Commit | Line | Data |
---|---|---|
a0d0e21e LW |
1 | /* regexec.c |
2 | */ | |
3 | ||
4 | /* | |
4ac71550 TC |
5 | * One Ring to rule them all, One Ring to find them |
6 | & | |
7 | * [p.v of _The Lord of the Rings_, opening poem] | |
8 | * [p.50 of _The Lord of the Rings_, I/iii: "The Shadow of the Past"] | |
9 | * [p.254 of _The Lord of the Rings_, II/ii: "The Council of Elrond"] | |
a0d0e21e LW |
10 | */ |
11 | ||
61296642 DM |
12 | /* This file contains functions for executing a regular expression. See |
13 | * also regcomp.c which funnily enough, contains functions for compiling | |
166f8a29 | 14 | * a regular expression. |
e4a054ea DM |
15 | * |
16 | * This file is also copied at build time to ext/re/re_exec.c, where | |
17 | * it's built with -DPERL_EXT_RE_BUILD -DPERL_EXT_RE_DEBUG -DPERL_EXT. | |
18 | * This causes the main functions to be compiled under new names and with | |
19 | * debugging support added, which makes "use re 'debug'" work. | |
166f8a29 DM |
20 | */ |
21 | ||
a687059c LW |
22 | /* NOTE: this is derived from Henry Spencer's regexp code, and should not |
23 | * confused with the original package (see point 3 below). Thanks, Henry! | |
24 | */ | |
25 | ||
26 | /* Additional note: this code is very heavily munged from Henry's version | |
27 | * in places. In some spots I've traded clarity for efficiency, so don't | |
28 | * blame Henry for some of the lack of readability. | |
29 | */ | |
30 | ||
e50aee73 AD |
31 | /* The names of the functions have been changed from regcomp and |
32 | * regexec to pregcomp and pregexec in order to avoid conflicts | |
33 | * with the POSIX routines of the same names. | |
34 | */ | |
35 | ||
b9d5759e | 36 | #ifdef PERL_EXT_RE_BUILD |
54df2634 | 37 | #include "re_top.h" |
9041c2e3 | 38 | #endif |
56953603 | 39 | |
a687059c | 40 | /* |
e50aee73 | 41 | * pregcomp and pregexec -- regsub and regerror are not used in perl |
a687059c LW |
42 | * |
43 | * Copyright (c) 1986 by University of Toronto. | |
44 | * Written by Henry Spencer. Not derived from licensed software. | |
45 | * | |
46 | * Permission is granted to anyone to use this software for any | |
47 | * purpose on any computer system, and to redistribute it freely, | |
48 | * subject to the following restrictions: | |
49 | * | |
50 | * 1. The author is not responsible for the consequences of use of | |
51 | * this software, no matter how awful, even if they arise | |
52 | * from defects in it. | |
53 | * | |
54 | * 2. The origin of this software must not be misrepresented, either | |
55 | * by explicit claim or by omission. | |
56 | * | |
57 | * 3. Altered versions must be plainly marked as such, and must not | |
58 | * be misrepresented as being the original software. | |
59 | * | |
60 | **** Alterations to Henry's code are... | |
61 | **** | |
4bb101f2 | 62 | **** Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
1129b882 NC |
63 | **** 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 |
64 | **** by Larry Wall and others | |
a687059c | 65 | **** |
9ef589d8 LW |
66 | **** You may distribute under the terms of either the GNU General Public |
67 | **** License or the Artistic License, as specified in the README file. | |
a687059c LW |
68 | * |
69 | * Beware that some of this code is subtly aware of the way operator | |
70 | * precedence is structured in regular expressions. Serious changes in | |
71 | * regular-expression syntax might require a total rethink. | |
72 | */ | |
73 | #include "EXTERN.h" | |
864dbfa3 | 74 | #define PERL_IN_REGEXEC_C |
a687059c | 75 | #include "perl.h" |
0f5d15d6 | 76 | |
54df2634 NC |
77 | #ifdef PERL_IN_XSUB_RE |
78 | # include "re_comp.h" | |
79 | #else | |
80 | # include "regcomp.h" | |
81 | #endif | |
a687059c | 82 | |
c277df42 IZ |
83 | #define RF_tainted 1 /* tainted information used? */ |
84 | #define RF_warned 2 /* warned about big count? */ | |
faec1544 | 85 | |
ab3bbdeb | 86 | #define RF_utf8 8 /* Pattern contains multibyte chars? */ |
a0ed51b3 | 87 | |
eb160463 | 88 | #define UTF ((PL_reg_flags & RF_utf8) != 0) |
ce862d02 IZ |
89 | |
90 | #define RS_init 1 /* eval environment created */ | |
91 | #define RS_set 2 /* replsv value is set */ | |
c277df42 | 92 | |
a687059c LW |
93 | #ifndef STATIC |
94 | #define STATIC static | |
95 | #endif | |
96 | ||
32fc9b6a | 97 | #define REGINCLASS(prog,p,c) (ANYOF_FLAGS(p) ? reginclass(prog,p,c,0,0) : ANYOF_BITMAP_TEST(p,*(c))) |
7d3e948e | 98 | |
c277df42 IZ |
99 | /* |
100 | * Forwards. | |
101 | */ | |
102 | ||
33b8afdf | 103 | #define CHR_SVLEN(sv) (do_utf8 ? sv_len_utf8(sv) : SvCUR(sv)) |
53c4c00c | 104 | #define CHR_DIST(a,b) (PL_reg_match_utf8 ? utf8_distance(a,b) : a - b) |
a0ed51b3 | 105 | |
3dab1dad YO |
106 | #define HOPc(pos,off) \ |
107 | (char *)(PL_reg_match_utf8 \ | |
52657f30 | 108 | ? reghop3((U8*)pos, off, (U8*)(off >= 0 ? PL_regeol : PL_bostr)) \ |
3dab1dad YO |
109 | : (U8*)(pos + off)) |
110 | #define HOPBACKc(pos, off) \ | |
07be1b83 YO |
111 | (char*)(PL_reg_match_utf8\ |
112 | ? reghopmaybe3((U8*)pos, -off, (U8*)PL_bostr) \ | |
113 | : (pos - off >= PL_bostr) \ | |
8e11feef | 114 | ? (U8*)pos - off \ |
3dab1dad | 115 | : NULL) |
efb30f32 | 116 | |
e7409c1b | 117 | #define HOP3(pos,off,lim) (PL_reg_match_utf8 ? reghop3((U8*)(pos), off, (U8*)(lim)) : (U8*)(pos + off)) |
1aa99e6b | 118 | #define HOP3c(pos,off,lim) ((char*)HOP3(pos,off,lim)) |
1aa99e6b | 119 | |
1a4fad37 AL |
120 | #define LOAD_UTF8_CHARCLASS(class,str) STMT_START { \ |
121 | if (!CAT2(PL_utf8_,class)) { bool ok; ENTER; save_re_context(); ok=CAT2(is_utf8_,class)((const U8*)str); assert(ok); LEAVE; } } STMT_END | |
122 | #define LOAD_UTF8_CHARCLASS_ALNUM() LOAD_UTF8_CHARCLASS(alnum,"a") | |
123 | #define LOAD_UTF8_CHARCLASS_DIGIT() LOAD_UTF8_CHARCLASS(digit,"0") | |
124 | #define LOAD_UTF8_CHARCLASS_SPACE() LOAD_UTF8_CHARCLASS(space," ") | |
125 | #define LOAD_UTF8_CHARCLASS_MARK() LOAD_UTF8_CHARCLASS(mark, "\xcd\x86") | |
51371543 | 126 | |
3dab1dad YO |
127 | /* TODO: Combine JUMPABLE and HAS_TEXT to cache OP(rn) */ |
128 | ||
5f80c4cf | 129 | /* for use after a quantifier and before an EXACT-like node -- japhy */ |
3e901dc0 YO |
130 | /* it would be nice to rework regcomp.sym to generate this stuff. sigh */ |
131 | #define JUMPABLE(rn) ( \ | |
132 | OP(rn) == OPEN || \ | |
133 | (OP(rn) == CLOSE && (!cur_eval || cur_eval->u.eval.close_paren != ARG(rn))) || \ | |
134 | OP(rn) == EVAL || \ | |
cca55fe3 JP |
135 | OP(rn) == SUSPEND || OP(rn) == IFMATCH || \ |
136 | OP(rn) == PLUS || OP(rn) == MINMOD || \ | |
ee9b8eae | 137 | OP(rn) == KEEPS || (PL_regkind[OP(rn)] == VERB) || \ |
3dab1dad | 138 | (PL_regkind[OP(rn)] == CURLY && ARG1(rn) > 0) \ |
e2d8ce26 | 139 | ) |
ee9b8eae | 140 | #define IS_EXACT(rn) (PL_regkind[OP(rn)] == EXACT) |
e2d8ce26 | 141 | |
ee9b8eae YO |
142 | #define HAS_TEXT(rn) ( IS_EXACT(rn) || PL_regkind[OP(rn)] == REF ) |
143 | ||
144 | #if 0 | |
145 | /* Currently these are only used when PL_regkind[OP(rn)] == EXACT so | |
146 | we don't need this definition. */ | |
147 | #define IS_TEXT(rn) ( OP(rn)==EXACT || OP(rn)==REF || OP(rn)==NREF ) | |
148 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF || OP(rn)==REFF || OP(rn)==NREFF ) | |
149 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL || OP(rn)==REFFL || OP(rn)==NREFFL ) | |
150 | ||
151 | #else | |
152 | /* ... so we use this as its faster. */ | |
153 | #define IS_TEXT(rn) ( OP(rn)==EXACT ) | |
154 | #define IS_TEXTF(rn) ( OP(rn)==EXACTF ) | |
155 | #define IS_TEXTFL(rn) ( OP(rn)==EXACTFL ) | |
156 | ||
157 | #endif | |
e2d8ce26 | 158 | |
a84d97b6 HS |
159 | /* |
160 | Search for mandatory following text node; for lookahead, the text must | |
161 | follow but for lookbehind (rn->flags != 0) we skip to the next step. | |
162 | */ | |
cca55fe3 | 163 | #define FIND_NEXT_IMPT(rn) STMT_START { \ |
3dab1dad YO |
164 | while (JUMPABLE(rn)) { \ |
165 | const OPCODE type = OP(rn); \ | |
166 | if (type == SUSPEND || PL_regkind[type] == CURLY) \ | |
e2d8ce26 | 167 | rn = NEXTOPER(NEXTOPER(rn)); \ |
3dab1dad | 168 | else if (type == PLUS) \ |
cca55fe3 | 169 | rn = NEXTOPER(rn); \ |
3dab1dad | 170 | else if (type == IFMATCH) \ |
a84d97b6 | 171 | rn = (rn->flags == 0) ? NEXTOPER(NEXTOPER(rn)) : rn + ARG(rn); \ |
e2d8ce26 | 172 | else rn += NEXT_OFF(rn); \ |
3dab1dad | 173 | } \ |
5f80c4cf | 174 | } STMT_END |
74750237 | 175 | |
c476f425 | 176 | |
acfe0abc | 177 | static void restore_pos(pTHX_ void *arg); |
51371543 | 178 | |
76e3520e | 179 | STATIC CHECKPOINT |
cea2e8a9 | 180 | S_regcppush(pTHX_ I32 parenfloor) |
a0d0e21e | 181 | { |
97aff369 | 182 | dVAR; |
a3b680e6 | 183 | const int retval = PL_savestack_ix; |
b1ce53c5 | 184 | #define REGCP_PAREN_ELEMS 4 |
a3b680e6 | 185 | const int paren_elems_to_push = (PL_regsize - parenfloor) * REGCP_PAREN_ELEMS; |
a0d0e21e | 186 | int p; |
40a82448 | 187 | GET_RE_DEBUG_FLAGS_DECL; |
a0d0e21e | 188 | |
e49a9654 IH |
189 | if (paren_elems_to_push < 0) |
190 | Perl_croak(aTHX_ "panic: paren_elems_to_push < 0"); | |
191 | ||
f0ab9afb | 192 | #define REGCP_OTHER_ELEMS 7 |
4b3c1a47 | 193 | SSGROW(paren_elems_to_push + REGCP_OTHER_ELEMS); |
7f69552c | 194 | |
3280af22 | 195 | for (p = PL_regsize; p > parenfloor; p--) { |
b1ce53c5 | 196 | /* REGCP_PARENS_ELEMS are pushed per pairs of parentheses. */ |
f0ab9afb NC |
197 | SSPUSHINT(PL_regoffs[p].end); |
198 | SSPUSHINT(PL_regoffs[p].start); | |
3280af22 | 199 | SSPUSHPTR(PL_reg_start_tmp[p]); |
a0d0e21e | 200 | SSPUSHINT(p); |
e7707071 | 201 | DEBUG_BUFFERS_r(PerlIO_printf(Perl_debug_log, |
40a82448 | 202 | " saving \\%"UVuf" %"IVdf"(%"IVdf")..%"IVdf"\n", |
f0ab9afb | 203 | (UV)p, (IV)PL_regoffs[p].start, |
40a82448 | 204 | (IV)(PL_reg_start_tmp[p] - PL_bostr), |
f0ab9afb | 205 | (IV)PL_regoffs[p].end |
40a82448 | 206 | )); |
a0d0e21e | 207 | } |
b1ce53c5 | 208 | /* REGCP_OTHER_ELEMS are pushed in any case, parentheses or no. */ |
f0ab9afb | 209 | SSPUSHPTR(PL_regoffs); |
3280af22 NIS |
210 | SSPUSHINT(PL_regsize); |
211 | SSPUSHINT(*PL_reglastparen); | |
a01268b5 | 212 | SSPUSHINT(*PL_reglastcloseparen); |
3280af22 | 213 | SSPUSHPTR(PL_reginput); |
41123dfd JH |
214 | #define REGCP_FRAME_ELEMS 2 |
215 | /* REGCP_FRAME_ELEMS are part of the REGCP_OTHER_ELEMS and | |
216 | * are needed for the regexp context stack bookkeeping. */ | |
217 | SSPUSHINT(paren_elems_to_push + REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); | |
b1ce53c5 | 218 | SSPUSHINT(SAVEt_REGCONTEXT); /* Magic cookie. */ |
41123dfd | 219 | |
a0d0e21e LW |
220 | return retval; |
221 | } | |
222 | ||
c277df42 | 223 | /* These are needed since we do not localize EVAL nodes: */ |
ab3bbdeb YO |
224 | #define REGCP_SET(cp) \ |
225 | DEBUG_STATE_r( \ | |
ab3bbdeb | 226 | PerlIO_printf(Perl_debug_log, \ |
e4f74956 | 227 | " Setting an EVAL scope, savestack=%"IVdf"\n", \ |
ab3bbdeb YO |
228 | (IV)PL_savestack_ix)); \ |
229 | cp = PL_savestack_ix | |
c3464db5 | 230 | |
ab3bbdeb | 231 | #define REGCP_UNWIND(cp) \ |
e4f74956 | 232 | DEBUG_STATE_r( \ |
ab3bbdeb | 233 | if (cp != PL_savestack_ix) \ |
e4f74956 YO |
234 | PerlIO_printf(Perl_debug_log, \ |
235 | " Clearing an EVAL scope, savestack=%"IVdf"..%"IVdf"\n", \ | |
ab3bbdeb YO |
236 | (IV)(cp), (IV)PL_savestack_ix)); \ |
237 | regcpblow(cp) | |
c277df42 | 238 | |
76e3520e | 239 | STATIC char * |
097eb12c | 240 | S_regcppop(pTHX_ const regexp *rex) |
a0d0e21e | 241 | { |
97aff369 | 242 | dVAR; |
3b6647e0 | 243 | U32 i; |
a0d0e21e | 244 | char *input; |
a3621e74 YO |
245 | GET_RE_DEBUG_FLAGS_DECL; |
246 | ||
7918f24d NC |
247 | PERL_ARGS_ASSERT_REGCPPOP; |
248 | ||
b1ce53c5 | 249 | /* Pop REGCP_OTHER_ELEMS before the parentheses loop starts. */ |
a0d0e21e | 250 | i = SSPOPINT; |
b1ce53c5 JH |
251 | assert(i == SAVEt_REGCONTEXT); /* Check that the magic cookie is there. */ |
252 | i = SSPOPINT; /* Parentheses elements to pop. */ | |
a0d0e21e | 253 | input = (char *) SSPOPPTR; |
a01268b5 | 254 | *PL_reglastcloseparen = SSPOPINT; |
3280af22 NIS |
255 | *PL_reglastparen = SSPOPINT; |
256 | PL_regsize = SSPOPINT; | |
f0ab9afb | 257 | PL_regoffs=(regexp_paren_pair *) SSPOPPTR; |
b1ce53c5 | 258 | |
6bda09f9 | 259 | |
b1ce53c5 | 260 | /* Now restore the parentheses context. */ |
41123dfd JH |
261 | for (i -= (REGCP_OTHER_ELEMS - REGCP_FRAME_ELEMS); |
262 | i > 0; i -= REGCP_PAREN_ELEMS) { | |
1df70142 | 263 | I32 tmps; |
097eb12c | 264 | U32 paren = (U32)SSPOPINT; |
3280af22 | 265 | PL_reg_start_tmp[paren] = (char *) SSPOPPTR; |
f0ab9afb | 266 | PL_regoffs[paren].start = SSPOPINT; |
cf93c79d | 267 | tmps = SSPOPINT; |
3280af22 | 268 | if (paren <= *PL_reglastparen) |
f0ab9afb | 269 | PL_regoffs[paren].end = tmps; |
e7707071 | 270 | DEBUG_BUFFERS_r( |
c3464db5 | 271 | PerlIO_printf(Perl_debug_log, |
b900a521 | 272 | " restoring \\%"UVuf" to %"IVdf"(%"IVdf")..%"IVdf"%s\n", |
f0ab9afb | 273 | (UV)paren, (IV)PL_regoffs[paren].start, |
b900a521 | 274 | (IV)(PL_reg_start_tmp[paren] - PL_bostr), |
f0ab9afb | 275 | (IV)PL_regoffs[paren].end, |
3280af22 | 276 | (paren > *PL_reglastparen ? "(no)" : "")); |
c277df42 | 277 | ); |
a0d0e21e | 278 | } |
e7707071 | 279 | DEBUG_BUFFERS_r( |
bb7a0f54 | 280 | if (*PL_reglastparen + 1 <= rex->nparens) { |
c3464db5 | 281 | PerlIO_printf(Perl_debug_log, |
faccc32b | 282 | " restoring \\%"IVdf"..\\%"IVdf" to undef\n", |
4f639d21 | 283 | (IV)(*PL_reglastparen + 1), (IV)rex->nparens); |
c277df42 IZ |
284 | } |
285 | ); | |
daf18116 | 286 | #if 1 |
dafc8851 JH |
287 | /* It would seem that the similar code in regtry() |
288 | * already takes care of this, and in fact it is in | |
289 | * a better location to since this code can #if 0-ed out | |
290 | * but the code in regtry() is needed or otherwise tests | |
291 | * requiring null fields (pat.t#187 and split.t#{13,14} | |
daf18116 JH |
292 | * (as of patchlevel 7877) will fail. Then again, |
293 | * this code seems to be necessary or otherwise | |
225593e1 DM |
294 | * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ |
295 | * --jhi updated by dapm */ | |
3b6647e0 | 296 | for (i = *PL_reglastparen + 1; i <= rex->nparens; i++) { |
097eb12c | 297 | if (i > PL_regsize) |
f0ab9afb NC |
298 | PL_regoffs[i].start = -1; |
299 | PL_regoffs[i].end = -1; | |
a0d0e21e | 300 | } |
dafc8851 | 301 | #endif |
a0d0e21e LW |
302 | return input; |
303 | } | |
304 | ||
02db2b7b | 305 | #define regcpblow(cp) LEAVE_SCOPE(cp) /* Ignores regcppush()ed data. */ |
a0d0e21e | 306 | |
a687059c | 307 | /* |
e50aee73 | 308 | * pregexec and friends |
a687059c LW |
309 | */ |
310 | ||
76234dfb | 311 | #ifndef PERL_IN_XSUB_RE |
a687059c | 312 | /* |
c277df42 | 313 | - pregexec - match a regexp against a string |
a687059c | 314 | */ |
c277df42 | 315 | I32 |
49d7dfbc | 316 | Perl_pregexec(pTHX_ REGEXP * const prog, char* stringarg, register char *strend, |
c3464db5 | 317 | char *strbeg, I32 minend, SV *screamer, U32 nosave) |
c277df42 IZ |
318 | /* strend: pointer to null at end of string */ |
319 | /* strbeg: real beginning of string */ | |
320 | /* minend: end of match must be >=minend after stringarg. */ | |
321 | /* nosave: For optimizations. */ | |
322 | { | |
7918f24d NC |
323 | PERL_ARGS_ASSERT_PREGEXEC; |
324 | ||
c277df42 | 325 | return |
9041c2e3 | 326 | regexec_flags(prog, stringarg, strend, strbeg, minend, screamer, NULL, |
c277df42 IZ |
327 | nosave ? 0 : REXEC_COPY_STR); |
328 | } | |
76234dfb | 329 | #endif |
22e551b9 | 330 | |
9041c2e3 | 331 | /* |
cad2e5aa JH |
332 | * Need to implement the following flags for reg_anch: |
333 | * | |
334 | * USE_INTUIT_NOML - Useful to call re_intuit_start() first | |
335 | * USE_INTUIT_ML | |
336 | * INTUIT_AUTORITATIVE_NOML - Can trust a positive answer | |
337 | * INTUIT_AUTORITATIVE_ML | |
338 | * INTUIT_ONCE_NOML - Intuit can match in one location only. | |
339 | * INTUIT_ONCE_ML | |
340 | * | |
341 | * Another flag for this function: SECOND_TIME (so that float substrs | |
342 | * with giant delta may be not rechecked). | |
343 | */ | |
344 | ||
345 | /* Assumptions: if ANCH_GPOS, then strpos is anchored. XXXX Check GPOS logic */ | |
346 | ||
3f7c398e | 347 | /* If SCREAM, then SvPVX_const(sv) should be compatible with strpos and strend. |
cad2e5aa JH |
348 | Otherwise, only SvCUR(sv) is used to get strbeg. */ |
349 | ||
350 | /* XXXX We assume that strpos is strbeg unless sv. */ | |
351 | ||
6eb5f6b9 JH |
352 | /* XXXX Some places assume that there is a fixed substring. |
353 | An update may be needed if optimizer marks as "INTUITable" | |
354 | RExen without fixed substrings. Similarly, it is assumed that | |
355 | lengths of all the strings are no more than minlen, thus they | |
356 | cannot come from lookahead. | |
40d049e4 YO |
357 | (Or minlen should take into account lookahead.) |
358 | NOTE: Some of this comment is not correct. minlen does now take account | |
359 | of lookahead/behind. Further research is required. -- demerphq | |
360 | ||
361 | */ | |
6eb5f6b9 | 362 | |
2c2d71f5 JH |
363 | /* A failure to find a constant substring means that there is no need to make |
364 | an expensive call to REx engine, thus we celebrate a failure. Similarly, | |
365 | finding a substring too deep into the string means that less calls to | |
30944b6d IZ |
366 | regtry() should be needed. |
367 | ||
368 | REx compiler's optimizer found 4 possible hints: | |
369 | a) Anchored substring; | |
370 | b) Fixed substring; | |
371 | c) Whether we are anchored (beginning-of-line or \G); | |
372 | d) First node (of those at offset 0) which may distingush positions; | |
6eb5f6b9 | 373 | We use a)b)d) and multiline-part of c), and try to find a position in the |
30944b6d IZ |
374 | string which does not contradict any of them. |
375 | */ | |
2c2d71f5 | 376 | |
6eb5f6b9 JH |
377 | /* Most of decisions we do here should have been done at compile time. |
378 | The nodes of the REx which we used for the search should have been | |
379 | deleted from the finite automaton. */ | |
380 | ||
cad2e5aa | 381 | char * |
288b8c02 | 382 | Perl_re_intuit_start(pTHX_ REGEXP * const rx, SV *sv, char *strpos, |
9f61653a | 383 | char *strend, const U32 flags, re_scream_pos_data *data) |
cad2e5aa | 384 | { |
97aff369 | 385 | dVAR; |
288b8c02 | 386 | struct regexp *const prog = (struct regexp *)SvANY(rx); |
b7953727 | 387 | register I32 start_shift = 0; |
cad2e5aa | 388 | /* Should be nonnegative! */ |
b7953727 | 389 | register I32 end_shift = 0; |
2c2d71f5 JH |
390 | register char *s; |
391 | register SV *check; | |
a1933d95 | 392 | char *strbeg; |
cad2e5aa | 393 | char *t; |
1de06328 | 394 | const bool do_utf8 = (sv && SvUTF8(sv)) ? 1 : 0; /* if no sv we have to assume bytes */ |
cad2e5aa | 395 | I32 ml_anch; |
bd61b366 SS |
396 | register char *other_last = NULL; /* other substr checked before this */ |
397 | char *check_at = NULL; /* check substr found at this pos */ | |
bbe252da | 398 | const I32 multiline = prog->extflags & RXf_PMf_MULTILINE; |
f8fc2ecf | 399 | RXi_GET_DECL(prog,progi); |
30944b6d | 400 | #ifdef DEBUGGING |
890ce7af | 401 | const char * const i_strpos = strpos; |
30944b6d | 402 | #endif |
a3621e74 YO |
403 | GET_RE_DEBUG_FLAGS_DECL; |
404 | ||
7918f24d NC |
405 | PERL_ARGS_ASSERT_RE_INTUIT_START; |
406 | ||
288b8c02 | 407 | RX_MATCH_UTF8_set(rx,do_utf8); |
cad2e5aa | 408 | |
3c8556c3 | 409 | if (RX_UTF8(rx)) { |
b8d68ded JH |
410 | PL_reg_flags |= RF_utf8; |
411 | } | |
ab3bbdeb | 412 | DEBUG_EXECUTE_r( |
efd26800 | 413 | debug_start_match(rx, do_utf8, strpos, strend, |
1de06328 YO |
414 | sv ? "Guessing start of match in sv for" |
415 | : "Guessing start of match in string for"); | |
2a782b5b | 416 | ); |
cad2e5aa | 417 | |
c344f387 JH |
418 | /* CHR_DIST() would be more correct here but it makes things slow. */ |
419 | if (prog->minlen > strend - strpos) { | |
a3621e74 | 420 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 | 421 | "String too short... [re_intuit_start]\n")); |
cad2e5aa | 422 | goto fail; |
2c2d71f5 | 423 | } |
1de06328 | 424 | |
a1933d95 | 425 | strbeg = (sv && SvPOK(sv)) ? strend - SvCUR(sv) : strpos; |
1aa99e6b | 426 | PL_regeol = strend; |
33b8afdf JH |
427 | if (do_utf8) { |
428 | if (!prog->check_utf8 && prog->check_substr) | |
429 | to_utf8_substr(prog); | |
430 | check = prog->check_utf8; | |
431 | } else { | |
432 | if (!prog->check_substr && prog->check_utf8) | |
433 | to_byte_substr(prog); | |
434 | check = prog->check_substr; | |
435 | } | |
1de06328 | 436 | if (check == &PL_sv_undef) { |
a3621e74 | 437 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
1de06328 | 438 | "Non-utf8 string cannot match utf8 check string\n")); |
33b8afdf JH |
439 | goto fail; |
440 | } | |
bbe252da YO |
441 | if (prog->extflags & RXf_ANCH) { /* Match at beg-of-str or after \n */ |
442 | ml_anch = !( (prog->extflags & RXf_ANCH_SINGLE) | |
443 | || ( (prog->extflags & RXf_ANCH_BOL) | |
7fba1cd6 | 444 | && !multiline ) ); /* Check after \n? */ |
cad2e5aa | 445 | |
7e25d62c | 446 | if (!ml_anch) { |
bbe252da YO |
447 | if ( !(prog->extflags & RXf_ANCH_GPOS) /* Checked by the caller */ |
448 | && !(prog->intflags & PREGf_IMPLICIT) /* not a real BOL */ | |
3f7c398e | 449 | /* SvCUR is not set on references: SvRV and SvPVX_const overlap */ |
7e25d62c JH |
450 | && sv && !SvROK(sv) |
451 | && (strpos != strbeg)) { | |
a3621e74 | 452 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not at start...\n")); |
7e25d62c JH |
453 | goto fail; |
454 | } | |
455 | if (prog->check_offset_min == prog->check_offset_max && | |
bbe252da | 456 | !(prog->extflags & RXf_CANY_SEEN)) { |
2c2d71f5 | 457 | /* Substring at constant offset from beg-of-str... */ |
cad2e5aa JH |
458 | I32 slen; |
459 | ||
1aa99e6b | 460 | s = HOP3c(strpos, prog->check_offset_min, strend); |
1de06328 | 461 | |
653099ff GS |
462 | if (SvTAIL(check)) { |
463 | slen = SvCUR(check); /* >= 1 */ | |
cad2e5aa | 464 | |
9041c2e3 | 465 | if ( strend - s > slen || strend - s < slen - 1 |
2c2d71f5 | 466 | || (strend - s == slen && strend[-1] != '\n')) { |
a3621e74 | 467 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String too long...\n")); |
2c2d71f5 | 468 | goto fail_finish; |
cad2e5aa JH |
469 | } |
470 | /* Now should match s[0..slen-2] */ | |
471 | slen--; | |
3f7c398e | 472 | if (slen && (*SvPVX_const(check) != *s |
cad2e5aa | 473 | || (slen > 1 |
3f7c398e | 474 | && memNE(SvPVX_const(check), s, slen)))) { |
2c2d71f5 | 475 | report_neq: |
a3621e74 | 476 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "String not equal...\n")); |
2c2d71f5 JH |
477 | goto fail_finish; |
478 | } | |
cad2e5aa | 479 | } |
3f7c398e | 480 | else if (*SvPVX_const(check) != *s |
653099ff | 481 | || ((slen = SvCUR(check)) > 1 |
3f7c398e | 482 | && memNE(SvPVX_const(check), s, slen))) |
2c2d71f5 | 483 | goto report_neq; |
c315bfe8 | 484 | check_at = s; |
2c2d71f5 | 485 | goto success_at_start; |
7e25d62c | 486 | } |
cad2e5aa | 487 | } |
2c2d71f5 | 488 | /* Match is anchored, but substr is not anchored wrt beg-of-str. */ |
cad2e5aa | 489 | s = strpos; |
2c2d71f5 | 490 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
1de06328 YO |
491 | end_shift = prog->check_end_shift; |
492 | ||
2c2d71f5 | 493 | if (!ml_anch) { |
a3b680e6 | 494 | const I32 end = prog->check_offset_max + CHR_SVLEN(check) |
653099ff | 495 | - (SvTAIL(check) != 0); |
a3b680e6 | 496 | const I32 eshift = CHR_DIST((U8*)strend, (U8*)s) - end; |
2c2d71f5 JH |
497 | |
498 | if (end_shift < eshift) | |
499 | end_shift = eshift; | |
500 | } | |
cad2e5aa | 501 | } |
2c2d71f5 | 502 | else { /* Can match at random position */ |
cad2e5aa JH |
503 | ml_anch = 0; |
504 | s = strpos; | |
1de06328 YO |
505 | start_shift = prog->check_offset_min; /* okay to underestimate on CC */ |
506 | end_shift = prog->check_end_shift; | |
507 | ||
508 | /* end shift should be non negative here */ | |
cad2e5aa JH |
509 | } |
510 | ||
bcdf7404 | 511 | #ifdef QDEBUGGING /* 7/99: reports of failure (with the older version) */ |
0033605d | 512 | if (end_shift < 0) |
1de06328 | 513 | Perl_croak(aTHX_ "panic: end_shift: %"IVdf" pattern:\n%s\n ", |
220fc49f | 514 | (IV)end_shift, RX_PRECOMP(prog)); |
2c2d71f5 JH |
515 | #endif |
516 | ||
2c2d71f5 JH |
517 | restart: |
518 | /* Find a possible match in the region s..strend by looking for | |
519 | the "check" substring in the region corrected by start/end_shift. */ | |
1de06328 YO |
520 | |
521 | { | |
522 | I32 srch_start_shift = start_shift; | |
523 | I32 srch_end_shift = end_shift; | |
524 | if (srch_start_shift < 0 && strbeg - s > srch_start_shift) { | |
525 | srch_end_shift -= ((strbeg - s) - srch_start_shift); | |
526 | srch_start_shift = strbeg - s; | |
527 | } | |
6bda09f9 | 528 | DEBUG_OPTIMISE_MORE_r({ |
1de06328 YO |
529 | PerlIO_printf(Perl_debug_log, "Check offset min: %"IVdf" Start shift: %"IVdf" End shift %"IVdf" Real End Shift: %"IVdf"\n", |
530 | (IV)prog->check_offset_min, | |
531 | (IV)srch_start_shift, | |
532 | (IV)srch_end_shift, | |
533 | (IV)prog->check_end_shift); | |
534 | }); | |
535 | ||
cad2e5aa | 536 | if (flags & REXEC_SCREAM) { |
cad2e5aa | 537 | I32 p = -1; /* Internal iterator of scream. */ |
a3b680e6 | 538 | I32 * const pp = data ? data->scream_pos : &p; |
cad2e5aa | 539 | |
2c2d71f5 JH |
540 | if (PL_screamfirst[BmRARE(check)] >= 0 |
541 | || ( BmRARE(check) == '\n' | |
85c508c3 | 542 | && (BmPREVIOUS(check) == SvCUR(check) - 1) |
2c2d71f5 | 543 | && SvTAIL(check) )) |
9041c2e3 | 544 | s = screaminstr(sv, check, |
1de06328 | 545 | srch_start_shift + (s - strbeg), srch_end_shift, pp, 0); |
cad2e5aa | 546 | else |
2c2d71f5 | 547 | goto fail_finish; |
4addbd3b | 548 | /* we may be pointing at the wrong string */ |
07bc277f | 549 | if (s && RXp_MATCH_COPIED(prog)) |
3f7c398e | 550 | s = strbeg + (s - SvPVX_const(sv)); |
cad2e5aa JH |
551 | if (data) |
552 | *data->scream_olds = s; | |
553 | } | |
1de06328 YO |
554 | else { |
555 | U8* start_point; | |
556 | U8* end_point; | |
bbe252da | 557 | if (prog->extflags & RXf_CANY_SEEN) { |
1de06328 YO |
558 | start_point= (U8*)(s + srch_start_shift); |
559 | end_point= (U8*)(strend - srch_end_shift); | |
560 | } else { | |
561 | start_point= HOP3(s, srch_start_shift, srch_start_shift < 0 ? strbeg : strend); | |
562 | end_point= HOP3(strend, -srch_end_shift, strbeg); | |
563 | } | |
6bda09f9 | 564 | DEBUG_OPTIMISE_MORE_r({ |
56570a2c | 565 | PerlIO_printf(Perl_debug_log, "fbm_instr len=%d str=<%.*s>\n", |
1de06328 | 566 | (int)(end_point - start_point), |
fc8cd66c | 567 | (int)(end_point - start_point) > 20 ? 20 : (int)(end_point - start_point), |
1de06328 YO |
568 | start_point); |
569 | }); | |
570 | ||
571 | s = fbm_instr( start_point, end_point, | |
7fba1cd6 | 572 | check, multiline ? FBMrf_MULTILINE : 0); |
1de06328 YO |
573 | } |
574 | } | |
cad2e5aa JH |
575 | /* Update the count-of-usability, remove useless subpatterns, |
576 | unshift s. */ | |
2c2d71f5 | 577 | |
ab3bbdeb YO |
578 | DEBUG_EXECUTE_r({ |
579 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
580 | SvPVX_const(check), RE_SV_DUMPLEN(check), 30); | |
581 | PerlIO_printf(Perl_debug_log, "%s %s substr %s%s%s", | |
2c2d71f5 | 582 | (s ? "Found" : "Did not find"), |
ab3bbdeb YO |
583 | (check == (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) |
584 | ? "anchored" : "floating"), | |
585 | quoted, | |
586 | RE_SV_TAIL(check), | |
587 | (s ? " at offset " : "...\n") ); | |
588 | }); | |
2c2d71f5 JH |
589 | |
590 | if (!s) | |
