)) | |
6eb5f6b9 JH |
1876 | goto got_it; |
1877 | goto phooey; | |
1878 | } | |
1879 | ||
1880 | /* Messy cases: unanchored match. */ | |
bbe252da | 1881 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) { |
6eb5f6b9 JH |
1882 | /* we have /x+whatever/ */ |
1883 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1884 | char ch; |
bf93d4cc GS |
1885 | #ifdef DEBUGGING |
1886 | int did_match = 0; | |
1887 | #endif | |
33b8afdf JH |
1888 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1889 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1890 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1891 | |
1aa99e6b | 1892 | if (do_utf8) { |
4cadc6a9 | 1893 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1894 | if (*s == ch) { |
a3621e74 | 1895 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1896 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1897 | s += UTF8SKIP(s); |
1898 | while (s < strend && *s == ch) | |
1899 | s += UTF8SKIP(s); | |
1900 | } | |
4cadc6a9 | 1901 | ); |
6eb5f6b9 JH |
1902 | } |
1903 | else { | |
4cadc6a9 | 1904 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1905 | if (*s == ch) { |
a3621e74 | 1906 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1907 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1908 | s++; |
1909 | while (s < strend && *s == ch) | |
1910 | s++; | |
1911 | } | |
4cadc6a9 | 1912 | ); |
6eb5f6b9 | 1913 | } |
a3621e74 | 1914 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1915 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1916 | "Did not find anchored character...\n") |
1917 | ); | |
6eb5f6b9 | 1918 | } |
a0714e2c SS |
1919 | else if (prog->anchored_substr != NULL |
1920 | || prog->anchored_utf8 != NULL | |
1921 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1922 | && prog->float_max_offset < strend - s)) { |
1923 | SV *must; | |
1924 | I32 back_max; | |
1925 | I32 back_min; | |
1926 | char *last; | |
6eb5f6b9 | 1927 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1928 | #ifdef DEBUGGING |
1929 | int did_match = 0; | |
1930 | #endif | |
33b8afdf JH |
1931 | if (prog->anchored_substr || prog->anchored_utf8) { |
1932 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1933 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1934 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1935 | back_max = back_min = prog->anchored_offset; | |
1936 | } else { | |
1937 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1938 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1939 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1940 | back_max = prog->float_max_offset; | |
1941 | back_min = prog->float_min_offset; | |
1942 | } | |
1de06328 YO |
1943 | |
1944 | ||
33b8afdf JH |
1945 | if (must == &PL_sv_undef) |
1946 | /* could not downgrade utf8 check substring, so must fail */ | |
1947 | goto phooey; | |
1948 | ||
1de06328 YO |
1949 | if (back_min<0) { |
1950 | last = strend; | |
1951 | } else { | |
1952 | last = HOP3c(strend, /* Cannot start after this */ | |
1953 | -(I32)(CHR_SVLEN(must) | |
1954 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
1955 | } | |
6eb5f6b9 JH |
1956 | if (s > PL_bostr) |
1957 | last1 = HOPc(s, -1); | |
1958 | else | |
1959 | last1 = s - 1; /* bogus */ | |
1960 | ||
a0288114 | 1961 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
1962 | check_substr==must. */ |
1963 | scream_pos = -1; | |
1964 | dontbother = end_shift; | |
1965 | strend = HOPc(strend, -dontbother); | |
1966 | while ( (s <= last) && | |
9041c2e3 | 1967 | ((flags & REXEC_SCREAM) |
1de06328 | 1968 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 1969 | end_shift, &scream_pos, 0)) |
1de06328 | 1970 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 1971 | (unsigned char*)strend, must, |
7fba1cd6 | 1972 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b HS |
1973 | /* we may be pointing at the wrong string */ |
1974 | if ((flags & REXEC_SCREAM) && RX_MATCH_COPIED(prog)) | |
3f7c398e | 1975 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 1976 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
1977 | if (HOPc(s, -back_max) > last1) { |
1978 | last1 = HOPc(s, -back_min); | |
1979 | s = HOPc(s, -back_max); | |
1980 | } | |
1981 | else { | |
52657f30 | 1982 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
1983 | |
1984 | last1 = HOPc(s, -back_min); | |
52657f30 | 1985 | s = t; |
6eb5f6b9 | 1986 | } |
1aa99e6b | 1987 | if (do_utf8) { |
6eb5f6b9 | 1988 | while (s <= last1) { |
24b23f37 | 1989 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1990 | goto got_it; |
1991 | s += UTF8SKIP(s); | |
1992 | } | |
1993 | } | |
1994 | else { | |
1995 | while (s <= last1) { | |
24b23f37 | 1996 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1997 | goto got_it; |
1998 | s++; | |
1999 | } | |
2000 | } | |
2001 | } | |
ab3bbdeb YO |
2002 | DEBUG_EXECUTE_r(if (!did_match) { |
2003 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
2004 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
2005 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 2006 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 2007 | ? "anchored" : "floating"), |
ab3bbdeb YO |
2008 | quoted, RE_SV_TAIL(must)); |
2009 | }); | |
6eb5f6b9 JH |
2010 | goto phooey; |
2011 | } | |
f8fc2ecf | 2012 | else if ( (c = progi->regstclass) ) { |
f14c76ed | 2013 | if (minlen) { |
f8fc2ecf | 2014 | const OPCODE op = OP(progi->regstclass); |
66e933ab | 2015 | /* don't bother with what can't match */ |
786e8c11 | 2016 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
2017 | strend = HOPc(strend, -(minlen - 1)); |
2018 | } | |
a3621e74 | 2019 | DEBUG_EXECUTE_r({ |
be8e71aa | 2020 | SV * const prop = sv_newmortal(); |
32fc9b6a | 2021 | regprop(prog, prop, c); |
0df25f3d | 2022 | { |
02daf0ab | 2023 | RE_PV_QUOTED_DECL(quoted,do_utf8,PERL_DEBUG_PAD_ZERO(1), |
ab3bbdeb | 2024 | s,strend-s,60); |
0df25f3d | 2025 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2026 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 2027 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 2028 | quoted, (int)(strend - s)); |
0df25f3d | 2029 | } |
ffc61ed2 | 2030 | }); |
3b0527fe | 2031 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 2032 | goto got_it; |
07be1b83 | 2033 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
2034 | } |
2035 | else { | |
2036 | dontbother = 0; | |
a0714e2c | 2037 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 2038 | /* Trim the end. */ |
d6a28714 | 2039 | char *last; |
33b8afdf JH |
2040 | SV* float_real; |
2041 | ||
2042 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
2043 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
2044 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
2045 | |
2046 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 2047 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
2048 | end_shift, &scream_pos, 1); /* last one */ |
2049 | if (!last) | |
ffc61ed2 | 2050 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b HS |
2051 | /* we may be pointing at the wrong string */ |
2052 | else if (RX_MATCH_COPIED(prog)) | |
3f7c398e | 2053 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 2054 | } |
d6a28714 JH |
2055 | else { |
2056 | STRLEN len; | |
cfd0369c | 2057 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 2058 | |
33b8afdf | 2059 | if (SvTAIL(float_real)) { |
d6a28714 JH |
2060 | if (memEQ(strend - len + 1, little, len - 1)) |
2061 | last = strend - len + 1; | |
7fba1cd6 | 2062 | else if (!multiline) |
9041c2e3 | 2063 | last = memEQ(strend - len, little, len) |
bd61b366 | 2064 | ? strend - len : NULL; |
b8c5462f | 2065 | else |
d6a28714 JH |
2066 | goto find_last; |
2067 | } else { | |
2068 | find_last: | |
9041c2e3 | 2069 | if (len) |
d6a28714 | 2070 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 2071 | else |
a0288114 | 2072 | last = strend; /* matching "$" */ |
b8c5462f | 2073 | } |
b8c5462f | 2074 | } |
bf93d4cc | 2075 | if (last == NULL) { |
6bda09f9 YO |
2076 | DEBUG_EXECUTE_r( |
2077 | PerlIO_printf(Perl_debug_log, | |
2078 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
2079 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
2080 | goto phooey; /* Should not happen! */ |
2081 | } | |
d6a28714 JH |
2082 | dontbother = strend - last + prog->float_min_offset; |
2083 | } | |
2084 | if (minlen && (dontbother < minlen)) | |
2085 | dontbother = minlen - 1; | |
2086 | strend -= dontbother; /* this one's always in bytes! */ | |
2087 | /* We don't know much -- general case. */ | |
1aa99e6b | 2088 | if (do_utf8) { |
d6a28714 | 2089 | for (;;) { |
24b23f37 | 2090 | if (regtry(®info, &s)) |
d6a28714 JH |
2091 | goto got_it; |
2092 | if (s >= strend) | |
2093 | break; | |
b8c5462f | 2094 | s += UTF8SKIP(s); |
d6a28714 JH |
2095 | }; |
2096 | } | |
2097 | else { | |
2098 | do { | |
24b23f37 | 2099 | if (regtry(®info, &s)) |
d6a28714 JH |
2100 | goto got_it; |
2101 | } while (s++ < strend); | |
2102 | } | |
2103 | } | |
2104 | ||
2105 | /* Failure. */ | |
2106 | goto phooey; | |
2107 | ||
2108 | got_it: | |
2109 | RX_MATCH_TAINTED_set(prog, PL_reg_flags & RF_tainted); | |
2110 | ||
19b95bf0 | 2111 | if (PL_reg_eval_set) |
4f639d21 | 2112 | restore_pos(aTHX_ prog); |
81714fb9 YO |
2113 | if (prog->paren_names) |
2114 | (void)hv_iterinit(prog->paren_names); | |
d6a28714 JH |
2115 | |
2116 | /* make sure $`, $&, $', and $digit will work later */ | |
2117 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
ed252734 | 2118 | RX_MATCH_COPY_FREE(prog); |
d6a28714 | 2119 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2120 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2121 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2122 | if ((SvIsCOW(sv) |
2123 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2124 | if (DEBUG_C_TEST) { | |
2125 | PerlIO_printf(Perl_debug_log, | |
2126 | "Copy on write: regexp capture, type %d\n", | |
2127 | (int) SvTYPE(sv)); | |
2128 | } | |
2129 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2130 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2131 | assert (SvPOKp(prog->saved_copy)); |
2132 | } else | |
2133 | #endif | |
2134 | { | |
2135 | RX_MATCH_COPIED_on(prog); | |
2136 | s = savepvn(strbeg, i); | |
2137 | prog->subbeg = s; | |
2138 | } | |
d6a28714 | 2139 | prog->sublen = i; |
d6a28714 JH |
2140 | } |
2141 | else { | |
2142 | prog->subbeg = strbeg; | |
2143 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2144 | } | |
2145 | } | |
9041c2e3 | 2146 | |
d6a28714 JH |
2147 | return 1; |
2148 | ||
2149 | phooey: | |
a3621e74 | 2150 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2151 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2152 | if (PL_reg_eval_set) |
4f639d21 | 2153 | restore_pos(aTHX_ prog); |
fae667d5 | 2154 | if (swap_on_fail) |
c74340f9 | 2155 | /* we failed :-( roll it back */ |
fae667d5 YO |
2156 | swap_match_buff(prog); |
2157 | ||
d6a28714 JH |
2158 | return 0; |
2159 | } | |
2160 | ||
6bda09f9 | 2161 | |
d6a28714 JH |
2162 | /* |
2163 | - regtry - try match at specific point | |
2164 | */ | |
2165 | STATIC I32 /* 0 failure, 1 success */ | |
24b23f37 | 2166 | S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) |
d6a28714 | 2167 | { |
97aff369 | 2168 | dVAR; |
d6a28714 | 2169 | CHECKPOINT lastcp; |
3b0527fe | 2170 | regexp *prog = reginfo->prog; |
f8fc2ecf | 2171 | RXi_GET_DECL(prog,progi); |
a3621e74 | 2172 | GET_RE_DEBUG_FLAGS_DECL; |
24b23f37 | 2173 | reginfo->cutpoint=NULL; |
d6a28714 | 2174 | |
bbe252da | 2175 | if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) { |
d6a28714 JH |
2176 | MAGIC *mg; |
2177 | ||
2178 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2179 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2180 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2181 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2182 | )); |
ea8d6ae1 | 2183 | SAVESTACK_CXPOS(); |
d6a28714 JH |
2184 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2185 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2186 | SAVETMPS; | |
2187 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2188 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2189 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2190 | ||
3b0527fe | 2191 | if (reginfo->sv) { |
d6a28714 | 2192 | /* Make $_ available to executed code. */ |
3b0527fe | 2193 | if (reginfo->sv != DEFSV) { |
59f00321 | 2194 | SAVE_DEFSV; |
3b0527fe | 2195 | DEFSV = reginfo->sv; |
b8c5462f | 2196 | } |
d6a28714 | 2197 | |
3b0527fe DM |
2198 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2199 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2200 | /* prepare for quick setting of pos */ |
d300d9fa | 2201 | #ifdef PERL_OLD_COPY_ON_WRITE |
51a9ea20 NC |
2202 | if (SvIsCOW(reginfo->sv)) |
2203 | sv_force_normal_flags(reginfo->sv, 0); | |
d300d9fa | 2204 | #endif |
3dab1dad | 2205 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2206 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2207 | mg->mg_len = -1; |
b8c5462f | 2208 | } |
d6a28714 JH |
2209 | PL_reg_magic = mg; |
2210 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2211 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2212 | } |
09687e5a | 2213 | if (!PL_reg_curpm) { |
a02a5408 | 2214 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2215 | #ifdef USE_ITHREADS |
2216 | { | |
be8e71aa | 2217 | SV* const repointer = newSViv(0); |
577e12cc | 2218 | /* so we know which PL_regex_padav element is PL_reg_curpm */ |
35061a7e | 2219 | SvFLAGS(repointer) |= SVf_BREAK; |
09687e5a AB |
2220 | av_push(PL_regex_padav,repointer); |
2221 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); | |
2222 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2223 | } | |
2224 | #endif | |
2225 | } | |
aaa362c4 | 2226 | PM_SETRE(PL_reg_curpm, prog); |
d6a28714 JH |
2227 | PL_reg_oldcurpm = PL_curpm; |
2228 | PL_curpm = PL_reg_curpm; | |
2229 | if (RX_MATCH_COPIED(prog)) { | |
2230 | /* Here is a serious problem: we cannot rewrite subbeg, | |
2231 | since it may be needed if this match fails. Thus | |
2232 | $` inside (?{}) could fail... */ | |
2233 | PL_reg_oldsaved = prog->subbeg; | |
2234 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2235 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2236 | PL_nrs = prog->saved_copy; |
2237 | #endif | |
d6a28714 JH |
2238 | RX_MATCH_COPIED_off(prog); |
2239 | } | |
2240 | else | |
bd61b366 | 2241 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2242 | prog->subbeg = PL_bostr; |
2243 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2244 | } | |
24b23f37 | 2245 | DEBUG_EXECUTE_r(PL_reg_starttry = *startpos); |
f0ab9afb | 2246 | prog->offs[0].start = *startpos - PL_bostr; |
24b23f37 | 2247 | PL_reginput = *startpos; |
d6a28714 | 2248 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2249 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2250 | prog->lastparen = 0; |
03994de8 | 2251 | prog->lastcloseparen = 0; |
d6a28714 | 2252 | PL_regsize = 0; |
f0ab9afb | 2253 | PL_regoffs = prog->offs; |
d6a28714 JH |
2254 | if (PL_reg_start_tmpl <= prog->nparens) { |
2255 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2256 | if(PL_reg_start_tmp) | |
2257 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2258 | else | |
a02a5408 | 2259 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2260 | } |
2261 | ||
2262 | /* XXXX What this code is doing here?!!! There should be no need | |
2263 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2264 | this! --ilya*/ |
dafc8851 JH |
2265 | |
2266 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2267 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 | 2268 | * PL_reglastparen), is not needed at all by the test suite |
225593e1 DM |
2269 | * (op/regexp, op/pat, op/split), but that code is needed otherwise |
2270 | * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ | |
2271 | * Meanwhile, this code *is* needed for the | |
daf18116 JH |
2272 | * above-mentioned test suite tests to succeed. The common theme |
2273 | * on those tests seems to be returning null fields from matches. | |
225593e1 | 2274 | * --jhi updated by dapm */ |
dafc8851 | 2275 | #if 1 |
d6a28714 | 2276 | if (prog->nparens) { |
f0ab9afb | 2277 | regexp_paren_pair *pp = PL_regoffs; |
097eb12c | 2278 | register I32 i; |
eb160463 | 2279 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
f0ab9afb NC |
2280 | ++pp; |
2281 | pp->start = -1; | |
2282 | pp->end = -1; | |
d6a28714 JH |
2283 | } |
2284 | } | |
dafc8851 | 2285 | #endif |
02db2b7b | 2286 | REGCP_SET(lastcp); |
f8fc2ecf | 2287 | if (regmatch(reginfo, progi->program + 1)) { |
f0ab9afb | 2288 | PL_regoffs[0].end = PL_reginput - PL_bostr; |
d6a28714 JH |
2289 | return 1; |
2290 | } | |
24b23f37 YO |
2291 | if (reginfo->cutpoint) |
2292 | *startpos= reginfo->cutpoint; | |
02db2b7b | 2293 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2294 | return 0; |
2295 | } | |
2296 | ||
02db2b7b | 2297 | |
8ba1375e MJD |
2298 | #define sayYES goto yes |
2299 | #define sayNO goto no | |
262b90c4 | 2300 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2301 | |
f9f4320a YO |
2302 | /* we dont use STMT_START/END here because it leads to |
2303 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2304 | #define CACHEsayNO \ | |
c476f425 DM |
2305 | if (ST.cache_mask) \ |
2306 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2307 | sayNO |
3298f257 | 2308 | |
a3621e74 | 2309 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2310 | 'failed...' are printed. It should be set such that messages |
2311 | are inline with the regop output that created them. | |
a3621e74 | 2312 | */ |
265c4333 | 2313 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2314 | |
2315 | ||
2316 | /* Make sure there is a test for this +1 options in re_tests */ | |
2317 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2318 | ||
40a82448 DM |
2319 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2320 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2321 | |
86545054 DM |
2322 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2323 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2324 | ||
5d9a96ca DM |
2325 | /* grab a new slab and return the first slot in it */ |
2326 | ||
2327 | STATIC regmatch_state * | |
2328 | S_push_slab(pTHX) | |
2329 | { | |
a35a87e7 | 2330 | #if PERL_VERSION < 9 && !defined(PERL_CORE) |
54df2634 NC |
2331 | dMY_CXT; |
2332 | #endif | |
5d9a96ca DM |
2333 | regmatch_slab *s = PL_regmatch_slab->next; |
2334 | if (!s) { | |
2335 | Newx(s, 1, regmatch_slab); | |
2336 | s->prev = PL_regmatch_slab; | |
2337 | s->next = NULL; | |
2338 | PL_regmatch_slab->next = s; | |
2339 | } | |
2340 | PL_regmatch_slab = s; | |
86545054 | 2341 | return SLAB_FIRST(s); |
5d9a96ca | 2342 | } |
5b47454d | 2343 | |
95b24440 | 2344 | |
40a82448 DM |
2345 | /* push a new state then goto it */ |
2346 | ||
2347 | #define PUSH_STATE_GOTO(state, node) \ | |
2348 | scan = node; \ | |
2349 | st->resume_state = state; \ | |
2350 | goto push_state; | |
2351 | ||
2352 | /* push a new state with success backtracking, then goto it */ | |
2353 | ||
2354 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2355 | scan = node; \ | |
2356 | st->resume_state = state; \ | |
2357 | goto push_yes_state; | |
2358 | ||
aa283a38 | 2359 | |
aa283a38 | 2360 | |
d6a28714 | 2361 | /* |
95b24440 | 2362 | |
bf1f174e DM |
2363 | regmatch() - main matching routine |
2364 | ||