591 | goto fail_finish; | |
2c2d71f5 | 592 | /* Finish the diagnostic message */ |
a3621e74 | 593 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%ld...\n", (long)(s - i_strpos)) ); |
2c2d71f5 | 594 | |
1de06328 YO |
595 | /* XXX dmq: first branch is for positive lookbehind... |
596 | Our check string is offset from the beginning of the pattern. | |
597 | So we need to do any stclass tests offset forward from that | |
598 | point. I think. :-( | |
599 | */ | |
600 | ||
601 | ||
602 | ||
603 | check_at=s; | |
604 | ||
605 | ||
2c2d71f5 JH |
606 | /* Got a candidate. Check MBOL anchoring, and the *other* substr. |
607 | Start with the other substr. | |
608 | XXXX no SCREAM optimization yet - and a very coarse implementation | |
a0288114 | 609 | XXXX /ttx+/ results in anchored="ttx", floating="x". floating will |
2c2d71f5 JH |
610 | *always* match. Probably should be marked during compile... |
611 | Probably it is right to do no SCREAM here... | |
612 | */ | |
613 | ||
1de06328 YO |
614 | if (do_utf8 ? (prog->float_utf8 && prog->anchored_utf8) |
615 | : (prog->float_substr && prog->anchored_substr)) | |
616 | { | |
30944b6d | 617 | /* Take into account the "other" substring. */ |
2c2d71f5 JH |
618 | /* XXXX May be hopelessly wrong for UTF... */ |
619 | if (!other_last) | |
6eb5f6b9 | 620 | other_last = strpos; |
33b8afdf | 621 | if (check == (do_utf8 ? prog->float_utf8 : prog->float_substr)) { |
30944b6d IZ |
622 | do_other_anchored: |
623 | { | |
890ce7af AL |
624 | char * const last = HOP3c(s, -start_shift, strbeg); |
625 | char *last1, *last2; | |
be8e71aa | 626 | char * const saved_s = s; |
33b8afdf | 627 | SV* must; |
2c2d71f5 | 628 | |
2c2d71f5 JH |
629 | t = s - prog->check_offset_max; |
630 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ | |
1d86a7f9 | 631 | && (!do_utf8 |
0ce71af7 | 632 | || ((t = (char*)reghopmaybe3((U8*)s, -(prog->check_offset_max), (U8*)strpos)) |
2c2d71f5 | 633 | && t > strpos))) |
6f207bd3 | 634 | NOOP; |
2c2d71f5 JH |
635 | else |
636 | t = strpos; | |
1aa99e6b | 637 | t = HOP3c(t, prog->anchored_offset, strend); |
6eb5f6b9 JH |
638 | if (t < other_last) /* These positions already checked */ |
639 | t = other_last; | |
1aa99e6b | 640 | last2 = last1 = HOP3c(strend, -prog->minlen, strbeg); |
2c2d71f5 JH |
641 | if (last < last1) |
642 | last1 = last; | |
1de06328 YO |
643 | /* XXXX It is not documented what units *_offsets are in. |
644 | We assume bytes, but this is clearly wrong. | |
645 | Meaning this code needs to be carefully reviewed for errors. | |
646 | dmq. | |
647 | */ | |
648 | ||
2c2d71f5 | 649 | /* On end-of-str: see comment below. */ |
33b8afdf JH |
650 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; |
651 | if (must == &PL_sv_undef) { | |
652 | s = (char*)NULL; | |
1de06328 | 653 | DEBUG_r(must = prog->anchored_utf8); /* for debug */ |
33b8afdf JH |
654 | } |
655 | else | |
656 | s = fbm_instr( | |
657 | (unsigned char*)t, | |
658 | HOP3(HOP3(last1, prog->anchored_offset, strend) | |
659 | + SvCUR(must), -(SvTAIL(must)!=0), strbeg), | |
660 | must, | |
7fba1cd6 | 661 | multiline ? FBMrf_MULTILINE : 0 |
33b8afdf | 662 | ); |
ab3bbdeb YO |
663 | DEBUG_EXECUTE_r({ |
664 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
665 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
666 | PerlIO_printf(Perl_debug_log, "%s anchored substr %s%s", | |
2c2d71f5 | 667 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
668 | quoted, RE_SV_TAIL(must)); |
669 | }); | |
670 | ||
671 | ||
2c2d71f5 JH |
672 | if (!s) { |
673 | if (last1 >= last2) { | |
a3621e74 | 674 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 JH |
675 | ", giving up...\n")); |
676 | goto fail_finish; | |
677 | } | |
a3621e74 | 678 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
2c2d71f5 | 679 | ", trying floating at offset %ld...\n", |
be8e71aa | 680 | (long)(HOP3c(saved_s, 1, strend) - i_strpos))); |
1aa99e6b IH |
681 | other_last = HOP3c(last1, prog->anchored_offset+1, strend); |
682 | s = HOP3c(last, 1, strend); | |
2c2d71f5 JH |
683 | goto restart; |
684 | } | |
685 | else { | |
a3621e74 | 686 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
30944b6d | 687 | (long)(s - i_strpos))); |
1aa99e6b IH |
688 | t = HOP3c(s, -prog->anchored_offset, strbeg); |
689 | other_last = HOP3c(s, 1, strend); | |
be8e71aa | 690 | s = saved_s; |
2c2d71f5 JH |
691 | if (t == strpos) |
692 | goto try_at_start; | |
2c2d71f5 JH |
693 | goto try_at_offset; |
694 | } | |
30944b6d | 695 | } |
2c2d71f5 JH |
696 | } |
697 | else { /* Take into account the floating substring. */ | |
33b8afdf | 698 | char *last, *last1; |
be8e71aa | 699 | char * const saved_s = s; |
33b8afdf JH |
700 | SV* must; |
701 | ||
702 | t = HOP3c(s, -start_shift, strbeg); | |
703 | last1 = last = | |
704 | HOP3c(strend, -prog->minlen + prog->float_min_offset, strbeg); | |
705 | if (CHR_DIST((U8*)last, (U8*)t) > prog->float_max_offset) | |
706 | last = HOP3c(t, prog->float_max_offset, strend); | |
707 | s = HOP3c(t, prog->float_min_offset, strend); | |
708 | if (s < other_last) | |
709 | s = other_last; | |
2c2d71f5 | 710 | /* XXXX It is not documented what units *_offsets are in. Assume bytes. */ |
33b8afdf JH |
711 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; |
712 | /* fbm_instr() takes into account exact value of end-of-str | |
713 | if the check is SvTAIL(ed). Since false positives are OK, | |
714 | and end-of-str is not later than strend we are OK. */ | |
715 | if (must == &PL_sv_undef) { | |
716 | s = (char*)NULL; | |
1de06328 | 717 | DEBUG_r(must = prog->float_utf8); /* for debug message */ |
33b8afdf JH |
718 | } |
719 | else | |
2c2d71f5 | 720 | s = fbm_instr((unsigned char*)s, |
33b8afdf JH |
721 | (unsigned char*)last + SvCUR(must) |
722 | - (SvTAIL(must)!=0), | |
7fba1cd6 | 723 | must, multiline ? FBMrf_MULTILINE : 0); |
ab3bbdeb YO |
724 | DEBUG_EXECUTE_r({ |
725 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
726 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
727 | PerlIO_printf(Perl_debug_log, "%s floating substr %s%s", | |
33b8afdf | 728 | (s ? "Found" : "Contradicts"), |
ab3bbdeb YO |
729 | quoted, RE_SV_TAIL(must)); |
730 | }); | |
33b8afdf JH |
731 | if (!s) { |
732 | if (last1 == last) { | |
a3621e74 | 733 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf JH |
734 | ", giving up...\n")); |
735 | goto fail_finish; | |
2c2d71f5 | 736 | } |
a3621e74 | 737 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
33b8afdf | 738 | ", trying anchored starting at offset %ld...\n", |
be8e71aa | 739 | (long)(saved_s + 1 - i_strpos))); |
33b8afdf JH |
740 | other_last = last; |
741 | s = HOP3c(t, 1, strend); | |
742 | goto restart; | |
743 | } | |
744 | else { | |
a3621e74 | 745 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, " at offset %ld...\n", |
33b8afdf JH |
746 | (long)(s - i_strpos))); |
747 | other_last = s; /* Fix this later. --Hugo */ | |
be8e71aa | 748 | s = saved_s; |
33b8afdf JH |
749 | if (t == strpos) |
750 | goto try_at_start; | |
751 | goto try_at_offset; | |
752 | } | |
2c2d71f5 | 753 | } |
cad2e5aa | 754 | } |
2c2d71f5 | 755 | |
1de06328 | 756 | |
9ef43ace | 757 | t= (char*)HOP3( s, -prog->check_offset_max, (prog->check_offset_max<0) ? strend : strpos); |
1de06328 | 758 | |
6bda09f9 | 759 | DEBUG_OPTIMISE_MORE_r( |
1de06328 YO |
760 | PerlIO_printf(Perl_debug_log, |
761 | "Check offset min:%"IVdf" max:%"IVdf" S:%"IVdf" t:%"IVdf" D:%"IVdf" end:%"IVdf"\n", | |
762 | (IV)prog->check_offset_min, | |
763 | (IV)prog->check_offset_max, | |
764 | (IV)(s-strpos), | |
765 | (IV)(t-strpos), | |
766 | (IV)(t-s), | |
767 | (IV)(strend-strpos) | |
768 | ) | |
769 | ); | |
770 | ||
2c2d71f5 | 771 | if (s - strpos > prog->check_offset_max /* signed-corrected t > strpos */ |
1d86a7f9 | 772 | && (!do_utf8 |
9ef43ace | 773 | || ((t = (char*)reghopmaybe3((U8*)s, -prog->check_offset_max, (U8*) ((prog->check_offset_max<0) ? strend : strpos))) |
1de06328 YO |
774 | && t > strpos))) |
775 | { | |
2c2d71f5 JH |
776 | /* Fixed substring is found far enough so that the match |
777 | cannot start at strpos. */ | |
778 | try_at_offset: | |
cad2e5aa | 779 | if (ml_anch && t[-1] != '\n') { |
30944b6d IZ |
780 | /* Eventually fbm_*() should handle this, but often |
781 | anchored_offset is not 0, so this check will not be wasted. */ | |
782 | /* XXXX In the code below we prefer to look for "^" even in | |
783 | presence of anchored substrings. And we search even | |
784 | beyond the found float position. These pessimizations | |
785 | are historical artefacts only. */ | |
786 | find_anchor: | |
2c2d71f5 | 787 | while (t < strend - prog->minlen) { |
cad2e5aa | 788 | if (*t == '\n') { |
4ee3650e | 789 | if (t < check_at - prog->check_offset_min) { |
33b8afdf | 790 | if (do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) { |
4ee3650e GS |
791 | /* Since we moved from the found position, |
792 | we definitely contradict the found anchored | |
30944b6d IZ |
793 | substr. Due to the above check we do not |
794 | contradict "check" substr. | |
795 | Thus we can arrive here only if check substr | |
796 | is float. Redo checking for "other"=="fixed". | |
797 | */ | |
9041c2e3 | 798 | strpos = t + 1; |
a3621e74 | 799 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld, rescanning for anchored from offset %ld...\n", |
e4584336 | 800 | PL_colors[0], PL_colors[1], (long)(strpos - i_strpos), (long)(strpos - i_strpos + prog->anchored_offset))); |
30944b6d IZ |
801 | goto do_other_anchored; |
802 | } | |
4ee3650e GS |
803 | /* We don't contradict the found floating substring. */ |
804 | /* XXXX Why not check for STCLASS? */ | |
cad2e5aa | 805 | s = t + 1; |
a3621e74 | 806 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m at offset %ld...\n", |
e4584336 | 807 | PL_colors[0], PL_colors[1], (long)(s - i_strpos))); |
cad2e5aa JH |
808 | goto set_useful; |
809 | } | |
4ee3650e GS |
810 | /* Position contradicts check-string */ |
811 | /* XXXX probably better to look for check-string | |
812 | than for "\n", so one should lower the limit for t? */ | |
a3621e74 | 813 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Found /%s^%s/m, restarting lookup for check-string at offset %ld...\n", |
e4584336 | 814 | PL_colors[0], PL_colors[1], (long)(t + 1 - i_strpos))); |
0e41cd87 | 815 | other_last = strpos = s = t + 1; |
cad2e5aa JH |
816 | goto restart; |
817 | } | |
818 | t++; | |
819 | } | |
a3621e74 | 820 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Did not find /%s^%s/m...\n", |
e4584336 | 821 | PL_colors[0], PL_colors[1])); |
2c2d71f5 | 822 | goto fail_finish; |
cad2e5aa | 823 | } |
f5952150 | 824 | else { |
a3621e74 | 825 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Starting position does not contradict /%s^%s/m...\n", |
e4584336 | 826 | PL_colors[0], PL_colors[1])); |
f5952150 | 827 | } |
cad2e5aa JH |
828 | s = t; |
829 | set_useful: | |
33b8afdf | 830 | ++BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr); /* hooray/5 */ |
cad2e5aa JH |
831 | } |
832 | else { | |
f5952150 | 833 | /* The found string does not prohibit matching at strpos, |
2c2d71f5 | 834 | - no optimization of calling REx engine can be performed, |
f5952150 GS |
835 | unless it was an MBOL and we are not after MBOL, |
836 | or a future STCLASS check will fail this. */ | |
2c2d71f5 JH |
837 | try_at_start: |
838 | /* Even in this situation we may use MBOL flag if strpos is offset | |
839 | wrt the start of the string. */ | |
05b4157f | 840 | if (ml_anch && sv && !SvROK(sv) /* See prev comment on SvROK */ |
a1933d95 | 841 | && (strpos != strbeg) && strpos[-1] != '\n' |
d506a20d | 842 | /* May be due to an implicit anchor of m{.*foo} */ |
bbe252da | 843 | && !(prog->intflags & PREGf_IMPLICIT)) |
d506a20d | 844 | { |
cad2e5aa JH |
845 | t = strpos; |
846 | goto find_anchor; | |
847 | } | |
a3621e74 | 848 | DEBUG_EXECUTE_r( if (ml_anch) |
f5952150 | 849 | PerlIO_printf(Perl_debug_log, "Position at offset %ld does not contradict /%s^%s/m...\n", |
70685ca0 | 850 | (long)(strpos - i_strpos), PL_colors[0], PL_colors[1]); |
30944b6d | 851 | ); |
2c2d71f5 | 852 | success_at_start: |
bbe252da | 853 | if (!(prog->intflags & PREGf_NAUGHTY) /* XXXX If strpos moved? */ |
33b8afdf JH |
854 | && (do_utf8 ? ( |
855 | prog->check_utf8 /* Could be deleted already */ | |
856 | && --BmUSEFUL(prog->check_utf8) < 0 | |
857 | && (prog->check_utf8 == prog->float_utf8) | |
858 | ) : ( | |
859 | prog->check_substr /* Could be deleted already */ | |
860 | && --BmUSEFUL(prog->check_substr) < 0 | |
861 | && (prog->check_substr == prog->float_substr) | |
862 | ))) | |
66e933ab | 863 | { |
cad2e5aa | 864 | /* If flags & SOMETHING - do not do it many times on the same match */ |
a3621e74 | 865 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "... Disabling check substring...\n")); |
33b8afdf JH |
866 | SvREFCNT_dec(do_utf8 ? prog->check_utf8 : prog->check_substr); |
867 | if (do_utf8 ? prog->check_substr : prog->check_utf8) | |
868 | SvREFCNT_dec(do_utf8 ? prog->check_substr : prog->check_utf8); | |
a0714e2c SS |
869 | prog->check_substr = prog->check_utf8 = NULL; /* disable */ |
870 | prog->float_substr = prog->float_utf8 = NULL; /* clear */ | |
871 | check = NULL; /* abort */ | |
cad2e5aa | 872 | s = strpos; |
3cf5c195 IZ |
873 | /* XXXX This is a remnant of the old implementation. It |
874 | looks wasteful, since now INTUIT can use many | |
6eb5f6b9 | 875 | other heuristics. */ |
bbe252da | 876 | prog->extflags &= ~RXf_USE_INTUIT; |
cad2e5aa JH |
877 | } |
878 | else | |
879 | s = strpos; | |
880 | } | |
881 | ||
6eb5f6b9 JH |
882 | /* Last resort... */ |
883 | /* XXXX BmUSEFUL already changed, maybe multiple change is meaningful... */ | |
1de06328 YO |
884 | /* trie stclasses are too expensive to use here, we are better off to |
885 | leave it to regmatch itself */ | |
f8fc2ecf | 886 | if (progi->regstclass && PL_regkind[OP(progi->regstclass)]!=TRIE) { |
6eb5f6b9 JH |
887 | /* minlen == 0 is possible if regstclass is \b or \B, |
888 | and the fixed substr is ''$. | |
889 | Since minlen is already taken into account, s+1 is before strend; | |
890 | accidentally, minlen >= 1 guaranties no false positives at s + 1 | |
891 | even for \b or \B. But (minlen? 1 : 0) below assumes that | |
892 | regstclass does not come from lookahead... */ | |
893 | /* If regstclass takes bytelength more than 1: If charlength==1, OK. | |
894 | This leaves EXACTF only, which is dealt with in find_byclass(). */ | |
f8fc2ecf YO |
895 | const U8* const str = (U8*)STRING(progi->regstclass); |
896 | const int cl_l = (PL_regkind[OP(progi->regstclass)] == EXACT | |
897 | ? CHR_DIST(str+STR_LEN(progi->regstclass), str) | |
66e933ab | 898 | : 1); |
1de06328 YO |
899 | char * endpos; |
900 | if (prog->anchored_substr || prog->anchored_utf8 || ml_anch) | |
901 | endpos= HOP3c(s, (prog->minlen ? cl_l : 0), strend); | |
902 | else if (prog->float_substr || prog->float_utf8) | |
903 | endpos= HOP3c(HOP3c(check_at, -start_shift, strbeg), cl_l, strend); | |
904 | else | |
905 | endpos= strend; | |
906 | ||
70685ca0 JH |
907 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "start_shift: %"IVdf" check_at: %"IVdf" s: %"IVdf" endpos: %"IVdf"\n", |
908 | (IV)start_shift, (IV)(check_at - strbeg), (IV)(s - strbeg), (IV)(endpos - strbeg))); | |
1de06328 | 909 | |
6eb5f6b9 | 910 | t = s; |
f8fc2ecf | 911 | s = find_byclass(prog, progi->regstclass, s, endpos, NULL); |
6eb5f6b9 JH |
912 | if (!s) { |
913 | #ifdef DEBUGGING | |
cbbf8932 | 914 | const char *what = NULL; |
6eb5f6b9 JH |
915 | #endif |
916 | if (endpos == strend) { | |
a3621e74 | 917 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
6eb5f6b9 JH |
918 | "Could not match STCLASS...\n") ); |
919 | goto fail; | |
920 | } | |
a3621e74 | 921 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab | 922 | "This position contradicts STCLASS...\n") ); |
bbe252da | 923 | if ((prog->extflags & RXf_ANCH) && !ml_anch) |
653099ff | 924 | goto fail; |
6eb5f6b9 | 925 | /* Contradict one of substrings */ |
33b8afdf JH |
926 | if (prog->anchored_substr || prog->anchored_utf8) { |
927 | if ((do_utf8 ? prog->anchored_utf8 : prog->anchored_substr) == check) { | |
a3621e74 | 928 | DEBUG_EXECUTE_r( what = "anchored" ); |
6eb5f6b9 | 929 | hop_and_restart: |
1aa99e6b | 930 | s = HOP3c(t, 1, strend); |
66e933ab GS |
931 | if (s + start_shift + end_shift > strend) { |
932 | /* XXXX Should be taken into account earlier? */ | |
a3621e74 | 933 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
66e933ab GS |
934 | "Could not match STCLASS...\n") ); |
935 | goto fail; | |
936 | } | |
5e39e1e5 HS |
937 | if (!check) |
938 | goto giveup; | |
a3621e74 | 939 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 940 | "Looking for %s substr starting at offset %ld...\n", |
6eb5f6b9 JH |
941 | what, (long)(s + start_shift - i_strpos)) ); |
942 | goto restart; | |
943 | } | |
66e933ab | 944 | /* Have both, check_string is floating */ |
6eb5f6b9 JH |
945 | if (t + start_shift >= check_at) /* Contradicts floating=check */ |
946 | goto retry_floating_check; | |
947 | /* Recheck anchored substring, but not floating... */ | |
9041c2e3 | 948 | s = check_at; |
5e39e1e5 HS |
949 | if (!check) |
950 | goto giveup; | |
a3621e74 | 951 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 952 | "Looking for anchored substr starting at offset %ld...\n", |
6eb5f6b9 JH |
953 | (long)(other_last - i_strpos)) ); |
954 | goto do_other_anchored; | |
955 | } | |
60e71179 GS |
956 | /* Another way we could have checked stclass at the |
957 | current position only: */ | |
958 | if (ml_anch) { | |
959 | s = t = t + 1; | |
5e39e1e5 HS |
960 | if (!check) |
961 | goto giveup; | |
a3621e74 | 962 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
f5952150 | 963 | "Looking for /%s^%s/m starting at offset %ld...\n", |
e4584336 | 964 | PL_colors[0], PL_colors[1], (long)(t - i_strpos)) ); |
60e71179 | 965 | goto try_at_offset; |
66e933ab | 966 | } |
33b8afdf | 967 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) /* Could have been deleted */ |
60e71179 | 968 | goto fail; |
6eb5f6b9 JH |
969 | /* Check is floating subtring. */ |
970 | retry_floating_check: | |
971 | t = check_at - start_shift; | |
a3621e74 | 972 | DEBUG_EXECUTE_r( what = "floating" ); |
6eb5f6b9 JH |
973 | goto hop_and_restart; |
974 | } | |
b7953727 | 975 | if (t != s) { |
a3621e74 | 976 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
6eb5f6b9 | 977 | "By STCLASS: moving %ld --> %ld\n", |
b7953727 JH |
978 | (long)(t - i_strpos), (long)(s - i_strpos)) |
979 | ); | |
980 | } | |
981 | else { | |
a3621e74 | 982 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
b7953727 JH |
983 | "Does not contradict STCLASS...\n"); |
984 | ); | |
985 | } | |
6eb5f6b9 | 986 | } |
5e39e1e5 | 987 | giveup: |
a3621e74 | 988 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%s%s:%s match at offset %ld\n", |
5e39e1e5 HS |
989 | PL_colors[4], (check ? "Guessed" : "Giving up"), |
990 | PL_colors[5], (long)(s - i_strpos)) ); | |
cad2e5aa | 991 | return s; |
2c2d71f5 JH |
992 | |
993 | fail_finish: /* Substring not found */ | |
33b8afdf JH |
994 | if (prog->check_substr || prog->check_utf8) /* could be removed already */ |
995 | BmUSEFUL(do_utf8 ? prog->check_utf8 : prog->check_substr) += 5; /* hooray */ | |
cad2e5aa | 996 | fail: |
a3621e74 | 997 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch rejected by optimizer%s\n", |
e4584336 | 998 | PL_colors[4], PL_colors[5])); |
bd61b366 | 999 | return NULL; |
cad2e5aa | 1000 | } |
9661b544 | 1001 | |
a0a388a1 YO |
1002 | #define DECL_TRIE_TYPE(scan) \ |
1003 | const enum { trie_plain, trie_utf8, trie_utf8_fold, trie_latin_utf8_fold } \ | |
1004 | trie_type = (scan->flags != EXACT) \ | |
1005 | ? (do_utf8 ? trie_utf8_fold : (UTF ? trie_latin_utf8_fold : trie_plain)) \ | |
1006 | : (do_utf8 ? trie_utf8 : trie_plain) | |
3b0527fe | 1007 | |
55eed653 NC |
1008 | #define REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, uscan, len, \ |
1009 | uvc, charid, foldlen, foldbuf, uniflags) STMT_START { \ | |
c012444f | 1010 | UV uvc_unfolded = 0; \ |
4cadc6a9 YO |
1011 | switch (trie_type) { \ |
1012 | case trie_utf8_fold: \ | |
1013 | if ( foldlen>0 ) { \ | |
c012444f | 1014 | uvc_unfolded = uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ |
4cadc6a9 YO |
1015 | foldlen -= len; \ |
1016 | uscan += len; \ | |
1017 | len=0; \ | |
1018 | } else { \ | |
c012444f | 1019 | uvc_unfolded = uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ |
4cadc6a9 YO |
1020 | uvc = to_uni_fold( uvc, foldbuf, &foldlen ); \ |
1021 | foldlen -= UNISKIP( uvc ); \ | |
1022 | uscan = foldbuf + UNISKIP( uvc ); \ | |
1023 | } \ | |
1024 | break; \ | |
a0a388a1 YO |
1025 | case trie_latin_utf8_fold: \ |
1026 | if ( foldlen>0 ) { \ | |
1027 | uvc = utf8n_to_uvuni( uscan, UTF8_MAXLEN, &len, uniflags ); \ | |
1028 | foldlen -= len; \ | |
1029 | uscan += len; \ | |
1030 | len=0; \ | |
1031 | } else { \ | |
1032 | len = 1; \ | |
1033 | uvc = to_uni_fold( *(U8*)uc, foldbuf, &foldlen ); \ | |
1034 | foldlen -= UNISKIP( uvc ); \ | |
1035 | uscan = foldbuf + UNISKIP( uvc ); \ | |
1036 | } \ | |
1037 | break; \ | |
4cadc6a9 YO |
1038 | case trie_utf8: \ |
1039 | uvc = utf8n_to_uvuni( (U8*)uc, UTF8_MAXLEN, &len, uniflags ); \ | |
1040 | break; \ | |
1041 | case trie_plain: \ | |
1042 | uvc = (UV)*uc; \ | |
1043 | len = 1; \ | |
1044 | } \ | |
1045 | \ | |
1046 | if (uvc < 256) { \ | |
1047 | charid = trie->charmap[ uvc ]; \ | |
1048 | } \ | |
1049 | else { \ | |
1050 | charid = 0; \ | |
55eed653 NC |
1051 | if (widecharmap) { \ |
1052 | SV** const svpp = hv_fetch(widecharmap, \ | |
4cadc6a9 YO |
1053 | (char*)&uvc, sizeof(UV), 0); \ |
1054 | if (svpp) \ | |
1055 | charid = (U16)SvIV(*svpp); \ | |
1056 | } \ | |
1057 | } \ | |
c012444f SR |
1058 | if (!charid && trie_type == trie_utf8_fold && !UTF) { \ |
1059 | charid = trie->charmap[uvc_unfolded]; \ | |
1060 | } \ | |
4cadc6a9 YO |
1061 | } STMT_END |
1062 | ||
a0a388a1 YO |
1063 | #define REXEC_FBC_EXACTISH_CHECK(CoNd) \ |
1064 | { \ | |
1065 | char *my_strend= (char *)strend; \ | |
4cadc6a9 YO |
1066 | if ( (CoNd) \ |
1067 | && (ln == len || \ | |
a0a388a1 | 1068 | !ibcmp_utf8(s, &my_strend, 0, do_utf8, \ |
4cadc6a9 | 1069 | m, NULL, ln, (bool)UTF)) \ |
a0a388a1 | 1070 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1071 | goto got_it; \ |
1072 | else { \ | |
1073 | U8 foldbuf[UTF8_MAXBYTES_CASE+1]; \ | |
1074 | uvchr_to_utf8(tmpbuf, c); \ | |
1075 | f = to_utf8_fold(tmpbuf, foldbuf, &foldlen); \ | |
1076 | if ( f != c \ | |
1077 | && (f == c1 || f == c2) \ | |
a0a388a1 YO |
1078 | && (ln == len || \ |
1079 | !ibcmp_utf8(s, &my_strend, 0, do_utf8,\ | |
1080 | m, NULL, ln, (bool)UTF)) \ | |
1081 | && (!reginfo || regtry(reginfo, &s)) ) \ | |
4cadc6a9 YO |
1082 | goto got_it; \ |
1083 | } \ | |
a0a388a1 YO |
1084 | } \ |
1085 | s += len | |
4cadc6a9 YO |
1086 | |
1087 | #define REXEC_FBC_EXACTISH_SCAN(CoNd) \ | |
1088 | STMT_START { \ | |
1089 | while (s <= e) { \ | |
1090 | if ( (CoNd) \ | |
1091 | && (ln == 1 || !(OP(c) == EXACTF \ | |
1092 | ? ibcmp(s, m, ln) \ | |
1093 | : ibcmp_locale(s, m, ln))) \ | |
24b23f37 | 1094 | && (!reginfo || regtry(reginfo, &s)) ) \ |
4cadc6a9 YO |
1095 | goto got_it; \ |
1096 | s++; \ | |
1097 | } \ | |
1098 | } STMT_END | |
1099 | ||
1100 | #define REXEC_FBC_UTF8_SCAN(CoDe) \ | |
1101 | STMT_START { \ | |
1102 | while (s + (uskip = UTF8SKIP(s)) <= strend) { \ | |
1103 | CoDe \ | |
1104 | s += uskip; \ | |
1105 | } \ | |
1106 | } STMT_END | |
1107 | ||
1108 | #define REXEC_FBC_SCAN(CoDe) \ | |
1109 | STMT_START { \ | |
1110 | while (s < strend) { \ | |
1111 | CoDe \ | |
1112 | s++; \ | |
1113 | } \ | |
1114 | } STMT_END | |
1115 | ||
1116 | #define REXEC_FBC_UTF8_CLASS_SCAN(CoNd) \ | |
1117 | REXEC_FBC_UTF8_SCAN( \ | |
1118 | if (CoNd) { \ | |
24b23f37 | 1119 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1120 | goto got_it; \ |
1121 | else \ | |
1122 | tmp = doevery; \ | |
1123 | } \ | |
1124 | else \ | |
1125 | tmp = 1; \ | |
1126 | ) | |
1127 | ||
1128 | #define REXEC_FBC_CLASS_SCAN(CoNd) \ | |
1129 | REXEC_FBC_SCAN( \ | |
1130 | if (CoNd) { \ | |
24b23f37 | 1131 | if (tmp && (!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1132 | goto got_it; \ |
1133 | else \ | |
1134 | tmp = doevery; \ | |
1135 | } \ | |
1136 | else \ | |
1137 | tmp = 1; \ | |
1138 | ) | |
1139 | ||
1140 | #define REXEC_FBC_TRYIT \ | |
24b23f37 | 1141 | if ((!reginfo || regtry(reginfo, &s))) \ |
4cadc6a9 YO |
1142 | goto got_it |
1143 | ||
e1d1eefb YO |
1144 | #define REXEC_FBC_CSCAN(CoNdUtF8,CoNd) \ |
1145 | if (do_utf8) { \ | |
1146 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1147 | } \ | |
1148 | else { \ | |
1149 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1150 | } \ | |
1151 | break | |
1152 | ||
4cadc6a9 YO |
1153 | #define REXEC_FBC_CSCAN_PRELOAD(UtFpReLoAd,CoNdUtF8,CoNd) \ |
1154 | if (do_utf8) { \ | |
1155 | UtFpReLoAd; \ | |
1156 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1157 | } \ | |
1158 | else { \ | |
1159 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1160 | } \ | |
1161 | break | |
1162 | ||
1163 | #define REXEC_FBC_CSCAN_TAINT(CoNdUtF8,CoNd) \ | |
1164 | PL_reg_flags |= RF_tainted; \ | |
1165 | if (do_utf8) { \ | |
1166 | REXEC_FBC_UTF8_CLASS_SCAN(CoNdUtF8); \ | |
1167 | } \ | |
1168 | else { \ | |
1169 | REXEC_FBC_CLASS_SCAN(CoNd); \ | |
1170 | } \ | |
1171 | break | |
1172 | ||
786e8c11 YO |
1173 | #define DUMP_EXEC_POS(li,s,doutf8) \ |
1174 | dump_exec_pos(li,s,(PL_regeol),(PL_bostr),(PL_reg_starttry),doutf8) | |
1175 | ||
1176 | /* We know what class REx starts with. Try to find this position... */ | |
1177 | /* if reginfo is NULL, its a dryrun */ | |
1178 | /* annoyingly all the vars in this routine have different names from their counterparts | |
1179 | in regmatch. /grrr */ | |
1180 | ||
3c3eec57 | 1181 | STATIC char * |
07be1b83 | 1182 | S_find_byclass(pTHX_ regexp * prog, const regnode *c, char *s, |
24b23f37 | 1183 | const char *strend, regmatch_info *reginfo) |
a687059c | 1184 | { |
27da23d5 | 1185 | dVAR; |
bbe252da | 1186 | const I32 doevery = (prog->intflags & PREGf_SKIP) == 0; |
6eb5f6b9 | 1187 | char *m; |
d8093b23 | 1188 | STRLEN ln; |
5dab1207 | 1189 | STRLEN lnc; |
078c425b | 1190 | register STRLEN uskip; |
d8093b23 G |
1191 | unsigned int c1; |
1192 | unsigned int c2; | |
6eb5f6b9 JH |
1193 | char *e; |
1194 | register I32 tmp = 1; /* Scratch variable? */ | |
a3b680e6 | 1195 | register const bool do_utf8 = PL_reg_match_utf8; |
f8fc2ecf | 1196 | RXi_GET_DECL(prog,progi); |
7918f24d NC |
1197 | |
1198 | PERL_ARGS_ASSERT_FIND_BYCLASS; | |
f8fc2ecf | 1199 | |
6eb5f6b9 JH |
1200 | /* We know what class it must start with. */ |
1201 | switch (OP(c)) { | |
6eb5f6b9 | 1202 | case ANYOF: |
388cc4de | 1203 | if (do_utf8) { |
4cadc6a9 | 1204 | REXEC_FBC_UTF8_CLASS_SCAN((ANYOF_FLAGS(c) & ANYOF_UNICODE) || |
388cc4de | 1205 | !UTF8_IS_INVARIANT((U8)s[0]) ? |
32fc9b6a | 1206 | reginclass(prog, c, (U8*)s, 0, do_utf8) : |
4cadc6a9 | 1207 | REGINCLASS(prog, c, (U8*)s)); |
388cc4de HS |
1208 | } |
1209 | else { | |
1210 | while (s < strend) { | |
1211 | STRLEN skip = 1; | |
1212 | ||
32fc9b6a | 1213 | if (REGINCLASS(prog, c, (U8*)s) || |
388cc4de HS |
1214 | (ANYOF_FOLD_SHARP_S(c, s, strend) && |