2365 | This is basically one big switch statement in a loop. We execute an op, | |
2366 | set 'next' to point the next op, and continue. If we come to a point which | |
2367 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2368 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2369 | state, and re-enter the loop at the state indicated. If there are no more | |
2370 | states to pop, we return failure. | |
2371 | ||
2372 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2373 | after successfully matching one A, we need to go back and try to | |
2374 | match another one; similarly for lookahead assertions: if the assertion | |
2375 | completes successfully, we backtrack to the state just before the assertion | |
2376 | and then carry on. In these cases, the pushed state is marked as | |
2377 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2378 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2379 | the nearest yes state, discarding any intermediate failure-only states. | |
2380 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2381 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2382 | it to free the inner regex. | |
2383 | ||
2384 | Note that failure backtracking rewinds the cursor position, while | |
2385 | success backtracking leaves it alone. | |
2386 | ||
2387 | A pattern is complete when the END op is executed, while a subpattern | |
2388 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2389 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2390 | behaviour. | |
2391 | ||
2392 | A common convention in this function is to use A and B to refer to the two | |
2393 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2394 | the subpattern to be matched possibly multiple times, while B is the entire | |
2395 | rest of the pattern. Variable and state names reflect this convention. | |
2396 | ||
2397 | The states in the main switch are the union of ops and failure/success of | |
2398 | substates associated with with that op. For example, IFMATCH is the op | |
2399 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2400 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2401 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2402 | just failed to match A. Resume states always come in pairs. The backtrack | |
2403 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2404 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2405 | on success or failure. | |
2406 | ||
2407 | The struct that holds a backtracking state is actually a big union, with | |
2408 | one variant for each major type of op. The variable st points to the | |
2409 | top-most backtrack struct. To make the code clearer, within each | |
2410 | block of code we #define ST to alias the relevant union. | |
2411 | ||
2412 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2413 | implementation: | |
2414 | ||
2415 | switch (state) { | |
2416 | .... | |
2417 | ||
2418 | #define ST st->u.ifmatch | |
2419 | ||
2420 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2421 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2422 | ... |
bf1f174e DM |
2423 | // push a yes backtrack state with a resume value of |
2424 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2425 | // first node of A: | |
2426 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2427 | // NOTREACHED | |
2428 | ||
2429 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2430 | next = B; | |
2431 | bar = ST.foo; // do something with the preserved value | |
2432 | break; | |
2433 | ||
2434 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2435 | ...; // do some housekeeping, then ... | |
2436 | sayNO; // propagate the failure | |
2437 | ||
2438 | #undef ST | |
95b24440 | 2439 | |
bf1f174e DM |
2440 | ... |
2441 | } | |
95b24440 | 2442 | |
bf1f174e DM |
2443 | For any old-timers reading this who are familiar with the old recursive |
2444 | approach, the code above is equivalent to: | |
95b24440 | 2445 | |
bf1f174e DM |
2446 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2447 | { | |
2448 | int foo = ... | |
95b24440 | 2449 | ... |
bf1f174e DM |
2450 | if (regmatch(A)) { |
2451 | next = B; | |
2452 | bar = foo; | |
2453 | break; | |
95b24440 | 2454 | } |
bf1f174e DM |
2455 | ...; // do some housekeeping, then ... |
2456 | sayNO; // propagate the failure | |
95b24440 | 2457 | } |
bf1f174e DM |
2458 | |
2459 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2460 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2461 | save, then do one of | |
2462 | ||
2463 | PUSH_STATE_GOTO(resume_state, node); | |
2464 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2465 | ||
2466 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2467 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2468 | On backtracking, the free slot is popped, and the saved state becomes the | |
2469 | new free state. An ST.foo field in this new top state can be temporarily | |
2470 | accessed to retrieve values, but once the main loop is re-entered, it | |
2471 | becomes available for reuse. | |
2472 | ||
2473 | Note that the depth of the backtrack stack constantly increases during the | |
2474 | left-to-right execution of the pattern, rather than going up and down with | |
2475 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2476 | end of the pattern, rather than at X in the following: | |
2477 | ||
2478 | /(((X)+)+)+....(Y)+....Z/ | |
2479 | ||
2480 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2481 | (?>A), which pop all the backtrack states associated with A before | |
2482 | continuing. | |
2483 | ||
2484 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2485 | PL_regmatch_state and st always point to the currently active state, | |
2486 | and PL_regmatch_slab points to the slab currently containing | |
2487 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2488 | allocated, and is never freed until interpreter destruction. When the slab | |
2489 | is full, a new one is allocated and chained to the end. At exit from | |
2490 | regmatch(), slabs allocated since entry are freed. | |
2491 | ||
2492 | */ | |
95b24440 | 2493 | |
40a82448 | 2494 | |
5bc10b2c | 2495 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2496 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2497 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2498 | PerlIO_printf(Perl_debug_log, \ | |
5d458dd8 | 2499 | " %*s"pp" %s%s%s%s%s\n", \ |
5bc10b2c | 2500 | depth*2, "", \ |
13d6edb4 | 2501 | PL_reg_name[st->resume_state], \ |
5d458dd8 YO |
2502 | ((st==yes_state||st==mark_state) ? "[" : ""), \ |
2503 | ((st==yes_state) ? "Y" : ""), \ | |
2504 | ((st==mark_state) ? "M" : ""), \ | |
2505 | ((st==yes_state||st==mark_state) ? "]" : "") \ | |
2506 | ); \ | |
265c4333 | 2507 | }); |
5bc10b2c | 2508 | |
40a82448 | 2509 | |
3dab1dad | 2510 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2511 | |
3df15adc | 2512 | #ifdef DEBUGGING |
5bc10b2c | 2513 | |
ab3bbdeb YO |
2514 | STATIC void |
2515 | S_debug_start_match(pTHX_ const regexp *prog, const bool do_utf8, | |
2516 | const char *start, const char *end, const char *blurb) | |
2517 | { | |
bbe252da | 2518 | const bool utf8_pat= prog->extflags & RXf_UTF8 ? 1 : 0; |
ab3bbdeb YO |
2519 | if (!PL_colorset) |
2520 | reginitcolors(); | |
2521 | { | |
2522 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
2523 | prog->precomp, prog->prelen, 60); | |
2524 | ||
2525 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2526 | start, end - start, 60); | |
2527 | ||
2528 | PerlIO_printf(Perl_debug_log, | |
2529 | "%s%s REx%s %s against %s\n", | |
2530 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2531 | ||
2532 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2533 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2534 | utf8_pat ? "pattern" : "", | |
2535 | utf8_pat && do_utf8 ? " and " : "", | |
2536 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2537 | ); |
2538 | } | |
2539 | } | |
3df15adc YO |
2540 | |
2541 | STATIC void | |
786e8c11 YO |
2542 | S_dump_exec_pos(pTHX_ const char *locinput, |
2543 | const regnode *scan, | |
2544 | const char *loc_regeol, | |
2545 | const char *loc_bostr, | |
2546 | const char *loc_reg_starttry, | |
2547 | const bool do_utf8) | |
07be1b83 | 2548 | { |
786e8c11 | 2549 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2550 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2551 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2552 | /* The part of the string before starttry has one color |
2553 | (pref0_len chars), between starttry and current | |
2554 | position another one (pref_len - pref0_len chars), | |
2555 | after the current position the third one. | |
2556 | We assume that pref0_len <= pref_len, otherwise we | |
2557 | decrease pref0_len. */ | |
786e8c11 YO |
2558 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2559 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2560 | int pref0_len; |
2561 | ||
2562 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) | |
2563 | pref_len++; | |
786e8c11 YO |
2564 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2565 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2566 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2567 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2568 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2569 | l--; | |
2570 | if (pref0_len < 0) | |
2571 | pref0_len = 0; | |
2572 | if (pref0_len > pref_len) | |
2573 | pref0_len = pref_len; | |
2574 | { | |
3df15adc | 2575 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2576 | |
ab3bbdeb | 2577 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2578 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2579 | |
ab3bbdeb | 2580 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2581 | (locinput - pref_len + pref0_len), |
1de06328 | 2582 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2583 | |
ab3bbdeb | 2584 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2585 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2586 | |
1de06328 | 2587 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2588 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2589 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2590 | (IV)(locinput - loc_bostr), |
07be1b83 | 2591 | len0, s0, |
07be1b83 | 2592 | len1, s1, |
07be1b83 | 2593 | (docolor ? "" : "> <"), |
07be1b83 | 2594 | len2, s2, |
f9f4320a | 2595 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2596 | ""); |
2597 | } | |
2598 | } | |
3df15adc | 2599 | |
07be1b83 YO |
2600 | #endif |
2601 | ||
0a4db386 YO |
2602 | /* reg_check_named_buff_matched() |
2603 | * Checks to see if a named buffer has matched. The data array of | |
2604 | * buffer numbers corresponding to the buffer is expected to reside | |
2605 | * in the regexp->data->data array in the slot stored in the ARG() of | |
2606 | * node involved. Note that this routine doesn't actually care about the | |
2607 | * name, that information is not preserved from compilation to execution. | |
2608 | * Returns the index of the leftmost defined buffer with the given name | |
2609 | * or 0 if non of the buffers matched. | |
2610 | */ | |
2611 | STATIC I32 | |
2612 | S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) { | |
2613 | I32 n; | |
f8fc2ecf YO |
2614 | RXi_GET_DECL(rex,rexi); |
2615 | SV *sv_dat=(SV*)rexi->data->data[ ARG( scan ) ]; | |
0a4db386 YO |
2616 | I32 *nums=(I32*)SvPVX(sv_dat); |
2617 | for ( n=0; n<SvIVX(sv_dat); n++ ) { | |
2618 | if ((I32)*PL_reglastparen >= nums[n] && | |
f0ab9afb | 2619 | PL_regoffs[nums[n]].end != -1) |
0a4db386 YO |
2620 | { |
2621 | return nums[n]; | |
2622 | } | |
2623 | } | |
2624 | return 0; | |
2625 | } | |
2626 | ||
2f554ef7 DM |
2627 | |
2628 | /* free all slabs above current one - called during LEAVE_SCOPE */ | |
2629 | ||
2630 | STATIC void | |
2631 | S_clear_backtrack_stack(pTHX_ void *p) | |
2632 | { | |
2633 | regmatch_slab *s = PL_regmatch_slab->next; | |
2634 | PERL_UNUSED_ARG(p); | |
2635 | ||
2636 | if (!s) | |
2637 | return; | |
2638 | PL_regmatch_slab->next = NULL; | |
2639 | while (s) { | |
2640 | regmatch_slab * const osl = s; | |
2641 | s = s->next; | |
2642 | Safefree(osl); | |
2643 | } | |
2644 | } | |
2645 | ||
2646 | ||
28d8d7f4 YO |
2647 | #define SETREX(Re1,Re2) \ |
2648 | if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \ | |
2649 | Re1 = (Re2) | |
2650 | ||
d6a28714 | 2651 | STATIC I32 /* 0 failure, 1 success */ |
24b23f37 | 2652 | S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2653 | { |
a35a87e7 | 2654 | #if PERL_VERSION < 9 && !defined(PERL_CORE) |
54df2634 NC |
2655 | dMY_CXT; |
2656 | #endif | |
27da23d5 | 2657 | dVAR; |
95b24440 | 2658 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2659 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
95b24440 | 2660 | |
3b0527fe | 2661 | regexp *rex = reginfo->prog; |
f8fc2ecf YO |
2662 | RXi_GET_DECL(rex,rexi); |
2663 | ||
2f554ef7 | 2664 | I32 oldsave; |
a3621e74 | 2665 | |
5d9a96ca DM |
2666 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2667 | register regmatch_state *st; | |
95b24440 | 2668 | |
5d9a96ca DM |
2669 | /* cache heavy used fields of st in registers */ |
2670 | register regnode *scan; | |
2671 | register regnode *next; | |
438e9bae | 2672 | register U32 n = 0; /* general value; init to avoid compiler warning */ |
24d3c4a9 | 2673 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ |
5d9a96ca | 2674 | register char *locinput = PL_reginput; |
5d9a96ca | 2675 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2676 | |
b69b0499 | 2677 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2678 | int depth = 0; /* depth of backtrack stack */ |
4b196cd4 YO |
2679 | U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */ |
2680 | const U32 max_nochange_depth = | |
2681 | (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ? | |
2682 | 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH; | |
2683 | ||
77cb431f DM |
2684 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2685 | subpattern */ | |
e2e6a0f1 YO |
2686 | /* mark_state piggy backs on the yes_state logic so that when we unwind |
2687 | the stack on success we can update the mark_state as we go */ | |
2688 | regmatch_state *mark_state = NULL; /* last mark state we have seen */ | |
5d458dd8 | 2689 | |
faec1544 | 2690 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2691 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2692 | U32 state_num; |
5d458dd8 YO |
2693 | bool no_final = 0; /* prevent failure from backtracking? */ |
2694 | bool do_cutgroup = 0; /* no_final only until next branch/trie entry */ | |
e2e6a0f1 | 2695 | char *startpoint = PL_reginput; |
5d458dd8 YO |
2696 | SV *popmark = NULL; /* are we looking for a mark? */ |
2697 | SV *sv_commit = NULL; /* last mark name seen in failure */ | |
2698 | SV *sv_yes_mark = NULL; /* last mark name we have seen | |
2699 | during a successfull match */ | |
2700 | U32 lastopen = 0; /* last open we saw */ | |
2701 | bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0; | |
19b95bf0 DM |
2702 | |
2703 | SV* const oreplsv = GvSV(PL_replgv); | |
58e23c8d | 2704 | |
5d458dd8 | 2705 | |
24d3c4a9 DM |
2706 | /* these three flags are set by various ops to signal information to |
2707 | * the very next op. They have a useful lifetime of exactly one loop | |
2708 | * iteration, and are not preserved or restored by state pushes/pops | |
2709 | */ | |
2710 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2711 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2712 | int logical = 0; /* the following EVAL is: | |
2713 | 0: (?{...}) | |
2714 | 1: (?(?{...})X|Y) | |
2715 | 2: (??{...}) | |
2716 | or the following IFMATCH/UNLESSM is: | |
2717 | false: plain (?=foo) | |
2718 | true: used as a condition: (?(?=foo)) | |
2719 | */ | |
2720 | ||
95b24440 | 2721 | #ifdef DEBUGGING |
e68ec53f | 2722 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2723 | #endif |
2724 | ||
3b57cd43 | 2725 | DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({ |
24b23f37 | 2726 | PerlIO_printf(Perl_debug_log,"regmatch start\n"); |
3b57cd43 | 2727 | })); |
5d9a96ca DM |
2728 | /* on first ever call to regmatch, allocate first slab */ |
2729 | if (!PL_regmatch_slab) { | |
2730 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2731 | PL_regmatch_slab->prev = NULL; | |
2732 | PL_regmatch_slab->next = NULL; | |
86545054 | 2733 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2734 | } |
2735 | ||
2f554ef7 DM |
2736 | oldsave = PL_savestack_ix; |
2737 | SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL); | |
2738 | SAVEVPTR(PL_regmatch_slab); | |
2739 | SAVEVPTR(PL_regmatch_state); | |
5d9a96ca DM |
2740 | |
2741 | /* grab next free state slot */ | |
2742 | st = ++PL_regmatch_state; | |
86545054 | 2743 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2744 | st = PL_regmatch_state = S_push_slab(aTHX); |
2745 | ||
d6a28714 JH |
2746 | /* Note that nextchr is a byte even in UTF */ |
2747 | nextchr = UCHARAT(locinput); | |
2748 | scan = prog; | |
2749 | while (scan != NULL) { | |
8ba1375e | 2750 | |
a3621e74 | 2751 | DEBUG_EXECUTE_r( { |
6136c704 | 2752 | SV * const prop = sv_newmortal(); |
1de06328 | 2753 | regnode *rnext=regnext(scan); |
786e8c11 | 2754 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2755 | regprop(rex, prop, scan); |
07be1b83 YO |
2756 | |
2757 | PerlIO_printf(Perl_debug_log, | |
2758 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
f8fc2ecf | 2759 | (IV)(scan - rexi->program), depth*2, "", |
07be1b83 | 2760 | SvPVX_const(prop), |
1de06328 | 2761 | (PL_regkind[OP(scan)] == END || !rnext) ? |
f8fc2ecf | 2762 | 0 : (IV)(rnext - rexi->program)); |
2a782b5b | 2763 | }); |
d6a28714 JH |
2764 | |
2765 | next = scan + NEXT_OFF(scan); | |
2766 | if (next == scan) | |
2767 | next = NULL; | |
40a82448 | 2768 | state_num = OP(scan); |
d6a28714 | 2769 | |
40a82448 DM |
2770 | reenter_switch: |
2771 | switch (state_num) { | |
d6a28714 | 2772 | case BOL: |
7fba1cd6 | 2773 | if (locinput == PL_bostr) |
d6a28714 | 2774 | { |
3b0527fe | 2775 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2776 | break; |
2777 | } | |
d6a28714 JH |
2778 | sayNO; |
2779 | case MBOL: | |
12d33761 HS |
2780 | if (locinput == PL_bostr || |
2781 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2782 | { |
b8c5462f JH |
2783 | break; |
2784 | } | |
d6a28714 JH |
2785 | sayNO; |
2786 | case SBOL: | |
c2a73568 | 2787 | if (locinput == PL_bostr) |
b8c5462f | 2788 | break; |
d6a28714 JH |
2789 | sayNO; |
2790 | case GPOS: | |
3b0527fe | 2791 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2792 | break; |
2793 | sayNO; | |
ee9b8eae YO |
2794 | |
2795 | case KEEPS: | |
2796 | /* update the startpoint */ | |
f0ab9afb | 2797 | st->u.keeper.val = PL_regoffs[0].start; |
ee9b8eae | 2798 | PL_reginput = locinput; |
f0ab9afb | 2799 | PL_regoffs[0].start = locinput - PL_bostr; |
ee9b8eae YO |
2800 | PUSH_STATE_GOTO(KEEPS_next, next); |
2801 | /*NOT-REACHED*/ | |
2802 | case KEEPS_next_fail: | |
2803 | /* rollback the start point change */ | |
f0ab9afb | 2804 | PL_regoffs[0].start = st->u.keeper.val; |
ee9b8eae YO |
2805 | sayNO_SILENT; |
2806 | /*NOT-REACHED*/ | |
d6a28714 | 2807 | case EOL: |
d6a28714 JH |
2808 | goto seol; |
2809 | case MEOL: | |
d6a28714 | 2810 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2811 | sayNO; |
b8c5462f | 2812 | break; |
d6a28714 JH |
2813 | case SEOL: |
2814 | seol: | |