1215 | /* The assignment of 2 is intentional: | |
1216 | * for the folded sharp s, the skip is 2. */ | |
1217 | (skip = SHARP_S_SKIP))) { | |
24b23f37 | 1218 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
388cc4de HS |
1219 | goto got_it; |
1220 | else | |
1221 | tmp = doevery; | |
1222 | } | |
1223 | else | |
1224 | tmp = 1; | |
1225 | s += skip; | |
1226 | } | |
a0d0e21e | 1227 | } |
6eb5f6b9 | 1228 | break; |
f33976b4 | 1229 | case CANY: |
4cadc6a9 | 1230 | REXEC_FBC_SCAN( |
24b23f37 | 1231 | if (tmp && (!reginfo || regtry(reginfo, &s))) |
f33976b4 DB |
1232 | goto got_it; |
1233 | else | |
1234 | tmp = doevery; | |
4cadc6a9 | 1235 | ); |
f33976b4 | 1236 | break; |
6eb5f6b9 | 1237 | case EXACTF: |
5dab1207 NIS |
1238 | m = STRING(c); |
1239 | ln = STR_LEN(c); /* length to match in octets/bytes */ | |
1240 | lnc = (I32) ln; /* length to match in characters */ | |
1aa99e6b | 1241 | if (UTF) { |
a2a2844f | 1242 | STRLEN ulen1, ulen2; |
5dab1207 | 1243 | U8 *sm = (U8 *) m; |
89ebb4a3 JH |
1244 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
1245 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
97dc7d3e RGS |
1246 | /* used by commented-out code below */ |
1247 | /*const U32 uniflags = UTF8_ALLOW_DEFAULT;*/ | |
a0a388a1 YO |
1248 | |
1249 | /* XXX: Since the node will be case folded at compile | |
1250 | time this logic is a little odd, although im not | |
1251 | sure that its actually wrong. --dmq */ | |
1252 | ||
1253 | c1 = to_utf8_lower((U8*)m, tmpbuf1, &ulen1); | |
1254 | c2 = to_utf8_upper((U8*)m, tmpbuf2, &ulen2); | |
1255 | ||
1256 | /* XXX: This is kinda strange. to_utf8_XYZ returns the | |
1257 | codepoint of the first character in the converted | |
1258 | form, yet originally we did the extra step. | |
1259 | No tests fail by commenting this code out however | |
1260 | so Ive left it out. -- dmq. | |
1261 | ||
89ebb4a3 | 1262 | c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXBYTES_CASE, |
041457d9 | 1263 | 0, uniflags); |
89ebb4a3 | 1264 | c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXBYTES_CASE, |
041457d9 | 1265 | 0, uniflags); |
a0a388a1 YO |
1266 | */ |
1267 | ||
5dab1207 NIS |
1268 | lnc = 0; |
1269 | while (sm < ((U8 *) m + ln)) { | |
1270 | lnc++; | |
1271 | sm += UTF8SKIP(sm); | |
1272 | } | |
1aa99e6b IH |
1273 | } |
1274 | else { | |
1275 | c1 = *(U8*)m; | |
1276 | c2 = PL_fold[c1]; | |
1277 | } | |
6eb5f6b9 JH |
1278 | goto do_exactf; |
1279 | case EXACTFL: | |
5dab1207 NIS |
1280 | m = STRING(c); |
1281 | ln = STR_LEN(c); | |
1282 | lnc = (I32) ln; | |
d8093b23 | 1283 | c1 = *(U8*)m; |
6eb5f6b9 JH |
1284 | c2 = PL_fold_locale[c1]; |
1285 | do_exactf: | |
db12adc6 | 1286 | e = HOP3c(strend, -((I32)lnc), s); |
b3c9acc1 | 1287 | |
3b0527fe | 1288 | if (!reginfo && e < s) |
6eb5f6b9 | 1289 | e = s; /* Due to minlen logic of intuit() */ |
1aa99e6b | 1290 | |
60a8b682 JH |
1291 | /* The idea in the EXACTF* cases is to first find the |
1292 | * first character of the EXACTF* node and then, if | |
1293 | * necessary, case-insensitively compare the full | |
1294 | * text of the node. The c1 and c2 are the first | |
1295 | * characters (though in Unicode it gets a bit | |
1296 | * more complicated because there are more cases | |
7f16dd3d JH |
1297 | * than just upper and lower: one needs to use |
1298 | * the so-called folding case for case-insensitive | |
1299 | * matching (called "loose matching" in Unicode). | |
1300 | * ibcmp_utf8() will do just that. */ | |
60a8b682 | 1301 | |
a0a388a1 | 1302 | if (do_utf8 || UTF) { |
575cac57 | 1303 | UV c, f; |
89ebb4a3 | 1304 | U8 tmpbuf [UTF8_MAXBYTES+1]; |
a0a388a1 YO |
1305 | STRLEN len = 1; |
1306 | STRLEN foldlen; | |
4ad0818d | 1307 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
09091399 | 1308 | if (c1 == c2) { |
5dab1207 NIS |
1309 | /* Upper and lower of 1st char are equal - |
1310 | * probably not a "letter". */ | |
1aa99e6b | 1311 | while (s <= e) { |
a0a388a1 YO |
1312 | if (do_utf8) { |
1313 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, | |
041457d9 | 1314 | uniflags); |
a0a388a1 YO |
1315 | } else { |
1316 | c = *((U8*)s); | |
1317 | } | |
4cadc6a9 | 1318 | REXEC_FBC_EXACTISH_CHECK(c == c1); |
1aa99e6b | 1319 | } |
09091399 JH |
1320 | } |
1321 | else { | |
1aa99e6b | 1322 | while (s <= e) { |
a0a388a1 YO |
1323 | if (do_utf8) { |
1324 | c = utf8n_to_uvchr((U8*)s, UTF8_MAXBYTES, &len, | |
041457d9 | 1325 | uniflags); |
a0a388a1 YO |
1326 | } else { |
1327 | c = *((U8*)s); | |
1328 | } | |
80aecb99 | 1329 | |
60a8b682 | 1330 | /* Handle some of the three Greek sigmas cases. |
8c01da3c JH |
1331 | * Note that not all the possible combinations |
1332 | * are handled here: some of them are handled | |
1333 | * by the standard folding rules, and some of | |
1334 | * them (the character class or ANYOF cases) | |
1335 | * are handled during compiletime in | |
1336 | * regexec.c:S_regclass(). */ | |
880bd946 JH |
1337 | if (c == (UV)UNICODE_GREEK_CAPITAL_LETTER_SIGMA || |
1338 | c == (UV)UNICODE_GREEK_SMALL_LETTER_FINAL_SIGMA) | |
1339 | c = (UV)UNICODE_GREEK_SMALL_LETTER_SIGMA; | |
80aecb99 | 1340 | |
4cadc6a9 | 1341 | REXEC_FBC_EXACTISH_CHECK(c == c1 || c == c2); |
1aa99e6b | 1342 | } |
09091399 | 1343 | } |
1aa99e6b IH |
1344 | } |
1345 | else { | |
a0a388a1 | 1346 | /* Neither pattern nor string are UTF8 */ |
1aa99e6b | 1347 | if (c1 == c2) |
4cadc6a9 | 1348 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1); |
1aa99e6b | 1349 | else |
4cadc6a9 | 1350 | REXEC_FBC_EXACTISH_SCAN(*(U8*)s == c1 || *(U8*)s == c2); |
b3c9acc1 IZ |
1351 | } |
1352 | break; | |
bbce6d69 | 1353 | case BOUNDL: |
3280af22 | 1354 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1355 | /* FALL THROUGH */ |
a0d0e21e | 1356 | case BOUND: |
ffc61ed2 | 1357 | if (do_utf8) { |
12d33761 | 1358 | if (s == PL_bostr) |
ffc61ed2 JH |
1359 | tmp = '\n'; |
1360 | else { | |
6136c704 | 1361 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1362 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1363 | } |
1364 | tmp = ((OP(c) == BOUND ? | |
9041c2e3 | 1365 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1366 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1367 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1368 | if (tmp == !(OP(c) == BOUND ? |
bb7a0f54 | 1369 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1370 | isALNUM_LC_utf8((U8*)s))) |
1371 | { | |
1372 | tmp = !tmp; | |
4cadc6a9 | 1373 | REXEC_FBC_TRYIT; |
a687059c | 1374 | } |
4cadc6a9 | 1375 | ); |
a0d0e21e | 1376 | } |
667bb95a | 1377 | else { |
12d33761 | 1378 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 | 1379 | tmp = ((OP(c) == BOUND ? isALNUM(tmp) : isALNUM_LC(tmp)) != 0); |
4cadc6a9 | 1380 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1381 | if (tmp == |
1382 | !(OP(c) == BOUND ? isALNUM(*s) : isALNUM_LC(*s))) { | |
1383 | tmp = !tmp; | |
4cadc6a9 | 1384 | REXEC_FBC_TRYIT; |
a0ed51b3 | 1385 | } |
4cadc6a9 | 1386 | ); |
a0ed51b3 | 1387 | } |
24b23f37 | 1388 | if ((!prog->minlen && tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1389 | goto got_it; |
1390 | break; | |
bbce6d69 | 1391 | case NBOUNDL: |
3280af22 | 1392 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 1393 | /* FALL THROUGH */ |
a0d0e21e | 1394 | case NBOUND: |
ffc61ed2 | 1395 | if (do_utf8) { |
12d33761 | 1396 | if (s == PL_bostr) |
ffc61ed2 JH |
1397 | tmp = '\n'; |
1398 | else { | |
6136c704 | 1399 | U8 * const r = reghop3((U8*)s, -1, (U8*)PL_bostr); |
4ad0818d | 1400 | tmp = utf8n_to_uvchr(r, UTF8SKIP(r), 0, UTF8_ALLOW_DEFAULT); |
ffc61ed2 JH |
1401 | } |
1402 | tmp = ((OP(c) == NBOUND ? | |
9041c2e3 | 1403 | isALNUM_uni(tmp) : isALNUM_LC_uvchr(UNI_TO_NATIVE(tmp))) != 0); |
1a4fad37 | 1404 | LOAD_UTF8_CHARCLASS_ALNUM(); |
4cadc6a9 | 1405 | REXEC_FBC_UTF8_SCAN( |
ffc61ed2 | 1406 | if (tmp == !(OP(c) == NBOUND ? |
bb7a0f54 | 1407 | (bool)swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8) : |
ffc61ed2 JH |
1408 | isALNUM_LC_utf8((U8*)s))) |
1409 | tmp = !tmp; | |
4cadc6a9 YO |
1410 | else REXEC_FBC_TRYIT; |
1411 | ); | |
a0d0e21e | 1412 | } |
667bb95a | 1413 | else { |
12d33761 | 1414 | tmp = (s != PL_bostr) ? UCHARAT(s - 1) : '\n'; |
ffc61ed2 JH |
1415 | tmp = ((OP(c) == NBOUND ? |
1416 | isALNUM(tmp) : isALNUM_LC(tmp)) != 0); | |
4cadc6a9 | 1417 | REXEC_FBC_SCAN( |
ffc61ed2 JH |
1418 | if (tmp == |
1419 | !(OP(c) == NBOUND ? isALNUM(*s) : isALNUM_LC(*s))) | |
1420 | tmp = !tmp; | |
4cadc6a9 YO |
1421 | else REXEC_FBC_TRYIT; |
1422 | ); | |
a0ed51b3 | 1423 | } |
24b23f37 | 1424 | if ((!prog->minlen && !tmp) && (!reginfo || regtry(reginfo, &s))) |
a0ed51b3 LW |
1425 | goto got_it; |
1426 | break; | |
a0d0e21e | 1427 | case ALNUM: |
4cadc6a9 YO |
1428 | REXEC_FBC_CSCAN_PRELOAD( |
1429 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1430 | swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1431 | isALNUM(*s) | |
1432 | ); | |
bbce6d69 | 1433 | case ALNUML: |
4cadc6a9 YO |
1434 | REXEC_FBC_CSCAN_TAINT( |
1435 | isALNUM_LC_utf8((U8*)s), | |
1436 | isALNUM_LC(*s) | |
1437 | ); | |
a0d0e21e | 1438 | case NALNUM: |
4cadc6a9 YO |
1439 | REXEC_FBC_CSCAN_PRELOAD( |
1440 | LOAD_UTF8_CHARCLASS_ALNUM(), | |
1441 | !swash_fetch(PL_utf8_alnum, (U8*)s, do_utf8), | |
1442 | !isALNUM(*s) | |
1443 | ); | |
bbce6d69 | 1444 | case NALNUML: |
4cadc6a9 YO |
1445 | REXEC_FBC_CSCAN_TAINT( |
1446 | !isALNUM_LC_utf8((U8*)s), | |
1447 | !isALNUM_LC(*s) | |
1448 | ); | |
a0d0e21e | 1449 | case SPACE: |
4cadc6a9 YO |
1450 | REXEC_FBC_CSCAN_PRELOAD( |
1451 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1452 | *s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8), | |
1453 | isSPACE(*s) | |
1454 | ); | |
bbce6d69 | 1455 | case SPACEL: |
4cadc6a9 YO |
1456 | REXEC_FBC_CSCAN_TAINT( |
1457 | *s == ' ' || isSPACE_LC_utf8((U8*)s), | |
1458 | isSPACE_LC(*s) | |
1459 | ); | |
a0d0e21e | 1460 | case NSPACE: |
4cadc6a9 YO |
1461 | REXEC_FBC_CSCAN_PRELOAD( |
1462 | LOAD_UTF8_CHARCLASS_SPACE(), | |
1463 | !(*s == ' ' || swash_fetch(PL_utf8_space,(U8*)s, do_utf8)), | |
1464 | !isSPACE(*s) | |
1465 | ); | |
bbce6d69 | 1466 | case NSPACEL: |
4cadc6a9 YO |
1467 | REXEC_FBC_CSCAN_TAINT( |
1468 | !(*s == ' ' || isSPACE_LC_utf8((U8*)s)), | |
1469 | !isSPACE_LC(*s) | |
1470 | ); | |
a0d0e21e | 1471 | case DIGIT: |
4cadc6a9 YO |
1472 | REXEC_FBC_CSCAN_PRELOAD( |
1473 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1474 | swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1475 | isDIGIT(*s) | |
1476 | ); | |
b8c5462f | 1477 | case DIGITL: |
4cadc6a9 YO |
1478 | REXEC_FBC_CSCAN_TAINT( |
1479 | isDIGIT_LC_utf8((U8*)s), | |
1480 | isDIGIT_LC(*s) | |
1481 | ); | |
a0d0e21e | 1482 | case NDIGIT: |
4cadc6a9 YO |
1483 | REXEC_FBC_CSCAN_PRELOAD( |
1484 | LOAD_UTF8_CHARCLASS_DIGIT(), | |
1485 | !swash_fetch(PL_utf8_digit,(U8*)s, do_utf8), | |
1486 | !isDIGIT(*s) | |
1487 | ); | |
b8c5462f | 1488 | case NDIGITL: |
4cadc6a9 YO |
1489 | REXEC_FBC_CSCAN_TAINT( |
1490 | !isDIGIT_LC_utf8((U8*)s), | |
1491 | !isDIGIT_LC(*s) | |
1492 | ); | |
e1d1eefb YO |
1493 | case LNBREAK: |
1494 | REXEC_FBC_CSCAN( | |
1495 | is_LNBREAK_utf8(s), | |
1496 | is_LNBREAK_latin1(s) | |
1497 | ); | |
1498 | case VERTWS: | |
1499 | REXEC_FBC_CSCAN( | |
1500 | is_VERTWS_utf8(s), | |
1501 | is_VERTWS_latin1(s) | |
1502 | ); | |
1503 | case NVERTWS: | |
1504 | REXEC_FBC_CSCAN( | |
1505 | !is_VERTWS_utf8(s), | |
1506 | !is_VERTWS_latin1(s) | |
1507 | ); | |
1508 | case HORIZWS: | |
1509 | REXEC_FBC_CSCAN( | |
1510 | is_HORIZWS_utf8(s), | |
1511 | is_HORIZWS_latin1(s) | |
1512 | ); | |
1513 | case NHORIZWS: | |
1514 | REXEC_FBC_CSCAN( | |
1515 | !is_HORIZWS_utf8(s), | |
1516 | !is_HORIZWS_latin1(s) | |
1517 | ); | |
1de06328 YO |
1518 | case AHOCORASICKC: |
1519 | case AHOCORASICK: | |
07be1b83 | 1520 | { |
a0a388a1 | 1521 | DECL_TRIE_TYPE(c); |
07be1b83 YO |
1522 | /* what trie are we using right now */ |
1523 | reg_ac_data *aho | |
f8fc2ecf | 1524 | = (reg_ac_data*)progi->data->data[ ARG( c ) ]; |
3251b653 NC |
1525 | reg_trie_data *trie |
1526 | = (reg_trie_data*)progi->data->data[ aho->trie ]; | |
85fbaab2 | 1527 | HV *widecharmap = MUTABLE_HV(progi->data->data[ aho->trie + 1 ]); |
07be1b83 YO |
1528 | |
1529 | const char *last_start = strend - trie->minlen; | |
6148ee25 | 1530 | #ifdef DEBUGGING |
07be1b83 | 1531 | const char *real_start = s; |
6148ee25 | 1532 | #endif |
07be1b83 | 1533 | STRLEN maxlen = trie->maxlen; |
be8e71aa YO |
1534 | SV *sv_points; |
1535 | U8 **points; /* map of where we were in the input string | |
786e8c11 | 1536 | when reading a given char. For ASCII this |
be8e71aa | 1537 | is unnecessary overhead as the relationship |
38a44b82 NC |
1538 | is always 1:1, but for Unicode, especially |
1539 | case folded Unicode this is not true. */ | |
f9e705e8 | 1540 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; |
786e8c11 YO |
1541 | U8 *bitmap=NULL; |
1542 | ||
07be1b83 YO |
1543 | |
1544 | GET_RE_DEBUG_FLAGS_DECL; | |
1545 | ||
be8e71aa YO |
1546 | /* We can't just allocate points here. We need to wrap it in |
1547 | * an SV so it gets freed properly if there is a croak while | |
1548 | * running the match */ | |
1549 | ENTER; | |
1550 | SAVETMPS; | |
1551 | sv_points=newSV(maxlen * sizeof(U8 *)); | |
1552 | SvCUR_set(sv_points, | |
1553 | maxlen * sizeof(U8 *)); | |
1554 | SvPOK_on(sv_points); | |
1555 | sv_2mortal(sv_points); | |
1556 | points=(U8**)SvPV_nolen(sv_points ); | |
1de06328 YO |
1557 | if ( trie_type != trie_utf8_fold |
1558 | && (trie->bitmap || OP(c)==AHOCORASICKC) ) | |
1559 | { | |
786e8c11 YO |
1560 | if (trie->bitmap) |
1561 | bitmap=(U8*)trie->bitmap; | |
1562 | else | |
1563 | bitmap=(U8*)ANYOF_BITMAP(c); | |
07be1b83 | 1564 | } |
786e8c11 YO |
1565 | /* this is the Aho-Corasick algorithm modified a touch |
1566 | to include special handling for long "unknown char" | |
1567 | sequences. The basic idea being that we use AC as long | |
1568 | as we are dealing with a possible matching char, when | |
1569 | we encounter an unknown char (and we have not encountered | |
1570 | an accepting state) we scan forward until we find a legal | |
1571 | starting char. | |
1572 | AC matching is basically that of trie matching, except | |
1573 | that when we encounter a failing transition, we fall back | |
1574 | to the current states "fail state", and try the current char | |
1575 | again, a process we repeat until we reach the root state, | |
1576 | state 1, or a legal transition. If we fail on the root state | |
1577 | then we can either terminate if we have reached an accepting | |
1578 | state previously, or restart the entire process from the beginning | |
1579 | if we have not. | |
1580 | ||
1581 | */ | |
07be1b83 YO |
1582 | while (s <= last_start) { |
1583 | const U32 uniflags = UTF8_ALLOW_DEFAULT; | |
1584 | U8 *uc = (U8*)s; | |
1585 | U16 charid = 0; | |
1586 | U32 base = 1; | |
1587 | U32 state = 1; | |
1588 | UV uvc = 0; | |
1589 | STRLEN len = 0; | |
1590 | STRLEN foldlen = 0; | |
1591 | U8 *uscan = (U8*)NULL; | |
1592 | U8 *leftmost = NULL; | |
786e8c11 YO |
1593 | #ifdef DEBUGGING |
1594 | U32 accepted_word= 0; | |
1595 | #endif | |
07be1b83 YO |
1596 | U32 pointpos = 0; |
1597 | ||
1598 | while ( state && uc <= (U8*)strend ) { | |
1599 | int failed=0; | |
786e8c11 YO |
1600 | U32 word = aho->states[ state ].wordnum; |
1601 | ||
1de06328 YO |
1602 | if( state==1 ) { |
1603 | if ( bitmap ) { | |
1604 | DEBUG_TRIE_EXECUTE_r( | |
1605 | if ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1606 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1607 | (char *)uc, do_utf8 ); | |
1608 | PerlIO_printf( Perl_debug_log, | |
1609 | " Scanning for legal start char...\n"); | |
1610 | } | |
1611 | ); | |
1612 | while ( uc <= (U8*)last_start && !BITMAP_TEST(bitmap,*uc) ) { | |
1613 | uc++; | |
786e8c11 | 1614 | } |
1de06328 | 1615 | s= (char *)uc; |
786e8c11 | 1616 | } |
786e8c11 YO |
1617 | if (uc >(U8*)last_start) break; |
1618 | } | |
1619 | ||
1620 | if ( word ) { | |
1621 | U8 *lpos= points[ (pointpos - trie->wordlen[word-1] ) % maxlen ]; | |
1622 | if (!leftmost || lpos < leftmost) { | |
1623 | DEBUG_r(accepted_word=word); | |
07be1b83 | 1624 | leftmost= lpos; |
786e8c11 | 1625 | } |
07be1b83 | 1626 | if (base==0) break; |
786e8c11 | 1627 | |
07be1b83 YO |
1628 | } |
1629 | points[pointpos++ % maxlen]= uc; | |
55eed653 NC |
1630 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
1631 | uscan, len, uvc, charid, foldlen, | |
1632 | foldbuf, uniflags); | |
786e8c11 YO |
1633 | DEBUG_TRIE_EXECUTE_r({ |
1634 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1635 | s, do_utf8 ); | |
07be1b83 | 1636 | PerlIO_printf(Perl_debug_log, |
786e8c11 YO |
1637 | " Charid:%3u CP:%4"UVxf" ", |
1638 | charid, uvc); | |
1639 | }); | |
07be1b83 YO |
1640 | |
1641 | do { | |
6148ee25 | 1642 | #ifdef DEBUGGING |
786e8c11 | 1643 | word = aho->states[ state ].wordnum; |
6148ee25 | 1644 | #endif |
07be1b83 YO |
1645 | base = aho->states[ state ].trans.base; |
1646 | ||
786e8c11 YO |
1647 | DEBUG_TRIE_EXECUTE_r({ |
1648 | if (failed) | |
1649 | dump_exec_pos( (char *)uc, c, strend, real_start, | |
1650 | s, do_utf8 ); | |
07be1b83 | 1651 | PerlIO_printf( Perl_debug_log, |
786e8c11 YO |
1652 | "%sState: %4"UVxf", word=%"UVxf, |
1653 | failed ? " Fail transition to " : "", | |
1654 | (UV)state, (UV)word); | |
1655 | }); | |
07be1b83 YO |
1656 | if ( base ) { |
1657 | U32 tmp; | |
1658 | if (charid && | |
1659 | (base + charid > trie->uniquecharcount ) | |
1660 | && (base + charid - 1 - trie->uniquecharcount | |
1661 | < trie->lasttrans) | |
1662 | && trie->trans[base + charid - 1 - | |
1663 | trie->uniquecharcount].check == state | |
1664 | && (tmp=trie->trans[base + charid - 1 - | |
1665 | trie->uniquecharcount ].next)) | |
1666 | { | |
786e8c11 YO |
1667 | DEBUG_TRIE_EXECUTE_r( |
1668 | PerlIO_printf( Perl_debug_log," - legal\n")); | |
07be1b83 YO |
1669 | state = tmp; |
1670 | break; | |
1671 | } | |
1672 | else { | |
786e8c11 YO |
1673 | DEBUG_TRIE_EXECUTE_r( |
1674 | PerlIO_printf( Perl_debug_log," - fail\n")); | |
1675 | failed = 1; | |
1676 | state = aho->fail[state]; | |
07be1b83 YO |
1677 | } |
1678 | } | |
1679 | else { | |
1680 | /* we must be accepting here */ | |
786e8c11 YO |
1681 | DEBUG_TRIE_EXECUTE_r( |
1682 | PerlIO_printf( Perl_debug_log," - accepting\n")); | |
1683 | failed = 1; | |
07be1b83 YO |
1684 | break; |
1685 | } | |
1686 | } while(state); | |
786e8c11 | 1687 | uc += len; |
07be1b83 YO |
1688 | if (failed) { |
1689 | if (leftmost) | |
1690 | break; | |
786e8c11 | 1691 | if (!state) state = 1; |
07be1b83 YO |
1692 | } |
1693 | } | |
1694 | if ( aho->states[ state ].wordnum ) { | |
1695 | U8 *lpos = points[ (pointpos - trie->wordlen[aho->states[ state ].wordnum-1]) % maxlen ]; | |
786e8c11 YO |
1696 | if (!leftmost || lpos < leftmost) { |
1697 | DEBUG_r(accepted_word=aho->states[ state ].wordnum); | |
07be1b83 | 1698 | leftmost = lpos; |
786e8c11 | 1699 | } |
07be1b83 | 1700 | } |
07be1b83 YO |
1701 | if (leftmost) { |
1702 | s = (char*)leftmost; | |
786e8c11 YO |
1703 | DEBUG_TRIE_EXECUTE_r({ |
1704 | PerlIO_printf( | |
70685ca0 JH |
1705 | Perl_debug_log,"Matches word #%"UVxf" at position %"IVdf". Trying full pattern...\n", |
1706 | (UV)accepted_word, (IV)(s - real_start) | |
786e8c11 YO |
1707 | ); |
1708 | }); | |
24b23f37 | 1709 | if (!reginfo || regtry(reginfo, &s)) { |
be8e71aa YO |
1710 | FREETMPS; |
1711 | LEAVE; | |
07be1b83 | 1712 | goto got_it; |
be8e71aa | 1713 | } |
07be1b83 | 1714 | s = HOPc(s,1); |
786e8c11 YO |
1715 | DEBUG_TRIE_EXECUTE_r({ |
1716 | PerlIO_printf( Perl_debug_log,"Pattern failed. Looking for new start point...\n"); | |
1717 | }); | |
07be1b83 | 1718 | } else { |
786e8c11 YO |
1719 | DEBUG_TRIE_EXECUTE_r( |
1720 | PerlIO_printf( Perl_debug_log,"No match.\n")); | |
07be1b83 YO |
1721 | break; |
1722 | } | |
1723 | } | |
be8e71aa YO |
1724 | FREETMPS; |
1725 | LEAVE; | |
07be1b83 YO |
1726 | } |
1727 | break; | |
b3c9acc1 | 1728 | default: |
3c3eec57 GS |
1729 | Perl_croak(aTHX_ "panic: unknown regstclass %d", (int)OP(c)); |
1730 | break; | |
d6a28714 | 1731 | } |
6eb5f6b9 JH |
1732 | return 0; |
1733 | got_it: | |
1734 | return s; | |
1735 | } | |
1736 | ||
fae667d5 | 1737 | |
6eb5f6b9 JH |
1738 | /* |
1739 | - regexec_flags - match a regexp against a string | |
1740 | */ | |
1741 | I32 | |
288b8c02 | 1742 | Perl_regexec_flags(pTHX_ REGEXP * const rx, char *stringarg, register char *strend, |
6eb5f6b9 JH |
1743 | char *strbeg, I32 minend, SV *sv, void *data, U32 flags) |
1744 | /* strend: pointer to null at end of string */ | |
1745 | /* strbeg: real beginning of string */ | |
1746 | /* minend: end of match must be >=minend after stringarg. */ | |
58e23c8d YO |
1747 | /* data: May be used for some additional optimizations. |
1748 | Currently its only used, with a U32 cast, for transmitting | |
1749 | the ganch offset when doing a /g match. This will change */ | |
6eb5f6b9 JH |
1750 | /* nosave: For optimizations. */ |
1751 | { | |
97aff369 | 1752 | dVAR; |
288b8c02 | 1753 | struct regexp *const prog = (struct regexp *)SvANY(rx); |
24b23f37 | 1754 | /*register*/ char *s; |
6eb5f6b9 | 1755 | register regnode *c; |
24b23f37 | 1756 | /*register*/ char *startpos = stringarg; |
6eb5f6b9 JH |
1757 | I32 minlen; /* must match at least this many chars */ |
1758 | I32 dontbother = 0; /* how many characters not to try at end */ | |
6eb5f6b9 JH |
1759 | I32 end_shift = 0; /* Same for the end. */ /* CC */ |
1760 | I32 scream_pos = -1; /* Internal iterator of scream. */ | |
ccac19ea | 1761 | char *scream_olds = NULL; |
f9f4320a | 1762 | const bool do_utf8 = (bool)DO_UTF8(sv); |
2757e526 | 1763 | I32 multiline; |
f8fc2ecf | 1764 | RXi_GET_DECL(prog,progi); |
3b0527fe | 1765 | regmatch_info reginfo; /* create some info to pass to regtry etc */ |
e9105d30 | 1766 | regexp_paren_pair *swap = NULL; |
a3621e74 YO |
1767 | GET_RE_DEBUG_FLAGS_DECL; |
1768 | ||
7918f24d | 1769 | PERL_ARGS_ASSERT_REGEXEC_FLAGS; |
9d4ba2ae | 1770 | PERL_UNUSED_ARG(data); |
6eb5f6b9 JH |
1771 | |
1772 | /* Be paranoid... */ | |
1773 | if (prog == NULL || startpos == NULL) { | |
1774 | Perl_croak(aTHX_ "NULL regexp parameter"); | |
1775 | return 0; | |
1776 | } | |
1777 | ||
bbe252da | 1778 | multiline = prog->extflags & RXf_PMf_MULTILINE; |
288b8c02 | 1779 | reginfo.prog = rx; /* Yes, sorry that this is confusing. */ |
2757e526 | 1780 | |
288b8c02 | 1781 | RX_MATCH_UTF8_set(rx, do_utf8); |
1de06328 | 1782 | DEBUG_EXECUTE_r( |
efd26800 | 1783 | debug_start_match(rx, do_utf8, startpos, strend, |
1de06328 YO |
1784 | "Matching"); |
1785 | ); | |
bac06658 | 1786 | |
6eb5f6b9 | 1787 | minlen = prog->minlen; |
1de06328 YO |
1788 | |
1789 | if (strend - startpos < (minlen+(prog->check_offset_min<0?prog->check_offset_min:0))) { | |
a3621e74 | 1790 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
a72c7584 JH |
1791 | "String too short [regexec_flags]...\n")); |
1792 | goto phooey; | |
1aa99e6b | 1793 | } |
6eb5f6b9 | 1794 | |
1de06328 | 1795 | |
6eb5f6b9 | 1796 | /* Check validity of program. */ |
f8fc2ecf | 1797 | if (UCHARAT(progi->program) != REG_MAGIC) { |
6eb5f6b9 JH |
1798 | Perl_croak(aTHX_ "corrupted regexp program"); |
1799 | } | |
1800 | ||
1801 | PL_reg_flags = 0; | |
1802 | PL_reg_eval_set = 0; | |
1803 | PL_reg_maxiter = 0; | |
1804 | ||
3c8556c3 | 1805 | if (RX_UTF8(rx)) |
6eb5f6b9 JH |
1806 | PL_reg_flags |= RF_utf8; |
1807 | ||
1808 | /* Mark beginning of line for ^ and lookbehind. */ | |
3b0527fe | 1809 | reginfo.bol = startpos; /* XXX not used ??? */ |
6eb5f6b9 | 1810 | PL_bostr = strbeg; |
3b0527fe | 1811 | reginfo.sv = sv; |
6eb5f6b9 JH |
1812 | |
1813 | /* Mark end of line for $ (and such) */ | |
1814 | PL_regeol = strend; | |
1815 | ||
1816 | /* see how far we have to get to not match where we matched before */ | |
3b0527fe | 1817 | reginfo.till = startpos+minend; |
6eb5f6b9 | 1818 | |
6eb5f6b9 JH |
1819 | /* If there is a "must appear" string, look for it. */ |
1820 | s = startpos; | |
1821 | ||
bbe252da | 1822 | if (prog->extflags & RXf_GPOS_SEEN) { /* Need to set reginfo->ganch */ |
6eb5f6b9 | 1823 | MAGIC *mg; |
2c296965 | 1824 | if (flags & REXEC_IGNOREPOS){ /* Means: check only at start */ |
58e23c8d | 1825 | reginfo.ganch = startpos + prog->gofs; |
2c296965 YO |
1826 | DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log, |
1827 | "GPOS IGNOREPOS: reginfo.ganch = startpos + %"UVxf"\n",prog->gofs)); | |
1828 | } else if (sv && SvTYPE(sv) >= SVt_PVMG | |
6eb5f6b9 | 1829 | && SvMAGIC(sv) |
14befaf4 DM |
1830 | && (mg = mg_find(sv, PERL_MAGIC_regex_global)) |
1831 | && mg->mg_len >= 0) { | |
3b0527fe | 1832 | reginfo.ganch = strbeg + mg->mg_len; /* Defined pos() */ |
2c296965 YO |
1833 | DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log, |
1834 | "GPOS MAGIC: reginfo.ganch = strbeg + %"IVdf"\n",mg->mg_len)); | |
1835 | ||
bbe252da | 1836 | if (prog->extflags & RXf_ANCH_GPOS) { |
3b0527fe | 1837 | if (s > reginfo.ganch) |
6eb5f6b9 | 1838 | goto phooey; |
58e23c8d | 1839 | s = reginfo.ganch - prog->gofs; |
2c296965 YO |
1840 | DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log, |
1841 | "GPOS ANCH_GPOS: s = ganch - %"UVxf"\n",prog->gofs)); | |
c584a96e YO |
1842 | if (s < strbeg) |
1843 | goto phooey; | |
6eb5f6b9 JH |
1844 | } |
1845 | } | |
58e23c8d | 1846 | else if (data) { |
70685ca0 | 1847 | reginfo.ganch = strbeg + PTR2UV(data); |
2c296965 YO |
1848 | DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log, |
1849 | "GPOS DATA: reginfo.ganch= strbeg + %"UVxf"\n",PTR2UV(data))); | |
1850 | ||
1851 | } else { /* pos() not defined */ | |
3b0527fe | 1852 | reginfo.ganch = strbeg; |
2c296965 YO |
1853 | DEBUG_GPOS_r(PerlIO_printf(Perl_debug_log, |
1854 | "GPOS: reginfo.ganch = strbeg\n")); | |
1855 | } | |
6eb5f6b9 | 1856 | } |
288b8c02 | 1857 | if (PL_curpm && (PM_GETRE(PL_curpm) == rx)) { |
e9105d30 GG |
1858 | /* We have to be careful. If the previous successful match |
1859 | was from this regex we don't want a subsequent partially | |
1860 | successful match to clobber the old results. | |
1861 | So when we detect this possibility we add a swap buffer | |
1862 | to the re, and switch the buffer each match. If we fail | |
1863 | we switch it back, otherwise we leave it swapped. | |
1864 | */ | |
1865 | swap = prog->offs; | |
1866 | /* do we need a save destructor here for eval dies? */ | |
1867 | Newxz(prog->offs, (prog->nparens + 1), regexp_paren_pair); | |
c74340f9 | 1868 | } |
a0714e2c | 1869 | if (!(flags & REXEC_CHECKED) && (prog->check_substr != NULL || prog->check_utf8 != NULL)) { |
6eb5f6b9 JH |
1870 | re_scream_pos_data d; |
1871 | ||
1872 | d.scream_olds = &scream_olds; | |
1873 | d.scream_pos = &scream_pos; | |
288b8c02 | 1874 | s = re_intuit_start(rx, sv, s, strend, flags, &d); |
3fa9c3d7 | 1875 | if (!s) { |
a3621e74 | 1876 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Not present...\n")); |
6eb5f6b9 | 1877 | goto phooey; /* not present */ |
3fa9c3d7 | 1878 | } |
6eb5f6b9 JH |
1879 | } |
1880 | ||
1de06328 | 1881 | |
6eb5f6b9 JH |
1882 | |
1883 | /* Simplest case: anchored match need be tried only once. */ | |
1884 | /* [unless only anchor is BOL and multiline is set] */ | |
bbe252da | 1885 | if (prog->extflags & (RXf_ANCH & ~RXf_ANCH_GPOS)) { |
24b23f37 | 1886 | if (s == startpos && regtry(®info, &startpos)) |