2815 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2816 | sayNO; |
d6a28714 | 2817 | if (PL_regeol - locinput > 1) |
b8c5462f | 2818 | sayNO; |
b8c5462f | 2819 | break; |
d6a28714 JH |
2820 | case EOS: |
2821 | if (PL_regeol != locinput) | |
b8c5462f | 2822 | sayNO; |
d6a28714 | 2823 | break; |
ffc61ed2 | 2824 | case SANY: |
d6a28714 | 2825 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2826 | sayNO; |
f33976b4 DB |
2827 | if (do_utf8) { |
2828 | locinput += PL_utf8skip[nextchr]; | |
2829 | if (locinput > PL_regeol) | |
2830 | sayNO; | |
2831 | nextchr = UCHARAT(locinput); | |
2832 | } | |
2833 | else | |
2834 | nextchr = UCHARAT(++locinput); | |
2835 | break; | |
2836 | case CANY: | |
2837 | if (!nextchr && locinput >= PL_regeol) | |
2838 | sayNO; | |
b8c5462f | 2839 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2840 | break; |
ffc61ed2 | 2841 | case REG_ANY: |
1aa99e6b IH |
2842 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2843 | sayNO; | |
2844 | if (do_utf8) { | |
b8c5462f | 2845 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2846 | if (locinput > PL_regeol) |
2847 | sayNO; | |
a0ed51b3 | 2848 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2849 | } |
1aa99e6b IH |
2850 | else |
2851 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2852 | break; |
166ba7cd DM |
2853 | |
2854 | #undef ST | |
2855 | #define ST st->u.trie | |
786e8c11 YO |
2856 | case TRIEC: |
2857 | /* In this case the charclass data is available inline so | |
2858 | we can fail fast without a lot of extra overhead. | |
2859 | */ | |
2860 | if (scan->flags == EXACT || !do_utf8) { | |
2861 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2862 | DEBUG_EXECUTE_r( | |
2863 | PerlIO_printf(Perl_debug_log, | |
2864 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2865 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2866 | ); |
2867 | sayNO_SILENT; | |
2868 | /* NOTREACHED */ | |
2869 | } | |
2870 | } | |
2871 | /* FALL THROUGH */ | |
5b47454d | 2872 | case TRIE: |
3dab1dad | 2873 | { |
07be1b83 | 2874 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
3dab1dad YO |
2875 | const enum { trie_plain, trie_utf8, trie_utf8_fold } |
2876 | trie_type = do_utf8 ? | |
2877 | (scan->flags == EXACT ? trie_utf8 : trie_utf8_fold) | |
2878 | : trie_plain; | |
2879 | ||
2880 | /* what trie are we using right now */ | |
be8e71aa | 2881 | reg_trie_data * const trie |
f8fc2ecf | 2882 | = (reg_trie_data*)rexi->data->data[ ARG( scan ) ]; |
55eed653 | 2883 | HV * widecharmap = (HV *)rexi->data->data[ ARG( scan ) + 1 ]; |
3dab1dad | 2884 | U32 state = trie->startstate; |
166ba7cd | 2885 | |
3dab1dad YO |
2886 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2887 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2888 | ) { | |
2889 | if (trie->states[ state ].wordnum) { | |
2890 | DEBUG_EXECUTE_r( | |
2891 | PerlIO_printf(Perl_debug_log, | |
2892 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2893 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2894 | ); |
2895 | break; | |
2896 | } else { | |
2897 | DEBUG_EXECUTE_r( | |
2898 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2899 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2900 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2901 | ); |
2902 | sayNO_SILENT; | |
2903 | } | |
2904 | } | |
166ba7cd | 2905 | |
786e8c11 YO |
2906 | { |
2907 | U8 *uc = ( U8* )locinput; | |
2908 | ||
2909 | STRLEN len = 0; | |
2910 | STRLEN foldlen = 0; | |
2911 | U8 *uscan = (U8*)NULL; | |
2912 | STRLEN bufflen=0; | |
2913 | SV *sv_accept_buff = NULL; | |
2914 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2915 | ||
2916 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2917 | ST.B = next; | |
2918 | ST.jump = trie->jump; | |
786e8c11 | 2919 | ST.me = scan; |
07be1b83 YO |
2920 | /* |
2921 | traverse the TRIE keeping track of all accepting states | |
2922 | we transition through until we get to a failing node. | |
2923 | */ | |
2924 | ||
a3621e74 | 2925 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 2926 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 2927 | UV uvc = 0; |
786e8c11 YO |
2928 | U16 charid; |
2929 | /* We use charid to hold the wordnum as we don't use it | |
2930 | for charid until after we have done the wordnum logic. | |
2931 | We define an alias just so that the wordnum logic reads | |
2932 | more naturally. */ | |
2933 | ||
2934 | #define got_wordnum charid | |
2935 | got_wordnum = trie->states[ state ].wordnum; | |
2936 | ||
2937 | if ( got_wordnum ) { | |
2938 | if ( ! ST.accepted ) { | |
5b47454d | 2939 | ENTER; |
6b173516 | 2940 | /* SAVETMPS; */ /* XXX is this necessary? dmq */ |
5b47454d DM |
2941 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; |
2942 | sv_accept_buff=newSV(bufflen * | |
2943 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 2944 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
2945 | SvPOK_on(sv_accept_buff); |
2946 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
2947 | SAVETMPS; |
2948 | ST.accept_buff = | |
5b47454d DM |
2949 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
2950 | } | |
786e8c11 | 2951 | do { |
166ba7cd | 2952 | if (ST.accepted >= bufflen) { |
5b47454d | 2953 | bufflen *= 2; |
166ba7cd | 2954 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
2955 | SvGROW(sv_accept_buff, |
2956 | bufflen * sizeof(reg_trie_accepted)); | |
2957 | } | |
2958 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
2959 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 2960 | |
786e8c11 YO |
2961 | |
2962 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
2963 | ST.accept_buff[ST.accepted].endpos = uc; | |
2964 | ++ST.accepted; | |
2965 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
2966 | } | |
2967 | #undef got_wordnum | |
a3621e74 | 2968 | |
07be1b83 | 2969 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 2970 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 2971 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 2972 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 2973 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 2974 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 2975 | }); |
a3621e74 YO |
2976 | |
2977 | if ( base ) { | |
55eed653 NC |
2978 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
2979 | uscan, len, uvc, charid, foldlen, | |
2980 | foldbuf, uniflags); | |
a3621e74 | 2981 | |
5b47454d DM |
2982 | if (charid && |
2983 | (base + charid > trie->uniquecharcount ) | |
2984 | && (base + charid - 1 - trie->uniquecharcount | |
2985 | < trie->lasttrans) | |
2986 | && trie->trans[base + charid - 1 - | |
2987 | trie->uniquecharcount].check == state) | |
2988 | { | |
2989 | state = trie->trans[base + charid - 1 - | |
2990 | trie->uniquecharcount ].next; | |
2991 | } | |
2992 | else { | |
2993 | state = 0; | |
2994 | } | |
2995 | uc += len; | |
2996 | ||
2997 | } | |
2998 | else { | |
a3621e74 YO |
2999 | state = 0; |
3000 | } | |
3001 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 3002 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 3003 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 3004 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
3005 | ); |
3006 | } | |
166ba7cd | 3007 | if (!ST.accepted ) |
a3621e74 | 3008 | sayNO; |
a3621e74 | 3009 | |
166ba7cd DM |
3010 | DEBUG_EXECUTE_r( |
3011 | PerlIO_printf( Perl_debug_log, | |
3012 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 3013 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
3014 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
3015 | ); | |
786e8c11 | 3016 | }} |
fae667d5 YO |
3017 | goto trie_first_try; /* jump into the fail handler */ |
3018 | /* NOTREACHED */ | |
166ba7cd | 3019 | case TRIE_next_fail: /* we failed - try next alterative */ |
fae667d5 YO |
3020 | if ( ST.jump) { |
3021 | REGCP_UNWIND(ST.cp); | |
3022 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
f0ab9afb | 3023 | PL_regoffs[n].end = -1; |
fae667d5 YO |
3024 | *PL_reglastparen = n; |
3025 | } | |
3026 | trie_first_try: | |
5d458dd8 YO |
3027 | if (do_cutgroup) { |
3028 | do_cutgroup = 0; | |
3029 | no_final = 0; | |
3030 | } | |
fae667d5 YO |
3031 | |
3032 | if ( ST.jump) { | |
3033 | ST.lastparen = *PL_reglastparen; | |
3034 | REGCP_SET(ST.cp); | |
3035 | } | |
166ba7cd DM |
3036 | if ( ST.accepted == 1 ) { |
3037 | /* only one choice left - just continue */ | |
3038 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
3039 | AV *const trie_words |
3040 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
3041 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 3042 | ST.accept_buff[ 0 ].wordnum-1, 0 ); |
de734bd5 A |
3043 | SV *sv= tmp ? sv_newmortal() : NULL; |
3044 | ||
166ba7cd DM |
3045 | PerlIO_printf( Perl_debug_log, |
3046 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 3047 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3048 | ST.accept_buff[ 0 ].wordnum, |
de734bd5 A |
3049 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3050 | PL_colors[0], PL_colors[1], | |
3051 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3052 | ) | |
3053 | : "not compiled under -Dr", | |
166ba7cd DM |
3054 | PL_colors[5] ); |
3055 | }); | |
3056 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
3057 | /* in this case we free tmps/leave before we call regmatch | |
3058 | as we wont be using accept_buff again. */ | |
5d458dd8 | 3059 | |
166ba7cd DM |
3060 | locinput = PL_reginput; |
3061 | nextchr = UCHARAT(locinput); | |
5d458dd8 YO |
3062 | if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum]) |
3063 | scan = ST.B; | |
3064 | else | |
3065 | scan = ST.me + ST.jump[ST.accept_buff[0].wordnum]; | |
3066 | if (!has_cutgroup) { | |
3067 | FREETMPS; | |
3068 | LEAVE; | |
3069 | } else { | |
3070 | ST.accepted--; | |
3071 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
3072 | } | |
786e8c11 | 3073 | |
166ba7cd DM |
3074 | continue; /* execute rest of RE */ |
3075 | } | |
fae667d5 YO |
3076 | |
3077 | if ( !ST.accepted-- ) { | |
5d458dd8 YO |
3078 | DEBUG_EXECUTE_r({ |
3079 | PerlIO_printf( Perl_debug_log, | |
3080 | "%*s %sTRIE failed...%s\n", | |
3081 | REPORT_CODE_OFF+depth*2, "", | |
3082 | PL_colors[4], | |
3083 | PL_colors[5] ); | |
3084 | }); | |
166ba7cd DM |
3085 | FREETMPS; |
3086 | LEAVE; | |
5d458dd8 | 3087 | sayNO_SILENT; |
fae667d5 YO |
3088 | /*NOTREACHED*/ |
3089 | } | |
166ba7cd | 3090 | |
a3621e74 | 3091 | /* |
166ba7cd DM |
3092 | There are at least two accepting states left. Presumably |
3093 | the number of accepting states is going to be low, | |
3094 | typically two. So we simply scan through to find the one | |
3095 | with lowest wordnum. Once we find it, we swap the last | |
3096 | state into its place and decrement the size. We then try to | |
3097 | match the rest of the pattern at the point where the word | |
3098 | ends. If we succeed, control just continues along the | |
3099 | regex; if we fail we return here to try the next accepting | |
3100 | state | |
3101 | */ | |
a3621e74 | 3102 | |
166ba7cd DM |
3103 | { |
3104 | U32 best = 0; | |
3105 | U32 cur; | |
3106 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
3107 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 3108 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 3109 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 3110 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
3111 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
3112 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
3113 | ); | |
3114 | ||
3115 | if (ST.accept_buff[cur].wordnum < | |
3116 | ST.accept_buff[best].wordnum) | |
3117 | best = cur; | |
a3621e74 | 3118 | } |
166ba7cd DM |
3119 | |
3120 | DEBUG_EXECUTE_r({ | |
2b8b4781 NC |
3121 | AV *const trie_words |
3122 | = (AV *) rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]; | |
3123 | SV ** const tmp = av_fetch( trie_words, | |
5c9f2f80 | 3124 | ST.accept_buff[ best ].wordnum - 1, 0 ); |
7f69552c | 3125 | regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ? |
265c4333 | 3126 | ST.B : |
7f69552c | 3127 | ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
de734bd5 A |
3128 | SV *sv= tmp ? sv_newmortal() : NULL; |
3129 | ||
265c4333 YO |
3130 | PerlIO_printf( Perl_debug_log, |
3131 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 3132 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3133 | ST.accept_buff[best].wordnum, |
de734bd5 A |
3134 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3135 | PL_colors[0], PL_colors[1], | |
3136 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3137 | ) : "not compiled under -Dr", | |
265c4333 | 3138 | REG_NODE_NUM(nextop), |
166ba7cd DM |
3139 | PL_colors[5] ); |
3140 | }); | |
3141 | ||
3142 | if ( best<ST.accepted ) { | |
3143 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
3144 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
3145 | ST.accept_buff[ ST.accepted ] = tmp; | |
3146 | best = ST.accepted; | |
a3621e74 | 3147 | } |
166ba7cd | 3148 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
7f69552c | 3149 | if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) { |
5d458dd8 | 3150 | scan = ST.B; |
786e8c11 | 3151 | } else { |
5d458dd8 | 3152 | scan = ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
5d458dd8 | 3153 | } |
6b173516 | 3154 | PUSH_YES_STATE_GOTO(TRIE_next, scan); |
786e8c11 | 3155 | /* NOTREACHED */ |
166ba7cd | 3156 | } |
166ba7cd | 3157 | /* NOTREACHED */ |
5d458dd8 YO |
3158 | case TRIE_next: |
3159 | FREETMPS; | |
3160 | LEAVE; | |
3161 | sayYES; | |
166ba7cd DM |
3162 | #undef ST |
3163 | ||
95b24440 DM |
3164 | case EXACT: { |
3165 | char *s = STRING(scan); | |
24d3c4a9 | 3166 | ln = STR_LEN(scan); |
eb160463 | 3167 | if (do_utf8 != UTF) { |
bc517b45 | 3168 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 3169 | char *l = locinput; |
24d3c4a9 | 3170 | const char * const e = s + ln; |
a72c7584 | 3171 | |
5ff6fc6d JH |
3172 | if (do_utf8) { |
3173 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 3174 | while (s < e) { |
a3b680e6 | 3175 | STRLEN ulen; |
1aa99e6b | 3176 | if (l >= PL_regeol) |
5ff6fc6d JH |
3177 | sayNO; |
3178 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 3179 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 3180 | uniflags)) |
5ff6fc6d | 3181 | sayNO; |
bc517b45 | 3182 | l += ulen; |
5ff6fc6d | 3183 | s ++; |
1aa99e6b | 3184 | } |
5ff6fc6d JH |
3185 | } |
3186 | else { | |
3187 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 3188 | while (s < e) { |
a3b680e6 | 3189 | STRLEN ulen; |
1aa99e6b IH |
3190 | if (l >= PL_regeol) |
3191 | sayNO; | |
5ff6fc6d | 3192 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 3193 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 3194 | uniflags)) |
1aa99e6b | 3195 | sayNO; |
bc517b45 | 3196 | s += ulen; |
a72c7584 | 3197 | l ++; |
1aa99e6b | 3198 | } |
5ff6fc6d | 3199 | } |
1aa99e6b IH |
3200 | locinput = l; |
3201 | nextchr = UCHARAT(locinput); | |
3202 | break; | |
3203 | } | |
bc517b45 | 3204 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
3205 | /* Inline the first character, for speed. */ |
3206 | if (UCHARAT(s) != nextchr) | |
3207 | sayNO; | |
24d3c4a9 | 3208 | if (PL_regeol - locinput < ln) |
d6a28714 | 3209 | sayNO; |
24d3c4a9 | 3210 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3211 | sayNO; |
24d3c4a9 | 3212 | locinput += ln; |
d6a28714 JH |
3213 | nextchr = UCHARAT(locinput); |
3214 | break; | |
95b24440 | 3215 | } |
d6a28714 | 3216 | case EXACTFL: |
b8c5462f JH |
3217 | PL_reg_flags |= RF_tainted; |
3218 | /* FALL THROUGH */ | |
95b24440 | 3219 | case EXACTF: { |
be8e71aa | 3220 | char * const s = STRING(scan); |
24d3c4a9 | 3221 | ln = STR_LEN(scan); |
d6a28714 | 3222 | |
d07ddd77 JH |
3223 | if (do_utf8 || UTF) { |
3224 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3225 | const char * const l = locinput; |
d07ddd77 | 3226 | char *e = PL_regeol; |
bc517b45 | 3227 | |
24d3c4a9 | 3228 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3229 | l, &e, 0, do_utf8)) { |
5486206c JH |
3230 | /* One more case for the sharp s: |
3231 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3232 | * the 0xC3 0x9F are the UTF-8 | |
3233 | * byte sequence for the U+00DF. */ | |
e1d1eefb | 3234 | |
5486206c | 3235 | if (!(do_utf8 && |
e1d1eefb | 3236 | toLOWER(s[0]) == 's' && |
24d3c4a9 | 3237 | ln >= 2 && |
5486206c JH |
3238 | toLOWER(s[1]) == 's' && |
3239 | (U8)l[0] == 0xC3 && | |
3240 | e - l >= 2 && | |
3241 | (U8)l[1] == 0x9F)) | |
3242 | sayNO; | |
3243 | } | |
d07ddd77 JH |
3244 | locinput = e; |
3245 | nextchr = UCHARAT(locinput); | |
3246 | break; | |
a0ed51b3 | 3247 | } |
d6a28714 | 3248 | |
bc517b45 JH |
3249 | /* Neither the target and the pattern are utf8. */ |
3250 | ||
d6a28714 JH |
3251 | /* Inline the first character, for speed. */ |
3252 | if (UCHARAT(s) != nextchr && | |
3253 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3254 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3255 | sayNO; |
24d3c4a9 | 3256 | if (PL_regeol - locinput < ln) |
b8c5462f | 3257 | sayNO; |
24d3c4a9 DM |
3258 | if (ln > 1 && (OP(scan) == EXACTF |
3259 | ? ibcmp(s, locinput, ln) | |
3260 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3261 | sayNO; |
24d3c4a9 | 3262 | locinput += ln; |
d6a28714 | 3263 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3264 | break; |
95b24440 | 3265 | } |
d6a28714 | 3266 | case ANYOF: |
ffc61ed2 | 3267 | if (do_utf8) { |
9e55ce06 JH |
3268 | STRLEN inclasslen = PL_regeol - locinput; |
3269 | ||
32fc9b6a | 3270 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3271 | goto anyof_fail; |
ffc61ed2 JH |
3272 | if (locinput >= PL_regeol) |
3273 | sayNO; | |
0f0076b4 | 3274 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3275 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3276 | break; |
ffc61ed2 JH |
3277 | } |
3278 | else { | |
3279 | if (nextchr < 0) | |
3280 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3281 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3282 | goto anyof_fail; |
ffc61ed2 JH |
3283 | if (!nextchr && locinput >= PL_regeol) |
3284 | sayNO; | |
3285 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3286 | break; |
3287 | } | |
262b90c4 | 3288 | anyof_fail: |
e0f9d4a8 JH |
3289 | /* If we might have the case of the German sharp s |
3290 | * in a casefolding Unicode character class. */ | |
3291 | ||
ebc501f0 JH |
3292 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3293 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3294 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3295 | } |
e0f9d4a8 JH |
3296 | else |
3297 | sayNO; | |
b8c5462f | 3298 | break; |
d6a28714 | 3299 | case ALNUML: |
b8c5462f JH |
3300 | PL_reg_flags |= RF_tainted; |
3301 | /* FALL THROUGH */ | |
d6a28714 | 3302 | case ALNUM: |
b8c5462f | 3303 | if (!nextchr) |
4633a7c4 | 3304 | sayNO; |
ffc61ed2 | 3305 | if (do_utf8) { |
1a4fad37 | 3306 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3307 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3308 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3309 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3310 | { |
3311 | sayNO; | |
a0ed51b3 | 3312 | } |
b8c5462f | 3313 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3314 | nextchr = UCHARAT(locinput); |
3315 | break; | |
3316 | } | |
ffc61ed2 | 3317 | if (!(OP(scan) == ALNUM |
d6a28714 | 3318 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3319 | sayNO; |