6eb5f6b9 | 1887 | goto got_it; |
bbe252da YO |
1888 | else if (multiline || (prog->intflags & PREGf_IMPLICIT) |
1889 | || (prog->extflags & RXf_ANCH_MBOL)) /* XXXX SBOL? */ | |
6eb5f6b9 JH |
1890 | { |
1891 | char *end; | |
1892 | ||
1893 | if (minlen) | |
1894 | dontbother = minlen - 1; | |
1aa99e6b | 1895 | end = HOP3c(strend, -dontbother, strbeg) - 1; |
6eb5f6b9 | 1896 | /* for multiline we only have to try after newlines */ |
33b8afdf | 1897 | if (prog->check_substr || prog->check_utf8) { |
6eb5f6b9 JH |
1898 | if (s == startpos) |
1899 | goto after_try; | |
1900 | while (1) { | |
24b23f37 | 1901 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1902 | goto got_it; |
1903 | after_try: | |
5339e136 | 1904 | if (s > end) |
6eb5f6b9 | 1905 | goto phooey; |
bbe252da | 1906 | if (prog->extflags & RXf_USE_INTUIT) { |
288b8c02 | 1907 | s = re_intuit_start(rx, sv, s + 1, strend, flags, NULL); |
6eb5f6b9 JH |
1908 | if (!s) |
1909 | goto phooey; | |
1910 | } | |
1911 | else | |
1912 | s++; | |
1913 | } | |
1914 | } else { | |
1915 | if (s > startpos) | |
1916 | s--; | |
1917 | while (s < end) { | |
1918 | if (*s++ == '\n') { /* don't need PL_utf8skip here */ | |
24b23f37 | 1919 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
1920 | goto got_it; |
1921 | } | |
1922 | } | |
1923 | } | |
1924 | } | |
1925 | goto phooey; | |
bbe252da | 1926 | } else if (RXf_GPOS_CHECK == (prog->extflags & RXf_GPOS_CHECK)) |
f9f4320a YO |
1927 | { |
1928 | /* the warning about reginfo.ganch being used without intialization | |
bbe252da | 1929 | is bogus -- we set it above, when prog->extflags & RXf_GPOS_SEEN |
f9f4320a | 1930 | and we only enter this block when the same bit is set. */ |
58e23c8d | 1931 | char *tmp_s = reginfo.ganch - prog->gofs; |
c584a96e YO |
1932 | |
1933 | if (tmp_s >= strbeg && regtry(®info, &tmp_s)) | |
6eb5f6b9 JH |
1934 | goto got_it; |
1935 | goto phooey; | |
1936 | } | |
1937 | ||
1938 | /* Messy cases: unanchored match. */ | |
bbe252da | 1939 | if ((prog->anchored_substr || prog->anchored_utf8) && prog->intflags & PREGf_SKIP) { |
6eb5f6b9 JH |
1940 | /* we have /x+whatever/ */ |
1941 | /* it must be a one character string (XXXX Except UTF?) */ | |
33b8afdf | 1942 | char ch; |
bf93d4cc GS |
1943 | #ifdef DEBUGGING |
1944 | int did_match = 0; | |
1945 | #endif | |
33b8afdf JH |
1946 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) |
1947 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
3f7c398e | 1948 | ch = SvPVX_const(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)[0]; |
bf93d4cc | 1949 | |
1aa99e6b | 1950 | if (do_utf8) { |
4cadc6a9 | 1951 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1952 | if (*s == ch) { |
a3621e74 | 1953 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1954 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1955 | s += UTF8SKIP(s); |
1956 | while (s < strend && *s == ch) | |
1957 | s += UTF8SKIP(s); | |
1958 | } | |
4cadc6a9 | 1959 | ); |
6eb5f6b9 JH |
1960 | } |
1961 | else { | |
4cadc6a9 | 1962 | REXEC_FBC_SCAN( |
6eb5f6b9 | 1963 | if (*s == ch) { |
a3621e74 | 1964 | DEBUG_EXECUTE_r( did_match = 1 ); |
24b23f37 | 1965 | if (regtry(®info, &s)) goto got_it; |
6eb5f6b9 JH |
1966 | s++; |
1967 | while (s < strend && *s == ch) | |
1968 | s++; | |
1969 | } | |
4cadc6a9 | 1970 | ); |
6eb5f6b9 | 1971 | } |
a3621e74 | 1972 | DEBUG_EXECUTE_r(if (!did_match) |
bf93d4cc | 1973 | PerlIO_printf(Perl_debug_log, |
b7953727 JH |
1974 | "Did not find anchored character...\n") |
1975 | ); | |
6eb5f6b9 | 1976 | } |
a0714e2c SS |
1977 | else if (prog->anchored_substr != NULL |
1978 | || prog->anchored_utf8 != NULL | |
1979 | || ((prog->float_substr != NULL || prog->float_utf8 != NULL) | |
33b8afdf JH |
1980 | && prog->float_max_offset < strend - s)) { |
1981 | SV *must; | |
1982 | I32 back_max; | |
1983 | I32 back_min; | |
1984 | char *last; | |
6eb5f6b9 | 1985 | char *last1; /* Last position checked before */ |
bf93d4cc GS |
1986 | #ifdef DEBUGGING |
1987 | int did_match = 0; | |
1988 | #endif | |
33b8afdf JH |
1989 | if (prog->anchored_substr || prog->anchored_utf8) { |
1990 | if (!(do_utf8 ? prog->anchored_utf8 : prog->anchored_substr)) | |
1991 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1992 | must = do_utf8 ? prog->anchored_utf8 : prog->anchored_substr; | |
1993 | back_max = back_min = prog->anchored_offset; | |
1994 | } else { | |
1995 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
1996 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
1997 | must = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
1998 | back_max = prog->float_max_offset; | |
1999 | back_min = prog->float_min_offset; | |
2000 | } | |
1de06328 YO |
2001 | |
2002 | ||
33b8afdf JH |
2003 | if (must == &PL_sv_undef) |
2004 | /* could not downgrade utf8 check substring, so must fail */ | |
2005 | goto phooey; | |
2006 | ||
1de06328 YO |
2007 | if (back_min<0) { |
2008 | last = strend; | |
2009 | } else { | |
2010 | last = HOP3c(strend, /* Cannot start after this */ | |
2011 | -(I32)(CHR_SVLEN(must) | |
2012 | - (SvTAIL(must) != 0) + back_min), strbeg); | |
2013 | } | |
6eb5f6b9 JH |
2014 | if (s > PL_bostr) |
2015 | last1 = HOPc(s, -1); | |
2016 | else | |
2017 | last1 = s - 1; /* bogus */ | |
2018 | ||
a0288114 | 2019 | /* XXXX check_substr already used to find "s", can optimize if |
6eb5f6b9 JH |
2020 | check_substr==must. */ |
2021 | scream_pos = -1; | |
2022 | dontbother = end_shift; | |
2023 | strend = HOPc(strend, -dontbother); | |
2024 | while ( (s <= last) && | |
9041c2e3 | 2025 | ((flags & REXEC_SCREAM) |
1de06328 | 2026 | ? (s = screaminstr(sv, must, HOP3c(s, back_min, (back_min<0 ? strbeg : strend)) - strbeg, |
6eb5f6b9 | 2027 | end_shift, &scream_pos, 0)) |
1de06328 | 2028 | : (s = fbm_instr((unsigned char*)HOP3(s, back_min, (back_min<0 ? strbeg : strend)), |
9041c2e3 | 2029 | (unsigned char*)strend, must, |
7fba1cd6 | 2030 | multiline ? FBMrf_MULTILINE : 0))) ) { |
4addbd3b | 2031 | /* we may be pointing at the wrong string */ |
07bc277f | 2032 | if ((flags & REXEC_SCREAM) && RXp_MATCH_COPIED(prog)) |
3f7c398e | 2033 | s = strbeg + (s - SvPVX_const(sv)); |
a3621e74 | 2034 | DEBUG_EXECUTE_r( did_match = 1 ); |
6eb5f6b9 JH |
2035 | if (HOPc(s, -back_max) > last1) { |
2036 | last1 = HOPc(s, -back_min); | |
2037 | s = HOPc(s, -back_max); | |
2038 | } | |
2039 | else { | |
52657f30 | 2040 | char * const t = (last1 >= PL_bostr) ? HOPc(last1, 1) : last1 + 1; |
6eb5f6b9 JH |
2041 | |
2042 | last1 = HOPc(s, -back_min); | |
52657f30 | 2043 | s = t; |
6eb5f6b9 | 2044 | } |
1aa99e6b | 2045 | if (do_utf8) { |
6eb5f6b9 | 2046 | while (s <= last1) { |
24b23f37 | 2047 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
2048 | goto got_it; |
2049 | s += UTF8SKIP(s); | |
2050 | } | |
2051 | } | |
2052 | else { | |
2053 | while (s <= last1) { | |
24b23f37 | 2054 | if (regtry(®info, &s)) |
6eb5f6b9 JH |
2055 | goto got_it; |
2056 | s++; | |
2057 | } | |
2058 | } | |
2059 | } | |
ab3bbdeb YO |
2060 | DEBUG_EXECUTE_r(if (!did_match) { |
2061 | RE_PV_QUOTED_DECL(quoted, do_utf8, PERL_DEBUG_PAD_ZERO(0), | |
2062 | SvPVX_const(must), RE_SV_DUMPLEN(must), 30); | |
2063 | PerlIO_printf(Perl_debug_log, "Did not find %s substr %s%s...\n", | |
33b8afdf | 2064 | ((must == prog->anchored_substr || must == prog->anchored_utf8) |
bf93d4cc | 2065 | ? "anchored" : "floating"), |
ab3bbdeb YO |
2066 | quoted, RE_SV_TAIL(must)); |
2067 | }); | |
6eb5f6b9 JH |
2068 | goto phooey; |
2069 | } | |
f8fc2ecf | 2070 | else if ( (c = progi->regstclass) ) { |
f14c76ed | 2071 | if (minlen) { |
f8fc2ecf | 2072 | const OPCODE op = OP(progi->regstclass); |
66e933ab | 2073 | /* don't bother with what can't match */ |
786e8c11 | 2074 | if (PL_regkind[op] != EXACT && op != CANY && PL_regkind[op] != TRIE) |
f14c76ed RGS |
2075 | strend = HOPc(strend, -(minlen - 1)); |
2076 | } | |
a3621e74 | 2077 | DEBUG_EXECUTE_r({ |
be8e71aa | 2078 | SV * const prop = sv_newmortal(); |
32fc9b6a | 2079 | regprop(prog, prop, c); |
0df25f3d | 2080 | { |
02daf0ab | 2081 | RE_PV_QUOTED_DECL(quoted,do_utf8,PERL_DEBUG_PAD_ZERO(1), |
ab3bbdeb | 2082 | s,strend-s,60); |
0df25f3d | 2083 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2084 | "Matching stclass %.*s against %s (%d chars)\n", |
e4f74956 | 2085 | (int)SvCUR(prop), SvPVX_const(prop), |
ab3bbdeb | 2086 | quoted, (int)(strend - s)); |
0df25f3d | 2087 | } |
ffc61ed2 | 2088 | }); |
3b0527fe | 2089 | if (find_byclass(prog, c, s, strend, ®info)) |
6eb5f6b9 | 2090 | goto got_it; |
07be1b83 | 2091 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "Contradicts stclass... [regexec_flags]\n")); |
d6a28714 JH |
2092 | } |
2093 | else { | |
2094 | dontbother = 0; | |
a0714e2c | 2095 | if (prog->float_substr != NULL || prog->float_utf8 != NULL) { |
33b8afdf | 2096 | /* Trim the end. */ |
d6a28714 | 2097 | char *last; |
33b8afdf JH |
2098 | SV* float_real; |
2099 | ||
2100 | if (!(do_utf8 ? prog->float_utf8 : prog->float_substr)) | |
2101 | do_utf8 ? to_utf8_substr(prog) : to_byte_substr(prog); | |
2102 | float_real = do_utf8 ? prog->float_utf8 : prog->float_substr; | |
d6a28714 JH |
2103 | |
2104 | if (flags & REXEC_SCREAM) { | |
33b8afdf | 2105 | last = screaminstr(sv, float_real, s - strbeg, |
d6a28714 JH |
2106 | end_shift, &scream_pos, 1); /* last one */ |
2107 | if (!last) | |
ffc61ed2 | 2108 | last = scream_olds; /* Only one occurrence. */ |
4addbd3b | 2109 | /* we may be pointing at the wrong string */ |
07bc277f | 2110 | else if (RXp_MATCH_COPIED(prog)) |
3f7c398e | 2111 | s = strbeg + (s - SvPVX_const(sv)); |
b8c5462f | 2112 | } |
d6a28714 JH |
2113 | else { |
2114 | STRLEN len; | |
cfd0369c | 2115 | const char * const little = SvPV_const(float_real, len); |
d6a28714 | 2116 | |
33b8afdf | 2117 | if (SvTAIL(float_real)) { |
d6a28714 JH |
2118 | if (memEQ(strend - len + 1, little, len - 1)) |
2119 | last = strend - len + 1; | |
7fba1cd6 | 2120 | else if (!multiline) |
9041c2e3 | 2121 | last = memEQ(strend - len, little, len) |
bd61b366 | 2122 | ? strend - len : NULL; |
b8c5462f | 2123 | else |
d6a28714 JH |
2124 | goto find_last; |
2125 | } else { | |
2126 | find_last: | |
9041c2e3 | 2127 | if (len) |
d6a28714 | 2128 | last = rninstr(s, strend, little, little + len); |
b8c5462f | 2129 | else |
a0288114 | 2130 | last = strend; /* matching "$" */ |
b8c5462f | 2131 | } |
b8c5462f | 2132 | } |
bf93d4cc | 2133 | if (last == NULL) { |
6bda09f9 YO |
2134 | DEBUG_EXECUTE_r( |
2135 | PerlIO_printf(Perl_debug_log, | |
2136 | "%sCan't trim the tail, match fails (should not happen)%s\n", | |
2137 | PL_colors[4], PL_colors[5])); | |
bf93d4cc GS |
2138 | goto phooey; /* Should not happen! */ |
2139 | } | |
d6a28714 JH |
2140 | dontbother = strend - last + prog->float_min_offset; |
2141 | } | |
2142 | if (minlen && (dontbother < minlen)) | |
2143 | dontbother = minlen - 1; | |
2144 | strend -= dontbother; /* this one's always in bytes! */ | |
2145 | /* We don't know much -- general case. */ | |
1aa99e6b | 2146 | if (do_utf8) { |
d6a28714 | 2147 | for (;;) { |
24b23f37 | 2148 | if (regtry(®info, &s)) |
d6a28714 JH |
2149 | goto got_it; |
2150 | if (s >= strend) | |
2151 | break; | |
b8c5462f | 2152 | s += UTF8SKIP(s); |
d6a28714 JH |
2153 | }; |
2154 | } | |
2155 | else { | |
2156 | do { | |
24b23f37 | 2157 | if (regtry(®info, &s)) |
d6a28714 JH |
2158 | goto got_it; |
2159 | } while (s++ < strend); | |
2160 | } | |
2161 | } | |
2162 | ||
2163 | /* Failure. */ | |
2164 | goto phooey; | |
2165 | ||
2166 | got_it: | |
e9105d30 | 2167 | Safefree(swap); |
288b8c02 | 2168 | RX_MATCH_TAINTED_set(rx, PL_reg_flags & RF_tainted); |
d6a28714 | 2169 | |
19b95bf0 | 2170 | if (PL_reg_eval_set) |
4f639d21 | 2171 | restore_pos(aTHX_ prog); |
5daac39c NC |
2172 | if (RXp_PAREN_NAMES(prog)) |
2173 | (void)hv_iterinit(RXp_PAREN_NAMES(prog)); | |
d6a28714 JH |
2174 | |
2175 | /* make sure $`, $&, $', and $digit will work later */ | |
2176 | if ( !(flags & REXEC_NOT_FIRST) ) { | |
288b8c02 | 2177 | RX_MATCH_COPY_FREE(rx); |
d6a28714 | 2178 | if (flags & REXEC_COPY_STR) { |
be8e71aa | 2179 | const I32 i = PL_regeol - startpos + (stringarg - strbeg); |
f8c7b90f | 2180 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2181 | if ((SvIsCOW(sv) |
2182 | || (SvFLAGS(sv) & CAN_COW_MASK) == CAN_COW_FLAGS)) { | |
2183 | if (DEBUG_C_TEST) { | |
2184 | PerlIO_printf(Perl_debug_log, | |
2185 | "Copy on write: regexp capture, type %d\n", | |
2186 | (int) SvTYPE(sv)); | |
2187 | } | |
2188 | prog->saved_copy = sv_setsv_cow(prog->saved_copy, sv); | |
d5263905 | 2189 | prog->subbeg = (char *)SvPVX_const(prog->saved_copy); |
ed252734 NC |
2190 | assert (SvPOKp(prog->saved_copy)); |
2191 | } else | |
2192 | #endif | |
2193 | { | |
288b8c02 | 2194 | RX_MATCH_COPIED_on(rx); |
ed252734 NC |
2195 | s = savepvn(strbeg, i); |
2196 | prog->subbeg = s; | |
2197 | } | |
d6a28714 | 2198 | prog->sublen = i; |
d6a28714 JH |
2199 | } |
2200 | else { | |
2201 | prog->subbeg = strbeg; | |
2202 | prog->sublen = PL_regeol - strbeg; /* strend may have been modified */ | |
2203 | } | |
2204 | } | |
9041c2e3 | 2205 | |
d6a28714 JH |
2206 | return 1; |
2207 | ||
2208 | phooey: | |
a3621e74 | 2209 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch failed%s\n", |
e4584336 | 2210 | PL_colors[4], PL_colors[5])); |
d6a28714 | 2211 | if (PL_reg_eval_set) |
4f639d21 | 2212 | restore_pos(aTHX_ prog); |
e9105d30 | 2213 | if (swap) { |
c74340f9 | 2214 | /* we failed :-( roll it back */ |
e9105d30 GG |
2215 | Safefree(prog->offs); |
2216 | prog->offs = swap; | |
2217 | } | |
2218 | ||
d6a28714 JH |
2219 | return 0; |
2220 | } | |
2221 | ||
6bda09f9 | 2222 | |
d6a28714 JH |
2223 | /* |
2224 | - regtry - try match at specific point | |
2225 | */ | |
2226 | STATIC I32 /* 0 failure, 1 success */ | |
24b23f37 | 2227 | S_regtry(pTHX_ regmatch_info *reginfo, char **startpos) |
d6a28714 | 2228 | { |
97aff369 | 2229 | dVAR; |
d6a28714 | 2230 | CHECKPOINT lastcp; |
288b8c02 NC |
2231 | REGEXP *const rx = reginfo->prog; |
2232 | regexp *const prog = (struct regexp *)SvANY(rx); | |
f8fc2ecf | 2233 | RXi_GET_DECL(prog,progi); |
a3621e74 | 2234 | GET_RE_DEBUG_FLAGS_DECL; |
7918f24d NC |
2235 | |
2236 | PERL_ARGS_ASSERT_REGTRY; | |
2237 | ||
24b23f37 | 2238 | reginfo->cutpoint=NULL; |
d6a28714 | 2239 | |
bbe252da | 2240 | if ((prog->extflags & RXf_EVAL_SEEN) && !PL_reg_eval_set) { |
d6a28714 JH |
2241 | MAGIC *mg; |
2242 | ||
2243 | PL_reg_eval_set = RS_init; | |
a3621e74 | 2244 | DEBUG_EXECUTE_r(DEBUG_s( |
b900a521 JH |
2245 | PerlIO_printf(Perl_debug_log, " setting stack tmpbase at %"IVdf"\n", |
2246 | (IV)(PL_stack_sp - PL_stack_base)); | |
d6a28714 | 2247 | )); |
ea8d6ae1 | 2248 | SAVESTACK_CXPOS(); |
d6a28714 JH |
2249 | cxstack[cxstack_ix].blk_oldsp = PL_stack_sp - PL_stack_base; |
2250 | /* Otherwise OP_NEXTSTATE will free whatever on stack now. */ | |
2251 | SAVETMPS; | |
2252 | /* Apparently this is not needed, judging by wantarray. */ | |
e8347627 | 2253 | /* SAVEI8(cxstack[cxstack_ix].blk_gimme); |
d6a28714 JH |
2254 | cxstack[cxstack_ix].blk_gimme = G_SCALAR; */ |
2255 | ||
3b0527fe | 2256 | if (reginfo->sv) { |
d6a28714 | 2257 | /* Make $_ available to executed code. */ |
3b0527fe | 2258 | if (reginfo->sv != DEFSV) { |
59f00321 | 2259 | SAVE_DEFSV; |
414bf5ae | 2260 | DEFSV_set(reginfo->sv); |
b8c5462f | 2261 | } |
d6a28714 | 2262 | |
3b0527fe DM |
2263 | if (!(SvTYPE(reginfo->sv) >= SVt_PVMG && SvMAGIC(reginfo->sv) |
2264 | && (mg = mg_find(reginfo->sv, PERL_MAGIC_regex_global)))) { | |
d6a28714 | 2265 | /* prepare for quick setting of pos */ |
d300d9fa | 2266 | #ifdef PERL_OLD_COPY_ON_WRITE |
51a9ea20 NC |
2267 | if (SvIsCOW(reginfo->sv)) |
2268 | sv_force_normal_flags(reginfo->sv, 0); | |
d300d9fa | 2269 | #endif |
3dab1dad | 2270 | mg = sv_magicext(reginfo->sv, NULL, PERL_MAGIC_regex_global, |
d300d9fa | 2271 | &PL_vtbl_mglob, NULL, 0); |
d6a28714 | 2272 | mg->mg_len = -1; |
b8c5462f | 2273 | } |
d6a28714 JH |
2274 | PL_reg_magic = mg; |
2275 | PL_reg_oldpos = mg->mg_len; | |
4f639d21 | 2276 | SAVEDESTRUCTOR_X(restore_pos, prog); |
d6a28714 | 2277 | } |
09687e5a | 2278 | if (!PL_reg_curpm) { |
a02a5408 | 2279 | Newxz(PL_reg_curpm, 1, PMOP); |
09687e5a AB |
2280 | #ifdef USE_ITHREADS |
2281 | { | |
14a49a24 | 2282 | SV* const repointer = &PL_sv_undef; |
92313705 NC |
2283 | /* this regexp is also owned by the new PL_reg_curpm, which |
2284 | will try to free it. */ | |
d2ece331 | 2285 | av_push(PL_regex_padav, repointer); |
09687e5a AB |
2286 | PL_reg_curpm->op_pmoffset = av_len(PL_regex_padav); |
2287 | PL_regex_pad = AvARRAY(PL_regex_padav); | |
2288 | } | |
2289 | #endif | |
2290 | } | |
86c29d75 NC |
2291 | #ifdef USE_ITHREADS |
2292 | /* It seems that non-ithreads works both with and without this code. | |
2293 | So for efficiency reasons it seems best not to have the code | |
2294 | compiled when it is not needed. */ | |
92313705 NC |
2295 | /* This is safe against NULLs: */ |
2296 | ReREFCNT_dec(PM_GETRE(PL_reg_curpm)); | |
2297 | /* PM_reg_curpm owns a reference to this regexp. */ | |
2298 | ReREFCNT_inc(rx); | |
86c29d75 | 2299 | #endif |
288b8c02 | 2300 | PM_SETRE(PL_reg_curpm, rx); |
d6a28714 JH |
2301 | PL_reg_oldcurpm = PL_curpm; |
2302 | PL_curpm = PL_reg_curpm; | |
07bc277f | 2303 | if (RXp_MATCH_COPIED(prog)) { |
d6a28714 JH |
2304 | /* Here is a serious problem: we cannot rewrite subbeg, |
2305 | since it may be needed if this match fails. Thus | |
2306 | $` inside (?{}) could fail... */ | |
2307 | PL_reg_oldsaved = prog->subbeg; | |
2308 | PL_reg_oldsavedlen = prog->sublen; | |
f8c7b90f | 2309 | #ifdef PERL_OLD_COPY_ON_WRITE |
ed252734 NC |
2310 | PL_nrs = prog->saved_copy; |
2311 | #endif | |
07bc277f | 2312 | RXp_MATCH_COPIED_off(prog); |
d6a28714 JH |
2313 | } |
2314 | else | |
bd61b366 | 2315 | PL_reg_oldsaved = NULL; |
d6a28714 JH |
2316 | prog->subbeg = PL_bostr; |
2317 | prog->sublen = PL_regeol - PL_bostr; /* strend may have been modified */ | |
2318 | } | |
24b23f37 | 2319 | DEBUG_EXECUTE_r(PL_reg_starttry = *startpos); |
f0ab9afb | 2320 | prog->offs[0].start = *startpos - PL_bostr; |
24b23f37 | 2321 | PL_reginput = *startpos; |
d6a28714 | 2322 | PL_reglastparen = &prog->lastparen; |
a01268b5 | 2323 | PL_reglastcloseparen = &prog->lastcloseparen; |
d6a28714 | 2324 | prog->lastparen = 0; |
03994de8 | 2325 | prog->lastcloseparen = 0; |
d6a28714 | 2326 | PL_regsize = 0; |
f0ab9afb | 2327 | PL_regoffs = prog->offs; |
d6a28714 JH |
2328 | if (PL_reg_start_tmpl <= prog->nparens) { |
2329 | PL_reg_start_tmpl = prog->nparens*3/2 + 3; | |
2330 | if(PL_reg_start_tmp) | |
2331 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
2332 | else | |
a02a5408 | 2333 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); |
d6a28714 JH |
2334 | } |
2335 | ||
2336 | /* XXXX What this code is doing here?!!! There should be no need | |
2337 | to do this again and again, PL_reglastparen should take care of | |
3dd2943c | 2338 | this! --ilya*/ |
dafc8851 JH |
2339 | |
2340 | /* Tests pat.t#187 and split.t#{13,14} seem to depend on this code. | |
2341 | * Actually, the code in regcppop() (which Ilya may be meaning by | |
daf18116 | 2342 | * PL_reglastparen), is not needed at all by the test suite |
225593e1 DM |
2343 | * (op/regexp, op/pat, op/split), but that code is needed otherwise |
2344 | * this erroneously leaves $1 defined: "1" =~ /^(?:(\d)x)?\d$/ | |
2345 | * Meanwhile, this code *is* needed for the | |
daf18116 JH |
2346 | * above-mentioned test suite tests to succeed. The common theme |
2347 | * on those tests seems to be returning null fields from matches. | |
225593e1 | 2348 | * --jhi updated by dapm */ |
dafc8851 | 2349 | #if 1 |
d6a28714 | 2350 | if (prog->nparens) { |
f0ab9afb | 2351 | regexp_paren_pair *pp = PL_regoffs; |
097eb12c | 2352 | register I32 i; |
eb160463 | 2353 | for (i = prog->nparens; i > (I32)*PL_reglastparen; i--) { |
f0ab9afb NC |
2354 | ++pp; |
2355 | pp->start = -1; | |
2356 | pp->end = -1; | |
d6a28714 JH |
2357 | } |
2358 | } | |
dafc8851 | 2359 | #endif |
02db2b7b | 2360 | REGCP_SET(lastcp); |
f8fc2ecf | 2361 | if (regmatch(reginfo, progi->program + 1)) { |
f0ab9afb | 2362 | PL_regoffs[0].end = PL_reginput - PL_bostr; |
d6a28714 JH |
2363 | return 1; |
2364 | } | |
24b23f37 YO |
2365 | if (reginfo->cutpoint) |
2366 | *startpos= reginfo->cutpoint; | |
02db2b7b | 2367 | REGCP_UNWIND(lastcp); |
d6a28714 JH |
2368 | return 0; |
2369 | } | |
2370 | ||
02db2b7b | 2371 | |
8ba1375e MJD |
2372 | #define sayYES goto yes |
2373 | #define sayNO goto no | |
262b90c4 | 2374 | #define sayNO_SILENT goto no_silent |
8ba1375e | 2375 | |
f9f4320a YO |
2376 | /* we dont use STMT_START/END here because it leads to |
2377 | "unreachable code" warnings, which are bogus, but distracting. */ | |
2378 | #define CACHEsayNO \ | |
c476f425 DM |
2379 | if (ST.cache_mask) \ |
2380 | PL_reg_poscache[ST.cache_offset] |= ST.cache_mask; \ | |
f9f4320a | 2381 | sayNO |
3298f257 | 2382 | |
a3621e74 | 2383 | /* this is used to determine how far from the left messages like |
265c4333 YO |
2384 | 'failed...' are printed. It should be set such that messages |
2385 | are inline with the regop output that created them. | |
a3621e74 | 2386 | */ |
265c4333 | 2387 | #define REPORT_CODE_OFF 32 |
a3621e74 YO |
2388 | |
2389 | ||
2390 | /* Make sure there is a test for this +1 options in re_tests */ | |
2391 | #define TRIE_INITAL_ACCEPT_BUFFLEN 4; | |
2392 | ||
40a82448 DM |
2393 | #define CHRTEST_UNINIT -1001 /* c1/c2 haven't been calculated yet */ |
2394 | #define CHRTEST_VOID -1000 /* the c1/c2 "next char" test should be skipped */ | |
9e137952 | 2395 | |
86545054 DM |
2396 | #define SLAB_FIRST(s) (&(s)->states[0]) |
2397 | #define SLAB_LAST(s) (&(s)->states[PERL_REGMATCH_SLAB_SLOTS-1]) | |
2398 | ||
5d9a96ca DM |
2399 | /* grab a new slab and return the first slot in it */ |
2400 | ||
2401 | STATIC regmatch_state * | |
2402 | S_push_slab(pTHX) | |
2403 | { | |
a35a87e7 | 2404 | #if PERL_VERSION < 9 && !defined(PERL_CORE) |
54df2634 NC |
2405 | dMY_CXT; |
2406 | #endif | |
5d9a96ca DM |
2407 | regmatch_slab *s = PL_regmatch_slab->next; |
2408 | if (!s) { | |
2409 | Newx(s, 1, regmatch_slab); | |
2410 | s->prev = PL_regmatch_slab; | |
2411 | s->next = NULL; | |
2412 | PL_regmatch_slab->next = s; | |
2413 | } | |
2414 | PL_regmatch_slab = s; | |
86545054 | 2415 | return SLAB_FIRST(s); |
5d9a96ca | 2416 | } |
5b47454d | 2417 | |
95b24440 | 2418 | |
40a82448 DM |
2419 | /* push a new state then goto it */ |
2420 | ||
2421 | #define PUSH_STATE_GOTO(state, node) \ | |
2422 | scan = node; \ | |
2423 | st->resume_state = state; \ | |
2424 | goto push_state; | |
2425 | ||
2426 | /* push a new state with success backtracking, then goto it */ | |
2427 | ||
2428 | #define PUSH_YES_STATE_GOTO(state, node) \ | |
2429 | scan = node; \ | |
2430 | st->resume_state = state; \ | |
2431 | goto push_yes_state; | |
2432 | ||
aa283a38 | 2433 | |
aa283a38 | 2434 | |
d6a28714 | 2435 | /* |
95b24440 | 2436 | |
bf1f174e DM |
2437 | regmatch() - main matching routine |
2438 | ||
2439 | This is basically one big switch statement in a loop. We execute an op, | |
2440 | set 'next' to point the next op, and continue. If we come to a point which | |
2441 | we may need to backtrack to on failure such as (A|B|C), we push a | |
2442 | backtrack state onto the backtrack stack. On failure, we pop the top | |
2443 | state, and re-enter the loop at the state indicated. If there are no more | |
2444 | states to pop, we return failure. | |
2445 | ||
2446 | Sometimes we also need to backtrack on success; for example /A+/, where | |
2447 | after successfully matching one A, we need to go back and try to | |
2448 | match another one; similarly for lookahead assertions: if the assertion | |
2449 | completes successfully, we backtrack to the state just before the assertion | |
2450 | and then carry on. In these cases, the pushed state is marked as | |
2451 | 'backtrack on success too'. This marking is in fact done by a chain of | |
2452 | pointers, each pointing to the previous 'yes' state. On success, we pop to | |
2453 | the nearest yes state, discarding any intermediate failure-only states. | |
2454 | Sometimes a yes state is pushed just to force some cleanup code to be | |
2455 | called at the end of a successful match or submatch; e.g. (??{$re}) uses | |
2456 | it to free the inner regex. | |
2457 | ||
2458 | Note that failure backtracking rewinds the cursor position, while | |
2459 | success backtracking leaves it alone. | |
2460 | ||
2461 | A pattern is complete when the END op is executed, while a subpattern | |
2462 | such as (?=foo) is complete when the SUCCESS op is executed. Both of these | |
2463 | ops trigger the "pop to last yes state if any, otherwise return true" | |
2464 | behaviour. | |
2465 | ||
2466 | A common convention in this function is to use A and B to refer to the two | |
2467 | subpatterns (or to the first nodes thereof) in patterns like /A*B/: so A is | |
2468 | the subpattern to be matched possibly multiple times, while B is the entire | |
2469 | rest of the pattern. Variable and state names reflect this convention. | |
2470 | ||
2471 | The states in the main switch are the union of ops and failure/success of | |
2472 | substates associated with with that op. For example, IFMATCH is the op | |
2473 | that does lookahead assertions /(?=A)B/ and so the IFMATCH state means | |
2474 | 'execute IFMATCH'; while IFMATCH_A is a state saying that we have just | |
2475 | successfully matched A and IFMATCH_A_fail is a state saying that we have | |
2476 | just failed to match A. Resume states always come in pairs. The backtrack | |
2477 | state we push is marked as 'IFMATCH_A', but when that is popped, we resume | |
2478 | at IFMATCH_A or IFMATCH_A_fail, depending on whether we are backtracking | |
2479 | on success or failure. | |
2480 | ||
2481 | The struct that holds a backtracking state is actually a big union, with | |
2482 | one variant for each major type of op. The variable st points to the | |
2483 | top-most backtrack struct. To make the code clearer, within each | |
2484 | block of code we #define ST to alias the relevant union. | |
2485 | ||
2486 | Here's a concrete example of a (vastly oversimplified) IFMATCH | |
2487 | implementation: | |
2488 | ||
2489 | switch (state) { | |
2490 | .... | |
2491 | ||
2492 | #define ST st->u.ifmatch | |
2493 | ||
2494 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B | |
2495 | ST.foo = ...; // some state we wish to save | |
95b24440 | 2496 | ... |
bf1f174e DM |
2497 | // push a yes backtrack state with a resume value of |
2498 | // IFMATCH_A/IFMATCH_A_fail, then continue execution at the | |
2499 | // first node of A: | |
2500 | PUSH_YES_STATE_GOTO(IFMATCH_A, A); | |
2501 | // NOTREACHED | |
2502 | ||
2503 | case IFMATCH_A: // we have successfully executed A; now continue with B | |
2504 | next = B; | |
2505 | bar = ST.foo; // do something with the preserved value | |
2506 | break; | |
2507 | ||
2508 | case IFMATCH_A_fail: // A failed, so the assertion failed | |
2509 | ...; // do some housekeeping, then ... | |
2510 | sayNO; // propagate the failure | |
2511 | ||
2512 | #undef ST | |
95b24440 | 2513 | |
bf1f174e DM |
2514 | ... |
2515 | } | |
95b24440 | 2516 | |
bf1f174e DM |
2517 | For any old-timers reading this who are familiar with the old recursive |
2518 | approach, the code above is equivalent to: | |
95b24440 | 2519 | |
bf1f174e DM |
2520 | case IFMATCH: // we are executing the IFMATCH op, (?=A)B |
2521 | { | |
2522 | int foo = ... | |
95b24440 | 2523 | ... |
bf1f174e DM |
2524 | if (regmatch(A)) { |
2525 | next = B; | |
2526 | bar = foo; | |
2527 | break; | |
95b24440 | 2528 | } |
bf1f174e DM |
2529 | ...; // do some housekeeping, then ... |
2530 | sayNO; // propagate the failure | |
95b24440 | 2531 | } |
bf1f174e DM |
2532 | |
2533 | The topmost backtrack state, pointed to by st, is usually free. If you | |
2534 | want to claim it, populate any ST.foo fields in it with values you wish to | |
2535 | save, then do one of | |
2536 | ||
2537 | PUSH_STATE_GOTO(resume_state, node); | |
2538 | PUSH_YES_STATE_GOTO(resume_state, node); | |
2539 | ||
2540 | which sets that backtrack state's resume value to 'resume_state', pushes a | |
2541 | new free entry to the top of the backtrack stack, then goes to 'node'. | |
2542 | On backtracking, the free slot is popped, and the saved state becomes the | |
2543 | new free state. An ST.foo field in this new top state can be temporarily | |
2544 | accessed to retrieve values, but once the main loop is re-entered, it | |
2545 | becomes available for reuse. | |
2546 | ||
2547 | Note that the depth of the backtrack stack constantly increases during the | |
2548 | left-to-right execution of the pattern, rather than going up and down with | |
2549 | the pattern nesting. For example the stack is at its maximum at Z at the | |
2550 | end of the pattern, rather than at X in the following: | |
2551 | ||
2552 | /(((X)+)+)+....(Y)+....Z/ | |
2553 | ||
2554 | The only exceptions to this are lookahead/behind assertions and the cut, | |
2555 | (?>A), which pop all the backtrack states associated with A before | |
2556 | continuing. | |
2557 | ||
2558 | Bascktrack state structs are allocated in slabs of about 4K in size. | |
2559 | PL_regmatch_state and st always point to the currently active state, | |
2560 | and PL_regmatch_slab points to the slab currently containing | |
2561 | PL_regmatch_state. The first time regmatch() is called, the first slab is | |
2562 | allocated, and is never freed until interpreter destruction. When the slab | |
2563 | is full, a new one is allocated and chained to the end. At exit from | |
2564 | regmatch(), slabs allocated since entry are freed. | |
2565 | ||
2566 | */ | |
95b24440 | 2567 | |
40a82448 | 2568 | |
5bc10b2c | 2569 | #define DEBUG_STATE_pp(pp) \ |
265c4333 | 2570 | DEBUG_STATE_r({ \ |
5bc10b2c DM |
2571 | DUMP_EXEC_POS(locinput, scan, do_utf8); \ |
2572 | PerlIO_printf(Perl_debug_log, \ | |
5d458dd8 | 2573 | " %*s"pp" %s%s%s%s%s\n", \ |
5bc10b2c | 2574 | depth*2, "", \ |
13d6edb4 | 2575 | PL_reg_name[st->resume_state], \ |
5d458dd8 YO |
2576 | ((st==yes_state||st==mark_state) ? "[" : ""), \ |
2577 | ((st==yes_state) ? "Y" : ""), \ | |
2578 | ((st==mark_state) ? "M" : ""), \ | |
2579 | ((st==yes_state||st==mark_state) ? "]" : "") \ | |
2580 | ); \ | |
265c4333 | 2581 | }); |
5bc10b2c | 2582 | |
40a82448 | 2583 | |
3dab1dad | 2584 | #define REG_NODE_NUM(x) ((x) ? (int)((x)-prog) : -1) |
95b24440 | 2585 | |
3df15adc | 2586 | #ifdef DEBUGGING |
5bc10b2c | 2587 | |
ab3bbdeb | 2588 | STATIC void |
efd26800 | 2589 | S_debug_start_match(pTHX_ const REGEXP *prog, const bool do_utf8, |
ab3bbdeb YO |
2590 | const char *start, const char *end, const char *blurb) |
2591 | { | |
efd26800 | 2592 | const bool utf8_pat = RX_UTF8(prog) ? 1 : 0; |
7918f24d NC |
2593 | |
2594 | PERL_ARGS_ASSERT_DEBUG_START_MATCH; | |
2595 | ||
ab3bbdeb YO |
2596 | if (!PL_colorset) |
2597 | reginitcolors(); | |
2598 | { | |
2599 | RE_PV_QUOTED_DECL(s0, utf8_pat, PERL_DEBUG_PAD_ZERO(0), | |
d2c6dc5e | 2600 | RX_PRECOMP_const(prog), RX_PRELEN(prog), 60); |
ab3bbdeb YO |
2601 | |
2602 | RE_PV_QUOTED_DECL(s1, do_utf8, PERL_DEBUG_PAD_ZERO(1), | |
2603 | start, end - start, 60); | |
2604 | ||
2605 | PerlIO_printf(Perl_debug_log, | |
2606 | "%s%s REx%s %s against %s\n", | |
2607 | PL_colors[4], blurb, PL_colors[5], s0, s1); | |
2608 | ||
2609 | if (do_utf8||utf8_pat) | |
1de06328 YO |
2610 | PerlIO_printf(Perl_debug_log, "UTF-8 %s%s%s...\n", |
2611 | utf8_pat ? "pattern" : "", | |
2612 | utf8_pat && do_utf8 ? " and " : "", | |
2613 | do_utf8 ? "string" : "" | |
ab3bbdeb YO |
2614 | ); |
2615 | } | |
2616 | } | |
3df15adc YO |
2617 | |
2618 | STATIC void | |
786e8c11 YO |
2619 | S_dump_exec_pos(pTHX_ const char *locinput, |
2620 | const regnode *scan, | |
2621 | const char *loc_regeol, | |
2622 | const char *loc_bostr, | |
2623 | const char *loc_reg_starttry, | |
2624 | const bool do_utf8) | |
07be1b83 | 2625 | { |
786e8c11 | 2626 | const int docolor = *PL_colors[0] || *PL_colors[2] || *PL_colors[4]; |
07be1b83 | 2627 | const int taill = (docolor ? 10 : 7); /* 3 chars for "> <" */ |
786e8c11 | 2628 | int l = (loc_regeol - locinput) > taill ? taill : (loc_regeol - locinput); |
07be1b83 YO |
2629 | /* The part of the string before starttry has one color |
2630 | (pref0_len chars), between starttry and current | |
2631 | position another one (pref_len - pref0_len chars), | |
2632 | after the current position the third one. | |
2633 | We assume that pref0_len <= pref_len, otherwise we | |
2634 | decrease pref0_len. */ | |
786e8c11 YO |
2635 | int pref_len = (locinput - loc_bostr) > (5 + taill) - l |
2636 | ? (5 + taill) - l : locinput - loc_bostr; | |
07be1b83 YO |
2637 | int pref0_len; |
2638 | ||
7918f24d NC |
2639 | PERL_ARGS_ASSERT_DUMP_EXEC_POS; |
2640 | ||
07be1b83 YO |
2641 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput - pref_len))) |
2642 | pref_len++; | |
786e8c11 YO |
2643 | pref0_len = pref_len - (locinput - loc_reg_starttry); |
2644 | if (l + pref_len < (5 + taill) && l < loc_regeol - locinput) | |
2645 | l = ( loc_regeol - locinput > (5 + taill) - pref_len | |
2646 | ? (5 + taill) - pref_len : loc_regeol - locinput); | |
07be1b83 YO |
2647 | while (do_utf8 && UTF8_IS_CONTINUATION(*(U8*)(locinput + l))) |
2648 | l--; | |
2649 | if (pref0_len < 0) | |
2650 | pref0_len = 0; | |
2651 | if (pref0_len > pref_len) | |
2652 | pref0_len = pref_len; | |
2653 | { | |
3df15adc | 2654 | const int is_uni = (do_utf8 && OP(scan) != CANY) ? 1 : 0; |
0df25f3d | 2655 | |
ab3bbdeb | 2656 | RE_PV_COLOR_DECL(s0,len0,is_uni,PERL_DEBUG_PAD(0), |
1de06328 | 2657 | (locinput - pref_len),pref0_len, 60, 4, 5); |
0df25f3d | 2658 | |
ab3bbdeb | 2659 | RE_PV_COLOR_DECL(s1,len1,is_uni,PERL_DEBUG_PAD(1), |
3df15adc | 2660 | (locinput - pref_len + pref0_len), |
1de06328 | 2661 | pref_len - pref0_len, 60, 2, 3); |
0df25f3d | 2662 | |
ab3bbdeb | 2663 | RE_PV_COLOR_DECL(s2,len2,is_uni,PERL_DEBUG_PAD(2), |
1de06328 | 2664 | locinput, loc_regeol - locinput, 10, 0, 1); |
0df25f3d | 2665 | |
1de06328 | 2666 | const STRLEN tlen=len0+len1+len2; |
3df15adc | 2667 | PerlIO_printf(Perl_debug_log, |
ab3bbdeb | 2668 | "%4"IVdf" <%.*s%.*s%s%.*s>%*s|", |
786e8c11 | 2669 | (IV)(locinput - loc_bostr), |
07be1b83 | 2670 | len0, s0, |
07be1b83 | 2671 | len1, s1, |
07be1b83 | 2672 | (docolor ? "" : "> <"), |
07be1b83 | 2673 | len2, s2, |
f9f4320a | 2674 | (int)(tlen > 19 ? 0 : 19 - tlen), |
07be1b83 YO |
2675 | ""); |
2676 | } | |
2677 | } | |
3df15adc | 2678 | |
07be1b83 YO |
2679 | #endif |
2680 | ||
0a4db386 YO |
2681 | /* reg_check_named_buff_matched() |
2682 | * Checks to see if a named buffer has matched. The data array of | |
2683 | * buffer numbers corresponding to the buffer is expected to reside | |
2684 | * in the regexp->data->data array in the slot stored in the ARG() of | |
2685 | * node involved. Note that this routine doesn't actually care about the | |
2686 | * name, that information is not preserved from compilation to execution. | |
2687 | * Returns the index of the leftmost defined buffer with the given name | |
2688 | * or 0 if non of the buffers matched. | |
2689 | */ | |
2690 | STATIC I32 | |
7918f24d NC |
2691 | S_reg_check_named_buff_matched(pTHX_ const regexp *rex, const regnode *scan) |
2692 | { | |
0a4db386 | 2693 | I32 n; |
f8fc2ecf | 2694 | RXi_GET_DECL(rex,rexi); |
ad64d0ec | 2695 | SV *sv_dat= MUTABLE_SV(rexi->data->data[ ARG( scan ) ]); |
0a4db386 | 2696 | I32 *nums=(I32*)SvPVX(sv_dat); |
7918f24d NC |
2697 | |
2698 | PERL_ARGS_ASSERT_REG_CHECK_NAMED_BUFF_MATCHED; | |
2699 | ||
0a4db386 YO |
2700 | for ( n=0; n<SvIVX(sv_dat); n++ ) { |
2701 | if ((I32)*PL_reglastparen >= nums[n] && | |
f0ab9afb | 2702 | PL_regoffs[nums[n]].end != -1) |
0a4db386 YO |
2703 | { |
2704 | return nums[n]; | |
2705 | } | |
2706 | } | |
2707 | return 0; | |
2708 | } | |
2709 | ||
2f554ef7 DM |
2710 | |
2711 | /* free all slabs above current one - called during LEAVE_SCOPE */ | |
2712 | ||
2713 | STATIC void | |
2714 | S_clear_backtrack_stack(pTHX_ void *p) | |
2715 | { | |
2716 | regmatch_slab *s = PL_regmatch_slab->next; | |
2717 | PERL_UNUSED_ARG(p); | |
2718 | ||
2719 | if (!s) | |
2720 | return; | |
2721 | PL_regmatch_slab->next = NULL; | |
2722 | while (s) { | |
2723 | regmatch_slab * const osl = s; | |
2724 | s = s->next; | |
2725 | Safefree(osl); | |
2726 | } | |
2727 | } | |
2728 | ||
2729 | ||
28d8d7f4 YO |
2730 | #define SETREX(Re1,Re2) \ |
2731 | if (PL_reg_eval_set) PM_SETRE((PL_reg_curpm), (Re2)); \ | |
2732 | Re1 = (Re2) | |
2733 | ||
d6a28714 | 2734 | STATIC I32 /* 0 failure, 1 success */ |
24b23f37 | 2735 | S_regmatch(pTHX_ regmatch_info *reginfo, regnode *prog) |
d6a28714 | 2736 | { |
a35a87e7 | 2737 | #if PERL_VERSION < 9 && !defined(PERL_CORE) |
54df2634 NC |
2738 | dMY_CXT; |
2739 | #endif | |
27da23d5 | 2740 | dVAR; |
95b24440 | 2741 | register const bool do_utf8 = PL_reg_match_utf8; |
4ad0818d | 2742 | const U32 uniflags = UTF8_ALLOW_DEFAULT; |
288b8c02 NC |
2743 | REGEXP *rex_sv = reginfo->prog; |
2744 | regexp *rex = (struct regexp *)SvANY(rex_sv); | |
f8fc2ecf | 2745 | RXi_GET_DECL(rex,rexi); |
2f554ef7 | 2746 | I32 oldsave; |
5d9a96ca DM |
2747 | /* the current state. This is a cached copy of PL_regmatch_state */ |
2748 | register regmatch_state *st; | |
5d9a96ca DM |
2749 | /* cache heavy used fields of st in registers */ |
2750 | register regnode *scan; | |
2751 | register regnode *next; | |
438e9bae | 2752 | register U32 n = 0; /* general value; init to avoid compiler warning */ |
24d3c4a9 | 2753 | register I32 ln = 0; /* len or last; init to avoid compiler warning */ |
5d9a96ca | 2754 | register char *locinput = PL_reginput; |
5d9a96ca | 2755 | register I32 nextchr; /* is always set to UCHARAT(locinput) */ |
24d3c4a9 | 2756 | |
b69b0499 | 2757 | bool result = 0; /* return value of S_regmatch */ |
24d3c4a9 | 2758 | int depth = 0; /* depth of backtrack stack */ |
4b196cd4 YO |
2759 | U32 nochange_depth = 0; /* depth of GOSUB recursion with nochange */ |
2760 | const U32 max_nochange_depth = | |
2761 | (3 * rex->nparens > MAX_RECURSE_EVAL_NOCHANGE_DEPTH) ? | |
2762 | 3 * rex->nparens : MAX_RECURSE_EVAL_NOCHANGE_DEPTH; | |
77cb431f DM |
2763 | regmatch_state *yes_state = NULL; /* state to pop to on success of |
2764 | subpattern */ | |
e2e6a0f1 YO |
2765 | /* mark_state piggy backs on the yes_state logic so that when we unwind |
2766 | the stack on success we can update the mark_state as we go */ | |
2767 | regmatch_state *mark_state = NULL; /* last mark state we have seen */ | |
faec1544 | 2768 | regmatch_state *cur_eval = NULL; /* most recent EVAL_AB state */ |
b8591aee | 2769 | struct regmatch_state *cur_curlyx = NULL; /* most recent curlyx */ |
40a82448 | 2770 | U32 state_num; |
5d458dd8 YO |
2771 | bool no_final = 0; /* prevent failure from backtracking? */ |
2772 | bool do_cutgroup = 0; /* no_final only until next branch/trie entry */ | |
e2e6a0f1 | 2773 | char *startpoint = PL_reginput; |
5d458dd8 YO |
2774 | SV *popmark = NULL; /* are we looking for a mark? */ |
2775 | SV *sv_commit = NULL; /* last mark name seen in failure */ | |
2776 | SV *sv_yes_mark = NULL; /* last mark name we have seen | |
2777 | during a successfull match */ | |
2778 | U32 lastopen = 0; /* last open we saw */ | |
2779 | bool has_cutgroup = RX_HAS_CUTGROUP(rex) ? 1 : 0; | |
19b95bf0 | 2780 | SV* const oreplsv = GvSV(PL_replgv); |
24d3c4a9 DM |
2781 | /* these three flags are set by various ops to signal information to |
2782 | * the very next op. They have a useful lifetime of exactly one loop | |
2783 | * iteration, and are not preserved or restored by state pushes/pops | |
2784 | */ | |
2785 | bool sw = 0; /* the condition value in (?(cond)a|b) */ | |
2786 | bool minmod = 0; /* the next "{n,m}" is a "{n,m}?" */ | |
2787 | int logical = 0; /* the following EVAL is: | |
2788 | 0: (?{...}) | |
2789 | 1: (?(?{...})X|Y) | |
2790 | 2: (??{...}) | |
2791 | or the following IFMATCH/UNLESSM is: | |
2792 | false: plain (?=foo) | |
2793 | true: used as a condition: (?(?=foo)) | |
2794 | */ | |
95b24440 | 2795 | #ifdef DEBUGGING |
e68ec53f | 2796 | GET_RE_DEBUG_FLAGS_DECL; |
d6a28714 JH |
2797 | #endif |
2798 | ||
7918f24d NC |
2799 | PERL_ARGS_ASSERT_REGMATCH; |
2800 | ||
3b57cd43 | 2801 | DEBUG_OPTIMISE_r( DEBUG_EXECUTE_r({ |
24b23f37 | 2802 | PerlIO_printf(Perl_debug_log,"regmatch start\n"); |
3b57cd43 | 2803 | })); |
5d9a96ca DM |
2804 | /* on first ever call to regmatch, allocate first slab */ |
2805 | if (!PL_regmatch_slab) { | |
2806 | Newx(PL_regmatch_slab, 1, regmatch_slab); | |
2807 | PL_regmatch_slab->prev = NULL; | |
2808 | PL_regmatch_slab->next = NULL; | |
86545054 | 2809 | PL_regmatch_state = SLAB_FIRST(PL_regmatch_slab); |
5d9a96ca DM |
2810 | } |
2811 | ||
2f554ef7 DM |
2812 | oldsave = PL_savestack_ix; |
2813 | SAVEDESTRUCTOR_X(S_clear_backtrack_stack, NULL); | |
2814 | SAVEVPTR(PL_regmatch_slab); | |
2815 | SAVEVPTR(PL_regmatch_state); | |
5d9a96ca DM |
2816 | |
2817 | /* grab next free state slot */ | |
2818 | st = ++PL_regmatch_state; | |
86545054 | 2819 | if (st > SLAB_LAST(PL_regmatch_slab)) |
5d9a96ca DM |
2820 | st = PL_regmatch_state = S_push_slab(aTHX); |
2821 | ||
d6a28714 JH |
2822 | /* Note that nextchr is a byte even in UTF */ |
2823 | nextchr = UCHARAT(locinput); | |
2824 | scan = prog; | |
2825 | while (scan != NULL) { | |
8ba1375e | 2826 | |
a3621e74 | 2827 | DEBUG_EXECUTE_r( { |
6136c704 | 2828 | SV * const prop = sv_newmortal(); |
1de06328 | 2829 | regnode *rnext=regnext(scan); |
786e8c11 | 2830 | DUMP_EXEC_POS( locinput, scan, do_utf8 ); |
32fc9b6a | 2831 | regprop(rex, prop, scan); |
07be1b83 YO |
2832 | |
2833 | PerlIO_printf(Perl_debug_log, | |
2834 | "%3"IVdf":%*s%s(%"IVdf")\n", | |
f8fc2ecf | 2835 | (IV)(scan - rexi->program), depth*2, "", |
07be1b83 | 2836 | SvPVX_const(prop), |
1de06328 | 2837 | (PL_regkind[OP(scan)] == END || !rnext) ? |
f8fc2ecf | 2838 | 0 : (IV)(rnext - rexi->program)); |
2a782b5b | 2839 | }); |
d6a28714 JH |
2840 | |
2841 | next = scan + NEXT_OFF(scan); | |
2842 | if (next == scan) | |
2843 | next = NULL; | |
40a82448 | 2844 | state_num = OP(scan); |
d6a28714 | 2845 | |
40a82448 | 2846 | reenter_switch: |
34a81e2b B |
2847 | |
2848 | assert(PL_reglastparen == &rex->lastparen); | |
2849 | assert(PL_reglastcloseparen == &rex->lastcloseparen); | |
2850 | assert(PL_regoffs == rex->offs); | |
2851 | ||
40a82448 | 2852 | switch (state_num) { |
d6a28714 | 2853 | case BOL: |
7fba1cd6 | 2854 | if (locinput == PL_bostr) |
d6a28714 | 2855 | { |
3b0527fe | 2856 | /* reginfo->till = reginfo->bol; */ |
b8c5462f JH |
2857 | break; |
2858 | } | |
d6a28714 JH |
2859 | sayNO; |
2860 | case MBOL: | |
12d33761 HS |
2861 | if (locinput == PL_bostr || |
2862 | ((nextchr || locinput < PL_regeol) && locinput[-1] == '\n')) | |
d6a28714 | 2863 | { |
b8c5462f JH |
2864 | break; |
2865 | } | |
d6a28714 JH |
2866 | sayNO; |
2867 | case SBOL: | |
c2a73568 | 2868 | if (locinput == PL_bostr) |
b8c5462f | 2869 | break; |
d6a28714 JH |
2870 | sayNO; |
2871 | case GPOS: | |
3b0527fe | 2872 | if (locinput == reginfo->ganch) |
d6a28714 JH |
2873 | break; |
2874 | sayNO; | |
ee9b8eae YO |
2875 | |
2876 | case KEEPS: | |
2877 | /* update the startpoint */ | |
f0ab9afb | 2878 | st->u.keeper.val = PL_regoffs[0].start; |
ee9b8eae | 2879 | PL_reginput = locinput; |
f0ab9afb | 2880 | PL_regoffs[0].start = locinput - PL_bostr; |
ee9b8eae YO |
2881 | PUSH_STATE_GOTO(KEEPS_next, next); |
2882 | /*NOT-REACHED*/ | |
2883 | case KEEPS_next_fail: | |
2884 | /* rollback the start point change */ | |
f0ab9afb | 2885 | PL_regoffs[0].start = st->u.keeper.val; |
ee9b8eae YO |
2886 | sayNO_SILENT; |
2887 | /*NOT-REACHED*/ | |
d6a28714 | 2888 | case EOL: |
d6a28714 JH |
2889 | goto seol; |
2890 | case MEOL: | |
d6a28714 | 2891 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') |
b8c5462f | 2892 | sayNO; |
b8c5462f | 2893 | break; |
d6a28714 JH |
2894 | case SEOL: |
2895 | seol: | |
2896 | if ((nextchr || locinput < PL_regeol) && nextchr != '\n') | |
b8c5462f | 2897 | sayNO; |
d6a28714 | 2898 | if (PL_regeol - locinput > 1) |
b8c5462f | 2899 | sayNO; |
b8c5462f | 2900 | break; |
d6a28714 JH |
2901 | case EOS: |
2902 | if (PL_regeol != locinput) | |
b8c5462f | 2903 | sayNO; |
d6a28714 | 2904 | break; |
ffc61ed2 | 2905 | case SANY: |
d6a28714 | 2906 | if (!nextchr && locinput >= PL_regeol) |
4633a7c4 | 2907 | sayNO; |
f33976b4 DB |
2908 | if (do_utf8) { |
2909 | locinput += PL_utf8skip[nextchr]; | |
2910 | if (locinput > PL_regeol) | |
2911 | sayNO; | |
2912 | nextchr = UCHARAT(locinput); | |
2913 | } | |
2914 | else | |
2915 | nextchr = UCHARAT(++locinput); | |
2916 | break; | |
2917 | case CANY: | |
2918 | if (!nextchr && locinput >= PL_regeol) | |
2919 | sayNO; | |
b8c5462f | 2920 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 2921 | break; |
ffc61ed2 | 2922 | case REG_ANY: |
1aa99e6b IH |
2923 | if ((!nextchr && locinput >= PL_regeol) || nextchr == '\n') |
2924 | sayNO; | |
2925 | if (do_utf8) { | |
b8c5462f | 2926 | locinput += PL_utf8skip[nextchr]; |
d6a28714 JH |
2927 | if (locinput > PL_regeol) |
2928 | sayNO; | |
a0ed51b3 | 2929 | nextchr = UCHARAT(locinput); |
a0ed51b3 | 2930 | } |
1aa99e6b IH |
2931 | else |
2932 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 2933 | break; |
166ba7cd DM |
2934 | |
2935 | #undef ST | |
2936 | #define ST st->u.trie | |
786e8c11 YO |
2937 | case TRIEC: |
2938 | /* In this case the charclass data is available inline so | |
2939 | we can fail fast without a lot of extra overhead. | |
2940 | */ | |
2941 | if (scan->flags == EXACT || !do_utf8) { | |
2942 | if(!ANYOF_BITMAP_TEST(scan, *locinput)) { | |
2943 | DEBUG_EXECUTE_r( | |
2944 | PerlIO_printf(Perl_debug_log, | |
2945 | "%*s %sfailed to match trie start class...%s\n", | |
5bc10b2c | 2946 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
786e8c11 YO |
2947 | ); |
2948 | sayNO_SILENT; | |
2949 | /* NOTREACHED */ | |
2950 | } | |
2951 | } | |
2952 | /* FALL THROUGH */ | |
5b47454d | 2953 | case TRIE: |
3dab1dad | 2954 | { |
07be1b83 | 2955 | /* what type of TRIE am I? (utf8 makes this contextual) */ |
a0a388a1 | 2956 | DECL_TRIE_TYPE(scan); |
3dab1dad YO |
2957 | |
2958 | /* what trie are we using right now */ | |
be8e71aa | 2959 | reg_trie_data * const trie |
f8fc2ecf | 2960 | = (reg_trie_data*)rexi->data->data[ ARG( scan ) ]; |
85fbaab2 | 2961 | HV * widecharmap = MUTABLE_HV(rexi->data->data[ ARG( scan ) + 1 ]); |
3dab1dad | 2962 | U32 state = trie->startstate; |
166ba7cd | 2963 | |
3dab1dad YO |
2964 | if (trie->bitmap && trie_type != trie_utf8_fold && |
2965 | !TRIE_BITMAP_TEST(trie,*locinput) | |
2966 | ) { | |
2967 | if (trie->states[ state ].wordnum) { | |
2968 | DEBUG_EXECUTE_r( | |
2969 | PerlIO_printf(Perl_debug_log, | |
2970 | "%*s %smatched empty string...%s\n", | |
5bc10b2c | 2971 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2972 | ); |
2973 | break; | |
2974 | } else { | |
2975 | DEBUG_EXECUTE_r( | |
2976 | PerlIO_printf(Perl_debug_log, | |
786e8c11 | 2977 | "%*s %sfailed to match trie start class...%s\n", |
5bc10b2c | 2978 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5]) |
3dab1dad YO |
2979 | ); |
2980 | sayNO_SILENT; | |
2981 | } | |
2982 | } | |
166ba7cd | 2983 | |
786e8c11 YO |
2984 | { |
2985 | U8 *uc = ( U8* )locinput; | |
2986 | ||
2987 | STRLEN len = 0; | |
2988 | STRLEN foldlen = 0; | |
2989 | U8 *uscan = (U8*)NULL; | |
2990 | STRLEN bufflen=0; | |
2991 | SV *sv_accept_buff = NULL; | |
2992 | U8 foldbuf[ UTF8_MAXBYTES_CASE + 1 ]; | |
2993 | ||
2994 | ST.accepted = 0; /* how many accepting states we have seen */ | |
2995 | ST.B = next; | |
2996 | ST.jump = trie->jump; | |
786e8c11 | 2997 | ST.me = scan; |
07be1b83 YO |
2998 | /* |
2999 | traverse the TRIE keeping track of all accepting states | |
3000 | we transition through until we get to a failing node. | |
3001 | */ | |
3002 | ||
a3621e74 | 3003 | while ( state && uc <= (U8*)PL_regeol ) { |
786e8c11 | 3004 | U32 base = trie->states[ state ].trans.base; |
f9f4320a | 3005 | UV uvc = 0; |
786e8c11 YO |
3006 | U16 charid; |
3007 | /* We use charid to hold the wordnum as we don't use it | |
3008 | for charid until after we have done the wordnum logic. | |
3009 | We define an alias just so that the wordnum logic reads | |
3010 | more naturally. */ | |
3011 | ||
3012 | #define got_wordnum charid | |
3013 | got_wordnum = trie->states[ state ].wordnum; | |
3014 | ||
3015 | if ( got_wordnum ) { | |
3016 | if ( ! ST.accepted ) { | |
5b47454d | 3017 | ENTER; |
6962fb1a | 3018 | SAVETMPS; /* XXX is this necessary? dmq */ |
5b47454d DM |
3019 | bufflen = TRIE_INITAL_ACCEPT_BUFFLEN; |
3020 | sv_accept_buff=newSV(bufflen * | |
3021 | sizeof(reg_trie_accepted) - 1); | |
786e8c11 | 3022 | SvCUR_set(sv_accept_buff, 0); |
5b47454d DM |
3023 | SvPOK_on(sv_accept_buff); |
3024 | sv_2mortal(sv_accept_buff); | |
166ba7cd DM |
3025 | SAVETMPS; |
3026 | ST.accept_buff = | |
5b47454d DM |
3027 | (reg_trie_accepted*)SvPV_nolen(sv_accept_buff ); |
3028 | } | |
786e8c11 | 3029 | do { |
166ba7cd | 3030 | if (ST.accepted >= bufflen) { |
5b47454d | 3031 | bufflen *= 2; |
166ba7cd | 3032 | ST.accept_buff =(reg_trie_accepted*) |
5b47454d DM |
3033 | SvGROW(sv_accept_buff, |
3034 | bufflen * sizeof(reg_trie_accepted)); | |
3035 | } | |
3036 | SvCUR_set(sv_accept_buff,SvCUR(sv_accept_buff) | |
3037 | + sizeof(reg_trie_accepted)); | |
a3621e74 | 3038 | |
786e8c11 YO |
3039 | |
3040 | ST.accept_buff[ST.accepted].wordnum = got_wordnum; | |
3041 | ST.accept_buff[ST.accepted].endpos = uc; | |
3042 | ++ST.accepted; | |
3043 | } while (trie->nextword && (got_wordnum= trie->nextword[got_wordnum])); | |
3044 | } | |
3045 | #undef got_wordnum | |
a3621e74 | 3046 | |
07be1b83 | 3047 | DEBUG_TRIE_EXECUTE_r({ |
786e8c11 | 3048 | DUMP_EXEC_POS( (char *)uc, scan, do_utf8 ); |
a3621e74 | 3049 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 3050 | "%*s %sState: %4"UVxf" Accepted: %4"UVxf" ", |
5bc10b2c | 3051 | 2+depth * 2, "", PL_colors[4], |
786e8c11 | 3052 | (UV)state, (UV)ST.accepted ); |
07be1b83 | 3053 | }); |
a3621e74 YO |
3054 | |
3055 | if ( base ) { | |
55eed653 NC |
3056 | REXEC_TRIE_READ_CHAR(trie_type, trie, widecharmap, uc, |
3057 | uscan, len, uvc, charid, foldlen, | |
3058 | foldbuf, uniflags); | |
a3621e74 | 3059 | |
5b47454d DM |
3060 | if (charid && |
3061 | (base + charid > trie->uniquecharcount ) | |
3062 | && (base + charid - 1 - trie->uniquecharcount | |
3063 | < trie->lasttrans) | |
3064 | && trie->trans[base + charid - 1 - | |
3065 | trie->uniquecharcount].check == state) | |
3066 | { | |
3067 | state = trie->trans[base + charid - 1 - | |
3068 | trie->uniquecharcount ].next; | |
3069 | } | |
3070 | else { | |
3071 | state = 0; | |
3072 | } | |
3073 | uc += len; | |
3074 | ||
3075 | } | |
3076 | else { | |
a3621e74 YO |
3077 | state = 0; |
3078 | } | |
3079 | DEBUG_TRIE_EXECUTE_r( | |
e4584336 | 3080 | PerlIO_printf( Perl_debug_log, |
786e8c11 | 3081 | "Charid:%3x CP:%4"UVxf" After State: %4"UVxf"%s\n", |
e4584336 | 3082 | charid, uvc, (UV)state, PL_colors[5] ); |
a3621e74 YO |
3083 | ); |
3084 | } | |
166ba7cd | 3085 | if (!ST.accepted ) |
a3621e74 | 3086 | sayNO; |
a3621e74 | 3087 | |
166ba7cd DM |
3088 | DEBUG_EXECUTE_r( |
3089 | PerlIO_printf( Perl_debug_log, | |
3090 | "%*s %sgot %"IVdf" possible matches%s\n", | |
5bc10b2c | 3091 | REPORT_CODE_OFF + depth * 2, "", |
166ba7cd DM |
3092 | PL_colors[4], (IV)ST.accepted, PL_colors[5] ); |
3093 | ); | |
786e8c11 | 3094 | }} |
fae667d5 YO |
3095 | goto trie_first_try; /* jump into the fail handler */ |
3096 | /* NOTREACHED */ | |
166ba7cd | 3097 | case TRIE_next_fail: /* we failed - try next alterative */ |
fae667d5 YO |
3098 | if ( ST.jump) { |
3099 | REGCP_UNWIND(ST.cp); | |
3100 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
f0ab9afb | 3101 | PL_regoffs[n].end = -1; |
fae667d5 YO |
3102 | *PL_reglastparen = n; |
3103 | } | |
3104 | trie_first_try: | |
5d458dd8 YO |
3105 | if (do_cutgroup) { |
3106 | do_cutgroup = 0; | |
3107 | no_final = 0; | |
3108 | } | |
fae667d5 YO |
3109 | |
3110 | if ( ST.jump) { | |
3111 | ST.lastparen = *PL_reglastparen; | |
3112 | REGCP_SET(ST.cp); | |
3113 | } | |
166ba7cd DM |
3114 | if ( ST.accepted == 1 ) { |
3115 | /* only one choice left - just continue */ | |
3116 | DEBUG_EXECUTE_r({ | |
2b8b4781 | 3117 | AV *const trie_words |
502c6561 | 3118 | = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]); |
2b8b4781 | 3119 | SV ** const tmp = av_fetch( trie_words, |
5c9f2f80 | 3120 | ST.accept_buff[ 0 ].wordnum-1, 0 ); |
de734bd5 A |
3121 | SV *sv= tmp ? sv_newmortal() : NULL; |
3122 | ||
166ba7cd DM |
3123 | PerlIO_printf( Perl_debug_log, |
3124 | "%*s %sonly one match left: #%d <%s>%s\n", | |
5bc10b2c | 3125 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3126 | ST.accept_buff[ 0 ].wordnum, |
de734bd5 A |
3127 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3128 | PL_colors[0], PL_colors[1], | |
3129 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3130 | ) | |
3131 | : "not compiled under -Dr", | |
166ba7cd DM |
3132 | PL_colors[5] ); |
3133 | }); | |
3134 | PL_reginput = (char *)ST.accept_buff[ 0 ].endpos; | |
3135 | /* in this case we free tmps/leave before we call regmatch | |
3136 | as we wont be using accept_buff again. */ | |
5d458dd8 | 3137 | |
166ba7cd DM |
3138 | locinput = PL_reginput; |
3139 | nextchr = UCHARAT(locinput); | |
5d458dd8 YO |
3140 | if ( !ST.jump || !ST.jump[ST.accept_buff[0].wordnum]) |
3141 | scan = ST.B; | |
3142 | else | |
3143 | scan = ST.me + ST.jump[ST.accept_buff[0].wordnum]; | |
3144 | if (!has_cutgroup) { | |
3145 | FREETMPS; | |
3146 | LEAVE; | |
3147 | } else { | |
3148 | ST.accepted--; | |
3149 | PUSH_YES_STATE_GOTO(TRIE_next, scan); | |
3150 | } | |
786e8c11 | 3151 | |
166ba7cd DM |
3152 | continue; /* execute rest of RE */ |
3153 | } | |
fae667d5 YO |
3154 | |
3155 | if ( !ST.accepted-- ) { | |
5d458dd8 YO |
3156 | DEBUG_EXECUTE_r({ |
3157 | PerlIO_printf( Perl_debug_log, | |
3158 | "%*s %sTRIE failed...%s\n", | |
3159 | REPORT_CODE_OFF+depth*2, "", | |
3160 | PL_colors[4], | |
3161 | PL_colors[5] ); | |
3162 | }); | |
166ba7cd DM |
3163 | FREETMPS; |
3164 | LEAVE; | |
5d458dd8 | 3165 | sayNO_SILENT; |
fae667d5 YO |
3166 | /*NOTREACHED*/ |
3167 | } | |
166ba7cd | 3168 | |
a3621e74 | 3169 | /* |
166ba7cd DM |
3170 | There are at least two accepting states left. Presumably |
3171 | the number of accepting states is going to be low, | |
3172 | typically two. So we simply scan through to find the one | |
3173 | with lowest wordnum. Once we find it, we swap the last | |