b8c5462f | 3320 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3321 | break; |
d6a28714 | 3322 | case NALNUML: |
b8c5462f JH |
3323 | PL_reg_flags |= RF_tainted; |
3324 | /* FALL THROUGH */ | |
d6a28714 JH |
3325 | case NALNUM: |
3326 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3327 | sayNO; |
ffc61ed2 | 3328 | if (do_utf8) { |
1a4fad37 | 3329 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3330 | if (OP(scan) == NALNUM |
bb7a0f54 | 3331 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3332 | : isALNUM_LC_utf8((U8*)locinput)) |
3333 | { | |
b8c5462f | 3334 | sayNO; |
d6a28714 | 3335 | } |
b8c5462f JH |
3336 | locinput += PL_utf8skip[nextchr]; |
3337 | nextchr = UCHARAT(locinput); | |
3338 | break; | |
3339 | } | |
ffc61ed2 | 3340 | if (OP(scan) == NALNUM |
d6a28714 | 3341 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3342 | sayNO; |
76e3520e | 3343 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3344 | break; |
d6a28714 JH |
3345 | case BOUNDL: |
3346 | case NBOUNDL: | |
3280af22 | 3347 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3348 | /* FALL THROUGH */ |
d6a28714 JH |
3349 | case BOUND: |
3350 | case NBOUND: | |
3351 | /* was last char in word? */ | |
ffc61ed2 | 3352 | if (do_utf8) { |
12d33761 | 3353 | if (locinput == PL_bostr) |
24d3c4a9 | 3354 | ln = '\n'; |
ffc61ed2 | 3355 | else { |
a3b680e6 | 3356 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3357 | |
24d3c4a9 | 3358 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3359 | } |
3360 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3361 | ln = isALNUM_uni(ln); |
1a4fad37 | 3362 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3363 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3364 | } |
3365 | else { | |
24d3c4a9 | 3366 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3367 | n = isALNUM_LC_utf8((U8*)locinput); |
3368 | } | |
a0ed51b3 | 3369 | } |
d6a28714 | 3370 | else { |
24d3c4a9 | 3371 | ln = (locinput != PL_bostr) ? |
12d33761 | 3372 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3373 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3374 | ln = isALNUM(ln); |
ffc61ed2 JH |
3375 | n = isALNUM(nextchr); |
3376 | } | |
3377 | else { | |
24d3c4a9 | 3378 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3379 | n = isALNUM_LC(nextchr); |
3380 | } | |
d6a28714 | 3381 | } |
24d3c4a9 | 3382 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3383 | OP(scan) == BOUNDL)) |
3384 | sayNO; | |
a0ed51b3 | 3385 | break; |
d6a28714 | 3386 | case SPACEL: |
3280af22 | 3387 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3388 | /* FALL THROUGH */ |
d6a28714 | 3389 | case SPACE: |
9442cb0e | 3390 | if (!nextchr) |
4633a7c4 | 3391 | sayNO; |
1aa99e6b | 3392 | if (do_utf8) { |
fd400ab9 | 3393 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3394 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3395 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3396 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3397 | : isSPACE_LC_utf8((U8*)locinput))) |
3398 | { | |
3399 | sayNO; | |
3400 | } | |
3401 | locinput += PL_utf8skip[nextchr]; | |
3402 | nextchr = UCHARAT(locinput); | |
3403 | break; | |
d6a28714 | 3404 | } |
ffc61ed2 JH |
3405 | if (!(OP(scan) == SPACE |
3406 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3407 | sayNO; | |
3408 | nextchr = UCHARAT(++locinput); | |
3409 | } | |
3410 | else { | |
3411 | if (!(OP(scan) == SPACE | |
3412 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3413 | sayNO; | |
3414 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3415 | } |
a0ed51b3 | 3416 | break; |
d6a28714 | 3417 | case NSPACEL: |
3280af22 | 3418 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3419 | /* FALL THROUGH */ |
d6a28714 | 3420 | case NSPACE: |
9442cb0e | 3421 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3422 | sayNO; |
1aa99e6b | 3423 | if (do_utf8) { |
1a4fad37 | 3424 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3425 | if (OP(scan) == NSPACE |
bb7a0f54 | 3426 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3427 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3428 | { |
3429 | sayNO; | |
3430 | } | |
3431 | locinput += PL_utf8skip[nextchr]; | |
3432 | nextchr = UCHARAT(locinput); | |
3433 | break; | |
a0ed51b3 | 3434 | } |
ffc61ed2 | 3435 | if (OP(scan) == NSPACE |
d6a28714 | 3436 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3437 | sayNO; |
76e3520e | 3438 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3439 | break; |
d6a28714 | 3440 | case DIGITL: |
a0ed51b3 LW |
3441 | PL_reg_flags |= RF_tainted; |
3442 | /* FALL THROUGH */ | |
d6a28714 | 3443 | case DIGIT: |
9442cb0e | 3444 | if (!nextchr) |
a0ed51b3 | 3445 | sayNO; |
1aa99e6b | 3446 | if (do_utf8) { |
1a4fad37 | 3447 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3448 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3449 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3450 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3451 | { |
a0ed51b3 | 3452 | sayNO; |
dfe13c55 | 3453 | } |
6f06b55f | 3454 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3455 | nextchr = UCHARAT(locinput); |
3456 | break; | |
3457 | } | |
ffc61ed2 | 3458 | if (!(OP(scan) == DIGIT |
9442cb0e | 3459 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3460 | sayNO; |
3461 | nextchr = UCHARAT(++locinput); | |
3462 | break; | |
d6a28714 | 3463 | case NDIGITL: |
b8c5462f JH |
3464 | PL_reg_flags |= RF_tainted; |
3465 | /* FALL THROUGH */ | |
d6a28714 | 3466 | case NDIGIT: |
9442cb0e | 3467 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3468 | sayNO; |
1aa99e6b | 3469 | if (do_utf8) { |
1a4fad37 | 3470 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3471 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3472 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3473 | : isDIGIT_LC_utf8((U8*)locinput)) |
3474 | { | |
a0ed51b3 | 3475 | sayNO; |
9442cb0e | 3476 | } |
6f06b55f | 3477 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3478 | nextchr = UCHARAT(locinput); |
3479 | break; | |
3480 | } | |
ffc61ed2 | 3481 | if (OP(scan) == NDIGIT |
9442cb0e | 3482 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3483 | sayNO; |
3484 | nextchr = UCHARAT(++locinput); | |
3485 | break; | |
3486 | case CLUMP: | |
b7c83a7e | 3487 | if (locinput >= PL_regeol) |
a0ed51b3 | 3488 | sayNO; |
b7c83a7e | 3489 | if (do_utf8) { |
1a4fad37 | 3490 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3491 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3492 | sayNO; | |
3493 | locinput += PL_utf8skip[nextchr]; | |
3494 | while (locinput < PL_regeol && | |
3495 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3496 | locinput += UTF8SKIP(locinput); | |
3497 | if (locinput > PL_regeol) | |
3498 | sayNO; | |
eb08e2da JH |
3499 | } |
3500 | else | |
3501 | locinput++; | |
a0ed51b3 LW |
3502 | nextchr = UCHARAT(locinput); |
3503 | break; | |
81714fb9 YO |
3504 | |
3505 | case NREFFL: | |
3506 | { | |
3507 | char *s; | |
ff1157ca | 3508 | char type; |
81714fb9 YO |
3509 | PL_reg_flags |= RF_tainted; |
3510 | /* FALL THROUGH */ | |
3511 | case NREF: | |
3512 | case NREFF: | |
ff1157ca | 3513 | type = OP(scan); |
0a4db386 YO |
3514 | n = reg_check_named_buff_matched(rex,scan); |
3515 | ||
3516 | if ( n ) { | |
3517 | type = REF + ( type - NREF ); | |
3518 | goto do_ref; | |
3519 | } else { | |
81714fb9 | 3520 | sayNO; |
0a4db386 YO |
3521 | } |
3522 | /* unreached */ | |
c8756f30 | 3523 | case REFFL: |
3280af22 | 3524 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3525 | /* FALL THROUGH */ |
c277df42 | 3526 | case REF: |
81714fb9 | 3527 | case REFF: |
c277df42 | 3528 | n = ARG(scan); /* which paren pair */ |
81714fb9 YO |
3529 | type = OP(scan); |
3530 | do_ref: | |
f0ab9afb | 3531 | ln = PL_regoffs[n].start; |
2c2d71f5 | 3532 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
3b6647e0 | 3533 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 | 3534 | sayNO; /* Do not match unless seen CLOSEn. */ |
f0ab9afb | 3535 | if (ln == PL_regoffs[n].end) |
a0d0e21e | 3536 | break; |
a0ed51b3 | 3537 | |
24d3c4a9 | 3538 | s = PL_bostr + ln; |
81714fb9 | 3539 | if (do_utf8 && type != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3540 | char *l = locinput; |
f0ab9afb | 3541 | const char *e = PL_bostr + PL_regoffs[n].end; |
a0ed51b3 LW |
3542 | /* |
3543 | * Note that we can't do the "other character" lookup trick as | |
3544 | * in the 8-bit case (no pun intended) because in Unicode we | |
3545 | * have to map both upper and title case to lower case. | |
3546 | */ | |
81714fb9 | 3547 | if (type == REFF) { |
a0ed51b3 | 3548 | while (s < e) { |
a3b680e6 AL |
3549 | STRLEN ulen1, ulen2; |
3550 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3551 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3552 | ||
a0ed51b3 LW |
3553 | if (l >= PL_regeol) |
3554 | sayNO; | |
a2a2844f JH |
3555 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3556 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3557 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3558 | sayNO; |
a2a2844f JH |
3559 | s += ulen1; |
3560 | l += ulen2; | |
a0ed51b3 LW |
3561 | } |
3562 | } | |
3563 | locinput = l; | |
3564 | nextchr = UCHARAT(locinput); | |
3565 | break; | |
3566 | } | |
3567 | ||
a0d0e21e | 3568 | /* Inline the first character, for speed. */ |
76e3520e | 3569 | if (UCHARAT(s) != nextchr && |
81714fb9 YO |
3570 | (type == REF || |
3571 | (UCHARAT(s) != (type == REFF | |
3572 | ? PL_fold : PL_fold_locale)[nextchr]))) | |
4633a7c4 | 3573 | sayNO; |
f0ab9afb | 3574 | ln = PL_regoffs[n].end - ln; |
24d3c4a9 | 3575 | if (locinput + ln > PL_regeol) |
4633a7c4 | 3576 | sayNO; |
81714fb9 | 3577 | if (ln > 1 && (type == REF |
24d3c4a9 | 3578 | ? memNE(s, locinput, ln) |
81714fb9 | 3579 | : (type == REFF |
24d3c4a9 DM |
3580 | ? ibcmp(s, locinput, ln) |
3581 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 3582 | sayNO; |
24d3c4a9 | 3583 | locinput += ln; |
76e3520e | 3584 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3585 | break; |
81714fb9 | 3586 | } |
a0d0e21e | 3587 | case NOTHING: |
c277df42 | 3588 | case TAIL: |
a0d0e21e LW |
3589 | break; |
3590 | case BACK: | |
3591 | break; | |
40a82448 DM |
3592 | |
3593 | #undef ST | |
3594 | #define ST st->u.eval | |
c277df42 | 3595 | { |
c277df42 | 3596 | SV *ret; |
6bda09f9 | 3597 | regexp *re; |
f8fc2ecf | 3598 | regexp_internal *rei; |
1a147d38 YO |
3599 | regnode *startpoint; |
3600 | ||
3601 | case GOSTART: | |
e7707071 YO |
3602 | case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */ |
3603 | if (cur_eval && cur_eval->locinput==locinput) { | |
24b23f37 | 3604 | if (cur_eval->u.eval.close_paren == (U32)ARG(scan)) |
1a147d38 | 3605 | Perl_croak(aTHX_ "Infinite recursion in regex"); |
4b196cd4 | 3606 | if ( ++nochange_depth > max_nochange_depth ) |
1a147d38 YO |
3607 | Perl_croak(aTHX_ |
3608 | "Pattern subroutine nesting without pos change" | |
3609 | " exceeded limit in regex"); | |
6bda09f9 YO |
3610 | } else { |
3611 | nochange_depth = 0; | |
1a147d38 | 3612 | } |
6bda09f9 | 3613 | re = rex; |
f8fc2ecf | 3614 | rei = rexi; |
6bda09f9 | 3615 | (void)ReREFCNT_inc(rex); |
1a147d38 | 3616 | if (OP(scan)==GOSUB) { |
6bda09f9 YO |
3617 | startpoint = scan + ARG2L(scan); |
3618 | ST.close_paren = ARG(scan); | |
3619 | } else { | |
f8fc2ecf | 3620 | startpoint = rei->program+1; |
6bda09f9 YO |
3621 | ST.close_paren = 0; |
3622 | } | |
3623 | goto eval_recurse_doit; | |
3624 | /* NOTREACHED */ | |
3625 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
3626 | if (cur_eval && cur_eval->locinput==locinput) { | |
4b196cd4 | 3627 | if ( ++nochange_depth > max_nochange_depth ) |
1a147d38 | 3628 | Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex"); |
6bda09f9 YO |
3629 | } else { |
3630 | nochange_depth = 0; | |
3631 | } | |
8e5e9ebe | 3632 | { |
4aabdb9b DM |
3633 | /* execute the code in the {...} */ |
3634 | dSP; | |
6136c704 | 3635 | SV ** const before = SP; |
4aabdb9b DM |
3636 | OP_4tree * const oop = PL_op; |
3637 | COP * const ocurcop = PL_curcop; | |
3638 | PAD *old_comppad; | |
4aabdb9b DM |
3639 | |
3640 | n = ARG(scan); | |
f8fc2ecf | 3641 | PL_op = (OP_4tree*)rexi->data->data[n]; |
24b23f37 YO |
3642 | DEBUG_STATE_r( PerlIO_printf(Perl_debug_log, |
3643 | " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); | |
f8fc2ecf | 3644 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]); |
f0ab9afb | 3645 | PL_regoffs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr; |
4aabdb9b | 3646 | |
2bf803e2 YO |
3647 | if (sv_yes_mark) { |
3648 | SV *sv_mrk = get_sv("REGMARK", 1); | |
3649 | sv_setsv(sv_mrk, sv_yes_mark); | |
3650 | } | |
3651 | ||
8e5e9ebe RGS |
3652 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3653 | SPAGAIN; | |
3654 | if (SP == before) | |
075aa684 | 3655 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3656 | else { |
3657 | ret = POPs; | |
3658 | PUTBACK; | |
3659 | } | |
4aabdb9b DM |
3660 | |
3661 | PL_op = oop; | |
3662 | PAD_RESTORE_LOCAL(old_comppad); | |
3663 | PL_curcop = ocurcop; | |
24d3c4a9 | 3664 | if (!logical) { |
4aabdb9b DM |
3665 | /* /(?{...})/ */ |
3666 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3667 | break; |
3668 | } | |
8e5e9ebe | 3669 | } |
24d3c4a9 DM |
3670 | if (logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3671 | logical = 0; | |
4aabdb9b | 3672 | { |
4f639d21 DM |
3673 | /* extract RE object from returned value; compiling if |
3674 | * necessary */ | |
3675 | ||
6136c704 | 3676 | MAGIC *mg = NULL; |
be8e71aa | 3677 | const SV *sv; |
faf82a0b AE |
3678 | if(SvROK(ret) && SvSMAGICAL(sv = SvRV(ret))) |
3679 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3680 | else if (SvSMAGICAL(ret)) { | |
3681 | if (SvGMAGICAL(ret)) | |
3682 | sv_unmagic(ret, PERL_MAGIC_qr); | |
3683 | else | |
3684 | mg = mg_find(ret, PERL_MAGIC_qr); | |
0f5d15d6 | 3685 | } |
faf82a0b | 3686 | |
0f5d15d6 | 3687 | if (mg) { |
28d8d7f4 | 3688 | re = reg_temp_copy((regexp *)mg->mg_obj); /*XXX:dmq*/ |
0f5d15d6 IZ |
3689 | } |
3690 | else { | |
c737faaf | 3691 | U32 pm_flags = 0; |
a3b680e6 | 3692 | const I32 osize = PL_regsize; |
0f5d15d6 | 3693 | |
c737faaf | 3694 | if (DO_UTF8(ret)) pm_flags |= RXf_UTF8; |
3ab4a224 | 3695 | re = CALLREGCOMP(ret, pm_flags); |
9041c2e3 | 3696 | if (!(SvFLAGS(ret) |
faf82a0b AE |
3697 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3698 | | SVs_GMG))) | |
14befaf4 DM |
3699 | sv_magic(ret,(SV*)ReREFCNT_inc(re), |
3700 | PERL_MAGIC_qr,0,0); | |
0f5d15d6 | 3701 | PL_regsize = osize; |
0f5d15d6 | 3702 | } |
4aabdb9b | 3703 | } |
28d8d7f4 YO |
3704 | RX_MATCH_COPIED_off(re); |
3705 | re->subbeg = rex->subbeg; | |
3706 | re->sublen = rex->sublen; | |
f8fc2ecf | 3707 | rei = RXi_GET(re); |
6bda09f9 YO |
3708 | DEBUG_EXECUTE_r( |
3709 | debug_start_match(re, do_utf8, locinput, PL_regeol, | |
3710 | "Matching embedded"); | |
3711 | ); | |
f8fc2ecf | 3712 | startpoint = rei->program + 1; |
1a147d38 | 3713 | ST.close_paren = 0; /* only used for GOSUB */ |
6bda09f9 YO |
3714 | /* borrowed from regtry */ |
3715 | if (PL_reg_start_tmpl <= re->nparens) { | |
3716 | PL_reg_start_tmpl = re->nparens*3/2 + 3; | |
3717 | if(PL_reg_start_tmp) | |
3718 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3719 | else | |
3720 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
dd5def09 | 3721 | } |
aa283a38 | 3722 | |
1a147d38 | 3723 | eval_recurse_doit: /* Share code with GOSUB below this line */ |
aa283a38 | 3724 | /* run the pattern returned from (??{...}) */ |
40a82448 DM |
3725 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3726 | REGCP_SET(ST.lastcp); | |
6bda09f9 | 3727 | |
f0ab9afb | 3728 | PL_regoffs = re->offs; /* essentially NOOP on GOSUB */ |
6bda09f9 | 3729 | |
4aabdb9b DM |
3730 | *PL_reglastparen = 0; |
3731 | *PL_reglastcloseparen = 0; | |
4aabdb9b | 3732 | PL_reginput = locinput; |
ae0beba1 | 3733 | PL_regsize = 0; |
4aabdb9b DM |
3734 | |
3735 | /* XXXX This is too dramatic a measure... */ | |
3736 | PL_reg_maxiter = 0; | |
3737 | ||
faec1544 | 3738 | ST.toggle_reg_flags = PL_reg_flags; |
bbe252da | 3739 | if (re->extflags & RXf_UTF8) |
faec1544 DM |
3740 | PL_reg_flags |= RF_utf8; |
3741 | else | |
3742 | PL_reg_flags &= ~RF_utf8; | |
3743 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3744 | ||
40a82448 | 3745 | ST.prev_rex = rex; |
faec1544 | 3746 | ST.prev_curlyx = cur_curlyx; |
28d8d7f4 | 3747 | SETREX(rex,re); |
f8fc2ecf | 3748 | rexi = rei; |
faec1544 | 3749 | cur_curlyx = NULL; |
40a82448 | 3750 | ST.B = next; |
faec1544 DM |
3751 | ST.prev_eval = cur_eval; |
3752 | cur_eval = st; | |
faec1544 | 3753 | /* now continue from first node in postoned RE */ |
6bda09f9 | 3754 | PUSH_YES_STATE_GOTO(EVAL_AB, startpoint); |
4aabdb9b | 3755 | /* NOTREACHED */ |
a0ed51b3 | 3756 | } |
24d3c4a9 DM |
3757 | /* logical is 1, /(?(?{...})X|Y)/ */ |
3758 | sw = (bool)SvTRUE(ret); | |
3759 | logical = 0; | |
c277df42 IZ |
3760 | break; |
3761 | } | |
40a82448 | 3762 | |
faec1544 DM |
3763 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3764 | /* note: this is called twice; first after popping B, then A */ | |
3765 | PL_reg_flags ^= ST.toggle_reg_flags; | |
40a82448 | 3766 | ReREFCNT_dec(rex); |
28d8d7f4 | 3767 | SETREX(rex,ST.prev_rex); |
f8fc2ecf | 3768 | rexi = RXi_GET(rex); |
faec1544 DM |
3769 | regcpblow(ST.cp); |
3770 | cur_eval = ST.prev_eval; | |
3771 | cur_curlyx = ST.prev_curlyx; | |
40a82448 DM |
3772 | /* XXXX This is too dramatic a measure... */ |
3773 | PL_reg_maxiter = 0; | |
e7707071 | 3774 | if ( nochange_depth ) |
4b196cd4 | 3775 | nochange_depth--; |
262b90c4 | 3776 | sayYES; |
40a82448 | 3777 | |
40a82448 | 3778 | |
faec1544 DM |
3779 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3780 | /* note: this is called twice; first after popping B, then A */ | |
3781 | PL_reg_flags ^= ST.toggle_reg_flags; | |
3782 | ReREFCNT_dec(rex); | |
28d8d7f4 | 3783 | SETREX(rex,ST.prev_rex); |
f8fc2ecf | 3784 | rexi = RXi_GET(rex); |
40a82448 DM |
3785 | PL_reginput = locinput; |
3786 | REGCP_UNWIND(ST.lastcp); | |
3787 | regcppop(rex); | |
faec1544 DM |
3788 | cur_eval = ST.prev_eval; |
3789 | cur_curlyx = ST.prev_curlyx; | |
3790 | /* XXXX This is too dramatic a measure... */ | |
3791 | PL_reg_maxiter = 0; | |
e7707071 | 3792 | if ( nochange_depth ) |
4b196cd4 | 3793 | nochange_depth--; |
40a82448 | 3794 | sayNO_SILENT; |
40a82448 DM |
3795 | #undef ST |
3796 | ||
a0d0e21e | 3797 | case OPEN: |
c277df42 | 3798 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3799 | PL_reg_start_tmp[n] = locinput; |
3800 | if (n > PL_regsize) | |
3801 | PL_regsize = n; | |
e2e6a0f1 | 3802 | lastopen = n; |
a0d0e21e LW |
3803 | break; |
3804 | case CLOSE: | |
c277df42 | 3805 | n = ARG(scan); /* which paren pair */ |
f0ab9afb NC |
3806 | PL_regoffs[n].start = PL_reg_start_tmp[n] - PL_bostr; |
3807 | PL_regoffs[n].end = locinput - PL_bostr; | |
7f69552c YO |
3808 | /*if (n > PL_regsize) |
3809 | PL_regsize = n;*/ | |
3b6647e0 | 3810 | if (n > *PL_reglastparen) |
3280af22 | 3811 | *PL_reglastparen = n; |
a01268b5 | 3812 | *PL_reglastcloseparen = n; |
3b6647e0 | 3813 | if (cur_eval && cur_eval->u.eval.close_paren == n) { |
6bda09f9 YO |
3814 | goto fake_end; |
3815 | } | |
a0d0e21e | 3816 | break; |
e2e6a0f1 YO |
3817 | case ACCEPT: |
3818 | if (ARG(scan)){ | |
3819 | regnode *cursor; | |
3820 | for (cursor=scan; | |