3174 | state into its place and decrement the size. We then try to | |
3175 | match the rest of the pattern at the point where the word | |
3176 | ends. If we succeed, control just continues along the | |
3177 | regex; if we fail we return here to try the next accepting | |
3178 | state | |
3179 | */ | |
a3621e74 | 3180 | |
166ba7cd DM |
3181 | { |
3182 | U32 best = 0; | |
3183 | U32 cur; | |
3184 | for( cur = 1 ; cur <= ST.accepted ; cur++ ) { | |
3185 | DEBUG_TRIE_EXECUTE_r( | |
f2278c82 | 3186 | PerlIO_printf( Perl_debug_log, |
166ba7cd | 3187 | "%*s %sgot %"IVdf" (%d) as best, looking at %"IVdf" (%d)%s\n", |
5bc10b2c | 3188 | REPORT_CODE_OFF + depth * 2, "", PL_colors[4], |
166ba7cd DM |
3189 | (IV)best, ST.accept_buff[ best ].wordnum, (IV)cur, |
3190 | ST.accept_buff[ cur ].wordnum, PL_colors[5] ); | |
3191 | ); | |
3192 | ||
3193 | if (ST.accept_buff[cur].wordnum < | |
3194 | ST.accept_buff[best].wordnum) | |
3195 | best = cur; | |
a3621e74 | 3196 | } |
166ba7cd DM |
3197 | |
3198 | DEBUG_EXECUTE_r({ | |
2b8b4781 | 3199 | AV *const trie_words |
502c6561 | 3200 | = MUTABLE_AV(rexi->data->data[ARG(ST.me)+TRIE_WORDS_OFFSET]); |
2b8b4781 | 3201 | SV ** const tmp = av_fetch( trie_words, |
5c9f2f80 | 3202 | ST.accept_buff[ best ].wordnum - 1, 0 ); |
7f69552c | 3203 | regnode *nextop=(!ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) ? |
265c4333 | 3204 | ST.B : |
7f69552c | 3205 | ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
de734bd5 A |
3206 | SV *sv= tmp ? sv_newmortal() : NULL; |
3207 | ||
265c4333 YO |
3208 | PerlIO_printf( Perl_debug_log, |
3209 | "%*s %strying alternation #%d <%s> at node #%d %s\n", | |
5bc10b2c | 3210 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], |
166ba7cd | 3211 | ST.accept_buff[best].wordnum, |
de734bd5 A |
3212 | tmp ? pv_pretty(sv, SvPV_nolen_const(*tmp), SvCUR(*tmp), 0, |
3213 | PL_colors[0], PL_colors[1], | |
3214 | (SvUTF8(*tmp) ? PERL_PV_ESCAPE_UNI : 0) | |
3215 | ) : "not compiled under -Dr", | |
265c4333 | 3216 | REG_NODE_NUM(nextop), |
166ba7cd DM |
3217 | PL_colors[5] ); |
3218 | }); | |
3219 | ||
3220 | if ( best<ST.accepted ) { | |
3221 | reg_trie_accepted tmp = ST.accept_buff[ best ]; | |
3222 | ST.accept_buff[ best ] = ST.accept_buff[ ST.accepted ]; | |
3223 | ST.accept_buff[ ST.accepted ] = tmp; | |
3224 | best = ST.accepted; | |
a3621e74 | 3225 | } |
166ba7cd | 3226 | PL_reginput = (char *)ST.accept_buff[ best ].endpos; |
7f69552c | 3227 | if ( !ST.jump || !ST.jump[ST.accept_buff[best].wordnum]) { |
5d458dd8 | 3228 | scan = ST.B; |
786e8c11 | 3229 | } else { |
5d458dd8 | 3230 | scan = ST.me + ST.jump[ST.accept_buff[best].wordnum]; |
5d458dd8 | 3231 | } |
6b173516 | 3232 | PUSH_YES_STATE_GOTO(TRIE_next, scan); |
786e8c11 | 3233 | /* NOTREACHED */ |
166ba7cd | 3234 | } |
166ba7cd | 3235 | /* NOTREACHED */ |
5d458dd8 | 3236 | case TRIE_next: |
6962fb1a YO |
3237 | /* we dont want to throw this away, see bug 57042*/ |
3238 | if (oreplsv != GvSV(PL_replgv)) | |
3239 | sv_setsv(oreplsv, GvSV(PL_replgv)); | |
5d458dd8 YO |
3240 | FREETMPS; |
3241 | LEAVE; | |
3242 | sayYES; | |
166ba7cd DM |
3243 | #undef ST |
3244 | ||
95b24440 DM |
3245 | case EXACT: { |
3246 | char *s = STRING(scan); | |
24d3c4a9 | 3247 | ln = STR_LEN(scan); |
eb160463 | 3248 | if (do_utf8 != UTF) { |
bc517b45 | 3249 | /* The target and the pattern have differing utf8ness. */ |
1aa99e6b | 3250 | char *l = locinput; |
24d3c4a9 | 3251 | const char * const e = s + ln; |
a72c7584 | 3252 | |
5ff6fc6d JH |
3253 | if (do_utf8) { |
3254 | /* The target is utf8, the pattern is not utf8. */ | |
1aa99e6b | 3255 | while (s < e) { |
a3b680e6 | 3256 | STRLEN ulen; |
1aa99e6b | 3257 | if (l >= PL_regeol) |
5ff6fc6d JH |
3258 | sayNO; |
3259 | if (NATIVE_TO_UNI(*(U8*)s) != | |
89ebb4a3 | 3260 | utf8n_to_uvuni((U8*)l, UTF8_MAXBYTES, &ulen, |
041457d9 | 3261 | uniflags)) |
5ff6fc6d | 3262 | sayNO; |
bc517b45 | 3263 | l += ulen; |
5ff6fc6d | 3264 | s ++; |
1aa99e6b | 3265 | } |
5ff6fc6d JH |
3266 | } |
3267 | else { | |
3268 | /* The target is not utf8, the pattern is utf8. */ | |
1aa99e6b | 3269 | while (s < e) { |
a3b680e6 | 3270 | STRLEN ulen; |
1aa99e6b IH |
3271 | if (l >= PL_regeol) |
3272 | sayNO; | |
5ff6fc6d | 3273 | if (NATIVE_TO_UNI(*((U8*)l)) != |
89ebb4a3 | 3274 | utf8n_to_uvuni((U8*)s, UTF8_MAXBYTES, &ulen, |
041457d9 | 3275 | uniflags)) |
1aa99e6b | 3276 | sayNO; |
bc517b45 | 3277 | s += ulen; |
a72c7584 | 3278 | l ++; |
1aa99e6b | 3279 | } |
5ff6fc6d | 3280 | } |
1aa99e6b IH |
3281 | locinput = l; |
3282 | nextchr = UCHARAT(locinput); | |
3283 | break; | |
3284 | } | |
bc517b45 | 3285 | /* The target and the pattern have the same utf8ness. */ |
d6a28714 JH |
3286 | /* Inline the first character, for speed. */ |
3287 | if (UCHARAT(s) != nextchr) | |
3288 | sayNO; | |
24d3c4a9 | 3289 | if (PL_regeol - locinput < ln) |
d6a28714 | 3290 | sayNO; |
24d3c4a9 | 3291 | if (ln > 1 && memNE(s, locinput, ln)) |
d6a28714 | 3292 | sayNO; |
24d3c4a9 | 3293 | locinput += ln; |
d6a28714 JH |
3294 | nextchr = UCHARAT(locinput); |
3295 | break; | |
95b24440 | 3296 | } |
d6a28714 | 3297 | case EXACTFL: |
b8c5462f JH |
3298 | PL_reg_flags |= RF_tainted; |
3299 | /* FALL THROUGH */ | |
95b24440 | 3300 | case EXACTF: { |
be8e71aa | 3301 | char * const s = STRING(scan); |
24d3c4a9 | 3302 | ln = STR_LEN(scan); |
d6a28714 | 3303 | |
d07ddd77 JH |
3304 | if (do_utf8 || UTF) { |
3305 | /* Either target or the pattern are utf8. */ | |
be8e71aa | 3306 | const char * const l = locinput; |
d07ddd77 | 3307 | char *e = PL_regeol; |
bc517b45 | 3308 | |
24d3c4a9 | 3309 | if (ibcmp_utf8(s, 0, ln, (bool)UTF, |
1feea2c7 | 3310 | l, &e, 0, do_utf8)) { |
5486206c JH |
3311 | /* One more case for the sharp s: |
3312 | * pack("U0U*", 0xDF) =~ /ss/i, | |
3313 | * the 0xC3 0x9F are the UTF-8 | |
3314 | * byte sequence for the U+00DF. */ | |
e1d1eefb | 3315 | |
5486206c | 3316 | if (!(do_utf8 && |
e1d1eefb | 3317 | toLOWER(s[0]) == 's' && |
24d3c4a9 | 3318 | ln >= 2 && |
5486206c JH |
3319 | toLOWER(s[1]) == 's' && |
3320 | (U8)l[0] == 0xC3 && | |
3321 | e - l >= 2 && | |
3322 | (U8)l[1] == 0x9F)) | |
3323 | sayNO; | |
3324 | } | |
d07ddd77 JH |
3325 | locinput = e; |
3326 | nextchr = UCHARAT(locinput); | |
3327 | break; | |
a0ed51b3 | 3328 | } |
d6a28714 | 3329 | |
bc517b45 JH |
3330 | /* Neither the target and the pattern are utf8. */ |
3331 | ||
d6a28714 JH |
3332 | /* Inline the first character, for speed. */ |
3333 | if (UCHARAT(s) != nextchr && | |
3334 | UCHARAT(s) != ((OP(scan) == EXACTF) | |
3335 | ? PL_fold : PL_fold_locale)[nextchr]) | |
a0ed51b3 | 3336 | sayNO; |
24d3c4a9 | 3337 | if (PL_regeol - locinput < ln) |
b8c5462f | 3338 | sayNO; |
24d3c4a9 DM |
3339 | if (ln > 1 && (OP(scan) == EXACTF |
3340 | ? ibcmp(s, locinput, ln) | |
3341 | : ibcmp_locale(s, locinput, ln))) | |
4633a7c4 | 3342 | sayNO; |
24d3c4a9 | 3343 | locinput += ln; |
d6a28714 | 3344 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3345 | break; |
95b24440 | 3346 | } |
d6a28714 | 3347 | case ANYOF: |
ffc61ed2 | 3348 | if (do_utf8) { |
9e55ce06 JH |
3349 | STRLEN inclasslen = PL_regeol - locinput; |
3350 | ||
32fc9b6a | 3351 | if (!reginclass(rex, scan, (U8*)locinput, &inclasslen, do_utf8)) |
262b90c4 | 3352 | goto anyof_fail; |
ffc61ed2 JH |
3353 | if (locinput >= PL_regeol) |
3354 | sayNO; | |
0f0076b4 | 3355 | locinput += inclasslen ? inclasslen : UTF8SKIP(locinput); |
b8c5462f | 3356 | nextchr = UCHARAT(locinput); |
e0f9d4a8 | 3357 | break; |
ffc61ed2 JH |
3358 | } |
3359 | else { | |
3360 | if (nextchr < 0) | |
3361 | nextchr = UCHARAT(locinput); | |
32fc9b6a | 3362 | if (!REGINCLASS(rex, scan, (U8*)locinput)) |
262b90c4 | 3363 | goto anyof_fail; |
ffc61ed2 JH |
3364 | if (!nextchr && locinput >= PL_regeol) |
3365 | sayNO; | |
3366 | nextchr = UCHARAT(++locinput); | |
e0f9d4a8 JH |
3367 | break; |
3368 | } | |
262b90c4 | 3369 | anyof_fail: |
e0f9d4a8 JH |
3370 | /* If we might have the case of the German sharp s |
3371 | * in a casefolding Unicode character class. */ | |
3372 | ||
ebc501f0 JH |
3373 | if (ANYOF_FOLD_SHARP_S(scan, locinput, PL_regeol)) { |
3374 | locinput += SHARP_S_SKIP; | |
e0f9d4a8 | 3375 | nextchr = UCHARAT(locinput); |
ffc61ed2 | 3376 | } |
e0f9d4a8 JH |
3377 | else |
3378 | sayNO; | |
b8c5462f | 3379 | break; |
d6a28714 | 3380 | case ALNUML: |
b8c5462f JH |
3381 | PL_reg_flags |= RF_tainted; |
3382 | /* FALL THROUGH */ | |
d6a28714 | 3383 | case ALNUM: |
b8c5462f | 3384 | if (!nextchr) |
4633a7c4 | 3385 | sayNO; |
ffc61ed2 | 3386 | if (do_utf8) { |
1a4fad37 | 3387 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3388 | if (!(OP(scan) == ALNUM |
bb7a0f54 | 3389 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 | 3390 | : isALNUM_LC_utf8((U8*)locinput))) |
b8c5462f JH |
3391 | { |
3392 | sayNO; | |
a0ed51b3 | 3393 | } |
b8c5462f | 3394 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3395 | nextchr = UCHARAT(locinput); |
3396 | break; | |
3397 | } | |
ffc61ed2 | 3398 | if (!(OP(scan) == ALNUM |
d6a28714 | 3399 | ? isALNUM(nextchr) : isALNUM_LC(nextchr))) |
bbce6d69 | 3400 | sayNO; |
b8c5462f | 3401 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3402 | break; |
d6a28714 | 3403 | case NALNUML: |
b8c5462f JH |
3404 | PL_reg_flags |= RF_tainted; |
3405 | /* FALL THROUGH */ | |
d6a28714 JH |
3406 | case NALNUM: |
3407 | if (!nextchr && locinput >= PL_regeol) | |
a0ed51b3 | 3408 | sayNO; |
ffc61ed2 | 3409 | if (do_utf8) { |
1a4fad37 | 3410 | LOAD_UTF8_CHARCLASS_ALNUM(); |
ffc61ed2 | 3411 | if (OP(scan) == NALNUM |
bb7a0f54 | 3412 | ? (bool)swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8) |
d6a28714 JH |
3413 | : isALNUM_LC_utf8((U8*)locinput)) |
3414 | { | |
b8c5462f | 3415 | sayNO; |
d6a28714 | 3416 | } |
b8c5462f JH |
3417 | locinput += PL_utf8skip[nextchr]; |
3418 | nextchr = UCHARAT(locinput); | |
3419 | break; | |
3420 | } | |
ffc61ed2 | 3421 | if (OP(scan) == NALNUM |
d6a28714 | 3422 | ? isALNUM(nextchr) : isALNUM_LC(nextchr)) |
b8c5462f | 3423 | sayNO; |
76e3520e | 3424 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3425 | break; |
d6a28714 JH |
3426 | case BOUNDL: |
3427 | case NBOUNDL: | |
3280af22 | 3428 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3429 | /* FALL THROUGH */ |
d6a28714 JH |
3430 | case BOUND: |
3431 | case NBOUND: | |
3432 | /* was last char in word? */ | |
ffc61ed2 | 3433 | if (do_utf8) { |
12d33761 | 3434 | if (locinput == PL_bostr) |
24d3c4a9 | 3435 | ln = '\n'; |
ffc61ed2 | 3436 | else { |
a3b680e6 | 3437 | const U8 * const r = reghop3((U8*)locinput, -1, (U8*)PL_bostr); |
9041c2e3 | 3438 | |
24d3c4a9 | 3439 | ln = utf8n_to_uvchr(r, UTF8SKIP(r), 0, uniflags); |
ffc61ed2 JH |
3440 | } |
3441 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { | |
24d3c4a9 | 3442 | ln = isALNUM_uni(ln); |
1a4fad37 | 3443 | LOAD_UTF8_CHARCLASS_ALNUM(); |
3568d838 | 3444 | n = swash_fetch(PL_utf8_alnum, (U8*)locinput, do_utf8); |
ffc61ed2 JH |
3445 | } |
3446 | else { | |
24d3c4a9 | 3447 | ln = isALNUM_LC_uvchr(UNI_TO_NATIVE(ln)); |
ffc61ed2 JH |
3448 | n = isALNUM_LC_utf8((U8*)locinput); |
3449 | } | |
a0ed51b3 | 3450 | } |
d6a28714 | 3451 | else { |
24d3c4a9 | 3452 | ln = (locinput != PL_bostr) ? |
12d33761 | 3453 | UCHARAT(locinput - 1) : '\n'; |
ffc61ed2 | 3454 | if (OP(scan) == BOUND || OP(scan) == NBOUND) { |
24d3c4a9 | 3455 | ln = isALNUM(ln); |
ffc61ed2 JH |
3456 | n = isALNUM(nextchr); |
3457 | } | |
3458 | else { | |
24d3c4a9 | 3459 | ln = isALNUM_LC(ln); |
ffc61ed2 JH |
3460 | n = isALNUM_LC(nextchr); |
3461 | } | |
d6a28714 | 3462 | } |
24d3c4a9 | 3463 | if (((!ln) == (!n)) == (OP(scan) == BOUND || |
ffc61ed2 JH |
3464 | OP(scan) == BOUNDL)) |
3465 | sayNO; | |
a0ed51b3 | 3466 | break; |
d6a28714 | 3467 | case SPACEL: |
3280af22 | 3468 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3469 | /* FALL THROUGH */ |
d6a28714 | 3470 | case SPACE: |
9442cb0e | 3471 | if (!nextchr) |
4633a7c4 | 3472 | sayNO; |
1aa99e6b | 3473 | if (do_utf8) { |
fd400ab9 | 3474 | if (UTF8_IS_CONTINUED(nextchr)) { |
1a4fad37 | 3475 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3476 | if (!(OP(scan) == SPACE |
bb7a0f54 | 3477 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
ffc61ed2 JH |
3478 | : isSPACE_LC_utf8((U8*)locinput))) |
3479 | { | |
3480 | sayNO; | |
3481 | } | |
3482 | locinput += PL_utf8skip[nextchr]; | |
3483 | nextchr = UCHARAT(locinput); | |
3484 | break; | |
d6a28714 | 3485 | } |
ffc61ed2 JH |
3486 | if (!(OP(scan) == SPACE |
3487 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3488 | sayNO; | |
3489 | nextchr = UCHARAT(++locinput); | |
3490 | } | |
3491 | else { | |
3492 | if (!(OP(scan) == SPACE | |
3493 | ? isSPACE(nextchr) : isSPACE_LC(nextchr))) | |
3494 | sayNO; | |
3495 | nextchr = UCHARAT(++locinput); | |
a0ed51b3 | 3496 | } |
a0ed51b3 | 3497 | break; |
d6a28714 | 3498 | case NSPACEL: |
3280af22 | 3499 | PL_reg_flags |= RF_tainted; |
bbce6d69 | 3500 | /* FALL THROUGH */ |
d6a28714 | 3501 | case NSPACE: |
9442cb0e | 3502 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3503 | sayNO; |
1aa99e6b | 3504 | if (do_utf8) { |
1a4fad37 | 3505 | LOAD_UTF8_CHARCLASS_SPACE(); |
ffc61ed2 | 3506 | if (OP(scan) == NSPACE |
bb7a0f54 | 3507 | ? (bool)swash_fetch(PL_utf8_space, (U8*)locinput, do_utf8) |
d6a28714 | 3508 | : isSPACE_LC_utf8((U8*)locinput)) |
b8c5462f JH |
3509 | { |
3510 | sayNO; | |
3511 | } | |
3512 | locinput += PL_utf8skip[nextchr]; | |
3513 | nextchr = UCHARAT(locinput); | |
3514 | break; | |
a0ed51b3 | 3515 | } |
ffc61ed2 | 3516 | if (OP(scan) == NSPACE |
d6a28714 | 3517 | ? isSPACE(nextchr) : isSPACE_LC(nextchr)) |
4633a7c4 | 3518 | sayNO; |
76e3520e | 3519 | nextchr = UCHARAT(++locinput); |
a0d0e21e | 3520 | break; |
d6a28714 | 3521 | case DIGITL: |
a0ed51b3 LW |
3522 | PL_reg_flags |= RF_tainted; |
3523 | /* FALL THROUGH */ | |
d6a28714 | 3524 | case DIGIT: |
9442cb0e | 3525 | if (!nextchr) |
a0ed51b3 | 3526 | sayNO; |
1aa99e6b | 3527 | if (do_utf8) { |
1a4fad37 | 3528 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3529 | if (!(OP(scan) == DIGIT |
bb7a0f54 | 3530 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e | 3531 | : isDIGIT_LC_utf8((U8*)locinput))) |
dfe13c55 | 3532 | { |
a0ed51b3 | 3533 | sayNO; |
dfe13c55 | 3534 | } |
6f06b55f | 3535 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3536 | nextchr = UCHARAT(locinput); |
3537 | break; | |
3538 | } | |
ffc61ed2 | 3539 | if (!(OP(scan) == DIGIT |
9442cb0e | 3540 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr))) |
a0ed51b3 LW |
3541 | sayNO; |
3542 | nextchr = UCHARAT(++locinput); | |
3543 | break; | |
d6a28714 | 3544 | case NDIGITL: |
b8c5462f JH |
3545 | PL_reg_flags |= RF_tainted; |
3546 | /* FALL THROUGH */ | |
d6a28714 | 3547 | case NDIGIT: |
9442cb0e | 3548 | if (!nextchr && locinput >= PL_regeol) |
b8c5462f | 3549 | sayNO; |
1aa99e6b | 3550 | if (do_utf8) { |
1a4fad37 | 3551 | LOAD_UTF8_CHARCLASS_DIGIT(); |
ffc61ed2 | 3552 | if (OP(scan) == NDIGIT |
bb7a0f54 | 3553 | ? (bool)swash_fetch(PL_utf8_digit, (U8*)locinput, do_utf8) |
9442cb0e GS |
3554 | : isDIGIT_LC_utf8((U8*)locinput)) |
3555 | { | |
a0ed51b3 | 3556 | sayNO; |
9442cb0e | 3557 | } |
6f06b55f | 3558 | locinput += PL_utf8skip[nextchr]; |
a0ed51b3 LW |
3559 | nextchr = UCHARAT(locinput); |
3560 | break; | |
3561 | } | |
ffc61ed2 | 3562 | if (OP(scan) == NDIGIT |
9442cb0e | 3563 | ? isDIGIT(nextchr) : isDIGIT_LC(nextchr)) |
a0ed51b3 LW |
3564 | sayNO; |
3565 | nextchr = UCHARAT(++locinput); | |
3566 | break; | |
3567 | case CLUMP: | |
b7c83a7e | 3568 | if (locinput >= PL_regeol) |
a0ed51b3 | 3569 | sayNO; |
b7c83a7e | 3570 | if (do_utf8) { |
1a4fad37 | 3571 | LOAD_UTF8_CHARCLASS_MARK(); |
b7c83a7e JH |
3572 | if (swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) |
3573 | sayNO; | |
3574 | locinput += PL_utf8skip[nextchr]; | |
3575 | while (locinput < PL_regeol && | |
3576 | swash_fetch(PL_utf8_mark,(U8*)locinput, do_utf8)) | |
3577 | locinput += UTF8SKIP(locinput); | |
3578 | if (locinput > PL_regeol) | |
3579 | sayNO; | |
eb08e2da JH |
3580 | } |
3581 | else | |
3582 | locinput++; | |
a0ed51b3 LW |
3583 | nextchr = UCHARAT(locinput); |
3584 | break; | |
81714fb9 YO |
3585 | |
3586 | case NREFFL: | |
3587 | { | |
3588 | char *s; | |
ff1157ca | 3589 | char type; |
81714fb9 YO |
3590 | PL_reg_flags |= RF_tainted; |
3591 | /* FALL THROUGH */ | |
3592 | case NREF: | |
3593 | case NREFF: | |
ff1157ca | 3594 | type = OP(scan); |
0a4db386 YO |
3595 | n = reg_check_named_buff_matched(rex,scan); |
3596 | ||
3597 | if ( n ) { | |
3598 | type = REF + ( type - NREF ); | |
3599 | goto do_ref; | |
3600 | } else { | |
81714fb9 | 3601 | sayNO; |
0a4db386 YO |
3602 | } |
3603 | /* unreached */ | |
c8756f30 | 3604 | case REFFL: |
3280af22 | 3605 | PL_reg_flags |= RF_tainted; |
c8756f30 | 3606 | /* FALL THROUGH */ |
c277df42 | 3607 | case REF: |
81714fb9 | 3608 | case REFF: |
c277df42 | 3609 | n = ARG(scan); /* which paren pair */ |
81714fb9 YO |
3610 | type = OP(scan); |
3611 | do_ref: | |
f0ab9afb | 3612 | ln = PL_regoffs[n].start; |
2c2d71f5 | 3613 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
3b6647e0 | 3614 | if (*PL_reglastparen < n || ln == -1) |
af3f8c16 | 3615 | sayNO; /* Do not match unless seen CLOSEn. */ |
f0ab9afb | 3616 | if (ln == PL_regoffs[n].end) |
a0d0e21e | 3617 | break; |
a0ed51b3 | 3618 | |
24d3c4a9 | 3619 | s = PL_bostr + ln; |
81714fb9 | 3620 | if (do_utf8 && type != REF) { /* REF can do byte comparison */ |
a0ed51b3 | 3621 | char *l = locinput; |
f0ab9afb | 3622 | const char *e = PL_bostr + PL_regoffs[n].end; |
a0ed51b3 LW |
3623 | /* |
3624 | * Note that we can't do the "other character" lookup trick as | |
3625 | * in the 8-bit case (no pun intended) because in Unicode we | |
3626 | * have to map both upper and title case to lower case. | |
3627 | */ | |
81714fb9 | 3628 | if (type == REFF) { |
a0ed51b3 | 3629 | while (s < e) { |
a3b680e6 AL |
3630 | STRLEN ulen1, ulen2; |
3631 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; | |
3632 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
3633 | ||
a0ed51b3 LW |
3634 | if (l >= PL_regeol) |
3635 | sayNO; | |
a2a2844f JH |
3636 | toLOWER_utf8((U8*)s, tmpbuf1, &ulen1); |
3637 | toLOWER_utf8((U8*)l, tmpbuf2, &ulen2); | |
7114a2d2 | 3638 | if (ulen1 != ulen2 || memNE((char *)tmpbuf1, (char *)tmpbuf2, ulen1)) |
a0ed51b3 | 3639 | sayNO; |
a2a2844f JH |
3640 | s += ulen1; |
3641 | l += ulen2; | |
a0ed51b3 LW |
3642 | } |
3643 | } | |
3644 | locinput = l; | |
3645 | nextchr = UCHARAT(locinput); | |
3646 | break; | |
3647 | } | |
3648 | ||
a0d0e21e | 3649 | /* Inline the first character, for speed. */ |
76e3520e | 3650 | if (UCHARAT(s) != nextchr && |
81714fb9 YO |
3651 | (type == REF || |
3652 | (UCHARAT(s) != (type == REFF | |
3653 | ? PL_fold : PL_fold_locale)[nextchr]))) | |
4633a7c4 | 3654 | sayNO; |
f0ab9afb | 3655 | ln = PL_regoffs[n].end - ln; |
24d3c4a9 | 3656 | if (locinput + ln > PL_regeol) |
4633a7c4 | 3657 | sayNO; |
81714fb9 | 3658 | if (ln > 1 && (type == REF |
24d3c4a9 | 3659 | ? memNE(s, locinput, ln) |
81714fb9 | 3660 | : (type == REFF |
24d3c4a9 DM |
3661 | ? ibcmp(s, locinput, ln) |
3662 | : ibcmp_locale(s, locinput, ln)))) | |
4633a7c4 | 3663 | sayNO; |
24d3c4a9 | 3664 | locinput += ln; |
76e3520e | 3665 | nextchr = UCHARAT(locinput); |
a0d0e21e | 3666 | break; |
81714fb9 | 3667 | } |
a0d0e21e | 3668 | case NOTHING: |
c277df42 | 3669 | case TAIL: |
a0d0e21e LW |
3670 | break; |
3671 | case BACK: | |
3672 | break; | |
40a82448 DM |
3673 | |
3674 | #undef ST | |
3675 | #define ST st->u.eval | |
c277df42 | 3676 | { |
c277df42 | 3677 | SV *ret; |
d2f13c59 | 3678 | REGEXP *re_sv; |
6bda09f9 | 3679 | regexp *re; |
f8fc2ecf | 3680 | regexp_internal *rei; |
1a147d38 YO |
3681 | regnode *startpoint; |
3682 | ||
3683 | case GOSTART: | |
e7707071 YO |
3684 | case GOSUB: /* /(...(?1))/ /(...(?&foo))/ */ |
3685 | if (cur_eval && cur_eval->locinput==locinput) { | |
24b23f37 | 3686 | if (cur_eval->u.eval.close_paren == (U32)ARG(scan)) |
1a147d38 | 3687 | Perl_croak(aTHX_ "Infinite recursion in regex"); |
4b196cd4 | 3688 | if ( ++nochange_depth > max_nochange_depth ) |
1a147d38 YO |
3689 | Perl_croak(aTHX_ |
3690 | "Pattern subroutine nesting without pos change" | |
3691 | " exceeded limit in regex"); | |
6bda09f9 YO |
3692 | } else { |
3693 | nochange_depth = 0; | |
1a147d38 | 3694 | } |
288b8c02 | 3695 | re_sv = rex_sv; |
6bda09f9 | 3696 | re = rex; |
f8fc2ecf | 3697 | rei = rexi; |
288b8c02 | 3698 | (void)ReREFCNT_inc(rex_sv); |
1a147d38 | 3699 | if (OP(scan)==GOSUB) { |
6bda09f9 YO |
3700 | startpoint = scan + ARG2L(scan); |
3701 | ST.close_paren = ARG(scan); | |
3702 | } else { | |
f8fc2ecf | 3703 | startpoint = rei->program+1; |
6bda09f9 YO |
3704 | ST.close_paren = 0; |
3705 | } | |
3706 | goto eval_recurse_doit; | |
3707 | /* NOTREACHED */ | |
3708 | case EVAL: /* /(?{A})B/ /(??{A})B/ and /(?(?{A})X|Y)B/ */ | |
3709 | if (cur_eval && cur_eval->locinput==locinput) { | |
4b196cd4 | 3710 | if ( ++nochange_depth > max_nochange_depth ) |
1a147d38 | 3711 | Perl_croak(aTHX_ "EVAL without pos change exceeded limit in regex"); |
6bda09f9 YO |
3712 | } else { |
3713 | nochange_depth = 0; | |
3714 | } | |
8e5e9ebe | 3715 | { |
4aabdb9b DM |
3716 | /* execute the code in the {...} */ |
3717 | dSP; | |
6136c704 | 3718 | SV ** const before = SP; |
4aabdb9b DM |
3719 | OP_4tree * const oop = PL_op; |
3720 | COP * const ocurcop = PL_curcop; | |
3721 | PAD *old_comppad; | |
d80618d2 | 3722 | char *saved_regeol = PL_regeol; |
4aabdb9b DM |
3723 | |
3724 | n = ARG(scan); | |
f8fc2ecf | 3725 | PL_op = (OP_4tree*)rexi->data->data[n]; |
24b23f37 YO |
3726 | DEBUG_STATE_r( PerlIO_printf(Perl_debug_log, |
3727 | " re_eval 0x%"UVxf"\n", PTR2UV(PL_op)) ); | |
f8fc2ecf | 3728 | PAD_SAVE_LOCAL(old_comppad, (PAD*)rexi->data->data[n + 2]); |
f0ab9afb | 3729 | PL_regoffs[0].end = PL_reg_magic->mg_len = locinput - PL_bostr; |
4aabdb9b | 3730 | |
2bf803e2 YO |
3731 | if (sv_yes_mark) { |
3732 | SV *sv_mrk = get_sv("REGMARK", 1); | |
3733 | sv_setsv(sv_mrk, sv_yes_mark); | |
3734 | } | |
3735 | ||
8e5e9ebe RGS |
3736 | CALLRUNOPS(aTHX); /* Scalar context. */ |
3737 | SPAGAIN; | |
3738 | if (SP == before) | |
075aa684 | 3739 | ret = &PL_sv_undef; /* protect against empty (?{}) blocks. */ |
8e5e9ebe RGS |
3740 | else { |
3741 | ret = POPs; | |
3742 | PUTBACK; | |
3743 | } | |
4aabdb9b DM |
3744 | |
3745 | PL_op = oop; | |
3746 | PAD_RESTORE_LOCAL(old_comppad); | |
3747 | PL_curcop = ocurcop; | |
d80618d2 | 3748 | PL_regeol = saved_regeol; |
24d3c4a9 | 3749 | if (!logical) { |
4aabdb9b DM |
3750 | /* /(?{...})/ */ |
3751 | sv_setsv(save_scalar(PL_replgv), ret); | |
4aabdb9b DM |
3752 | break; |
3753 | } | |
8e5e9ebe | 3754 | } |
24d3c4a9 DM |
3755 | if (logical == 2) { /* Postponed subexpression: /(??{...})/ */ |
3756 | logical = 0; | |
4aabdb9b | 3757 | { |
4f639d21 DM |
3758 | /* extract RE object from returned value; compiling if |
3759 | * necessary */ | |
6136c704 | 3760 | MAGIC *mg = NULL; |
288b8c02 | 3761 | REGEXP *rx = NULL; |
5c35adbb NC |
3762 | |
3763 | if (SvROK(ret)) { | |
288b8c02 | 3764 | SV *const sv = SvRV(ret); |
5c35adbb NC |
3765 | |
3766 | if (SvTYPE(sv) == SVt_REGEXP) { | |
d2f13c59 | 3767 | rx = (REGEXP*) sv; |
5c35adbb NC |
3768 | } else if (SvSMAGICAL(sv)) { |
3769 | mg = mg_find(sv, PERL_MAGIC_qr); | |
3770 | assert(mg); | |
3771 | } | |
3772 | } else if (SvTYPE(ret) == SVt_REGEXP) { | |
d2f13c59 | 3773 | rx = (REGEXP*) ret; |
5c35adbb | 3774 | } else if (SvSMAGICAL(ret)) { |
124ee91a NC |
3775 | if (SvGMAGICAL(ret)) { |
3776 | /* I don't believe that there is ever qr magic | |
3777 | here. */ | |
3778 | assert(!mg_find(ret, PERL_MAGIC_qr)); | |
faf82a0b | 3779 | sv_unmagic(ret, PERL_MAGIC_qr); |
124ee91a NC |
3780 | } |
3781 | else { | |
faf82a0b | 3782 | mg = mg_find(ret, PERL_MAGIC_qr); |
124ee91a NC |
3783 | /* testing suggests mg only ends up non-NULL for |
3784 | scalars who were upgraded and compiled in the | |
3785 | else block below. In turn, this is only | |
3786 | triggered in the "postponed utf8 string" tests | |
3787 | in t/op/pat.t */ | |
3788 | } | |
0f5d15d6 | 3789 | } |
faf82a0b | 3790 | |
0f5d15d6 | 3791 | if (mg) { |
d2f13c59 | 3792 | rx = (REGEXP *) mg->mg_obj; /*XXX:dmq*/ |
e2560c33 | 3793 | assert(rx); |
0f5d15d6 | 3794 | } |
288b8c02 NC |
3795 | if (rx) { |
3796 | rx = reg_temp_copy(rx); | |
3797 | } | |
0f5d15d6 | 3798 | else { |
c737faaf | 3799 | U32 pm_flags = 0; |
a3b680e6 | 3800 | const I32 osize = PL_regsize; |
0f5d15d6 | 3801 | |
b9ad30b4 NC |
3802 | if (DO_UTF8(ret)) { |
3803 | assert (SvUTF8(ret)); | |
3804 | } else if (SvUTF8(ret)) { | |
3805 | /* Not doing UTF-8, despite what the SV says. Is | |
3806 | this only if we're trapped in use 'bytes'? */ | |
3807 | /* Make a copy of the octet sequence, but without | |
3808 | the flag on, as the compiler now honours the | |
3809 | SvUTF8 flag on ret. */ | |
3810 | STRLEN len; | |
3811 | const char *const p = SvPV(ret, len); | |
3812 | ret = newSVpvn_flags(p, len, SVs_TEMP); | |
3813 | } | |
288b8c02 | 3814 | rx = CALLREGCOMP(ret, pm_flags); |
9041c2e3 | 3815 | if (!(SvFLAGS(ret) |
faf82a0b | 3816 | & (SVs_TEMP | SVs_PADTMP | SVf_READONLY |
3ce3ed55 | 3817 | | SVs_GMG))) { |
a2794585 NC |
3818 | /* This isn't a first class regexp. Instead, it's |
3819 | caching a regexp onto an existing, Perl visible | |
3820 | scalar. */ | |
ad64d0ec | 3821 | sv_magic(ret, MUTABLE_SV(rx), PERL_MAGIC_qr, 0, 0); |
3ce3ed55 | 3822 | } |
0f5d15d6 | 3823 | PL_regsize = osize; |
0f5d15d6 | 3824 | } |
288b8c02 NC |
3825 | re_sv = rx; |
3826 | re = (struct regexp *)SvANY(rx); | |
4aabdb9b | 3827 | } |
07bc277f | 3828 | RXp_MATCH_COPIED_off(re); |
28d8d7f4 YO |
3829 | re->subbeg = rex->subbeg; |
3830 | re->sublen = rex->sublen; | |
f8fc2ecf | 3831 | rei = RXi_GET(re); |
6bda09f9 | 3832 | DEBUG_EXECUTE_r( |
efd26800 | 3833 | debug_start_match(re_sv, do_utf8, locinput, PL_regeol, |
6bda09f9 YO |
3834 | "Matching embedded"); |
3835 | ); | |
f8fc2ecf | 3836 | startpoint = rei->program + 1; |
1a147d38 | 3837 | ST.close_paren = 0; /* only used for GOSUB */ |
6bda09f9 YO |
3838 | /* borrowed from regtry */ |
3839 | if (PL_reg_start_tmpl <= re->nparens) { | |
3840 | PL_reg_start_tmpl = re->nparens*3/2 + 3; | |
3841 | if(PL_reg_start_tmp) | |
3842 | Renew(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
3843 | else | |
3844 | Newx(PL_reg_start_tmp, PL_reg_start_tmpl, char*); | |
dd5def09 | 3845 | } |
aa283a38 | 3846 | |
1a147d38 | 3847 | eval_recurse_doit: /* Share code with GOSUB below this line */ |
aa283a38 | 3848 | /* run the pattern returned from (??{...}) */ |
40a82448 DM |
3849 | ST.cp = regcppush(0); /* Save *all* the positions. */ |
3850 | REGCP_SET(ST.lastcp); | |
6bda09f9 | 3851 | |
f0ab9afb | 3852 | PL_regoffs = re->offs; /* essentially NOOP on GOSUB */ |
6bda09f9 | 3853 | |
0357f1fd ML |
3854 | /* see regtry, specifically PL_reglast(?:close)?paren is a pointer! (i dont know why) :dmq */ |
3855 | PL_reglastparen = &re->lastparen; | |
3856 | PL_reglastcloseparen = &re->lastcloseparen; | |
3857 | re->lastparen = 0; | |
3858 | re->lastcloseparen = 0; | |
3859 | ||
4aabdb9b | 3860 | PL_reginput = locinput; |
ae0beba1 | 3861 | PL_regsize = 0; |
4aabdb9b DM |
3862 | |
3863 | /* XXXX This is too dramatic a measure... */ | |
3864 | PL_reg_maxiter = 0; | |
3865 | ||