3821 | cursor && OP(cursor)!=END; | |
3822 | cursor=regnext(cursor)) | |
3823 | { | |
3824 | if ( OP(cursor)==CLOSE ){ | |
3825 | n = ARG(cursor); | |
3826 | if ( n <= lastopen ) { | |
f0ab9afb NC |
3827 | PL_regoffs[n].start |
3828 | = PL_reg_start_tmp[n] - PL_bostr; | |
3829 | PL_regoffs[n].end = locinput - PL_bostr; | |
e2e6a0f1 YO |
3830 | /*if (n > PL_regsize) |
3831 | PL_regsize = n;*/ | |
3b6647e0 | 3832 | if (n > *PL_reglastparen) |
e2e6a0f1 YO |
3833 | *PL_reglastparen = n; |
3834 | *PL_reglastcloseparen = n; | |
3b6647e0 RB |
3835 | if ( n == ARG(scan) || (cur_eval && |
3836 | cur_eval->u.eval.close_paren == n)) | |
e2e6a0f1 YO |
3837 | break; |
3838 | } | |
3839 | } | |
3840 | } | |
3841 | } | |
3842 | goto fake_end; | |
3843 | /*NOTREACHED*/ | |
c277df42 IZ |
3844 | case GROUPP: |
3845 | n = ARG(scan); /* which paren pair */ | |
f0ab9afb | 3846 | sw = (bool)(*PL_reglastparen >= n && PL_regoffs[n].end != -1); |
c277df42 | 3847 | break; |
0a4db386 YO |
3848 | case NGROUPP: |
3849 | /* reg_check_named_buff_matched returns 0 for no match */ | |
3850 | sw = (bool)(0 < reg_check_named_buff_matched(rex,scan)); | |
3851 | break; | |
1a147d38 | 3852 | case INSUBP: |
0a4db386 | 3853 | n = ARG(scan); |
3b6647e0 | 3854 | sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n)); |
0a4db386 YO |
3855 | break; |
3856 | case DEFINEP: | |
3857 | sw = 0; | |
3858 | break; | |
c277df42 | 3859 | case IFTHEN: |
2c2d71f5 | 3860 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 3861 | if (sw) |
c277df42 IZ |
3862 | next = NEXTOPER(NEXTOPER(scan)); |
3863 | else { | |
3864 | next = scan + ARG(scan); | |
3865 | if (OP(next) == IFTHEN) /* Fake one. */ | |
3866 | next = NEXTOPER(NEXTOPER(next)); | |
3867 | } | |
3868 | break; | |
3869 | case LOGICAL: | |
24d3c4a9 | 3870 | logical = scan->flags; |
c277df42 | 3871 | break; |
c476f425 | 3872 | |
2ab05381 | 3873 | /******************************************************************* |
2ab05381 | 3874 | |
c476f425 DM |
3875 | The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/ |
3876 | pattern, where A and B are subpatterns. (For simple A, CURLYM or | |
3877 | STAR/PLUS/CURLY/CURLYN are used instead.) | |
2ab05381 | 3878 | |
c476f425 | 3879 | A*B is compiled as <CURLYX><A><WHILEM><B> |
2ab05381 | 3880 | |
c476f425 DM |
3881 | On entry to the subpattern, CURLYX is called. This pushes a CURLYX |
3882 | state, which contains the current count, initialised to -1. It also sets | |
3883 | cur_curlyx to point to this state, with any previous value saved in the | |
3884 | state block. | |
2ab05381 | 3885 | |
c476f425 DM |
3886 | CURLYX then jumps straight to the WHILEM op, rather than executing A, |
3887 | since the pattern may possibly match zero times (i.e. it's a while {} loop | |
3888 | rather than a do {} while loop). | |
2ab05381 | 3889 | |
c476f425 DM |
3890 | Each entry to WHILEM represents a successful match of A. The count in the |
3891 | CURLYX block is incremented, another WHILEM state is pushed, and execution | |
3892 | passes to A or B depending on greediness and the current count. | |
2ab05381 | 3893 | |
c476f425 DM |
3894 | For example, if matching against the string a1a2a3b (where the aN are |
3895 | substrings that match /A/), then the match progresses as follows: (the | |
3896 | pushed states are interspersed with the bits of strings matched so far): | |
2ab05381 | 3897 | |
c476f425 DM |
3898 | <CURLYX cnt=-1> |
3899 | <CURLYX cnt=0><WHILEM> | |
3900 | <CURLYX cnt=1><WHILEM> a1 <WHILEM> | |
3901 | <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM> | |
3902 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> | |
3903 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b | |
2ab05381 | 3904 | |
c476f425 DM |
3905 | (Contrast this with something like CURLYM, which maintains only a single |
3906 | backtrack state: | |
2ab05381 | 3907 | |
c476f425 DM |
3908 | <CURLYM cnt=0> a1 |
3909 | a1 <CURLYM cnt=1> a2 | |
3910 | a1 a2 <CURLYM cnt=2> a3 | |
3911 | a1 a2 a3 <CURLYM cnt=3> b | |
3912 | ) | |
2ab05381 | 3913 | |
c476f425 DM |
3914 | Each WHILEM state block marks a point to backtrack to upon partial failure |
3915 | of A or B, and also contains some minor state data related to that | |
3916 | iteration. The CURLYX block, pointed to by cur_curlyx, contains the | |
3917 | overall state, such as the count, and pointers to the A and B ops. | |
2ab05381 | 3918 | |
c476f425 DM |
3919 | This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx |
3920 | must always point to the *current* CURLYX block, the rules are: | |
2ab05381 | 3921 | |
c476f425 DM |
3922 | When executing CURLYX, save the old cur_curlyx in the CURLYX state block, |
3923 | and set cur_curlyx to point the new block. | |
2ab05381 | 3924 | |
c476f425 DM |
3925 | When popping the CURLYX block after a successful or unsuccessful match, |
3926 | restore the previous cur_curlyx. | |
2ab05381 | 3927 | |
c476f425 DM |
3928 | When WHILEM is about to execute B, save the current cur_curlyx, and set it |
3929 | to the outer one saved in the CURLYX block. | |
2ab05381 | 3930 | |
c476f425 DM |
3931 | When popping the WHILEM block after a successful or unsuccessful B match, |
3932 | restore the previous cur_curlyx. | |
2ab05381 | 3933 | |
c476f425 DM |
3934 | Here's an example for the pattern (AI* BI)*BO |
3935 | I and O refer to inner and outer, C and W refer to CURLYX and WHILEM: | |
2ab05381 | 3936 | |
c476f425 DM |
3937 | cur_ |
3938 | curlyx backtrack stack | |
3939 | ------ --------------- | |
3940 | NULL | |
3941 | CO <CO prev=NULL> <WO> | |
3942 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3943 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi | |
3944 | NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo | |
2ab05381 | 3945 | |
c476f425 DM |
3946 | At this point the pattern succeeds, and we work back down the stack to |
3947 | clean up, restoring as we go: | |
95b24440 | 3948 | |
c476f425 DM |
3949 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi |
3950 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
3951 | CO <CO prev=NULL> <WO> | |
3952 | NULL | |
a0374537 | 3953 | |
c476f425 DM |
3954 | *******************************************************************/ |
3955 | ||
3956 | #define ST st->u.curlyx | |
3957 | ||
3958 | case CURLYX: /* start of /A*B/ (for complex A) */ | |
3959 | { | |
3960 | /* No need to save/restore up to this paren */ | |
3961 | I32 parenfloor = scan->flags; | |
3962 | ||
3963 | assert(next); /* keep Coverity happy */ | |
3964 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
3965 | next += ARG(next); | |
3966 | ||
3967 | /* XXXX Probably it is better to teach regpush to support | |
3968 | parenfloor > PL_regsize... */ | |
3969 | if (parenfloor > (I32)*PL_reglastparen) | |
3970 | parenfloor = *PL_reglastparen; /* Pessimization... */ | |
3971 | ||
3972 | ST.prev_curlyx= cur_curlyx; | |
3973 | cur_curlyx = st; | |
3974 | ST.cp = PL_savestack_ix; | |
3975 | ||
3976 | /* these fields contain the state of the current curly. | |
3977 | * they are accessed by subsequent WHILEMs */ | |
3978 | ST.parenfloor = parenfloor; | |
3979 | ST.min = ARG1(scan); | |
3980 | ST.max = ARG2(scan); | |
3981 | ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
3982 | ST.B = next; | |
24d3c4a9 DM |
3983 | ST.minmod = minmod; |
3984 | minmod = 0; | |
c476f425 DM |
3985 | ST.count = -1; /* this will be updated by WHILEM */ |
3986 | ST.lastloc = NULL; /* this will be updated by WHILEM */ | |
3987 | ||
3988 | PL_reginput = locinput; | |
3989 | PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next)); | |
5f66b61c | 3990 | /* NOTREACHED */ |
c476f425 | 3991 | } |
a0d0e21e | 3992 | |
c476f425 | 3993 | case CURLYX_end: /* just finished matching all of A*B */ |
c476f425 DM |
3994 | cur_curlyx = ST.prev_curlyx; |
3995 | sayYES; | |
3996 | /* NOTREACHED */ | |
a0d0e21e | 3997 | |
c476f425 DM |
3998 | case CURLYX_end_fail: /* just failed to match all of A*B */ |
3999 | regcpblow(ST.cp); | |
4000 | cur_curlyx = ST.prev_curlyx; | |
4001 | sayNO; | |
4002 | /* NOTREACHED */ | |
4633a7c4 | 4003 | |
a0d0e21e | 4004 | |
c476f425 DM |
4005 | #undef ST |
4006 | #define ST st->u.whilem | |
4007 | ||
4008 | case WHILEM: /* just matched an A in /A*B/ (for complex A) */ | |
4009 | { | |
4010 | /* see the discussion above about CURLYX/WHILEM */ | |
c476f425 DM |
4011 | I32 n; |
4012 | assert(cur_curlyx); /* keep Coverity happy */ | |
4013 | n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */ | |
4014 | ST.save_lastloc = cur_curlyx->u.curlyx.lastloc; | |
4015 | ST.cache_offset = 0; | |
4016 | ST.cache_mask = 0; | |
4017 | ||
4018 | PL_reginput = locinput; | |
4019 | ||
4020 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4021 | "%*s whilem: matched %ld out of %ld..%ld\n", | |
4022 | REPORT_CODE_OFF+depth*2, "", (long)n, | |
4023 | (long)cur_curlyx->u.curlyx.min, | |
4024 | (long)cur_curlyx->u.curlyx.max) | |
4025 | ); | |
a0d0e21e | 4026 | |
c476f425 | 4027 | /* First just match a string of min A's. */ |
a0d0e21e | 4028 | |
c476f425 DM |
4029 | if (n < cur_curlyx->u.curlyx.min) { |
4030 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4031 | PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A); | |
4032 | /* NOTREACHED */ | |
4033 | } | |
4034 | ||
4035 | /* If degenerate A matches "", assume A done. */ | |
4036 | ||
4037 | if (locinput == cur_curlyx->u.curlyx.lastloc) { | |
4038 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4039 | "%*s whilem: empty match detected, trying continuation...\n", | |
4040 | REPORT_CODE_OFF+depth*2, "") | |
4041 | ); | |
4042 | goto do_whilem_B_max; | |
4043 | } | |
4044 | ||
4045 | /* super-linear cache processing */ | |
4046 | ||
4047 | if (scan->flags) { | |
a0d0e21e | 4048 | |
2c2d71f5 | 4049 | if (!PL_reg_maxiter) { |
c476f425 DM |
4050 | /* start the countdown: Postpone detection until we |
4051 | * know the match is not *that* much linear. */ | |
2c2d71f5 | 4052 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
66bf836d DM |
4053 | /* possible overflow for long strings and many CURLYX's */ |
4054 | if (PL_reg_maxiter < 0) | |
4055 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
4056 | PL_reg_leftiter = PL_reg_maxiter; |
4057 | } | |
c476f425 | 4058 | |
2c2d71f5 | 4059 | if (PL_reg_leftiter-- == 0) { |
c476f425 | 4060 | /* initialise cache */ |
3298f257 | 4061 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 4062 | if (PL_reg_poscache) { |
eb160463 | 4063 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
4064 | Renew(PL_reg_poscache, size, char); |
4065 | PL_reg_poscache_size = size; | |
4066 | } | |
4067 | Zero(PL_reg_poscache, size, char); | |
4068 | } | |
4069 | else { | |
4070 | PL_reg_poscache_size = size; | |
a02a5408 | 4071 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 4072 | } |
c476f425 DM |
4073 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
4074 | "%swhilem: Detected a super-linear match, switching on caching%s...\n", | |
4075 | PL_colors[4], PL_colors[5]) | |
4076 | ); | |
2c2d71f5 | 4077 | } |
c476f425 | 4078 | |
2c2d71f5 | 4079 | if (PL_reg_leftiter < 0) { |
c476f425 DM |
4080 | /* have we already failed at this position? */ |
4081 | I32 offset, mask; | |
4082 | offset = (scan->flags & 0xf) - 1 | |
4083 | + (locinput - PL_bostr) * (scan->flags>>4); | |
4084 | mask = 1 << (offset % 8); | |
4085 | offset /= 8; | |
4086 | if (PL_reg_poscache[offset] & mask) { | |
4087 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4088 | "%*s whilem: (cache) already tried at this position...\n", | |
4089 | REPORT_CODE_OFF+depth*2, "") | |
2c2d71f5 | 4090 | ); |
3298f257 | 4091 | sayNO; /* cache records failure */ |
2c2d71f5 | 4092 | } |
c476f425 DM |
4093 | ST.cache_offset = offset; |
4094 | ST.cache_mask = mask; | |
2c2d71f5 | 4095 | } |
c476f425 | 4096 | } |
2c2d71f5 | 4097 | |
c476f425 | 4098 | /* Prefer B over A for minimal matching. */ |
a687059c | 4099 | |
c476f425 DM |
4100 | if (cur_curlyx->u.curlyx.minmod) { |
4101 | ST.save_curlyx = cur_curlyx; | |
4102 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4103 | ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor); | |
4104 | REGCP_SET(ST.lastcp); | |
4105 | PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B); | |
4106 | /* NOTREACHED */ | |
4107 | } | |
a0d0e21e | 4108 | |
c476f425 DM |
4109 | /* Prefer A over B for maximal matching. */ |
4110 | ||
4111 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ | |
4112 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4113 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4114 | REGCP_SET(ST.lastcp); | |
4115 | PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A); | |
4116 | /* NOTREACHED */ | |
4117 | } | |
4118 | goto do_whilem_B_max; | |
4119 | } | |
4120 | /* NOTREACHED */ | |
4121 | ||
4122 | case WHILEM_B_min: /* just matched B in a minimal match */ | |
4123 | case WHILEM_B_max: /* just matched B in a maximal match */ | |
4124 | cur_curlyx = ST.save_curlyx; | |
4125 | sayYES; | |
4126 | /* NOTREACHED */ | |
4127 | ||
4128 | case WHILEM_B_max_fail: /* just failed to match B in a maximal match */ | |
4129 | cur_curlyx = ST.save_curlyx; | |
4130 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4131 | cur_curlyx->u.curlyx.count--; | |
4132 | CACHEsayNO; | |
4133 | /* NOTREACHED */ | |
4134 | ||
4135 | case WHILEM_A_min_fail: /* just failed to match A in a minimal match */ | |
4136 | REGCP_UNWIND(ST.lastcp); | |
4137 | regcppop(rex); | |
4138 | /* FALL THROUGH */ | |
4139 | case WHILEM_A_pre_fail: /* just failed to match even minimal A */ | |
4140 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4141 | cur_curlyx->u.curlyx.count--; | |
4142 | CACHEsayNO; | |
4143 | /* NOTREACHED */ | |
4144 | ||
4145 | case WHILEM_A_max_fail: /* just failed to match A in a maximal match */ | |
4146 | REGCP_UNWIND(ST.lastcp); | |
4147 | regcppop(rex); /* Restore some previous $<digit>s? */ | |
4148 | PL_reginput = locinput; | |
4149 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4150 | "%*s whilem: failed, trying continuation...\n", | |
4151 | REPORT_CODE_OFF+depth*2, "") | |
4152 | ); | |
4153 | do_whilem_B_max: | |
4154 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4155 | && ckWARN(WARN_REGEXP) | |
4156 | && !(PL_reg_flags & RF_warned)) | |
4157 | { | |
4158 | PL_reg_flags |= RF_warned; | |
4159 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", | |
4160 | "Complex regular subexpression recursion", | |
4161 | REG_INFTY - 1); | |
4162 | } | |
4163 | ||
4164 | /* now try B */ | |
4165 | ST.save_curlyx = cur_curlyx; | |
4166 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4167 | PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B); | |
4168 | /* NOTREACHED */ | |
4169 | ||
4170 | case WHILEM_B_min_fail: /* just failed to match B in a minimal match */ | |
4171 | cur_curlyx = ST.save_curlyx; | |
4172 | REGCP_UNWIND(ST.lastcp); | |
4173 | regcppop(rex); | |
4174 | ||
4175 | if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) { | |
4176 | /* Maximum greed exceeded */ | |
4177 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4178 | && ckWARN(WARN_REGEXP) | |
4179 | && !(PL_reg_flags & RF_warned)) | |
4180 | { | |
3280af22 | 4181 | PL_reg_flags |= RF_warned; |
c476f425 DM |
4182 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), |
4183 | "%s limit (%d) exceeded", | |
4184 | "Complex regular subexpression recursion", | |
4185 | REG_INFTY - 1); | |
a0d0e21e | 4186 | } |
c476f425 | 4187 | cur_curlyx->u.curlyx.count--; |
3ab3c9b4 | 4188 | CACHEsayNO; |
a0d0e21e | 4189 | } |
c476f425 DM |
4190 | |
4191 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4192 | "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "") | |
4193 | ); | |
4194 | /* Try grabbing another A and see if it helps. */ | |
4195 | PL_reginput = locinput; | |
4196 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4197 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4198 | REGCP_SET(ST.lastcp); | |
4199 | PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A); | |
5f66b61c | 4200 | /* NOTREACHED */ |
40a82448 DM |
4201 | |
4202 | #undef ST | |
4203 | #define ST st->u.branch | |
4204 | ||
4205 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
4206 | next = scan + ARG(scan); |
4207 | if (next == scan) | |
4208 | next = NULL; | |
40a82448 DM |
4209 | scan = NEXTOPER(scan); |
4210 | /* FALL THROUGH */ | |
c277df42 | 4211 | |
40a82448 DM |
4212 | case BRANCH: /* /(...|A|...)/ */ |
4213 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
40a82448 DM |
4214 | ST.lastparen = *PL_reglastparen; |
4215 | ST.next_branch = next; | |
4216 | REGCP_SET(ST.cp); | |
4217 | PL_reginput = locinput; | |
02db2b7b | 4218 | |
40a82448 | 4219 | /* Now go into the branch */ |
5d458dd8 YO |
4220 | if (has_cutgroup) { |
4221 | PUSH_YES_STATE_GOTO(BRANCH_next, scan); | |
4222 | } else { | |
4223 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
4224 | } | |
40a82448 | 4225 | /* NOTREACHED */ |
5d458dd8 YO |
4226 | case CUTGROUP: |
4227 | PL_reginput = locinput; | |
4228 | sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL : | |
f8fc2ecf | 4229 | (SV*)rexi->data->data[ ARG( scan ) ]; |
5d458dd8 YO |
4230 | PUSH_STATE_GOTO(CUTGROUP_next,next); |
4231 | /* NOTREACHED */ | |
4232 | case CUTGROUP_next_fail: | |
4233 | do_cutgroup = 1; | |
4234 | no_final = 1; | |
4235 | if (st->u.mark.mark_name) | |
4236 | sv_commit = st->u.mark.mark_name; | |
4237 | sayNO; | |
4238 | /* NOTREACHED */ | |
4239 | case BRANCH_next: | |
4240 | sayYES; | |
4241 | /* NOTREACHED */ | |
40a82448 | 4242 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ |
5d458dd8 YO |
4243 | if (do_cutgroup) { |
4244 | do_cutgroup = 0; | |
4245 | no_final = 0; | |
4246 | } | |
40a82448 DM |
4247 | REGCP_UNWIND(ST.cp); |
4248 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
f0ab9afb | 4249 | PL_regoffs[n].end = -1; |
40a82448 | 4250 | *PL_reglastparen = n; |
0a4db386 | 4251 | /*dmq: *PL_reglastcloseparen = n; */ |
40a82448 DM |
4252 | scan = ST.next_branch; |
4253 | /* no more branches? */ | |
5d458dd8 YO |
4254 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) { |
4255 | DEBUG_EXECUTE_r({ | |
4256 | PerlIO_printf( Perl_debug_log, | |
4257 | "%*s %sBRANCH failed...%s\n", | |
4258 | REPORT_CODE_OFF+depth*2, "", | |
4259 | PL_colors[4], | |
4260 | PL_colors[5] ); | |
4261 | }); | |
4262 | sayNO_SILENT; | |
4263 | } | |
40a82448 DM |
4264 | continue; /* execute next BRANCH[J] op */ |
4265 | /* NOTREACHED */ | |
4266 | ||
a0d0e21e | 4267 | case MINMOD: |
24d3c4a9 | 4268 | minmod = 1; |
a0d0e21e | 4269 | break; |
40a82448 DM |
4270 | |
4271 | #undef ST | |
4272 | #define ST st->u.curlym | |
4273 | ||
4274 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
4275 | ||
4276 | /* This is an optimisation of CURLYX that enables us to push | |
4277 | * only a single backtracking state, no matter now many matches | |
4278 | * there are in {m,n}. It relies on the pattern being constant | |
4279 | * length, with no parens to influence future backrefs | |
4280 | */ | |
4281 | ||
4282 | ST.me = scan; | |
dc45a647 | 4283 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
4284 | |
4285 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
4286 | if (ST.me->flags) { | |
3b6647e0 | 4287 | U32 paren = ST.me->flags; |
40a82448 DM |
4288 | if (paren > PL_regsize) |
4289 | PL_regsize = paren; | |
3b6647e0 | 4290 | if (paren > *PL_reglastparen) |
40a82448 | 4291 | *PL_reglastparen = paren; |