faec1544 | 3866 | ST.toggle_reg_flags = PL_reg_flags; |
3c8556c3 | 3867 | if (RX_UTF8(re_sv)) |
faec1544 DM |
3868 | PL_reg_flags |= RF_utf8; |
3869 | else | |
3870 | PL_reg_flags &= ~RF_utf8; | |
3871 | ST.toggle_reg_flags ^= PL_reg_flags; /* diff of old and new */ | |
3872 | ||
288b8c02 | 3873 | ST.prev_rex = rex_sv; |
faec1544 | 3874 | ST.prev_curlyx = cur_curlyx; |
288b8c02 NC |
3875 | SETREX(rex_sv,re_sv); |
3876 | rex = re; | |
f8fc2ecf | 3877 | rexi = rei; |
faec1544 | 3878 | cur_curlyx = NULL; |
40a82448 | 3879 | ST.B = next; |
faec1544 DM |
3880 | ST.prev_eval = cur_eval; |
3881 | cur_eval = st; | |
faec1544 | 3882 | /* now continue from first node in postoned RE */ |
6bda09f9 | 3883 | PUSH_YES_STATE_GOTO(EVAL_AB, startpoint); |
4aabdb9b | 3884 | /* NOTREACHED */ |
a0ed51b3 | 3885 | } |
24d3c4a9 DM |
3886 | /* logical is 1, /(?(?{...})X|Y)/ */ |
3887 | sw = (bool)SvTRUE(ret); | |
3888 | logical = 0; | |
c277df42 IZ |
3889 | break; |
3890 | } | |
40a82448 | 3891 | |
faec1544 DM |
3892 | case EVAL_AB: /* cleanup after a successful (??{A})B */ |
3893 | /* note: this is called twice; first after popping B, then A */ | |
3894 | PL_reg_flags ^= ST.toggle_reg_flags; | |
288b8c02 NC |
3895 | ReREFCNT_dec(rex_sv); |
3896 | SETREX(rex_sv,ST.prev_rex); | |
3897 | rex = (struct regexp *)SvANY(rex_sv); | |
f8fc2ecf | 3898 | rexi = RXi_GET(rex); |
faec1544 DM |
3899 | regcpblow(ST.cp); |
3900 | cur_eval = ST.prev_eval; | |
3901 | cur_curlyx = ST.prev_curlyx; | |
34a81e2b B |
3902 | |
3903 | /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ | |
0357f1fd ML |
3904 | PL_reglastparen = &rex->lastparen; |
3905 | PL_reglastcloseparen = &rex->lastcloseparen; | |
34a81e2b B |
3906 | /* also update PL_regoffs */ |
3907 | PL_regoffs = rex->offs; | |
0357f1fd | 3908 | |
40a82448 DM |
3909 | /* XXXX This is too dramatic a measure... */ |
3910 | PL_reg_maxiter = 0; | |
e7707071 | 3911 | if ( nochange_depth ) |
4b196cd4 | 3912 | nochange_depth--; |
262b90c4 | 3913 | sayYES; |
40a82448 | 3914 | |
40a82448 | 3915 | |
faec1544 DM |
3916 | case EVAL_AB_fail: /* unsuccessfully ran A or B in (??{A})B */ |
3917 | /* note: this is called twice; first after popping B, then A */ | |
3918 | PL_reg_flags ^= ST.toggle_reg_flags; | |
288b8c02 NC |
3919 | ReREFCNT_dec(rex_sv); |
3920 | SETREX(rex_sv,ST.prev_rex); | |
3921 | rex = (struct regexp *)SvANY(rex_sv); | |
f8fc2ecf | 3922 | rexi = RXi_GET(rex); |
34a81e2b | 3923 | /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ |
0357f1fd ML |
3924 | PL_reglastparen = &rex->lastparen; |
3925 | PL_reglastcloseparen = &rex->lastcloseparen; | |
3926 | ||
40a82448 DM |
3927 | PL_reginput = locinput; |
3928 | REGCP_UNWIND(ST.lastcp); | |
3929 | regcppop(rex); | |
faec1544 DM |
3930 | cur_eval = ST.prev_eval; |
3931 | cur_curlyx = ST.prev_curlyx; | |
3932 | /* XXXX This is too dramatic a measure... */ | |
3933 | PL_reg_maxiter = 0; | |
e7707071 | 3934 | if ( nochange_depth ) |
4b196cd4 | 3935 | nochange_depth--; |
40a82448 | 3936 | sayNO_SILENT; |
40a82448 DM |
3937 | #undef ST |
3938 | ||
a0d0e21e | 3939 | case OPEN: |
c277df42 | 3940 | n = ARG(scan); /* which paren pair */ |
3280af22 NIS |
3941 | PL_reg_start_tmp[n] = locinput; |
3942 | if (n > PL_regsize) | |
3943 | PL_regsize = n; | |
e2e6a0f1 | 3944 | lastopen = n; |
a0d0e21e LW |
3945 | break; |
3946 | case CLOSE: | |
c277df42 | 3947 | n = ARG(scan); /* which paren pair */ |
f0ab9afb NC |
3948 | PL_regoffs[n].start = PL_reg_start_tmp[n] - PL_bostr; |
3949 | PL_regoffs[n].end = locinput - PL_bostr; | |
7f69552c YO |
3950 | /*if (n > PL_regsize) |
3951 | PL_regsize = n;*/ | |
3b6647e0 | 3952 | if (n > *PL_reglastparen) |
3280af22 | 3953 | *PL_reglastparen = n; |
a01268b5 | 3954 | *PL_reglastcloseparen = n; |
3b6647e0 | 3955 | if (cur_eval && cur_eval->u.eval.close_paren == n) { |
6bda09f9 YO |
3956 | goto fake_end; |
3957 | } | |
a0d0e21e | 3958 | break; |
e2e6a0f1 YO |
3959 | case ACCEPT: |
3960 | if (ARG(scan)){ | |
3961 | regnode *cursor; | |
3962 | for (cursor=scan; | |
3963 | cursor && OP(cursor)!=END; | |
3964 | cursor=regnext(cursor)) | |
3965 | { | |
3966 | if ( OP(cursor)==CLOSE ){ | |
3967 | n = ARG(cursor); | |
3968 | if ( n <= lastopen ) { | |
f0ab9afb NC |
3969 | PL_regoffs[n].start |
3970 | = PL_reg_start_tmp[n] - PL_bostr; | |
3971 | PL_regoffs[n].end = locinput - PL_bostr; | |
e2e6a0f1 YO |
3972 | /*if (n > PL_regsize) |
3973 | PL_regsize = n;*/ | |
3b6647e0 | 3974 | if (n > *PL_reglastparen) |
e2e6a0f1 YO |
3975 | *PL_reglastparen = n; |
3976 | *PL_reglastcloseparen = n; | |
3b6647e0 RB |
3977 | if ( n == ARG(scan) || (cur_eval && |
3978 | cur_eval->u.eval.close_paren == n)) | |
e2e6a0f1 YO |
3979 | break; |
3980 | } | |
3981 | } | |
3982 | } | |
3983 | } | |
3984 | goto fake_end; | |
3985 | /*NOTREACHED*/ | |
c277df42 IZ |
3986 | case GROUPP: |
3987 | n = ARG(scan); /* which paren pair */ | |
f0ab9afb | 3988 | sw = (bool)(*PL_reglastparen >= n && PL_regoffs[n].end != -1); |
c277df42 | 3989 | break; |
0a4db386 YO |
3990 | case NGROUPP: |
3991 | /* reg_check_named_buff_matched returns 0 for no match */ | |
3992 | sw = (bool)(0 < reg_check_named_buff_matched(rex,scan)); | |
3993 | break; | |
1a147d38 | 3994 | case INSUBP: |
0a4db386 | 3995 | n = ARG(scan); |
3b6647e0 | 3996 | sw = (cur_eval && (!n || cur_eval->u.eval.close_paren == n)); |
0a4db386 YO |
3997 | break; |
3998 | case DEFINEP: | |
3999 | sw = 0; | |
4000 | break; | |
c277df42 | 4001 | case IFTHEN: |
2c2d71f5 | 4002 | PL_reg_leftiter = PL_reg_maxiter; /* Void cache */ |
24d3c4a9 | 4003 | if (sw) |
c277df42 IZ |
4004 | next = NEXTOPER(NEXTOPER(scan)); |
4005 | else { | |
4006 | next = scan + ARG(scan); | |
4007 | if (OP(next) == IFTHEN) /* Fake one. */ | |
4008 | next = NEXTOPER(NEXTOPER(next)); | |
4009 | } | |
4010 | break; | |
4011 | case LOGICAL: | |
24d3c4a9 | 4012 | logical = scan->flags; |
c277df42 | 4013 | break; |
c476f425 | 4014 | |
2ab05381 | 4015 | /******************************************************************* |
2ab05381 | 4016 | |
c476f425 DM |
4017 | The CURLYX/WHILEM pair of ops handle the most generic case of the /A*B/ |
4018 | pattern, where A and B are subpatterns. (For simple A, CURLYM or | |
4019 | STAR/PLUS/CURLY/CURLYN are used instead.) | |
2ab05381 | 4020 | |
c476f425 | 4021 | A*B is compiled as <CURLYX><A><WHILEM><B> |
2ab05381 | 4022 | |
c476f425 DM |
4023 | On entry to the subpattern, CURLYX is called. This pushes a CURLYX |
4024 | state, which contains the current count, initialised to -1. It also sets | |
4025 | cur_curlyx to point to this state, with any previous value saved in the | |
4026 | state block. | |
2ab05381 | 4027 | |
c476f425 DM |
4028 | CURLYX then jumps straight to the WHILEM op, rather than executing A, |
4029 | since the pattern may possibly match zero times (i.e. it's a while {} loop | |
4030 | rather than a do {} while loop). | |
2ab05381 | 4031 | |
c476f425 DM |
4032 | Each entry to WHILEM represents a successful match of A. The count in the |
4033 | CURLYX block is incremented, another WHILEM state is pushed, and execution | |
4034 | passes to A or B depending on greediness and the current count. | |
2ab05381 | 4035 | |
c476f425 DM |
4036 | For example, if matching against the string a1a2a3b (where the aN are |
4037 | substrings that match /A/), then the match progresses as follows: (the | |
4038 | pushed states are interspersed with the bits of strings matched so far): | |
2ab05381 | 4039 | |
c476f425 DM |
4040 | <CURLYX cnt=-1> |
4041 | <CURLYX cnt=0><WHILEM> | |
4042 | <CURLYX cnt=1><WHILEM> a1 <WHILEM> | |
4043 | <CURLYX cnt=2><WHILEM> a1 <WHILEM> a2 <WHILEM> | |
4044 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> | |
4045 | <CURLYX cnt=3><WHILEM> a1 <WHILEM> a2 <WHILEM> a3 <WHILEM> b | |
2ab05381 | 4046 | |
c476f425 DM |
4047 | (Contrast this with something like CURLYM, which maintains only a single |
4048 | backtrack state: | |
2ab05381 | 4049 | |
c476f425 DM |
4050 | <CURLYM cnt=0> a1 |
4051 | a1 <CURLYM cnt=1> a2 | |
4052 | a1 a2 <CURLYM cnt=2> a3 | |
4053 | a1 a2 a3 <CURLYM cnt=3> b | |
4054 | ) | |
2ab05381 | 4055 | |
c476f425 DM |
4056 | Each WHILEM state block marks a point to backtrack to upon partial failure |
4057 | of A or B, and also contains some minor state data related to that | |
4058 | iteration. The CURLYX block, pointed to by cur_curlyx, contains the | |
4059 | overall state, such as the count, and pointers to the A and B ops. | |
2ab05381 | 4060 | |
c476f425 DM |
4061 | This is complicated slightly by nested CURLYX/WHILEM's. Since cur_curlyx |
4062 | must always point to the *current* CURLYX block, the rules are: | |
2ab05381 | 4063 | |
c476f425 DM |
4064 | When executing CURLYX, save the old cur_curlyx in the CURLYX state block, |
4065 | and set cur_curlyx to point the new block. | |
2ab05381 | 4066 | |
c476f425 DM |
4067 | When popping the CURLYX block after a successful or unsuccessful match, |
4068 | restore the previous cur_curlyx. | |
2ab05381 | 4069 | |
c476f425 DM |
4070 | When WHILEM is about to execute B, save the current cur_curlyx, and set it |
4071 | to the outer one saved in the CURLYX block. | |
2ab05381 | 4072 | |
c476f425 DM |
4073 | When popping the WHILEM block after a successful or unsuccessful B match, |
4074 | restore the previous cur_curlyx. | |
2ab05381 | 4075 | |
c476f425 DM |
4076 | Here's an example for the pattern (AI* BI)*BO |
4077 | I and O refer to inner and outer, C and W refer to CURLYX and WHILEM: | |
2ab05381 | 4078 | |
c476f425 DM |
4079 | cur_ |
4080 | curlyx backtrack stack | |
4081 | ------ --------------- | |
4082 | NULL | |
4083 | CO <CO prev=NULL> <WO> | |
4084 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
4085 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi | |
4086 | NULL <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi <WO prev=CO> bo | |
2ab05381 | 4087 | |
c476f425 DM |
4088 | At this point the pattern succeeds, and we work back down the stack to |
4089 | clean up, restoring as we go: | |
95b24440 | 4090 | |
c476f425 DM |
4091 | CO <CO prev=NULL> <WO> <CI prev=CO> <WI> ai <WI prev=CI> bi |
4092 | CI <CO prev=NULL> <WO> <CI prev=CO> <WI> ai | |
4093 | CO <CO prev=NULL> <WO> | |
4094 | NULL | |
a0374537 | 4095 | |
c476f425 DM |
4096 | *******************************************************************/ |
4097 | ||
4098 | #define ST st->u.curlyx | |
4099 | ||
4100 | case CURLYX: /* start of /A*B/ (for complex A) */ | |
4101 | { | |
4102 | /* No need to save/restore up to this paren */ | |
4103 | I32 parenfloor = scan->flags; | |
4104 | ||
4105 | assert(next); /* keep Coverity happy */ | |
4106 | if (OP(PREVOPER(next)) == NOTHING) /* LONGJMP */ | |
4107 | next += ARG(next); | |
4108 | ||
4109 | /* XXXX Probably it is better to teach regpush to support | |
4110 | parenfloor > PL_regsize... */ | |
4111 | if (parenfloor > (I32)*PL_reglastparen) | |
4112 | parenfloor = *PL_reglastparen; /* Pessimization... */ | |
4113 | ||
4114 | ST.prev_curlyx= cur_curlyx; | |
4115 | cur_curlyx = st; | |
4116 | ST.cp = PL_savestack_ix; | |
4117 | ||
4118 | /* these fields contain the state of the current curly. | |
4119 | * they are accessed by subsequent WHILEMs */ | |
4120 | ST.parenfloor = parenfloor; | |
4121 | ST.min = ARG1(scan); | |
4122 | ST.max = ARG2(scan); | |
4123 | ST.A = NEXTOPER(scan) + EXTRA_STEP_2ARGS; | |
4124 | ST.B = next; | |
24d3c4a9 DM |
4125 | ST.minmod = minmod; |
4126 | minmod = 0; | |
c476f425 DM |
4127 | ST.count = -1; /* this will be updated by WHILEM */ |
4128 | ST.lastloc = NULL; /* this will be updated by WHILEM */ | |
4129 | ||
4130 | PL_reginput = locinput; | |
4131 | PUSH_YES_STATE_GOTO(CURLYX_end, PREVOPER(next)); | |
5f66b61c | 4132 | /* NOTREACHED */ |
c476f425 | 4133 | } |
a0d0e21e | 4134 | |
c476f425 | 4135 | case CURLYX_end: /* just finished matching all of A*B */ |
c476f425 DM |
4136 | cur_curlyx = ST.prev_curlyx; |
4137 | sayYES; | |
4138 | /* NOTREACHED */ | |
a0d0e21e | 4139 | |
c476f425 DM |
4140 | case CURLYX_end_fail: /* just failed to match all of A*B */ |
4141 | regcpblow(ST.cp); | |
4142 | cur_curlyx = ST.prev_curlyx; | |
4143 | sayNO; | |
4144 | /* NOTREACHED */ | |
4633a7c4 | 4145 | |
a0d0e21e | 4146 | |
c476f425 DM |
4147 | #undef ST |
4148 | #define ST st->u.whilem | |
4149 | ||
4150 | case WHILEM: /* just matched an A in /A*B/ (for complex A) */ | |
4151 | { | |
4152 | /* see the discussion above about CURLYX/WHILEM */ | |
c476f425 DM |
4153 | I32 n; |
4154 | assert(cur_curlyx); /* keep Coverity happy */ | |
4155 | n = ++cur_curlyx->u.curlyx.count; /* how many A's matched */ | |
4156 | ST.save_lastloc = cur_curlyx->u.curlyx.lastloc; | |
4157 | ST.cache_offset = 0; | |
4158 | ST.cache_mask = 0; | |
4159 | ||
4160 | PL_reginput = locinput; | |
4161 | ||
4162 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4163 | "%*s whilem: matched %ld out of %ld..%ld\n", | |
4164 | REPORT_CODE_OFF+depth*2, "", (long)n, | |
4165 | (long)cur_curlyx->u.curlyx.min, | |
4166 | (long)cur_curlyx->u.curlyx.max) | |
4167 | ); | |
a0d0e21e | 4168 | |
c476f425 | 4169 | /* First just match a string of min A's. */ |
a0d0e21e | 4170 | |
c476f425 DM |
4171 | if (n < cur_curlyx->u.curlyx.min) { |
4172 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4173 | PUSH_STATE_GOTO(WHILEM_A_pre, cur_curlyx->u.curlyx.A); | |
4174 | /* NOTREACHED */ | |
4175 | } | |
4176 | ||
4177 | /* If degenerate A matches "", assume A done. */ | |
4178 | ||
4179 | if (locinput == cur_curlyx->u.curlyx.lastloc) { | |
4180 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4181 | "%*s whilem: empty match detected, trying continuation...\n", | |
4182 | REPORT_CODE_OFF+depth*2, "") | |
4183 | ); | |
4184 | goto do_whilem_B_max; | |
4185 | } | |
4186 | ||
4187 | /* super-linear cache processing */ | |
4188 | ||
4189 | if (scan->flags) { | |
a0d0e21e | 4190 | |
2c2d71f5 | 4191 | if (!PL_reg_maxiter) { |
c476f425 DM |
4192 | /* start the countdown: Postpone detection until we |
4193 | * know the match is not *that* much linear. */ | |
2c2d71f5 | 4194 | PL_reg_maxiter = (PL_regeol - PL_bostr + 1) * (scan->flags>>4); |
66bf836d DM |
4195 | /* possible overflow for long strings and many CURLYX's */ |
4196 | if (PL_reg_maxiter < 0) | |
4197 | PL_reg_maxiter = I32_MAX; | |
2c2d71f5 JH |
4198 | PL_reg_leftiter = PL_reg_maxiter; |
4199 | } | |
c476f425 | 4200 | |
2c2d71f5 | 4201 | if (PL_reg_leftiter-- == 0) { |
c476f425 | 4202 | /* initialise cache */ |
3298f257 | 4203 | const I32 size = (PL_reg_maxiter + 7)/8; |
2c2d71f5 | 4204 | if (PL_reg_poscache) { |
eb160463 | 4205 | if ((I32)PL_reg_poscache_size < size) { |
2c2d71f5 JH |
4206 | Renew(PL_reg_poscache, size, char); |
4207 | PL_reg_poscache_size = size; | |
4208 | } | |
4209 | Zero(PL_reg_poscache, size, char); | |
4210 | } | |
4211 | else { | |
4212 | PL_reg_poscache_size = size; | |
a02a5408 | 4213 | Newxz(PL_reg_poscache, size, char); |
2c2d71f5 | 4214 | } |
c476f425 DM |
4215 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, |
4216 | "%swhilem: Detected a super-linear match, switching on caching%s...\n", | |
4217 | PL_colors[4], PL_colors[5]) | |
4218 | ); | |
2c2d71f5 | 4219 | } |
c476f425 | 4220 | |
2c2d71f5 | 4221 | if (PL_reg_leftiter < 0) { |
c476f425 DM |
4222 | /* have we already failed at this position? */ |
4223 | I32 offset, mask; | |
4224 | offset = (scan->flags & 0xf) - 1 | |
4225 | + (locinput - PL_bostr) * (scan->flags>>4); | |
4226 | mask = 1 << (offset % 8); | |
4227 | offset /= 8; | |
4228 | if (PL_reg_poscache[offset] & mask) { | |
4229 | DEBUG_EXECUTE_r( PerlIO_printf(Perl_debug_log, | |
4230 | "%*s whilem: (cache) already tried at this position...\n", | |
4231 | REPORT_CODE_OFF+depth*2, "") | |
2c2d71f5 | 4232 | ); |
3298f257 | 4233 | sayNO; /* cache records failure */ |
2c2d71f5 | 4234 | } |
c476f425 DM |
4235 | ST.cache_offset = offset; |
4236 | ST.cache_mask = mask; | |
2c2d71f5 | 4237 | } |
c476f425 | 4238 | } |
2c2d71f5 | 4239 | |
c476f425 | 4240 | /* Prefer B over A for minimal matching. */ |
a687059c | 4241 | |
c476f425 DM |
4242 | if (cur_curlyx->u.curlyx.minmod) { |
4243 | ST.save_curlyx = cur_curlyx; | |
4244 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4245 | ST.cp = regcppush(ST.save_curlyx->u.curlyx.parenfloor); | |
4246 | REGCP_SET(ST.lastcp); | |
4247 | PUSH_YES_STATE_GOTO(WHILEM_B_min, ST.save_curlyx->u.curlyx.B); | |
4248 | /* NOTREACHED */ | |
4249 | } | |
a0d0e21e | 4250 | |
c476f425 DM |
4251 | /* Prefer A over B for maximal matching. */ |
4252 | ||
4253 | if (n < cur_curlyx->u.curlyx.max) { /* More greed allowed? */ | |
4254 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4255 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4256 | REGCP_SET(ST.lastcp); | |
4257 | PUSH_STATE_GOTO(WHILEM_A_max, cur_curlyx->u.curlyx.A); | |
4258 | /* NOTREACHED */ | |
4259 | } | |
4260 | goto do_whilem_B_max; | |
4261 | } | |
4262 | /* NOTREACHED */ | |
4263 | ||
4264 | case WHILEM_B_min: /* just matched B in a minimal match */ | |
4265 | case WHILEM_B_max: /* just matched B in a maximal match */ | |
4266 | cur_curlyx = ST.save_curlyx; | |
4267 | sayYES; | |
4268 | /* NOTREACHED */ | |
4269 | ||
4270 | case WHILEM_B_max_fail: /* just failed to match B in a maximal match */ | |
4271 | cur_curlyx = ST.save_curlyx; | |
4272 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4273 | cur_curlyx->u.curlyx.count--; | |
4274 | CACHEsayNO; | |
4275 | /* NOTREACHED */ | |
4276 | ||
4277 | case WHILEM_A_min_fail: /* just failed to match A in a minimal match */ | |
4278 | REGCP_UNWIND(ST.lastcp); | |
4279 | regcppop(rex); | |
4280 | /* FALL THROUGH */ | |
4281 | case WHILEM_A_pre_fail: /* just failed to match even minimal A */ | |
4282 | cur_curlyx->u.curlyx.lastloc = ST.save_lastloc; | |
4283 | cur_curlyx->u.curlyx.count--; | |
4284 | CACHEsayNO; | |
4285 | /* NOTREACHED */ | |
4286 | ||
4287 | case WHILEM_A_max_fail: /* just failed to match A in a maximal match */ | |
4288 | REGCP_UNWIND(ST.lastcp); | |
4289 | regcppop(rex); /* Restore some previous $<digit>s? */ | |
4290 | PL_reginput = locinput; | |
4291 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4292 | "%*s whilem: failed, trying continuation...\n", | |
4293 | REPORT_CODE_OFF+depth*2, "") | |
4294 | ); | |
4295 | do_whilem_B_max: | |
4296 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4297 | && ckWARN(WARN_REGEXP) | |
4298 | && !(PL_reg_flags & RF_warned)) | |
4299 | { | |
4300 | PL_reg_flags |= RF_warned; | |
4301 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), "%s limit (%d) exceeded", | |
4302 | "Complex regular subexpression recursion", | |
4303 | REG_INFTY - 1); | |
4304 | } | |
4305 | ||
4306 | /* now try B */ | |
4307 | ST.save_curlyx = cur_curlyx; | |
4308 | cur_curlyx = cur_curlyx->u.curlyx.prev_curlyx; | |
4309 | PUSH_YES_STATE_GOTO(WHILEM_B_max, ST.save_curlyx->u.curlyx.B); | |
4310 | /* NOTREACHED */ | |
4311 | ||
4312 | case WHILEM_B_min_fail: /* just failed to match B in a minimal match */ | |
4313 | cur_curlyx = ST.save_curlyx; | |
4314 | REGCP_UNWIND(ST.lastcp); | |
4315 | regcppop(rex); | |
4316 | ||
4317 | if (cur_curlyx->u.curlyx.count >= cur_curlyx->u.curlyx.max) { | |
4318 | /* Maximum greed exceeded */ | |
4319 | if (cur_curlyx->u.curlyx.count >= REG_INFTY | |
4320 | && ckWARN(WARN_REGEXP) | |
4321 | && !(PL_reg_flags & RF_warned)) | |
4322 | { | |
3280af22 | 4323 | PL_reg_flags |= RF_warned; |
c476f425 DM |
4324 | Perl_warner(aTHX_ packWARN(WARN_REGEXP), |
4325 | "%s limit (%d) exceeded", | |
4326 | "Complex regular subexpression recursion", | |
4327 | REG_INFTY - 1); | |
a0d0e21e | 4328 | } |
c476f425 | 4329 | cur_curlyx->u.curlyx.count--; |
3ab3c9b4 | 4330 | CACHEsayNO; |
a0d0e21e | 4331 | } |
c476f425 DM |
4332 | |
4333 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, | |
4334 | "%*s trying longer...\n", REPORT_CODE_OFF+depth*2, "") | |
4335 | ); | |
4336 | /* Try grabbing another A and see if it helps. */ | |
4337 | PL_reginput = locinput; | |
4338 | cur_curlyx->u.curlyx.lastloc = locinput; | |
4339 | ST.cp = regcppush(cur_curlyx->u.curlyx.parenfloor); | |
4340 | REGCP_SET(ST.lastcp); | |
4341 | PUSH_STATE_GOTO(WHILEM_A_min, ST.save_curlyx->u.curlyx.A); | |
5f66b61c | 4342 | /* NOTREACHED */ |
40a82448 DM |
4343 | |
4344 | #undef ST | |
4345 | #define ST st->u.branch | |
4346 | ||
4347 | case BRANCHJ: /* /(...|A|...)/ with long next pointer */ | |
c277df42 IZ |
4348 | next = scan + ARG(scan); |
4349 | if (next == scan) | |
4350 | next = NULL; | |
40a82448 DM |
4351 | scan = NEXTOPER(scan); |
4352 | /* FALL THROUGH */ | |
c277df42 | 4353 | |
40a82448 DM |
4354 | case BRANCH: /* /(...|A|...)/ */ |
4355 | scan = NEXTOPER(scan); /* scan now points to inner node */ | |
40a82448 DM |
4356 | ST.lastparen = *PL_reglastparen; |
4357 | ST.next_branch = next; | |
4358 | REGCP_SET(ST.cp); | |
4359 | PL_reginput = locinput; | |
02db2b7b | 4360 | |
40a82448 | 4361 | /* Now go into the branch */ |
5d458dd8 YO |
4362 | if (has_cutgroup) { |
4363 | PUSH_YES_STATE_GOTO(BRANCH_next, scan); | |
4364 | } else { | |
4365 | PUSH_STATE_GOTO(BRANCH_next, scan); | |
4366 | } | |
40a82448 | 4367 | /* NOTREACHED */ |
5d458dd8 YO |
4368 | case CUTGROUP: |
4369 | PL_reginput = locinput; | |
4370 | sv_yes_mark = st->u.mark.mark_name = scan->flags ? NULL : | |
ad64d0ec | 4371 | MUTABLE_SV(rexi->data->data[ ARG( scan ) ]); |
5d458dd8 YO |
4372 | PUSH_STATE_GOTO(CUTGROUP_next,next); |
4373 | /* NOTREACHED */ | |
4374 | case CUTGROUP_next_fail: | |
4375 | do_cutgroup = 1; | |
4376 | no_final = 1; | |
4377 | if (st->u.mark.mark_name) | |
4378 | sv_commit = st->u.mark.mark_name; | |
4379 | sayNO; | |
4380 | /* NOTREACHED */ | |
4381 | case BRANCH_next: | |
4382 | sayYES; | |
4383 | /* NOTREACHED */ | |
40a82448 | 4384 | case BRANCH_next_fail: /* that branch failed; try the next, if any */ |
5d458dd8 YO |
4385 | if (do_cutgroup) { |
4386 | do_cutgroup = 0; | |
4387 | no_final = 0; | |
4388 | } | |
40a82448 DM |
4389 | REGCP_UNWIND(ST.cp); |
4390 | for (n = *PL_reglastparen; n > ST.lastparen; n--) | |
f0ab9afb | 4391 | PL_regoffs[n].end = -1; |
40a82448 | 4392 | *PL_reglastparen = n; |
0a4db386 | 4393 | /*dmq: *PL_reglastcloseparen = n; */ |
40a82448 DM |
4394 | scan = ST.next_branch; |
4395 | /* no more branches? */ | |
5d458dd8 YO |
4396 | if (!scan || (OP(scan) != BRANCH && OP(scan) != BRANCHJ)) { |
4397 | DEBUG_EXECUTE_r({ | |
4398 | PerlIO_printf( Perl_debug_log, | |
4399 | "%*s %sBRANCH failed...%s\n", | |
4400 | REPORT_CODE_OFF+depth*2, "", | |
4401 | PL_colors[4], | |
4402 | PL_colors[5] ); | |
4403 | }); | |
4404 | sayNO_SILENT; | |
4405 | } | |
40a82448 DM |
4406 | continue; /* execute next BRANCH[J] op */ |
4407 | /* NOTREACHED */ | |
4408 | ||
a0d0e21e | 4409 | case MINMOD: |
24d3c4a9 | 4410 | minmod = 1; |
a0d0e21e | 4411 | break; |
40a82448 DM |
4412 | |
4413 | #undef ST | |
4414 | #define ST st->u.curlym | |
4415 | ||
4416 | case CURLYM: /* /A{m,n}B/ where A is fixed-length */ | |
4417 | ||
4418 | /* This is an optimisation of CURLYX that enables us to push | |
84d2fa14 | 4419 | * only a single backtracking state, no matter how many matches |
40a82448 DM |
4420 | * there are in {m,n}. It relies on the pattern being constant |
4421 | * length, with no parens to influence future backrefs | |
4422 | */ | |
4423 | ||
4424 | ST.me = scan; | |
dc45a647 | 4425 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; |
40a82448 DM |
4426 | |
4427 | /* if paren positive, emulate an OPEN/CLOSE around A */ | |
4428 | if (ST.me->flags) { | |
3b6647e0 | 4429 | U32 paren = ST.me->flags; |
40a82448 DM |
4430 | if (paren > PL_regsize) |
4431 | PL_regsize = paren; | |
3b6647e0 | 4432 | if (paren > *PL_reglastparen) |
40a82448 | 4433 | *PL_reglastparen = paren; |
c277df42 | 4434 | scan += NEXT_OFF(scan); /* Skip former OPEN. */ |
6407bf3b | 4435 | } |
40a82448 DM |
4436 | ST.A = scan; |
4437 | ST.B = next; | |
4438 | ST.alen = 0; | |
4439 | ST.count = 0; | |
24d3c4a9 DM |
4440 | ST.minmod = minmod; |
4441 | minmod = 0; | |
40a82448 DM |
4442 | ST.c1 = CHRTEST_UNINIT; |
4443 | REGCP_SET(ST.cp); | |
6407bf3b | 4444 | |
40a82448 DM |
4445 | if (!(ST.minmod ? ARG1(ST.me) : ARG2(ST.me))) /* min/max */ |
4446 | goto curlym_do_B; | |
4447 | ||
4448 | curlym_do_A: /* execute the A in /A{m,n}B/ */ | |
6407bf3b | 4449 | PL_reginput = locinput; |
40a82448 DM |
4450 | PUSH_YES_STATE_GOTO(CURLYM_A, ST.A); /* match A */ |
4451 | /* NOTREACHED */ | |
5f80c4cf | 4452 | |
40a82448 DM |
4453 | case CURLYM_A: /* we've just matched an A */ |
4454 | locinput = st->locinput; | |
4455 | nextchr = UCHARAT(locinput); | |
4456 | ||
4457 | ST.count++; | |
4458 | /* after first match, determine A's length: u.curlym.alen */ | |
4459 | if (ST.count == 1) { | |
4460 | if (PL_reg_match_utf8) { | |
4461 | char *s = locinput; | |
4462 | while (s < PL_reginput) { | |
4463 | ST.alen++; | |
4464 | s += UTF8SKIP(s); | |
4465 | } | |
4466 | } | |
4467 | else { | |
4468 | ST.alen = PL_reginput - locinput; | |
4469 | } | |
4470 | if (ST.alen == 0) | |
4471 | ST.count = ST.minmod ? ARG1(ST.me) : ARG2(ST.me); | |
4472 | } | |
0cadcf80 DM |
4473 | DEBUG_EXECUTE_r( |
4474 | PerlIO_printf(Perl_debug_log, | |
40a82448 | 4475 | "%*s CURLYM now matched %"IVdf" times, len=%"IVdf"...\n", |
5bc10b2c | 4476 | (int)(REPORT_CODE_OFF+(depth*2)), "", |
40a82448 | 4477 | (IV) ST.count, (IV)ST.alen) |
0cadcf80 DM |
4478 | ); |
4479 | ||
40a82448 | 4480 | locinput = PL_reginput; |
0a4db386 YO |
4481 | |
4482 | if (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4483 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4484 | goto fake_end; |
4485 | ||
c966426a DM |
4486 | { |
4487 | I32 max = (ST.minmod ? ARG1(ST.me) : ARG2(ST.me)); | |
4488 | if ( max == REG_INFTY || ST.count < max ) | |
4489 | goto curlym_do_A; /* try to match another A */ | |
4490 | } | |
40a82448 | 4491 | goto curlym_do_B; /* try to match B */ |
5f80c4cf | 4492 | |
40a82448 DM |
4493 | case CURLYM_A_fail: /* just failed to match an A */ |
4494 | REGCP_UNWIND(ST.cp); | |
0a4db386 YO |
4495 | |
4496 | if (ST.minmod || ST.count < ARG1(ST.me) /* min*/ | |
4497 | || (cur_eval && cur_eval->u.eval.close_paren && | |
24b23f37 | 4498 | cur_eval->u.eval.close_paren == (U32)ST.me->flags)) |
40a82448 | 4499 | sayNO; |
0cadcf80 | 4500 | |
40a82448 DM |
4501 | curlym_do_B: /* execute the B in /A{m,n}B/ */ |
4502 | PL_reginput = locinput; | |
4503 | if (ST.c1 == CHRTEST_UNINIT) { | |
4504 | /* calculate c1 and c2 for possible match of 1st char | |
4505 | * following curly */ | |
4506 | ST.c1 = ST.c2 = CHRTEST_VOID; | |
4507 | if (HAS_TEXT(ST.B) || JUMPABLE(ST.B)) { | |
4508 | regnode *text_node = ST.B; | |
4509 | if (! HAS_TEXT(text_node)) | |
4510 | FIND_NEXT_IMPT(text_node); | |
ee9b8eae YO |
4511 | /* this used to be |
4512 | ||
4513 | (HAS_TEXT(text_node) && PL_regkind[OP(text_node)] == EXACT) | |
4514 | ||
4515 | But the former is redundant in light of the latter. | |
4516 | ||
4517 | if this changes back then the macro for | |
4518 | IS_TEXT and friends need to change. | |
4519 | */ | |
4520 | if (PL_regkind[OP(text_node)] == EXACT) | |
40a82448 | 4521 | { |
ee9b8eae | 4522 | |
40a82448 DM |
4523 | ST.c1 = (U8)*STRING(text_node); |
4524 | ST.c2 = | |
ee9b8eae | 4525 | (IS_TEXTF(text_node)) |
40a82448 | 4526 | ? PL_fold[ST.c1] |
ee9b8eae | 4527 | : (IS_TEXTFL(text_node)) |
40a82448 DM |