c277df42 | 4292 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 4293 | } |
40a82448 DM |
4294 | ST.A = scan; |
4295 | ST.B = next; | |
4296 | ST.alen = 0; | |
4297 | ST.count = 0; | |
24d3c4a9 DM |
4298 | ST.minmod = minmod; |
4299 | minmod = 0; | |
40a82448 DM |
4300 | ST.c1 = CHRTEST_UNINIT; |
4301 | REGCP_SET(ST.cp); | |
6407bf3b | 4302 | |
40a82448 DM |
4303 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
4304 | goto curlym_do_B; | |
4305 | ||
4306 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 4307 | PL_reginput = locinput; |
40a82448 DM |
4308 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
4309 | /* NOTREACHED */ | |
5f80c4cf | 4310 | |
40a82448 DM |
4311 | case CURLYM_A: /* we've just matched an A */ |
4312 | locinput = st->locinput; | |
4313 | nextchr = UCHARAT(locinput); | |
4314 | ||
4315 | ST.count++; | |
4316 | /* after first match, determine A's length: u.curlym.alen */ | |
4317 | if (ST.count == 1) { | |
4318 | if (PL_reg_match_utf8) { | |
4319 | char *s = locinput; | |
4320 | while (s < PL_reginput) { | |
4321 | ST.alen++; | |
4322 | s += UTF8SKIP(s); | |
4323 | } | |
4324 | } | |
4325 | else { | |
4326 | ST.alen = PL_reginput - locinput; | |
4327 | } | |
4328 | if (ST.alen == 0) | |
4329 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
4330 | } | |
0cadcf80 DM |
4331 | DEBUG_EXECUTE_r( |
4332 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 4333 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 4334 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 4335 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
4336 | ); |
4337 | ||
40a82448 | 4338 | locinput = PL_reginput; |
0a4db386 YO |
4339 | |
4340 | if (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4341 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4342 | goto fake_end; |
4343 | ||
4344 | if ( ST.count < (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)) ) | |
40a82448 DM |
4345 | goto curlym_do_A; /* try to match another A */ |
4346 | goto curlym_do_B; /* try to match B */ | |
5f80c4cf | 4347 | |
40a82448 DM |
4348 | case CURLYM_A_fail: /* just failed to match an A */ |
4349 | REGCP_UNWIND(ST.cp); | |
0a4db386 YO |
4350 | |
4351 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ | |
4352 | || (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4353 | cur_eval->u.eval.close_paren == (U32)ST.me->flags)) |
40a82448 | 4354 | sayNO; |
0cadcf80 | 4355 | |
40a82448 DM |
4356 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
4357 | PL_reginput = locinput; | |
4358 | if (ST.c1 == CHRTEST_UNINIT) { | |
4359 | /* calculate c1 and c2 for possible match of 1st char | |
4360 | * following curly */ | |
4361 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
4362 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
4363 | regnode *text_node = ST.B; | |
4364 | if (! HAS_TEXT(text_node)) | |
4365 | FIND_NEXT_IMPT(text_node); | |
ee9b8eae YO |
4366 | /* this used to be |
4367 | ||
4368 | (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT) | |
4369 | ||
4370 | But the former is redundant in light of the latter. | |
4371 | ||
4372 | if this changes back then the macro for | |
4373 | IS_TEXT and friends need to change. | |
4374 | */ | |
4375 | if (PL_regkind[OP(text_node)] == EXACT) | |
40a82448 | 4376 | { |
ee9b8eae | 4377 | |
40a82448 DM |
4378 | ST.c1 = (U8)*STRING(text_node); |
4379 | ST.c2 = | |
ee9b8eae | 4380 | (IS_TEXTF(text_node)) |
40a82448 | 4381 | ? PL_fold[ST.c1] |
ee9b8eae | 4382 | : (IS_TEXTFL(text_node)) |
40a82448 DM |
4383 | ? PL_fold_locale[ST.c1] |
4384 | : ST.c1; | |
c277df42 | 4385 | } |
c277df42 | 4386 | } |
40a82448 DM |
4387 | } |
4388 | ||
4389 | DEBUG_EXECUTE_r( | |
4390 | PerlIO_printf(Perl_debug_log, | |
4391 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 4392 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
4393 | "", (IV)ST.count) |
4394 | ); | |
4395 | if (ST.c1 != CHRTEST_VOID | |
4396 | && UCHARAT(PL_reginput) != ST.c1 | |
4397 | && UCHARAT(PL_reginput) != ST.c2) | |
4398 | { | |
4399 | /* simulate B failing */ | |
3e901dc0 YO |
4400 | DEBUG_OPTIMISE_r( |
4401 | PerlIO_printf(Perl_debug_log, | |
4402 | "%*s CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n", | |
4403 | (int)(REPORT_CODE_OFF+(depth*2)),"", | |
4404 | (IV)ST.c1,(IV)ST.c2 | |
4405 | )); | |
40a82448 DM |
4406 | state_num = CURLYM_B_fail; |
4407 | goto reenter_switch; | |
4408 | } | |
4409 | ||
4410 | if (ST.me->flags) { | |
4411 | /* mark current A as captured */ | |
4412 | I32 paren = ST.me->flags; | |
4413 | if (ST.count) { | |
f0ab9afb | 4414 | PL_regoffs[paren].start |
40a82448 | 4415 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; |
f0ab9afb | 4416 | PL_regoffs[paren].end = PL_reginput - PL_bostr; |
0a4db386 | 4417 | /*dmq: *PL_reglastcloseparen = paren; */ |
c277df42 | 4418 | } |
40a82448 | 4419 | else |
f0ab9afb | 4420 | PL_regoffs[paren].end = -1; |
0a4db386 | 4421 | if (cur_eval && cur_eval->u.eval.close_paren && |
24b23f37 | 4422 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4423 | { |
4424 | if (ST.count) | |
4425 | goto fake_end; | |
4426 | else | |
4427 | sayNO; | |
4428 | } | |
c277df42 | 4429 | } |
0a4db386 | 4430 | |
40a82448 | 4431 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4432 | /* NOTREACHED */ |
40a82448 DM |
4433 | |
4434 | case CURLYM_B_fail: /* just failed to match a B */ | |
4435 | REGCP_UNWIND(ST.cp); | |
4436 | if (ST.minmod) { | |
4437 | if (ST.count == ARG2(ST.me) /* max */) | |
4438 | sayNO; | |
4439 | goto curlym_do_A; /* try to match a further A */ | |
4440 | } | |
4441 | /* backtrack one A */ | |
4442 | if (ST.count == ARG1(ST.me) /* min */) | |
4443 | sayNO; | |
4444 | ST.count--; | |
4445 | locinput = HOPc(locinput, -ST.alen); | |
4446 | goto curlym_do_B; /* try to match B */ | |
4447 | ||
c255a977 DM |
4448 | #undef ST |
4449 | #define ST st->u.curly | |
40a82448 | 4450 | |
c255a977 DM |
4451 | #define CURLY_SETPAREN(paren, success) \ |
4452 | if (paren) { \ | |
4453 | if (success) { \ | |
f0ab9afb NC |
4454 | PL_regoffs[paren].start = HOPc(locinput, -1) - PL_bostr; \ |
4455 | PL_regoffs[paren].end = locinput - PL_bostr; \ | |
0a4db386 | 4456 | *PL_reglastcloseparen = paren; \ |
c255a977 DM |
4457 | } \ |
4458 | else \ | |
f0ab9afb | 4459 | PL_regoffs[paren].end = -1; \ |
c255a977 DM |
4460 | } |
4461 | ||
4462 | case STAR: /* /A*B/ where A is width 1 */ | |
4463 | ST.paren = 0; | |
4464 | ST.min = 0; | |
4465 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4466 | scan = NEXTOPER(scan); |
4467 | goto repeat; | |
c255a977 DM |
4468 | case PLUS: /* /A+B/ where A is width 1 */ |
4469 | ST.paren = 0; | |
4470 | ST.min = 1; | |
4471 | ST.max = REG_INFTY; | |
c277df42 | 4472 | scan = NEXTOPER(scan); |
c255a977 DM |
4473 | goto repeat; |
4474 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4475 | ST.paren = scan->flags; /* Which paren to set */ | |
4476 | if (ST.paren > PL_regsize) | |
4477 | PL_regsize = ST.paren; | |
3b6647e0 | 4478 | if (ST.paren > *PL_reglastparen) |
c255a977 DM |
4479 | *PL_reglastparen = ST.paren; |
4480 | ST.min = ARG1(scan); /* min to match */ | |
4481 | ST.max = ARG2(scan); /* max to match */ | |
0a4db386 | 4482 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4483 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4484 | ST.min=1; |
4485 | ST.max=1; | |
4486 | } | |
c255a977 DM |
4487 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
4488 | goto repeat; | |
4489 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4490 | ST.paren = 0; | |
4491 | ST.min = ARG1(scan); /* min to match */ | |
4492 | ST.max = ARG2(scan); /* max to match */ | |
4493 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4494 | repeat: |
a0d0e21e LW |
4495 | /* |
4496 | * Lookahead to avoid useless match attempts | |
4497 | * when we know what character comes next. | |
c255a977 | 4498 | * |
5f80c4cf JP |
4499 | * Used to only do .*x and .*?x, but now it allows |
4500 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4501 | * of the quantifier and the EXACT-like node. -- japhy | |
4502 | */ | |
4503 | ||
c255a977 DM |
4504 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4505 | sayNO; | |
cca55fe3 | 4506 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4507 | U8 *s; |
4508 | regnode *text_node = next; | |
4509 | ||
3dab1dad YO |
4510 | if (! HAS_TEXT(text_node)) |
4511 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4512 | |
9e137952 | 4513 | if (! HAS_TEXT(text_node)) |
c255a977 | 4514 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4515 | else { |
ee9b8eae | 4516 | if ( PL_regkind[OP(text_node)] != EXACT ) { |
c255a977 | 4517 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4518 | goto assume_ok_easy; |
cca55fe3 | 4519 | } |
be8e71aa YO |
4520 | else |
4521 | s = (U8*)STRING(text_node); | |
ee9b8eae YO |
4522 | |
4523 | /* Currently we only get here when | |
4524 | ||
4525 | PL_rekind[OP(text_node)] == EXACT | |
4526 | ||
4527 | if this changes back then the macro for IS_TEXT and | |
4528 | friends need to change. */ | |
5f80c4cf | 4529 | if (!UTF) { |
c255a977 | 4530 | ST.c2 = ST.c1 = *s; |
ee9b8eae | 4531 | if (IS_TEXTF(text_node)) |
c255a977 | 4532 | ST.c2 = PL_fold[ST.c1]; |
ee9b8eae | 4533 | else if (IS_TEXTFL(text_node)) |
c255a977 | 4534 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4535 | } |
5f80c4cf | 4536 | else { /* UTF */ |
ee9b8eae | 4537 | if (IS_TEXTF(text_node)) { |
a2a2844f | 4538 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4539 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4540 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4541 | |
4542 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4543 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4544 | #ifdef EBCDIC |
4545 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4546 | ckWARN(WARN_UTF8) ? | |
4547 | 0 : UTF8_ALLOW_ANY); | |
4548 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4549 | ckWARN(WARN_UTF8) ? | |
4550 | 0 : UTF8_ALLOW_ANY); | |
4551 | #else | |
c255a977 | 4552 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4553 | uniflags); |
c255a977 | 4554 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4555 | uniflags); |
4556 | #endif | |
5f80c4cf JP |
4557 | } |
4558 | else { | |
c255a977 | 4559 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4560 | uniflags); |
5f80c4cf | 4561 | } |
1aa99e6b IH |
4562 | } |
4563 | } | |
bbce6d69 | 4564 | } |
a0d0e21e | 4565 | else |
c255a977 | 4566 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4567 | assume_ok_easy: |
c255a977 DM |
4568 | |
4569 | ST.A = scan; | |
4570 | ST.B = next; | |
3280af22 | 4571 | PL_reginput = locinput; |
24d3c4a9 DM |
4572 | if (minmod) { |
4573 | minmod = 0; | |
e2e6a0f1 | 4574 | if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min) |
4633a7c4 | 4575 | sayNO; |
c255a977 | 4576 | ST.count = ST.min; |
a0ed51b3 | 4577 | locinput = PL_reginput; |
c255a977 DM |
4578 | REGCP_SET(ST.cp); |
4579 | if (ST.c1 == CHRTEST_VOID) | |
4580 | goto curly_try_B_min; | |
4581 | ||
4582 | ST.oldloc = locinput; | |
4583 | ||
4584 | /* set ST.maxpos to the furthest point along the | |
4585 | * string that could possibly match */ | |
4586 | if (ST.max == REG_INFTY) { | |
4587 | ST.maxpos = PL_regeol - 1; | |
4588 | if (do_utf8) | |
4589 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4590 | ST.maxpos--; | |
4591 | } | |
4592 | else if (do_utf8) { | |
4593 | int m = ST.max - ST.min; | |
4594 | for (ST.maxpos = locinput; | |
4595 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4596 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4597 | } | |
4598 | else { | |
4599 | ST.maxpos = locinput + ST.max - ST.min; | |
4600 | if (ST.maxpos >= PL_regeol) | |
4601 | ST.maxpos = PL_regeol - 1; | |
4602 | } | |
4603 | goto curly_try_B_min_known; | |
4604 | ||
4605 | } | |
4606 | else { | |
e2e6a0f1 | 4607 | ST.count = regrepeat(rex, ST.A, ST.max, depth); |
c255a977 DM |
4608 | locinput = PL_reginput; |
4609 | if (ST.count < ST.min) | |
4610 | sayNO; | |
4611 | if ((ST.count > ST.min) | |
4612 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4613 | { | |
4614 | /* A{m,n} must come at the end of the string, there's | |
4615 | * no point in backing off ... */ | |
4616 | ST.min = ST.count; | |
4617 | /* ...except that $ and \Z can match before *and* after | |
4618 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4619 | We may back off by one in this case. */ | |
4620 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4621 | ST.min--; | |
4622 | } | |
4623 | REGCP_SET(ST.cp); | |
4624 | goto curly_try_B_max; | |
4625 | } | |
4626 | /* NOTREACHED */ | |
4627 | ||
4628 | ||
4629 | case CURLY_B_min_known_fail: | |
4630 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4631 | if (ST.paren && ST.count) | |
f0ab9afb | 4632 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4633 | |
4634 | PL_reginput = locinput; /* Could be reset... */ | |
4635 | REGCP_UNWIND(ST.cp); | |
4636 | /* Couldn't or didn't -- move forward. */ | |
4637 | ST.oldloc = locinput; | |
4638 | if (do_utf8) | |
4639 | locinput += UTF8SKIP(locinput); | |
4640 | else | |
4641 | locinput++; | |
4642 | ST.count++; | |
4643 | curly_try_B_min_known: | |
4644 | /* find the next place where 'B' could work, then call B */ | |
4645 | { | |
4646 | int n; | |
4647 | if (do_utf8) { | |
4648 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4649 | if (ST.c1 == ST.c2) { | |
4650 | STRLEN len; | |
4651 | /* set n to utf8_distance(oldloc, locinput) */ | |
4652 | while (locinput <= ST.maxpos && | |
4653 | utf8n_to_uvchr((U8*)locinput, | |
4654 | UTF8_MAXBYTES, &len, | |
4655 | uniflags) != (UV)ST.c1) { | |
4656 | locinput += len; | |
4657 | n++; | |
4658 | } | |
1aa99e6b IH |
4659 | } |
4660 | else { | |
c255a977 DM |
4661 | /* set n to utf8_distance(oldloc, locinput) */ |
4662 | while (locinput <= ST.maxpos) { | |
4663 | STRLEN len; | |
4664 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4665 | UTF8_MAXBYTES, &len, | |
4666 | uniflags); | |
4667 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4668 | break; | |
4669 | locinput += len; | |
4670 | n++; | |
1aa99e6b | 4671 | } |
0fe9bf95 IZ |
4672 | } |
4673 | } | |
c255a977 DM |
4674 | else { |
4675 | if (ST.c1 == ST.c2) { | |
4676 | while (locinput <= ST.maxpos && | |
4677 | UCHARAT(locinput) != ST.c1) | |
4678 | locinput++; | |
bbce6d69 | 4679 | } |
c255a977 DM |
4680 | else { |
4681 | while (locinput <= ST.maxpos | |
4682 | && UCHARAT(locinput) != ST.c1 | |
4683 | && UCHARAT(locinput) != ST.c2) | |
4684 | locinput++; | |
a0ed51b3 | 4685 | } |
c255a977 DM |
4686 | n = locinput - ST.oldloc; |
4687 | } | |
4688 | if (locinput > ST.maxpos) | |
4689 | sayNO; | |
4690 | /* PL_reginput == oldloc now */ | |
4691 | if (n) { | |
4692 | ST.count += n; | |
e2e6a0f1 | 4693 | if (regrepeat(rex, ST.A, n, depth) < n) |
4633a7c4 | 4694 | sayNO; |
a0d0e21e | 4695 | } |
c255a977 DM |
4696 | PL_reginput = locinput; |
4697 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4698 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4699 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4700 | goto fake_end; |
4701 | } | |
c255a977 | 4702 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); |
a0d0e21e | 4703 | } |
c255a977 DM |
4704 | /* NOTREACHED */ |
4705 | ||
4706 | ||
4707 | case CURLY_B_min_fail: | |
4708 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4709 | if (ST.paren && ST.count) | |
f0ab9afb | 4710 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4711 | |
4712 | REGCP_UNWIND(ST.cp); | |
4713 | /* failed -- move forward one */ | |
4714 | PL_reginput = locinput; | |
e2e6a0f1 | 4715 | if (regrepeat(rex, ST.A, 1, depth)) { |
c255a977 | 4716 | ST.count++; |
a0ed51b3 | 4717 | locinput = PL_reginput; |
c255a977 DM |
4718 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4719 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4720 | { |
c255a977 DM |
4721 | curly_try_B_min: |
4722 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4723 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4724 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4725 | goto fake_end; |
4726 | } | |
c255a977 | 4727 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); |
a0d0e21e LW |
4728 | } |
4729 | } | |
4633a7c4 | 4730 | sayNO; |
c255a977 DM |
4731 | /* NOTREACHED */ |
4732 | ||
4733 | ||
4734 | curly_try_B_max: | |
4735 | /* a successful greedy match: now try to match B */ | |
40d049e4 | 4736 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4737 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
40d049e4 YO |
4738 | goto fake_end; |
4739 | } | |
c255a977 DM |
4740 | { |
4741 | UV c = 0; | |
4742 | if (ST.c1 != CHRTEST_VOID) | |
4743 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4744 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4745 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4746 | /* If it could work, try it. */ |
4747 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4748 | CURLY_SETPAREN(ST.paren, ST.count); | |
4749 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4750 | /* NOTREACHED */ | |
4751 | } | |
4752 | } | |
4753 | /* FALL THROUGH */ | |
4754 | case CURLY_B_max_fail: | |
4755 | /* failed to find B in a greedy match */ | |
4756 | if (ST.paren && ST.count) | |
f0ab9afb | 4757 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4758 | |
4759 | REGCP_UNWIND(ST.cp); | |
4760 | /* back up. */ | |
4761 | if (--ST.count < ST.min) | |
4762 | sayNO; | |
4763 | PL_reginput = locinput = HOPc(locinput, -1); | |
4764 | goto curly_try_B_max; | |
4765 | ||
4766 | #undef ST | |
4767 | ||
a0d0e21e | 4768 | case END: |
6bda09f9 | 4769 | fake_end: |
faec1544 DM |
4770 | if (cur_eval) { |
4771 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4772 | I32 tmpix; | |
faec1544 DM |
4773 | st->u.eval.toggle_reg_flags |
4774 | = cur_eval->u.eval.toggle_reg_flags; | |
4775 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4776 | ||
4777 | st->u.eval.prev_rex = rex; /* inner */ | |
28d8d7f4 | 4778 | SETREX(rex,cur_eval->u.eval.prev_rex); |
f8fc2ecf | 4779 | rexi = RXi_GET(rex); |
faec1544 DM |
4780 | cur_curlyx = cur_eval->u.eval.prev_curlyx; |
4781 | ReREFCNT_inc(rex); | |
4782 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ | |
4783 | REGCP_SET(st->u.eval.lastcp); | |
4784 | PL_reginput = locinput; | |
4785 | ||
4786 | /* Restore parens of the outer rex without popping the | |
4787 | * savestack */ | |
4788 | tmpix = PL_savestack_ix; | |
4789 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4790 | regcppop(rex); | |
4791 | PL_savestack_ix = tmpix; | |
4792 | ||
4793 | st->u.eval.prev_eval = cur_eval; | |
4794 | cur_eval = cur_eval->u.eval.prev_eval; | |
4795 | DEBUG_EXECUTE_r( | |
2a49f0f5 JH |
4796 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n", |
4797 | REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval));); | |
e7707071 YO |
4798 | if ( nochange_depth ) |
4799 | nochange_depth--; | |
4800 | ||
4801 | PUSH_YES_STATE_GOTO(EVAL_AB, | |
faec1544 DM |
4802 | st->u.eval.prev_eval->u.eval.B); /* match B */ |
4803 | } | |
4804 | ||
3b0527fe | 4805 | if (locinput < reginfo->till) { |
a3621e74 | 4806 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4807 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4808 | PL_colors[4], | |
4809 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4810 | (long)(reginfo->till - PL_reg_starttry), |
7821416a | 4811 | PL_colors[5])); |
58e23c8d | 4812 | |
262b90c4 | 4813 | sayNO_SILENT; /* Cannot match: too short. */ |
7821416a IZ |
4814 | } |
4815 | PL_reginput = locinput; /* put where regtry can find it */ | |
262b90c4 | 4816 | sayYES; /* Success! */ |
dad79028 DM |
4817 | |
4818 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4819 | DEBUG_EXECUTE_r( | |
4820 | PerlIO_printf(Perl_debug_log, | |
4821 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4822 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4823 | PL_reginput = locinput; /* put where regtry can find it */ |
262b90c4 | 4824 | sayYES; /* Success! */ |
dad79028 | 4825 | |
40a82448 DM |
4826 | #undef ST |
4827 | #define ST st->u.ifmatch | |
4828 | ||
4829 | case SUSPEND: /* (?>A) */ | |
4830 | ST.wanted = 1; | |
9fe1d20c | 4831 | PL_reginput = locinput; |
9041c2e3 | 4832 | goto do_ifmatch; |
dad79028 | 4833 | |
40a82448 DM |
4834 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4835 | ST.wanted = 0; | |
dad79028 DM |
4836 | goto ifmatch_trivial_fail_test; |
4837 | ||
40a82448 DM |
4838 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4839 | ST.wanted = 1; | |
dad79028 | 4840 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4841 | if (scan->flags) { |
52657f30 | 4842 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4843 | if (!s) { |
4844 | /* trivial fail */ | |
24d3c4a9 DM |
4845 | if (logical) { |
4846 | logical = 0; | |
4847 | sw = 1 - (bool)ST.wanted; | |
dad79028 | 4848 | } |
40a82448 | 4849 | else if (ST.wanted) |
dad79028 DM |
4850 | sayNO; |
4851 | next = scan + ARG(scan); | |
4852 | if (next == scan) | |
4853 | next = NULL; | |
4854 | break; | |
4855 | } | |
efb30f32 | 4856 | PL_reginput = s; |
a0ed51b3 LW |
4857 | } |
4858 | else | |
4859 | PL_reginput = locinput; | |
4860 | ||
c277df42 | 4861 | do_ifmatch: |
40a82448 | 4862 | ST.me = scan; |
24d3c4a9 | 4863 | ST.logical = logical; |
40a82448 DM |
4864 | /* execute body of (?...A) */ |
4865 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
4866 | /* NOTREACHED */ | |
4867 | ||
4868 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
4869 | ST.wanted = !ST.wanted; | |
4870 | /* FALL THROUGH */ | |
4871 | ||
4872 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
24d3c4a9 DM |
4873 | if (ST.logical) { |
4874 | sw = (bool)ST.wanted; | |
40a82448 DM |
4875 | } |
4876 | else if (!ST.wanted) | |
4877 | sayNO; | |
4878 | ||
4879 | if (OP(ST.me) == SUSPEND) | |
4880 | locinput = PL_reginput; | |
4881 | else { | |
4882 | locinput = PL_reginput = st->locinput; | |
4883 | nextchr = UCHARAT(locinput); | |
4884 | } | |
4885 | scan = ST.me + ARG(ST.me); | |
4886 | if (scan == ST.me) | |
4887 | scan = NULL; | |
4888 | continue; /* execute B */ | |
4889 | ||
4890 | #undef ST | |
dad79028 | 4891 | |
c277df42 | 4892 | case LONGJMP: |
c277df42 IZ |
4893 | next = scan + ARG(scan); |
4894 | if (next == scan) | |
4895 | next = NULL; | |
a0d0e21e | 4896 | break; |
54612592 | 4897 | case COMMIT: |
e2e6a0f1 YO |
4898 | reginfo->cutpoint = PL_regeol; |
4899 | /* FALLTHROUGH */ | |
5d458dd8 | 4900 | case PRUNE: |
24b23f37 | 4901 | PL_reginput = locinput; |
e2e6a0f1 | 4902 | if (!scan->flags) |
f8fc2ecf | 4903 | sv_yes_mark = sv_commit = (SV*)rexi->data->data[ ARG( scan ) ]; |
54612592 YO |
4904 | PUSH_STATE_GOTO(COMMIT_next,next); |
4905 | /* NOTREACHED */ | |
4906 | case COMMIT_next_fail: | |
4907 | no_final = 1; | |
4908 | /* FALLTHROUGH */ | |
7f69552c YO |
4909 | case OPFAIL: |
4910 | sayNO; | |
e2e6a0f1 YO |
4911 | /* NOTREACHED */ |
4912 | ||
4913 | #define ST st->u.mark | |
4914 | case MARKPOINT: | |
4915 | ST.prev_mark = mark_state; | |
5d458dd8 | 4916 | ST.mark_name = sv_commit = sv_yes_mark |
f8fc2ecf | 4917 | = (SV*)rexi->data->data[ ARG( scan ) ]; |
e2e6a0f1 YO |
4918 | mark_state = st; |
4919 | ST.mark_loc = PL_reginput = locinput; | |
4920 | PUSH_YES_STATE_GOTO(MARKPOINT_next,next); | |
4921 | /* NOTREACHED */ | |
4922 | case MARKPOINT_next: | |
4923 | mark_state = ST.prev_mark; | |
4924 | sayYES; | |
4925 | /* NOTREACHED */ | |
4926 | case MARKPOINT_next_fail: | |
5d458dd8 | 4927 | if (popmark && sv_eq(ST.mark_name,popmark)) |
e2e6a0f1 YO |
4928 | { |
4929 | if (ST.mark_loc > startpoint) | |
4930 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
4931 | popmark = NULL; /* we found our mark */ | |
4932 | sv_commit = ST.mark_name; | |
4933 | ||
4934 | DEBUG_EXECUTE_r({ | |
5d458dd8 | 4935 | PerlIO_printf(Perl_debug_log, |
e2e6a0f1 YO |
4936 | "%*s %ssetting cutpoint to mark:%"SVf"...%s\n", |
4937 | REPORT_CODE_OFF+depth*2, "", | |
be2597df | 4938 | PL_colors[4], SVfARG(sv_commit), PL_colors[5]); |
e2e6a0f1 YO |
4939 | }); |
4940 | } | |
4941 | mark_state = ST.prev_mark; | |
5d458dd8 YO |
4942 | sv_yes_mark = mark_state ? |
4943 | mark_state->u.mark.mark_name : NULL; | |
e2e6a0f1 YO |
4944 | sayNO; |
4945 | /* NOTREACHED */ | |
5d458dd8 YO |
4946 | case SKIP: |
4947 | PL_reginput = locinput; | |
4948 | if (scan->flags) { | |
2bf803e2 | 4949 | /* (*SKIP) : if we fail we cut here*/ |
5d458dd8 | 4950 | ST.mark_name = NULL; |
e2e6a0f1 | 4951 | ST.mark_loc = locinput; |
5d458dd8 YO |
4952 | PUSH_STATE_GOTO(SKIP_next,next); |
4953 | } else { | |
2bf803e2 | 4954 | /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was, |
5d458dd8 YO |
4955 | otherwise do nothing. Meaning we need to scan |
4956 | */ | |
4957 | regmatch_state *cur = mark_state; | |
f8fc2ecf | 4958 | SV *find = (SV*)rexi->data->data[ ARG( scan ) ]; |
5d458dd8 YO |
4959 | |
4960 | while (cur) { | |
4961 | if ( sv_eq( cur->u.mark.mark_name, | |
4962 | find ) ) | |
4963 | { | |
4964 | ST.mark_name = find; | |
4965 | PUSH_STATE_GOTO( SKIP_next, next ); | |
4966 | } | |
4967 | cur = cur->u.mark.prev_mark; | |
4968 | } | |
e2e6a0f1 | 4969 | } |
2bf803e2 | 4970 | /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */ |
5d458dd8 YO |
4971 | break; |
4972 | case SKIP_next_fail: | |
4973 | if (ST.mark_name) { | |
4974 | /* (*CUT:NAME) - Set up to search for the name as we | |
4975 | collapse the stack*/ | |
4976 | popmark = ST.mark_name; | |
4977 | } else { | |
4978 | /* (*CUT) - No name, we cut here.*/ | |
e2e6a0f1 YO |
4979 | if (ST.mark_loc > startpoint) |
4980 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
5d458dd8 YO |
4981 | /* but we set sv_commit to latest mark_name if there |
4982 | is one so they can test to see how things lead to this | |
4983 | cut */ | |
4984 | if (mark_state) | |
4985 | sv_commit=mark_state->u.mark.mark_name; | |
4986 | } | |
e2e6a0f1 YO |
4987 | no_final = 1; |
4988 | sayNO; | |
4989 | /* NOTREACHED */ | |
4990 | #undef ST | |
32e6a07c YO |
4991 | case FOLDCHAR: |
4992 | n = ARG(scan); | |
81d4fa0f | 4993 | if ( n == (U32)what_len_TRICKYFOLD(locinput,do_utf8,ln) ) { |
e64b1bd1 YO |
4994 | locinput += ln; |
4995 | } else if ( 0xDF == n && !do_utf8 && !UTF ) { | |
4996 | sayNO; | |
4997 | } else { | |
4998 | U8 folded[UTF8_MAXBYTES_CASE+1]; | |
4999 | STRLEN foldlen; | |
5000 | const char * const l = locinput; | |
5001 | char *e = PL_regeol; | |
5002 | to_uni_fold(n, folded, &foldlen); | |
5003 | ||
59fe32ea | 5004 | if (ibcmp_utf8((const char*) folded, 0, foldlen, 1, |
e64b1bd1 | 5005 | l, &e, 0, do_utf8)) { |
32e6a07c | 5006 | sayNO; |
e64b1bd1 YO |
5007 | } |
5008 | locinput = e; | |
32e6a07c YO |
5009 | } |
5010 | nextchr = UCHARAT(locinput); | |
5011 | break; | |
e1d1eefb YO |
5012 | case LNBREAK: |
5013 | if ((n=is_LNBREAK(locinput,do_utf8))) { | |
5014 | locinput += n; | |
5015 | nextchr = UCHARAT(locinput); | |
5016 | } else | |
5017 | sayNO; | |
5018 | break; | |
5019 | ||
5020 | #define CASE_CLASS(nAmE) \ | |
5021 | case nAmE: \ | |
5022 | if ((n=is_##nAmE(locinput,do_utf8))) { \ | |
5023 | locinput += n; \ | |
5024 | nextchr = UCHARAT(locinput); \ | |
5025 | } else \ | |
5026 | sayNO; \ | |
5027 | break; \ | |
5028 | case N##nAmE: \ | |
5029 | if ((n=is_##nAmE(locinput,do_utf8))) { \ | |
5030 | sayNO; \ | |
5031 | } else { \ | |
5032 | locinput += UTF8SKIP(locinput); \ | |
5033 | nextchr = UCHARAT(locinput); \ | |
5034 | } \ | |
5035 | break | |
5036 | ||
5037 | CASE_CLASS(VERTWS); | |
5038 | CASE_CLASS(HORIZWS); | |
5039 | #undef CASE_CLASS | |
5040 | ||
a0d0e21e | 5041 | default: |
b900a521 | 5042 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 5043 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 5044 | Perl_croak(aTHX_ "regexp memory corruption"); |
5d458dd8 YO |
5045 | |
5046 | } /* end switch */ | |
95b24440 | 5047 | |
5d458dd8 YO |
5048 | /* switch break jumps here */ |
5049 | scan = next; /* prepare to execute the next op and ... */ | |
5050 | continue; /* ... jump back to the top, reusing st */ | |
95b24440 DM |
5051 | /* NOTREACHED */ |
5052 | ||
40a82448 DM |
5053 | push_yes_state: |
5054 | /* push a state that backtracks on success */ | |
5055 | st->u.yes.prev_yes_state = yes_state; | |
5056 | yes_state = st; | |
5057 | /* FALL THROUGH */ | |
5058 | push_state: | |
5059 | /* push a new regex state, then continue at scan */ | |
5060 | { | |
5061 | regmatch_state *newst; | |
5062 | ||
24b23f37 YO |
5063 | DEBUG_STACK_r({ |
5064 | regmatch_state *cur = st; | |
5065 | regmatch_state *curyes = yes_state; | |
5066 | int curd = depth; | |
5067 | regmatch_slab *slab = PL_regmatch_slab; | |
5068 | for (;curd > -1;cur--,curd--) { | |
5069 | if (cur < SLAB_FIRST(slab)) { | |
5070 | slab = slab->prev; | |
5071 | cur = SLAB_LAST(slab); | |
5072 | } | |
5073 | PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n", | |
5074 | REPORT_CODE_OFF + 2 + depth * 2,"", | |
13d6edb4 | 5075 | curd, PL_reg_name[cur->resume_state], |
24b23f37 YO |
5076 | (curyes == cur) ? "yes" : "" |
5077 | ); | |
5078 | if (curyes == cur) | |
5079 | curyes = cur->u.yes.prev_yes_state; | |
5080 | } | |
5081 | } else | |
5082 | DEBUG_STATE_pp("push") | |
5083 | ); | |
40a82448 | 5084 | depth++; |
40a82448 DM |
5085 | st->locinput = locinput; |
5086 | newst = st+1; | |
5087 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
5088 | newst = S_push_slab(aTHX); | |
5089 | PL_regmatch_state = newst; | |
786e8c11 | 5090 | |
40a82448 DM |
5091 | locinput = PL_reginput; |
5092 | nextchr = UCHARAT(locinput); | |
5093 | st = newst; | |
5094 | continue; | |
5095 | /* NOTREACHED */ | |
5096 | } | |
a0d0e21e | 5097 | } |
a687059c | 5098 | |
a0d0e21e LW |
5099 | /* |
5100 | * We get here only if there's trouble -- normally "case END" is | |
5101 | * the terminating point. | |
5102 | */ | |
cea2e8a9 | 5103 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 5104 | /*NOTREACHED*/ |
4633a7c4 LW |
5105 | sayNO; |
5106 | ||
262b90c4 | 5107 | yes: |
77cb431f DM |
5108 | if (yes_state) { |
5109 | /* we have successfully completed a subexpression, but we must now | |
5110 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 5111 | assert(st != yes_state); |
5bc10b2c DM |
5112 | #ifdef DEBUGGING |
5113 | while (st != yes_state) { | |
5114 | st--; | |
5115 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
5116 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5117 | st = SLAB_LAST(PL_regmatch_slab); | |
5118 | } | |
e2e6a0f1 | 5119 | DEBUG_STATE_r({ |
54612592 YO |
5120 | if (no_final) { |
5121 | DEBUG_STATE_pp("pop (no final)"); | |
5122 | } else { | |
5123 | DEBUG_STATE_pp("pop (yes)"); | |
5124 | } | |
e2e6a0f1 | 5125 | }); |
5bc10b2c DM |
5126 | depth--; |
5127 | } | |
5128 | #else | |
77cb431f DM |
5129 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
5130 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
5131 | { | |
5132 | /* not in this slab, pop slab */ | |
5133 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
5134 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5135 | st = SLAB_LAST(PL_regmatch_slab); | |
5136 | } | |
5137 | depth -= (st - yes_state); | |
5bc10b2c | 5138 | #endif |
77cb431f DM |
5139 | st = yes_state; |
5140 | yes_state = st->u.yes.prev_yes_state; | |
5141 | PL_regmatch_state = st; | |
24b23f37 | 5142 | |
5d458dd8 YO |
5143 | if (no_final) { |
5144 | locinput= st->locinput; | |
5145 | nextchr = UCHARAT(locinput); | |
5146 | } | |
54612592 | 5147 | state_num = st->resume_state + no_final; |
24d3c4a9 | 5148 | goto reenter_switch; |
77cb431f DM |
5149 | } |
5150 | ||
a3621e74 | 5151 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 5152 | PL_colors[4], PL_colors[5])); |
02db2b7b | 5153 | |
19b95bf0 DM |
5154 | if (PL_reg_eval_set) { |
5155 | /* each successfully executed (?{...}) block does the equivalent of | |
5156 | * local $^R = do {...} | |
5157 | * When popping the save stack, all these locals would be undone; | |
5158 | * bypass this by setting the outermost saved $^R to the latest | |
5159 | * value */ | |
5160 | if (oreplsv != GvSV(PL_replgv)) | |
5161 | sv_setsv(oreplsv, GvSV(PL_replgv)); | |
5162 | } | |
95b24440 | 5163 | result = 1; |
aa283a38 | 5164 | goto final_exit; |
4633a7c4 LW |
5165 | |
5166 | no: | |
a3621e74 | 5167 | DEBUG_EXECUTE_r( |
7821416a | 5168 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 5169 | "%*s %sfailed...%s\n", |
5bc10b2c | 5170 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 5171 | PL_colors[4], PL_colors[5]) |
7821416a | 5172 | ); |
aa283a38 | 5173 | |
262b90c4 | 5174 | no_silent: |
54612592 YO |
5175 | if (no_final) { |
5176 | if (yes_state) { | |
5177 | goto yes; | |
5178 | } else { | |
5179 | goto final_exit; | |
5180 | } | |
5181 | } | |
aa283a38 DM |
5182 | if (depth) { |
5183 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
5184 | st--; |
5185 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
5186 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5187 | st = SLAB_LAST(PL_regmatch_slab); | |
5188 | } | |
5189 | PL_regmatch_state = st; | |
40a82448 DM |
5190 | locinput= st->locinput; |
5191 | nextchr = UCHARAT(locinput); | |
5192 | ||
5bc10b2c DM |
5193 | DEBUG_STATE_pp("pop"); |
5194 | depth--; | |
262b90c4 DM |
5195 | if (yes_state == st) |
5196 | yes_state = st->u.yes.prev_yes_state; | |
5bc10b2c | 5197 | |
24d3c4a9 DM |
5198 | state_num = st->resume_state + 1; /* failure = success + 1 */ |
5199 | goto reenter_switch; | |
95b24440 | 5200 | } |
24d3c4a9 | 5201 | result = 0; |
aa283a38 | 5202 | |
262b90c4 | 5203 | final_exit: |
bbe252da | 5204 | if (rex->intflags & PREGf_VERBARG_SEEN) { |
5d458dd8 YO |
5205 | SV *sv_err = get_sv("REGERROR", 1); |
5206 | SV *sv_mrk = get_sv("REGMARK", 1); | |
5207 | if (result) { | |
e2e6a0f1 | 5208 | sv_commit = &PL_sv_no; |
5d458dd8 YO |
5209 | if (!sv_yes_mark) |
5210 | sv_yes_mark = &PL_sv_yes; | |
5211 | } else { | |
5212 | if (!sv_commit) | |
5213 | sv_commit = &PL_sv_yes; | |
5214 | sv_yes_mark = &PL_sv_no; | |
5215 | } | |
5216 | sv_setsv(sv_err, sv_commit); | |
5217 | sv_setsv(sv_mrk, sv_yes_mark); | |
e2e6a0f1 | 5218 | } |
19b95bf0 | 5219 | |
2f554ef7 DM |
5220 | /* clean up; in particular, free all slabs above current one */ |
5221 | LEAVE_SCOPE(oldsave); | |
5d9a96ca | 5222 | |
95b24440 | 5223 | return result; |
a687059c LW |
5224 | } |
5225 | ||
5226 | /* | |
5227 | - regrepeat - repeatedly match something simple, report how many | |
5228 | */ | |
5229 | /* | |
5230 | * [This routine now assumes that it will only match on things of length 1. | |
5231 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 5232 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 5233 | */ |
76e3520e | 5234 | STATIC I32 |
e2e6a0f1 | 5235 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth) |
a687059c | 5236 | { |
27da23d5 | 5237 | dVAR; |
a0d0e21e | 5238 | register char *scan; |
a0d0e21e | 5239 | register I32 c; |
3280af22 | 5240 | register char *loceol = PL_regeol; |
a0ed51b3 | 5241 | register I32 hardcount = 0; |
53c4c00c | 5242 | register bool do_utf8 = PL_reg_match_utf8; |
4f55667c SP |
5243 | #ifndef DEBUGGING |
5244 | PERL_UNUSED_ARG(depth); | |
5245 | #endif | |
a0d0e21e | 5246 | |
3280af22 | 5247 | scan = PL_reginput; |
faf11cac HS |
5248 | if (max == REG_INFTY) |
5249 | max = I32_MAX; | |
5250 | else if (max < loceol - scan) | |
7f596f4c | 5251 | loceol = scan + max; |
a0d0e21e | 5252 | switch (OP(p)) { |
22c35a8c | 5253 | case REG_ANY: |
1aa99e6b | 5254 | if (do_utf8) { |
ffc61ed2 | 5255 | loceol = PL_regeol; |
1aa99e6b | 5256 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
5257 | scan += UTF8SKIP(scan); |
5258 | hardcount++; | |
5259 | } | |
5260 | } else { | |
5261 | while (scan < loceol && *scan != '\n') | |
5262 | scan++; | |
a0ed51b3 LW |
5263 | } |
5264 | break; | |
ffc61ed2 | 5265 | case SANY: |
def8e4ea JH |
5266 | if (do_utf8) { |
5267 | loceol = PL_regeol; | |
a0804c9e | 5268 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
5269 | scan += UTF8SKIP(scan); |
5270 | hardcount++; | |
5271 | } | |
5272 | } | |
5273 | else | |
5274 | scan = loceol; | |
a0ed51b3 | 5275 | break; |
f33976b4 DB |
5276 | case CANY: |
5277 | scan = loceol; | |
5278 | break; | |
090f7165 JH |
5279 | case EXACT: /* length of string is 1 */ |
5280 | c = (U8)*STRING(p); | |
5281 | while (scan < loceol && UCHARAT(scan) == c) | |
5282 | scan++; | |
bbce6d69 | 5283 | break; |
5284 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 5285 | c = (U8)*STRING(p); |
bbce6d69 | 5286 | while (scan < loceol && |
22c35a8c | 5287 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 5288 | scan++; |
5289 | break; | |
5290 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 5291 | PL_reg_flags |= RF_tainted; |
cd439c50 | 5292 | c = (U8)*STRING(p); |
bbce6d69 | 5293 | while (scan < loceol && |
22c35a8c | 5294 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
5295 | scan++; |
5296 | break; | |
5297 | case ANYOF: | |
ffc61ed2 JH |
5298 | if (do_utf8) { |
5299 | loceol = PL_regeol; | |
cfc92286 | 5300 | while (hardcount < max && scan < loceol && |
32fc9b6a | 5301 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
5302 | scan += UTF8SKIP(scan); |
5303 | hardcount++; | |
5304 | } | |
5305 | } else { | |
32fc9b6a | 5306 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
5307 | scan++; |
5308 | } | |
a0d0e21e LW |
5309 | break; |
5310 | case ALNUM: | |
1aa99e6b | 5311 | if (do_utf8) { |
ffc61ed2 | 5312 | loceol = PL_regeol; |
1a4fad37 | 5313 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5314 | while (hardcount < max && scan < loceol && |
3568d838 | 5315 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5316 | scan += UTF8SKIP(scan); |
5317 | hardcount++; | |
5318 | } | |
5319 | } else { | |
5320 | while (scan < loceol && isALNUM(*scan)) | |
5321 | scan++; | |
a0ed51b3 LW |
5322 | } |
5323 | break; | |
bbce6d69 | 5324 | case ALNUML: |
3280af22 | 5325 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5326 | if (do_utf8) { |
ffc61ed2 | 5327 | loceol = PL_regeol; |
1aa99e6b IH |
5328 | while (hardcount < max && scan < loceol && |
5329 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5330 | scan += UTF8SKIP(scan); |
5331 | hardcount++; | |
5332 | } | |
5333 | } else { | |
5334 | while (scan < loceol && isALNUM_LC(*scan)) | |
5335 | scan++; | |
a0ed51b3 LW |
5336 | } |
5337 | break; | |
a0d0e21e | 5338 | case NALNUM: |
1aa99e6b | 5339 | if (do_utf8) { |
ffc61ed2 | 5340 | loceol = PL_regeol; |
1a4fad37 | 5341 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5342 | while (hardcount < max && scan < loceol && |
3568d838 | 5343 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5344 | scan += UTF8SKIP(scan); |
5345 | hardcount++; | |
5346 | } | |
5347 | } else { | |
5348 | while (scan < loceol && !isALNUM(*scan)) | |
5349 | scan++; | |
a0ed51b3 LW |
5350 | } |
5351 | break; | |
bbce6d69 | 5352 | case NALNUML: |
3280af22 | 5353 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5354 | if (do_utf8) { |
ffc61ed2 | 5355 | loceol = PL_regeol; |
1aa99e6b IH |
5356 | while (hardcount < max && scan < loceol && |
5357 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5358 | scan += UTF8SKIP(scan); |
5359 | hardcount++; | |
5360 | } | |
5361 | } else { | |
5362 | while (scan < loceol && !isALNUM_LC(*scan)) | |
5363 | scan++; | |
a0ed51b3 LW |
5364 | } |
5365 | break; | |
a0d0e21e | 5366 | case SPACE: |
1aa99e6b | 5367 | if (do_utf8) { |
ffc61ed2 | 5368 | loceol = PL_regeol; |
1a4fad37 | 5369 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5370 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5371 | (*scan == ' ' || |
5372 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5373 | scan += UTF8SKIP(scan); |
5374 | hardcount++; | |
5375 | } | |
5376 | } else { | |
5377 | while (scan < loceol && isSPACE(*scan)) | |
5378 | scan++; | |
a0ed51b3 LW |
5379 | } |
5380 | break; | |
bbce6d69 | 5381 | case SPACEL: |
3280af22 | 5382 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5383 | if (do_utf8) { |
ffc61ed2 | 5384 | loceol = PL_regeol; |
1aa99e6b | 5385 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5386 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5387 | scan += UTF8SKIP(scan); | |
5388 | hardcount++; | |
5389 | } | |
5390 | } else { | |
5391 | while (scan < loceol && isSPACE_LC(*scan)) | |
5392 | scan++; | |
a0ed51b3 LW |
5393 | } |
5394 | break; | |
a0d0e21e | 5395 | case NSPACE: |
1aa99e6b | 5396 | if (do_utf8) { |
ffc61ed2 | 5397 | loceol = PL_regeol; |
1a4fad37 | 5398 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5399 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5400 | !(*scan == ' ' || |
5401 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5402 | scan += UTF8SKIP(scan); |
5403 | hardcount++; | |
5404 | } | |
5405 | } else { | |
5406 | while (scan < loceol && !isSPACE(*scan)) | |
5407 | scan++; | |
5408 | break; | |
a0ed51b3 | 5409 | } |
bbce6d69 | 5410 | case NSPACEL: |
3280af22 | 5411 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5412 | if (do_utf8) { |
ffc61ed2 | 5413 | loceol = PL_regeol; |
1aa99e6b | 5414 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5415 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5416 | scan += UTF8SKIP(scan); | |
5417 | hardcount++; | |
5418 | } | |
5419 | } else { | |
5420 | while (scan < loceol && !isSPACE_LC(*scan)) | |
5421 | scan++; | |
a0ed51b3 LW |
5422 | } |
5423 | break; | |
a0d0e21e | 5424 | case DIGIT: |
1aa99e6b | 5425 | if (do_utf8) { |
ffc61ed2 | 5426 | loceol = PL_regeol; |
1a4fad37 | 5427 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5428 | while (hardcount < max && scan < loceol && |
3568d838 | 5429 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5430 | scan += UTF8SKIP(scan); |
5431 | hardcount++; | |
5432 | } | |
5433 | } else { | |
5434 | while (scan < loceol && isDIGIT(*scan)) | |
5435 | scan++; | |
a0ed51b3 LW |
5436 | } |
5437 | break; | |
a0d0e21e | 5438 | case NDIGIT: |
1aa99e6b | 5439 | if (do_utf8) { |
ffc61ed2 | 5440 | loceol = PL_regeol; |
1a4fad37 | 5441 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5442 | while (hardcount < max && scan < loceol && |
3568d838 | 5443 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5444 | scan += UTF8SKIP(scan); |
5445 | hardcount++; | |
5446 | } | |
5447 | } else { | |
5448 | while (scan < loceol && !isDIGIT(*scan)) | |
5449 | scan++; | |
a0ed51b3 | 5450 | } |
e1d1eefb YO |
5451 | case LNBREAK: |
5452 | if (do_utf8) { | |
5453 | loceol = PL_regeol; | |
5454 | while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) { | |
5455 | scan += c; | |
5456 | hardcount++; | |
5457 | } | |
5458 | } else { | |
5459 | /* | |
5460 | LNBREAK can match two latin chars, which is ok, | |
5461 | because we have a null terminated string, but we | |
5462 | have to use hardcount in this situation | |
5463 | */ | |
5464 | while (scan < loceol && (c=is_LNBREAK_latin1(scan))) { | |
5465 | scan+=c; | |
5466 | hardcount++; | |
5467 | } | |
5468 | } | |
5469 | break; | |
5470 | case HORIZWS: | |
5471 | if (do_utf8) { | |
5472 | loceol = PL_regeol; | |
5473 | while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) { | |
5474 | scan += c; | |
5475 | hardcount++; | |
5476 | } | |
5477 | } else { | |
5478 | while (scan < loceol && is_HORIZWS_latin1(scan)) | |
5479 | scan++; | |
5480 | } | |
a0ed51b3 | 5481 | break; |
e1d1eefb YO |
5482 | case NHORIZWS: |
5483 | if (do_utf8) { | |
5484 | loceol = PL_regeol; | |
5485 | while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) { | |
5486 | scan += UTF8SKIP(scan); | |
5487 | hardcount++; | |
5488 | } | |
5489 | } else { | |
5490 | while (scan < loceol && !is_HORIZWS_latin1(scan)) | |
5491 | scan++; | |
5492 | ||
5493 | } | |
5494 | break; | |
5495 | case VERTWS: | |
5496 | if (do_utf8) { | |
5497 | loceol = PL_regeol; | |
5498 | while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) { | |
5499 | scan += c; | |
5500 | hardcount++; | |
5501 | } | |
5502 | } else { | |
5503 | while (scan < loceol && is_VERTWS_latin1(scan)) | |
5504 | scan++; | |
5505 | ||
5506 | } | |
5507 | break; | |
5508 | case NVERTWS: | |
5509 | if (do_utf8) { | |
5510 | loceol = PL_regeol; | |
5511 | while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) { | |
5512 | scan += UTF8SKIP(scan); | |
5513 | hardcount++; | |
5514 | } | |
5515 | } else { | |
5516 | while (scan < loceol && !is_VERTWS_latin1(scan)) | |
5517 | scan++; | |
5518 | ||
5519 | } | |
5520 | break; | |
5521 | ||
a0d0e21e LW |
5522 | default: /* Called on something of 0 width. */ |
5523 | break; /* So match right here or not at all. */ | |
5524 | } | |
a687059c | 5525 | |
a0ed51b3 LW |
5526 | if (hardcount) |
5527 | c = hardcount; | |
5528 | else | |
5529 | c = scan - PL_reginput; | |
3280af22 | 5530 | PL_reginput = scan; |
a687059c | 5531 | |
a3621e74 | 5532 | DEBUG_r({ |
e68ec53f | 5533 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 5534 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
5535 | SV * const prop = sv_newmortal(); |
5536 | regprop(prog, prop, p); | |
5537 | PerlIO_printf(Perl_debug_log, | |
be8e71aa | 5538 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
e2e6a0f1 | 5539 | REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max); |
a3621e74 | 5540 | }); |
be8e71aa | 5541 | }); |
9041c2e3 | 5542 | |
a0d0e21e | 5543 | return(c); |
a687059c LW |
5544 | } |
5545 | ||
c277df42 | 5546 | |
be8e71aa | 5547 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 5548 | /* |
ffc61ed2 JH |
5549 | - regclass_swash - prepare the utf8 swash |
5550 | */ | |
5551 | ||
5552 | SV * | |
32fc9b6a | 5553 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 5554 | { |
97aff369 | 5555 | dVAR; |
9e55ce06 JH |
5556 | SV *sw = NULL; |
5557 | SV *si = NULL; | |
5558 | SV *alt = NULL; | |
f8fc2ecf YO |
5559 | RXi_GET_DECL(prog,progi); |
5560 | const struct reg_data * const data = prog ? progi->data : NULL; | |
ffc61ed2 | 5561 | |
4f639d21 | 5562 | if (data && data->count) { |
a3b680e6 | 5563 | const U32 n = ARG(node); |
ffc61ed2 | 5564 | |
4f639d21 DM |
5565 | if (data->what[n] == 's') { |
5566 | SV * const rv = (SV*)data->data[n]; | |
890ce7af | 5567 | AV * const av = (AV*)SvRV((SV*)rv); |
2d03de9c | 5568 | SV **const ary = AvARRAY(av); |
9e55ce06 | 5569 | SV **a, **b; |
9041c2e3 | 5570 | |
711a919c | 5571 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
5572 | * documentation of these array elements. */ |
5573 | ||
b11f357e | 5574 | si = *ary; |
fe5bfecd JH |
5575 | a = SvROK(ary[1]) ? &ary[1] : NULL; |
5576 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : NULL; | |
b11f357e | 5577 | |
ffc61ed2 JH |
5578 | if (a) |
5579 | sw = *a; | |
5580 | else if (si && doinit) { | |
5581 | sw = swash_init("utf8", "", si, 1, 0); | |
5582 | (void)av_store(av, 1, sw); | |
5583 | } | |
9e55ce06 JH |
5584 | if (b) |
5585 | alt = *b; | |
ffc61ed2 JH |
5586 | } |
5587 | } | |
5588 | ||
9e55ce06 JH |
5589 | if (listsvp) |
5590 | *listsvp = si; | |
5591 | if (altsvp) | |
5592 | *altsvp = alt; | |
ffc61ed2 JH |
5593 | |
5594 | return sw; | |
5595 | } | |
76234dfb | 5596 | #endif |
ffc61ed2 JH |
5597 | |
5598 | /* | |
ba7b4546 | 5599 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
5600 | |
5601 | The n is the ANYOF regnode, the p is the target string, lenp | |
5602 | is pointer to the maximum length of how far to go in the p | |
5603 | (if the lenp is zero, UTF8SKIP(p) is used), | |
5604 | do_utf8 tells whether the target string is in UTF-8. | |
5605 | ||
bbce6d69 | 5606 | */ |
5607 | ||
76e3520e | 5608 | STATIC bool |
32fc9b6a | 5609 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 5610 | { |
27da23d5 | 5611 | dVAR; |
a3b680e6 | 5612 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 5613 | bool match = FALSE; |
cc07378b | 5614 | UV c = *p; |
ae9ddab8 | 5615 | STRLEN len = 0; |
9e55ce06 | 5616 | STRLEN plen; |
1aa99e6b | 5617 | |
19f67299 TS |
5618 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5619 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5620 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5621 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5622 | if (len == (STRLEN)-1) |
5623 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5624 | } |
bbce6d69 | 5625 | |
0f0076b4 | 5626 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5627 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5628 | if (lenp) |
5629 | *lenp = 0; | |
ffc61ed2 | 5630 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5631 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5632 | match = TRUE; | |
bbce6d69 | 5633 | } |
3568d838 | 5634 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5635 | match = TRUE; |
ffc61ed2 | 5636 | if (!match) { |
9e55ce06 | 5637 | AV *av; |
32fc9b6a | 5638 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5639 | |
5640 | if (sw) { | |
3568d838 | 5641 | if (swash_fetch(sw, p, do_utf8)) |
ffc61ed2 JH |
5642 | match = TRUE; |
5643 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5644 | if (!match && lenp && av) { |
5645 | I32 i; | |
9e55ce06 | 5646 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5647 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5648 | STRLEN len; |
890ce7af | 5649 | const char * const s = SvPV_const(sv, len); |
9e55ce06 | 5650 | |
061b10df | 5651 | if (len <= plen && memEQ(s, (char*)p, len)) { |
9e55ce06 JH |
5652 | *lenp = len; |
5653 | match = TRUE; | |
5654 | break; | |
5655 | } | |
5656 | } | |
5657 | } | |
5658 | if (!match) { | |
89ebb4a3 | 5659 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 JH |
5660 | STRLEN tmplen; |
5661 | ||
9e55ce06 JH |
5662 | to_utf8_fold(p, tmpbuf, &tmplen); |
5663 | if (swash_fetch(sw, tmpbuf, do_utf8)) | |
5664 | match = TRUE; | |
5665 | } | |
ffc61ed2 JH |
5666 | } |
5667 | } | |
bbce6d69 | 5668 | } |
9e55ce06 | 5669 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5670 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5671 | } |
1aa99e6b | 5672 | if (!match && c < 256) { |
ffc61ed2 JH |
5673 | if (ANYOF_BITMAP_TEST(n, c)) |
5674 | match = TRUE; | |
5675 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5676 | U8 f; |
a0ed51b3 | 5677 | |
ffc61ed2 JH |
5678 | if (flags & ANYOF_LOCALE) { |
5679 | PL_reg_flags |= RF_tainted; | |
5680 | f = PL_fold_locale[c]; | |
5681 | } | |
5682 | else | |
5683 | f = PL_fold[c]; | |
5684 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5685 | match = TRUE; | |
5686 | } | |
5687 | ||
5688 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5689 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5690 | if ( |
5691 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5692 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5693 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5694 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5695 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5696 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5697 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5698 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5699 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5700 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5701 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5702 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5703 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5704 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5705 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5706 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5707 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5708 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5709 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5710 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5711 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5712 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5713 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5714 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5715 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5716 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5717 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5718 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5719 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5720 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5721 | ) /* How's that for a conditional? */ | |
5722 | { | |
5723 | match = TRUE; | |
5724 | } | |
a0ed51b3 | 5725 | } |
a0ed51b3 LW |
5726 | } |
5727 | ||
a0ed51b3 LW |
5728 | return (flags & ANYOF_INVERT) ? !match : match; |
5729 | } | |
161b471a | 5730 | |
dfe13c55 | 5731 | STATIC U8 * |
0ce71af7 | 5732 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5733 | { |
97aff369 | 5734 | dVAR; |
a0ed51b3 | 5735 | if (off >= 0) { |
1aa99e6b | 5736 | while (off-- && s < lim) { |
ffc61ed2 | 5737 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5738 | s += UTF8SKIP(s); |
ffc61ed2 | 5739 | } |
a0ed51b3 LW |
5740 | } |
5741 | else { | |
1de06328 YO |
5742 | while (off++ && s > lim) { |
5743 | s--; | |
5744 | if (UTF8_IS_CONTINUED(*s)) { | |
5745 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5746 | s--; | |
a0ed51b3 | 5747 | } |
1de06328 | 5748 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5749 | } |
5750 | } | |
5751 | return s; | |
5752 | } | |
161b471a | 5753 | |
f9f4320a YO |
5754 | #ifdef XXX_dmq |
5755 | /* there are a bunch of places where we use two reghop3's that should | |
5756 | be replaced with this routine. but since thats not done yet | |
5757 | we ifdef it out - dmq | |
5758 | */ | |
dfe13c55 | 5759 | STATIC U8 * |
1de06328 YO |
5760 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5761 | { | |
5762 | dVAR; | |
5763 | if (off >= 0) { | |
5764 | while (off-- && s < rlim) { | |
5765 | /* XXX could check well-formedness here */ | |
5766 | s += UTF8SKIP(s); | |
5767 | } | |
5768 | } | |
5769 | else { | |
5770 | while (off++ && s > llim) { | |
5771 | s--; | |
5772 | if (UTF8_IS_CONTINUED(*s)) { | |
5773 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5774 | s--; | |
5775 | } | |
5776 | /* XXX could check well-formedness here */ | |
5777 | } | |
5778 | } | |
5779 | return s; | |
5780 | } | |
f9f4320a | 5781 | #endif |
1de06328 YO |
5782 | |
5783 | STATIC U8 * | |
0ce71af7 | 5784 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5785 | { |
97aff369 | 5786 | dVAR; |
a0ed51b3 | 5787 | if (off >= 0) { |
1aa99e6b | 5788 | while (off-- && s < lim) { |
ffc61ed2 | 5789 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5790 | s += UTF8SKIP(s); |
ffc61ed2 | 5791 | } |
a0ed51b3 | 5792 | if (off >= 0) |
3dab1dad | 5793 | return NULL; |
a0ed51b3 LW |
5794 | } |
5795 | else { | |
1de06328 YO |
5796 | while (off++ && s > lim) { |
5797 | s--; | |
5798 | if (UTF8_IS_CONTINUED(*s)) { | |
5799 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5800 | s--; | |
a0ed51b3 | 5801 | } |
1de06328 | 5802 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5803 | } |
5804 | if (off <= 0) | |
3dab1dad | 5805 | return NULL; |
a0ed51b3 LW |
5806 | } |
5807 | return s; | |
5808 | } | |
51371543 | 5809 | |
51371543 | 5810 | static void |
acfe0abc | 5811 | restore_pos(pTHX_ void *arg) |
51371543 | 5812 | { |
97aff369 | 5813 | dVAR; |
097eb12c | 5814 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5815 | if (PL_reg_eval_set) { |
5816 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5817 | rex->subbeg = PL_reg_oldsaved; |
5818 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5819 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5820 | rex->saved_copy = PL_nrs; |
ed252734 | 5821 | #endif |
4f639d21 | 5822 | RX_MATCH_COPIED_on(rex); |
51371543 GS |
5823 | } |
5824 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
5825 | PL_reg_eval_set = 0; | |
5826 | PL_curpm = PL_reg_oldcurpm; | |
5827 | } | |
5828 | } | |
33b8afdf JH |
5829 | |
5830 | STATIC void | |
5831 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
5832 | { | |
a1cac82e NC |
5833 | int i = 1; |
5834 | do { | |
5835 | if (prog->substrs->data[i].substr | |
5836 | && !prog->substrs->data[i].utf8_substr) { | |
5837 | SV* const sv = newSVsv(prog->substrs->data[i].substr); | |
5838 | prog->substrs->data[i].utf8_substr = sv; | |
5839 | sv_utf8_upgrade(sv); | |
610460f9 NC |
5840 | if (SvVALID(prog->substrs->data[i].substr)) { |
5841 | const U8 flags = BmFLAGS(prog->substrs->data[i].substr); | |
5842 | if (flags & FBMcf_TAIL) { | |
5843 | /* Trim the trailing \n that fbm_compile added last | |
5844 | time. */ | |
5845 | SvCUR_set(sv, SvCUR(sv) - 1); | |
5846 | /* Whilst this makes the SV technically "invalid" (as its | |
5847 | buffer is no longer followed by "\0") when fbm_compile() | |
5848 | adds the "\n" back, a "\0" is restored. */ | |
5849 | } | |
5850 | fbm_compile(sv, flags); | |
5851 | } | |
a1cac82e NC |
5852 | if (prog->substrs->data[i].substr == prog->check_substr) |
5853 | prog->check_utf8 = sv; | |
5854 | } | |
5855 | } while (i--); | |
33b8afdf JH |
5856 | } |
5857 | ||
5858 | STATIC void | |
5859 | S_to_byte_substr(pTHX_ register regexp *prog) | |
5860 | { | |
97aff369 | 5861 | dVAR; |
a1cac82e NC |
5862 | int i = 1; |
5863 | do { | |
5864 | if (prog->substrs->data[i].utf8_substr | |
5865 | && !prog->substrs->data[i].substr) { | |
5866 | SV* sv = newSVsv(prog->substrs->data[i].utf8_substr); | |
5867 | if (sv_utf8_downgrade(sv, TRUE)) { | |
610460f9 NC |
5868 | if (SvVALID(prog->substrs->data[i].utf8_substr)) { |
5869 | const U8 flags | |
5870 | = BmFLAGS(prog->substrs->data[i].utf8_substr); | |
5871 | if (flags & FBMcf_TAIL) { | |
5872 | /* Trim the trailing \n that fbm_compile added last | |
5873 | time. */ | |
5874 | SvCUR_set(sv, SvCUR(sv) - 1); | |
5875 | } | |
5876 | fbm_compile(sv, flags); | |
5877 | } | |
a1cac82e NC |
5878 | } else { |
5879 | SvREFCNT_dec(sv); | |
5880 | sv = &PL_sv_undef; | |
5881 | } | |
5882 | prog->substrs->data[i].substr = sv; | |
5883 | if (prog->substrs->data[i].utf8_substr == prog->check_utf8) | |
5884 | prog->check_substr = sv; | |
33b8afdf | 5885 | } |
a1cac82e | 5886 | } while (i--); |
33b8afdf | 5887 | } |
66610fdd RGS |
5888 | |
5889 | /* | |
5890 | * Local variables: | |
5891 | * c-indentation-style: bsd | |
5892 | * c-basic-offset: 4 | |
5893 | * indent-tabs-mode: t | |
5894 | * End: | |
5895 | * | |
37442d52 RGS |
5896 | * ex: set ts=8 sts=4 sw=4 noet: |
5897 | */ |