4528 | ? PL_fold_locale[ST.c1] |
4529 | : ST.c1; | |
c277df42 | 4530 | } |
c277df42 | 4531 | } |
40a82448 DM |
4532 | } |
4533 | ||
4534 | DEBUG_EXECUTE_r( | |
4535 | PerlIO_printf(Perl_debug_log, | |
4536 | "%*s CURLYM trying tail with matches=%"IVdf"...\n", | |
5bc10b2c | 4537 | (int)(REPORT_CODE_OFF+(depth*2)), |
40a82448 DM |
4538 | "", (IV)ST.count) |
4539 | ); | |
4540 | if (ST.c1 != CHRTEST_VOID | |
4541 | && UCHARAT(PL_reginput) != ST.c1 | |
4542 | && UCHARAT(PL_reginput) != ST.c2) | |
4543 | { | |
4544 | /* simulate B failing */ | |
3e901dc0 YO |
4545 | DEBUG_OPTIMISE_r( |
4546 | PerlIO_printf(Perl_debug_log, | |
4547 | "%*s CURLYM Fast bail c1=%"IVdf" c2=%"IVdf"\n", | |
4548 | (int)(REPORT_CODE_OFF+(depth*2)),"", | |
4549 | (IV)ST.c1,(IV)ST.c2 | |
4550 | )); | |
40a82448 DM |
4551 | state_num = CURLYM_B_fail; |
4552 | goto reenter_switch; | |
4553 | } | |
4554 | ||
4555 | if (ST.me->flags) { | |
4556 | /* mark current A as captured */ | |
4557 | I32 paren = ST.me->flags; | |
4558 | if (ST.count) { | |
f0ab9afb | 4559 | PL_regoffs[paren].start |
40a82448 | 4560 | = HOPc(PL_reginput, -ST.alen) - PL_bostr; |
f0ab9afb | 4561 | PL_regoffs[paren].end = PL_reginput - PL_bostr; |
0a4db386 | 4562 | /*dmq: *PL_reglastcloseparen = paren; */ |
c277df42 | 4563 | } |
40a82448 | 4564 | else |
f0ab9afb | 4565 | PL_regoffs[paren].end = -1; |
0a4db386 | 4566 | if (cur_eval && cur_eval->u.eval.close_paren && |
24b23f37 | 4567 | cur_eval->u.eval.close_paren == (U32)ST.me->flags) |
0a4db386 YO |
4568 | { |
4569 | if (ST.count) | |
4570 | goto fake_end; | |
4571 | else | |
4572 | sayNO; | |
4573 | } | |
c277df42 | 4574 | } |
0a4db386 | 4575 | |
40a82448 | 4576 | PUSH_STATE_GOTO(CURLYM_B, ST.B); /* match B */ |
5f66b61c | 4577 | /* NOTREACHED */ |
40a82448 DM |
4578 | |
4579 | case CURLYM_B_fail: /* just failed to match a B */ | |
4580 | REGCP_UNWIND(ST.cp); | |
4581 | if (ST.minmod) { | |
84d2fa14 HS |
4582 | I32 max = ARG2(ST.me); |
4583 | if (max != REG_INFTY && ST.count == max) | |
40a82448 DM |
4584 | sayNO; |
4585 | goto curlym_do_A; /* try to match a further A */ | |
4586 | } | |
4587 | /* backtrack one A */ | |
4588 | if (ST.count == ARG1(ST.me) /* min */) | |
4589 | sayNO; | |
4590 | ST.count--; | |
4591 | locinput = HOPc(locinput, -ST.alen); | |
4592 | goto curlym_do_B; /* try to match B */ | |
4593 | ||
c255a977 DM |
4594 | #undef ST |
4595 | #define ST st->u.curly | |
40a82448 | 4596 | |
c255a977 DM |
4597 | #define CURLY_SETPAREN(paren, success) \ |
4598 | if (paren) { \ | |
4599 | if (success) { \ | |
f0ab9afb NC |
4600 | PL_regoffs[paren].start = HOPc(locinput, -1) - PL_bostr; \ |
4601 | PL_regoffs[paren].end = locinput - PL_bostr; \ | |
0a4db386 | 4602 | *PL_reglastcloseparen = paren; \ |
c255a977 DM |
4603 | } \ |
4604 | else \ | |
f0ab9afb | 4605 | PL_regoffs[paren].end = -1; \ |
c255a977 DM |
4606 | } |
4607 | ||
4608 | case STAR: /* /A*B/ where A is width 1 */ | |
4609 | ST.paren = 0; | |
4610 | ST.min = 0; | |
4611 | ST.max = REG_INFTY; | |
a0d0e21e LW |
4612 | scan = NEXTOPER(scan); |
4613 | goto repeat; | |
c255a977 DM |
4614 | case PLUS: /* /A+B/ where A is width 1 */ |
4615 | ST.paren = 0; | |
4616 | ST.min = 1; | |
4617 | ST.max = REG_INFTY; | |
c277df42 | 4618 | scan = NEXTOPER(scan); |
c255a977 DM |
4619 | goto repeat; |
4620 | case CURLYN: /* /(A){m,n}B/ where A is width 1 */ | |
4621 | ST.paren = scan->flags; /* Which paren to set */ | |
4622 | if (ST.paren > PL_regsize) | |
4623 | PL_regsize = ST.paren; | |
3b6647e0 | 4624 | if (ST.paren > *PL_reglastparen) |
c255a977 DM |
4625 | *PL_reglastparen = ST.paren; |
4626 | ST.min = ARG1(scan); /* min to match */ | |
4627 | ST.max = ARG2(scan); /* max to match */ | |
0a4db386 | 4628 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4629 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4630 | ST.min=1; |
4631 | ST.max=1; | |
4632 | } | |
c255a977 DM |
4633 | scan = regnext(NEXTOPER(scan) + NODE_STEP_REGNODE); |
4634 | goto repeat; | |
4635 | case CURLY: /* /A{m,n}B/ where A is width 1 */ | |
4636 | ST.paren = 0; | |
4637 | ST.min = ARG1(scan); /* min to match */ | |
4638 | ST.max = ARG2(scan); /* max to match */ | |
4639 | scan = NEXTOPER(scan) + NODE_STEP_REGNODE; | |
c277df42 | 4640 | repeat: |
a0d0e21e LW |
4641 | /* |
4642 | * Lookahead to avoid useless match attempts | |
4643 | * when we know what character comes next. | |
c255a977 | 4644 | * |
5f80c4cf JP |
4645 | * Used to only do .*x and .*?x, but now it allows |
4646 | * for )'s, ('s and (?{ ... })'s to be in the way | |
4647 | * of the quantifier and the EXACT-like node. -- japhy | |
4648 | */ | |
4649 | ||
c255a977 DM |
4650 | if (ST.min > ST.max) /* XXX make this a compile-time check? */ |
4651 | sayNO; | |
cca55fe3 | 4652 | if (HAS_TEXT(next) || JUMPABLE(next)) { |
5f80c4cf JP |
4653 | U8 *s; |
4654 | regnode *text_node = next; | |
4655 | ||
3dab1dad YO |
4656 | if (! HAS_TEXT(text_node)) |
4657 | FIND_NEXT_IMPT(text_node); | |
5f80c4cf | 4658 | |
9e137952 | 4659 | if (! HAS_TEXT(text_node)) |
c255a977 | 4660 | ST.c1 = ST.c2 = CHRTEST_VOID; |
5f80c4cf | 4661 | else { |
ee9b8eae | 4662 | if ( PL_regkind[OP(text_node)] != EXACT ) { |
c255a977 | 4663 | ST.c1 = ST.c2 = CHRTEST_VOID; |
44a68960 | 4664 | goto assume_ok_easy; |
cca55fe3 | 4665 | } |
be8e71aa YO |
4666 | else |
4667 | s = (U8*)STRING(text_node); | |
ee9b8eae YO |
4668 | |
4669 | /* Currently we only get here when | |
4670 | ||
4671 | PL_rekind[OP(text_node)] == EXACT | |
4672 | ||
4673 | if this changes back then the macro for IS_TEXT and | |
4674 | friends need to change. */ | |
5f80c4cf | 4675 | if (!UTF) { |
c255a977 | 4676 | ST.c2 = ST.c1 = *s; |
ee9b8eae | 4677 | if (IS_TEXTF(text_node)) |
c255a977 | 4678 | ST.c2 = PL_fold[ST.c1]; |
ee9b8eae | 4679 | else if (IS_TEXTFL(text_node)) |
c255a977 | 4680 | ST.c2 = PL_fold_locale[ST.c1]; |
1aa99e6b | 4681 | } |
5f80c4cf | 4682 | else { /* UTF */ |
ee9b8eae | 4683 | if (IS_TEXTF(text_node)) { |
a2a2844f | 4684 | STRLEN ulen1, ulen2; |
89ebb4a3 JH |
4685 | U8 tmpbuf1[UTF8_MAXBYTES_CASE+1]; |
4686 | U8 tmpbuf2[UTF8_MAXBYTES_CASE+1]; | |
a2a2844f JH |
4687 | |
4688 | to_utf8_lower((U8*)s, tmpbuf1, &ulen1); | |
4689 | to_utf8_upper((U8*)s, tmpbuf2, &ulen2); | |
e294cc5d JH |
4690 | #ifdef EBCDIC |
4691 | ST.c1 = utf8n_to_uvchr(tmpbuf1, UTF8_MAXLEN, 0, | |
4692 | ckWARN(WARN_UTF8) ? | |
4693 | 0 : UTF8_ALLOW_ANY); | |
4694 | ST.c2 = utf8n_to_uvchr(tmpbuf2, UTF8_MAXLEN, 0, | |
4695 | ckWARN(WARN_UTF8) ? | |
4696 | 0 : UTF8_ALLOW_ANY); | |
4697 | #else | |
c255a977 | 4698 | ST.c1 = utf8n_to_uvuni(tmpbuf1, UTF8_MAXBYTES, 0, |
e294cc5d | 4699 | uniflags); |
c255a977 | 4700 | ST.c2 = utf8n_to_uvuni(tmpbuf2, UTF8_MAXBYTES, 0, |
e294cc5d JH |
4701 | uniflags); |
4702 | #endif | |
5f80c4cf JP |
4703 | } |
4704 | else { | |
c255a977 | 4705 | ST.c2 = ST.c1 = utf8n_to_uvchr(s, UTF8_MAXBYTES, 0, |
041457d9 | 4706 | uniflags); |
5f80c4cf | 4707 | } |
1aa99e6b IH |
4708 | } |
4709 | } | |
bbce6d69 | 4710 | } |
a0d0e21e | 4711 | else |
c255a977 | 4712 | ST.c1 = ST.c2 = CHRTEST_VOID; |
cca55fe3 | 4713 | assume_ok_easy: |
c255a977 DM |
4714 | |
4715 | ST.A = scan; | |
4716 | ST.B = next; | |
3280af22 | 4717 | PL_reginput = locinput; |
24d3c4a9 DM |
4718 | if (minmod) { |
4719 | minmod = 0; | |
e2e6a0f1 | 4720 | if (ST.min && regrepeat(rex, ST.A, ST.min, depth) < ST.min) |
4633a7c4 | 4721 | sayNO; |
c255a977 | 4722 | ST.count = ST.min; |
a0ed51b3 | 4723 | locinput = PL_reginput; |
c255a977 DM |
4724 | REGCP_SET(ST.cp); |
4725 | if (ST.c1 == CHRTEST_VOID) | |
4726 | goto curly_try_B_min; | |
4727 | ||
4728 | ST.oldloc = locinput; | |
4729 | ||
4730 | /* set ST.maxpos to the furthest point along the | |
4731 | * string that could possibly match */ | |
4732 | if (ST.max == REG_INFTY) { | |
4733 | ST.maxpos = PL_regeol - 1; | |
4734 | if (do_utf8) | |
4735 | while (UTF8_IS_CONTINUATION(*(U8*)ST.maxpos)) | |
4736 | ST.maxpos--; | |
4737 | } | |
4738 | else if (do_utf8) { | |
4739 | int m = ST.max - ST.min; | |
4740 | for (ST.maxpos = locinput; | |
4741 | m >0 && ST.maxpos + UTF8SKIP(ST.maxpos) <= PL_regeol; m--) | |
4742 | ST.maxpos += UTF8SKIP(ST.maxpos); | |
4743 | } | |
4744 | else { | |
4745 | ST.maxpos = locinput + ST.max - ST.min; | |
4746 | if (ST.maxpos >= PL_regeol) | |
4747 | ST.maxpos = PL_regeol - 1; | |
4748 | } | |
4749 | goto curly_try_B_min_known; | |
4750 | ||
4751 | } | |
4752 | else { | |
e2e6a0f1 | 4753 | ST.count = regrepeat(rex, ST.A, ST.max, depth); |
c255a977 DM |
4754 | locinput = PL_reginput; |
4755 | if (ST.count < ST.min) | |
4756 | sayNO; | |
4757 | if ((ST.count > ST.min) | |
4758 | && (PL_regkind[OP(ST.B)] == EOL) && (OP(ST.B) != MEOL)) | |
4759 | { | |
4760 | /* A{m,n} must come at the end of the string, there's | |
4761 | * no point in backing off ... */ | |
4762 | ST.min = ST.count; | |
4763 | /* ...except that $ and \Z can match before *and* after | |
4764 | newline at the end. Consider "\n\n" =~ /\n+\Z\n/. | |
4765 | We may back off by one in this case. */ | |
4766 | if (UCHARAT(PL_reginput - 1) == '\n' && OP(ST.B) != EOS) | |
4767 | ST.min--; | |
4768 | } | |
4769 | REGCP_SET(ST.cp); | |
4770 | goto curly_try_B_max; | |
4771 | } | |
4772 | /* NOTREACHED */ | |
4773 | ||
4774 | ||
4775 | case CURLY_B_min_known_fail: | |
4776 | /* failed to find B in a non-greedy match where c1,c2 valid */ | |
4777 | if (ST.paren && ST.count) | |
f0ab9afb | 4778 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4779 | |
4780 | PL_reginput = locinput; /* Could be reset... */ | |
4781 | REGCP_UNWIND(ST.cp); | |
4782 | /* Couldn't or didn't -- move forward. */ | |
4783 | ST.oldloc = locinput; | |
4784 | if (do_utf8) | |
4785 | locinput += UTF8SKIP(locinput); | |
4786 | else | |
4787 | locinput++; | |
4788 | ST.count++; | |
4789 | curly_try_B_min_known: | |
4790 | /* find the next place where 'B' could work, then call B */ | |
4791 | { | |
4792 | int n; | |
4793 | if (do_utf8) { | |
4794 | n = (ST.oldloc == locinput) ? 0 : 1; | |
4795 | if (ST.c1 == ST.c2) { | |
4796 | STRLEN len; | |
4797 | /* set n to utf8_distance(oldloc, locinput) */ | |
4798 | while (locinput <= ST.maxpos && | |
4799 | utf8n_to_uvchr((U8*)locinput, | |
4800 | UTF8_MAXBYTES, &len, | |
4801 | uniflags) != (UV)ST.c1) { | |
4802 | locinput += len; | |
4803 | n++; | |
4804 | } | |
1aa99e6b IH |
4805 | } |
4806 | else { | |
c255a977 DM |
4807 | /* set n to utf8_distance(oldloc, locinput) */ |
4808 | while (locinput <= ST.maxpos) { | |
4809 | STRLEN len; | |
4810 | const UV c = utf8n_to_uvchr((U8*)locinput, | |
4811 | UTF8_MAXBYTES, &len, | |
4812 | uniflags); | |
4813 | if (c == (UV)ST.c1 || c == (UV)ST.c2) | |
4814 | break; | |
4815 | locinput += len; | |
4816 | n++; | |
1aa99e6b | 4817 | } |
0fe9bf95 IZ |
4818 | } |
4819 | } | |
c255a977 DM |
4820 | else { |
4821 | if (ST.c1 == ST.c2) { | |
4822 | while (locinput <= ST.maxpos && | |
4823 | UCHARAT(locinput) != ST.c1) | |
4824 | locinput++; | |
bbce6d69 | 4825 | } |
c255a977 DM |
4826 | else { |
4827 | while (locinput <= ST.maxpos | |
4828 | && UCHARAT(locinput) != ST.c1 | |
4829 | && UCHARAT(locinput) != ST.c2) | |
4830 | locinput++; | |
a0ed51b3 | 4831 | } |
c255a977 DM |
4832 | n = locinput - ST.oldloc; |
4833 | } | |
4834 | if (locinput > ST.maxpos) | |
4835 | sayNO; | |
4836 | /* PL_reginput == oldloc now */ | |
4837 | if (n) { | |
4838 | ST.count += n; | |
e2e6a0f1 | 4839 | if (regrepeat(rex, ST.A, n, depth) < n) |
4633a7c4 | 4840 | sayNO; |
a0d0e21e | 4841 | } |
c255a977 DM |
4842 | PL_reginput = locinput; |
4843 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4844 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4845 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4846 | goto fake_end; |
4847 | } | |
c255a977 | 4848 | PUSH_STATE_GOTO(CURLY_B_min_known, ST.B); |
a0d0e21e | 4849 | } |
c255a977 DM |
4850 | /* NOTREACHED */ |
4851 | ||
4852 | ||
4853 | case CURLY_B_min_fail: | |
4854 | /* failed to find B in a non-greedy match where c1,c2 invalid */ | |
4855 | if (ST.paren && ST.count) | |
f0ab9afb | 4856 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4857 | |
4858 | REGCP_UNWIND(ST.cp); | |
4859 | /* failed -- move forward one */ | |
4860 | PL_reginput = locinput; | |
e2e6a0f1 | 4861 | if (regrepeat(rex, ST.A, 1, depth)) { |
c255a977 | 4862 | ST.count++; |
a0ed51b3 | 4863 | locinput = PL_reginput; |
c255a977 DM |
4864 | if (ST.count <= ST.max || (ST.max == REG_INFTY && |
4865 | ST.count > 0)) /* count overflow ? */ | |
15272685 | 4866 | { |
c255a977 DM |
4867 | curly_try_B_min: |
4868 | CURLY_SETPAREN(ST.paren, ST.count); | |
0a4db386 | 4869 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4870 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
0a4db386 YO |
4871 | goto fake_end; |
4872 | } | |
c255a977 | 4873 | PUSH_STATE_GOTO(CURLY_B_min, ST.B); |
a0d0e21e LW |
4874 | } |
4875 | } | |
4633a7c4 | 4876 | sayNO; |
c255a977 DM |
4877 | /* NOTREACHED */ |
4878 | ||
4879 | ||
4880 | curly_try_B_max: | |
4881 | /* a successful greedy match: now try to match B */ | |
40d049e4 | 4882 | if (cur_eval && cur_eval->u.eval.close_paren && |
86413ec0 | 4883 | cur_eval->u.eval.close_paren == (U32)ST.paren) { |
40d049e4 YO |
4884 | goto fake_end; |
4885 | } | |
c255a977 DM |
4886 | { |
4887 | UV c = 0; | |
4888 | if (ST.c1 != CHRTEST_VOID) | |
4889 | c = do_utf8 ? utf8n_to_uvchr((U8*)PL_reginput, | |
4890 | UTF8_MAXBYTES, 0, uniflags) | |
466787eb | 4891 | : (UV) UCHARAT(PL_reginput); |
c255a977 DM |
4892 | /* If it could work, try it. */ |
4893 | if (ST.c1 == CHRTEST_VOID || c == (UV)ST.c1 || c == (UV)ST.c2) { | |
4894 | CURLY_SETPAREN(ST.paren, ST.count); | |
4895 | PUSH_STATE_GOTO(CURLY_B_max, ST.B); | |
4896 | /* NOTREACHED */ | |
4897 | } | |
4898 | } | |
4899 | /* FALL THROUGH */ | |
4900 | case CURLY_B_max_fail: | |
4901 | /* failed to find B in a greedy match */ | |
4902 | if (ST.paren && ST.count) | |
f0ab9afb | 4903 | PL_regoffs[ST.paren].end = -1; |
c255a977 DM |
4904 | |
4905 | REGCP_UNWIND(ST.cp); | |
4906 | /* back up. */ | |
4907 | if (--ST.count < ST.min) | |
4908 | sayNO; | |
4909 | PL_reginput = locinput = HOPc(locinput, -1); | |
4910 | goto curly_try_B_max; | |
4911 | ||
4912 | #undef ST | |
4913 | ||
a0d0e21e | 4914 | case END: |
6bda09f9 | 4915 | fake_end: |
faec1544 DM |
4916 | if (cur_eval) { |
4917 | /* we've just finished A in /(??{A})B/; now continue with B */ | |
4918 | I32 tmpix; | |
faec1544 DM |
4919 | st->u.eval.toggle_reg_flags |
4920 | = cur_eval->u.eval.toggle_reg_flags; | |
4921 | PL_reg_flags ^= st->u.eval.toggle_reg_flags; | |
4922 | ||
288b8c02 NC |
4923 | st->u.eval.prev_rex = rex_sv; /* inner */ |
4924 | SETREX(rex_sv,cur_eval->u.eval.prev_rex); | |
4925 | rex = (struct regexp *)SvANY(rex_sv); | |
f8fc2ecf | 4926 | rexi = RXi_GET(rex); |
faec1544 | 4927 | cur_curlyx = cur_eval->u.eval.prev_curlyx; |
288b8c02 | 4928 | ReREFCNT_inc(rex_sv); |
faec1544 | 4929 | st->u.eval.cp = regcppush(0); /* Save *all* the positions. */ |
34a81e2b B |
4930 | |
4931 | /* rex was changed so update the pointer in PL_reglastparen and PL_reglastcloseparen */ | |
4932 | PL_reglastparen = &rex->lastparen; | |
4933 | PL_reglastcloseparen = &rex->lastcloseparen; | |
4934 | ||
faec1544 DM |
4935 | REGCP_SET(st->u.eval.lastcp); |
4936 | PL_reginput = locinput; | |
4937 | ||
4938 | /* Restore parens of the outer rex without popping the | |
4939 | * savestack */ | |
4940 | tmpix = PL_savestack_ix; | |
4941 | PL_savestack_ix = cur_eval->u.eval.lastcp; | |
4942 | regcppop(rex); | |
4943 | PL_savestack_ix = tmpix; | |
4944 | ||
4945 | st->u.eval.prev_eval = cur_eval; | |
4946 | cur_eval = cur_eval->u.eval.prev_eval; | |
4947 | DEBUG_EXECUTE_r( | |
2a49f0f5 JH |
4948 | PerlIO_printf(Perl_debug_log, "%*s EVAL trying tail ... %"UVxf"\n", |
4949 | REPORT_CODE_OFF+depth*2, "",PTR2UV(cur_eval));); | |
e7707071 YO |
4950 | if ( nochange_depth ) |
4951 | nochange_depth--; | |
4952 | ||
4953 | PUSH_YES_STATE_GOTO(EVAL_AB, | |
faec1544 DM |
4954 | st->u.eval.prev_eval->u.eval.B); /* match B */ |
4955 | } | |
4956 | ||
3b0527fe | 4957 | if (locinput < reginfo->till) { |
a3621e74 | 4958 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, |
7821416a IZ |
4959 | "%sMatch possible, but length=%ld is smaller than requested=%ld, failing!%s\n", |
4960 | PL_colors[4], | |
4961 | (long)(locinput - PL_reg_starttry), | |
3b0527fe | 4962 | (long)(reginfo->till - PL_reg_starttry), |
7821416a | 4963 | PL_colors[5])); |
58e23c8d | 4964 | |
262b90c4 | 4965 | sayNO_SILENT; /* Cannot match: too short. */ |
7821416a IZ |
4966 | } |
4967 | PL_reginput = locinput; /* put where regtry can find it */ | |
262b90c4 | 4968 | sayYES; /* Success! */ |
dad79028 DM |
4969 | |
4970 | case SUCCEED: /* successful SUSPEND/UNLESSM/IFMATCH/CURLYM */ | |
4971 | DEBUG_EXECUTE_r( | |
4972 | PerlIO_printf(Perl_debug_log, | |
4973 | "%*s %ssubpattern success...%s\n", | |
5bc10b2c | 4974 | REPORT_CODE_OFF+depth*2, "", PL_colors[4], PL_colors[5])); |
3280af22 | 4975 | PL_reginput = locinput; /* put where regtry can find it */ |
262b90c4 | 4976 | sayYES; /* Success! */ |
dad79028 | 4977 | |
40a82448 DM |
4978 | #undef ST |
4979 | #define ST st->u.ifmatch | |
4980 | ||
4981 | case SUSPEND: /* (?>A) */ | |
4982 | ST.wanted = 1; | |
9fe1d20c | 4983 | PL_reginput = locinput; |
9041c2e3 | 4984 | goto do_ifmatch; |
dad79028 | 4985 | |
40a82448 DM |
4986 | case UNLESSM: /* -ve lookaround: (?!A), or with flags, (?<!A) */ |
4987 | ST.wanted = 0; | |
dad79028 DM |
4988 | goto ifmatch_trivial_fail_test; |
4989 | ||
40a82448 DM |
4990 | case IFMATCH: /* +ve lookaround: (?=A), or with flags, (?<=A) */ |
4991 | ST.wanted = 1; | |
dad79028 | 4992 | ifmatch_trivial_fail_test: |
a0ed51b3 | 4993 | if (scan->flags) { |
52657f30 | 4994 | char * const s = HOPBACKc(locinput, scan->flags); |
dad79028 DM |
4995 | if (!s) { |
4996 | /* trivial fail */ | |
24d3c4a9 DM |
4997 | if (logical) { |
4998 | logical = 0; | |
4999 | sw = 1 - (bool)ST.wanted; | |
dad79028 | 5000 | } |
40a82448 | 5001 | else if (ST.wanted) |
dad79028 DM |
5002 | sayNO; |
5003 | next = scan + ARG(scan); | |
5004 | if (next == scan) | |
5005 | next = NULL; | |
5006 | break; | |
5007 | } | |
efb30f32 | 5008 | PL_reginput = s; |
a0ed51b3 LW |
5009 | } |
5010 | else | |
5011 | PL_reginput = locinput; | |
5012 | ||
c277df42 | 5013 | do_ifmatch: |
40a82448 | 5014 | ST.me = scan; |
24d3c4a9 | 5015 | ST.logical = logical; |
24d786f4 YO |
5016 | logical = 0; /* XXX: reset state of logical once it has been saved into ST */ |
5017 | ||
40a82448 DM |
5018 | /* execute body of (?...A) */ |
5019 | PUSH_YES_STATE_GOTO(IFMATCH_A, NEXTOPER(NEXTOPER(scan))); | |
5020 | /* NOTREACHED */ | |
5021 | ||
5022 | case IFMATCH_A_fail: /* body of (?...A) failed */ | |
5023 | ST.wanted = !ST.wanted; | |
5024 | /* FALL THROUGH */ | |
5025 | ||
5026 | case IFMATCH_A: /* body of (?...A) succeeded */ | |
24d3c4a9 DM |
5027 | if (ST.logical) { |
5028 | sw = (bool)ST.wanted; | |
40a82448 DM |
5029 | } |
5030 | else if (!ST.wanted) | |
5031 | sayNO; | |
5032 | ||
5033 | if (OP(ST.me) == SUSPEND) | |
5034 | locinput = PL_reginput; | |
5035 | else { | |
5036 | locinput = PL_reginput = st->locinput; | |
5037 | nextchr = UCHARAT(locinput); | |
5038 | } | |
5039 | scan = ST.me + ARG(ST.me); | |
5040 | if (scan == ST.me) | |
5041 | scan = NULL; | |
5042 | continue; /* execute B */ | |
5043 | ||
5044 | #undef ST | |
dad79028 | 5045 | |
c277df42 | 5046 | case LONGJMP: |
c277df42 IZ |
5047 | next = scan + ARG(scan); |
5048 | if (next == scan) | |
5049 | next = NULL; | |
a0d0e21e | 5050 | break; |
54612592 | 5051 | case COMMIT: |
e2e6a0f1 YO |
5052 | reginfo->cutpoint = PL_regeol; |
5053 | /* FALLTHROUGH */ | |
5d458dd8 | 5054 | case PRUNE: |
24b23f37 | 5055 | PL_reginput = locinput; |
e2e6a0f1 | 5056 | if (!scan->flags) |
ad64d0ec | 5057 | sv_yes_mark = sv_commit = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]); |
54612592 YO |
5058 | PUSH_STATE_GOTO(COMMIT_next,next); |
5059 | /* NOTREACHED */ | |
5060 | case COMMIT_next_fail: | |
5061 | no_final = 1; | |
5062 | /* FALLTHROUGH */ | |
7f69552c YO |
5063 | case OPFAIL: |
5064 | sayNO; | |
e2e6a0f1 YO |
5065 | /* NOTREACHED */ |
5066 | ||
5067 | #define ST st->u.mark | |
5068 | case MARKPOINT: | |
5069 | ST.prev_mark = mark_state; | |
5d458dd8 | 5070 | ST.mark_name = sv_commit = sv_yes_mark |
ad64d0ec | 5071 | = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]); |
e2e6a0f1 YO |
5072 | mark_state = st; |
5073 | ST.mark_loc = PL_reginput = locinput; | |
5074 | PUSH_YES_STATE_GOTO(MARKPOINT_next,next); | |
5075 | /* NOTREACHED */ | |
5076 | case MARKPOINT_next: | |
5077 | mark_state = ST.prev_mark; | |
5078 | sayYES; | |
5079 | /* NOTREACHED */ | |
5080 | case MARKPOINT_next_fail: | |
5d458dd8 | 5081 | if (popmark && sv_eq(ST.mark_name,popmark)) |
e2e6a0f1 YO |
5082 | { |
5083 | if (ST.mark_loc > startpoint) | |
5084 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
5085 | popmark = NULL; /* we found our mark */ | |
5086 | sv_commit = ST.mark_name; | |
5087 | ||
5088 | DEBUG_EXECUTE_r({ | |
5d458dd8 | 5089 | PerlIO_printf(Perl_debug_log, |
e2e6a0f1 YO |
5090 | "%*s %ssetting cutpoint to mark:%"SVf"...%s\n", |
5091 | REPORT_CODE_OFF+depth*2, "", | |
be2597df | 5092 | PL_colors[4], SVfARG(sv_commit), PL_colors[5]); |
e2e6a0f1 YO |
5093 | }); |
5094 | } | |
5095 | mark_state = ST.prev_mark; | |
5d458dd8 YO |
5096 | sv_yes_mark = mark_state ? |
5097 | mark_state->u.mark.mark_name : NULL; | |
e2e6a0f1 YO |
5098 | sayNO; |
5099 | /* NOTREACHED */ | |
5d458dd8 YO |
5100 | case SKIP: |
5101 | PL_reginput = locinput; | |
5102 | if (scan->flags) { | |
2bf803e2 | 5103 | /* (*SKIP) : if we fail we cut here*/ |
5d458dd8 | 5104 | ST.mark_name = NULL; |
e2e6a0f1 | 5105 | ST.mark_loc = locinput; |
5d458dd8 YO |
5106 | PUSH_STATE_GOTO(SKIP_next,next); |
5107 | } else { | |
2bf803e2 | 5108 | /* (*SKIP:NAME) : if there is a (*MARK:NAME) fail where it was, |
5d458dd8 YO |
5109 | otherwise do nothing. Meaning we need to scan |
5110 | */ | |
5111 | regmatch_state *cur = mark_state; | |
ad64d0ec | 5112 | SV *find = MUTABLE_SV(rexi->data->data[ ARG( scan ) ]); |
5d458dd8 YO |
5113 | |
5114 | while (cur) { | |
5115 | if ( sv_eq( cur->u.mark.mark_name, | |
5116 | find ) ) | |
5117 | { | |
5118 | ST.mark_name = find; | |
5119 | PUSH_STATE_GOTO( SKIP_next, next ); | |
5120 | } | |
5121 | cur = cur->u.mark.prev_mark; | |
5122 | } | |
e2e6a0f1 | 5123 | } |
2bf803e2 | 5124 | /* Didn't find our (*MARK:NAME) so ignore this (*SKIP:NAME) */ |
5d458dd8 YO |
5125 | break; |
5126 | case SKIP_next_fail: | |
5127 | if (ST.mark_name) { | |
5128 | /* (*CUT:NAME) - Set up to search for the name as we | |
5129 | collapse the stack*/ | |
5130 | popmark = ST.mark_name; | |
5131 | } else { | |
5132 | /* (*CUT) - No name, we cut here.*/ | |
e2e6a0f1 YO |
5133 | if (ST.mark_loc > startpoint) |
5134 | reginfo->cutpoint = HOPBACKc(ST.mark_loc, 1); | |
5d458dd8 YO |
5135 | /* but we set sv_commit to latest mark_name if there |
5136 | is one so they can test to see how things lead to this | |
5137 | cut */ | |
5138 | if (mark_state) | |
5139 | sv_commit=mark_state->u.mark.mark_name; | |
5140 | } | |
e2e6a0f1 YO |
5141 | no_final = 1; |
5142 | sayNO; | |
5143 | /* NOTREACHED */ | |
5144 | #undef ST | |
32e6a07c YO |
5145 | case FOLDCHAR: |
5146 | n = ARG(scan); | |
81d4fa0f | 5147 | if ( n == (U32)what_len_TRICKYFOLD(locinput,do_utf8,ln) ) { |
e64b1bd1 YO |
5148 | locinput += ln; |
5149 | } else if ( 0xDF == n && !do_utf8 && !UTF ) { | |
5150 | sayNO; | |
5151 | } else { | |
5152 | U8 folded[UTF8_MAXBYTES_CASE+1]; | |
5153 | STRLEN foldlen; | |
5154 | const char * const l = locinput; | |
5155 | char *e = PL_regeol; | |
5156 | to_uni_fold(n, folded, &foldlen); | |
5157 | ||
59fe32ea | 5158 | if (ibcmp_utf8((const char*) folded, 0, foldlen, 1, |
e64b1bd1 | 5159 | l, &e, 0, do_utf8)) { |
32e6a07c | 5160 | sayNO; |
e64b1bd1 YO |
5161 | } |
5162 | locinput = e; | |
32e6a07c YO |
5163 | } |
5164 | nextchr = UCHARAT(locinput); | |
5165 | break; | |
e1d1eefb YO |
5166 | case LNBREAK: |
5167 | if ((n=is_LNBREAK(locinput,do_utf8))) { | |
5168 | locinput += n; | |
5169 | nextchr = UCHARAT(locinput); | |
5170 | } else | |
5171 | sayNO; | |
5172 | break; | |
5173 | ||
5174 | #define CASE_CLASS(nAmE) \ | |
5175 | case nAmE: \ | |
5176 | if ((n=is_##nAmE(locinput,do_utf8))) { \ | |
5177 | locinput += n; \ | |
5178 | nextchr = UCHARAT(locinput); \ | |
5179 | } else \ | |
5180 | sayNO; \ | |
5181 | break; \ | |
5182 | case N##nAmE: \ | |
5183 | if ((n=is_##nAmE(locinput,do_utf8))) { \ | |
5184 | sayNO; \ | |
5185 | } else { \ | |
5186 | locinput += UTF8SKIP(locinput); \ | |
5187 | nextchr = UCHARAT(locinput); \ | |
5188 | } \ | |
5189 | break | |
5190 | ||
5191 | CASE_CLASS(VERTWS); | |
5192 | CASE_CLASS(HORIZWS); | |
5193 | #undef CASE_CLASS | |
5194 | ||
a0d0e21e | 5195 | default: |
b900a521 | 5196 | PerlIO_printf(Perl_error_log, "%"UVxf" %d\n", |
d7d93a81 | 5197 | PTR2UV(scan), OP(scan)); |
cea2e8a9 | 5198 | Perl_croak(aTHX_ "regexp memory corruption"); |
5d458dd8 YO |
5199 | |
5200 | } /* end switch */ | |
95b24440 | 5201 | |
5d458dd8 YO |
5202 | /* switch break jumps here */ |
5203 | scan = next; /* prepare to execute the next op and ... */ | |
5204 | continue; /* ... jump back to the top, reusing st */ | |
95b24440 DM |
5205 | /* NOTREACHED */ |
5206 | ||
40a82448 DM |
5207 | push_yes_state: |
5208 | /* push a state that backtracks on success */ | |
5209 | st->u.yes.prev_yes_state = yes_state; | |
5210 | yes_state = st; | |
5211 | /* FALL THROUGH */ | |
5212 | push_state: | |
5213 | /* push a new regex state, then continue at scan */ | |
5214 | { | |
5215 | regmatch_state *newst; | |
5216 | ||
24b23f37 YO |
5217 | DEBUG_STACK_r({ |
5218 | regmatch_state *cur = st; | |
5219 | regmatch_state *curyes = yes_state; | |
5220 | int curd = depth; | |
5221 | regmatch_slab *slab = PL_regmatch_slab; | |
5222 | for (;curd > -1;cur--,curd--) { | |
5223 | if (cur < SLAB_FIRST(slab)) { | |
5224 | slab = slab->prev; | |
5225 | cur = SLAB_LAST(slab); | |
5226 | } | |
5227 | PerlIO_printf(Perl_error_log, "%*s#%-3d %-10s %s\n", | |
5228 | REPORT_CODE_OFF + 2 + depth * 2,"", | |
13d6edb4 | 5229 | curd, PL_reg_name[cur->resume_state], |
24b23f37 YO |
5230 | (curyes == cur) ? "yes" : "" |
5231 | ); | |
5232 | if (curyes == cur) | |
5233 | curyes = cur->u.yes.prev_yes_state; | |
5234 | } | |
5235 | } else | |
5236 | DEBUG_STATE_pp("push") | |
5237 | ); | |
40a82448 | 5238 | depth++; |
40a82448 DM |
5239 | st->locinput = locinput; |
5240 | newst = st+1; | |
5241 | if (newst > SLAB_LAST(PL_regmatch_slab)) | |
5242 | newst = S_push_slab(aTHX); | |
5243 | PL_regmatch_state = newst; | |
786e8c11 | 5244 | |
40a82448 DM |
5245 | locinput = PL_reginput; |
5246 | nextchr = UCHARAT(locinput); | |
5247 | st = newst; | |
5248 | continue; | |
5249 | /* NOTREACHED */ | |
5250 | } | |
a0d0e21e | 5251 | } |
a687059c | 5252 | |
a0d0e21e LW |
5253 | /* |
5254 | * We get here only if there's trouble -- normally "case END" is | |
5255 | * the terminating point. | |
5256 | */ | |
cea2e8a9 | 5257 | Perl_croak(aTHX_ "corrupted regexp pointers"); |
a0d0e21e | 5258 | /*NOTREACHED*/ |
4633a7c4 LW |
5259 | sayNO; |
5260 | ||
262b90c4 | 5261 | yes: |
77cb431f DM |
5262 | if (yes_state) { |
5263 | /* we have successfully completed a subexpression, but we must now | |
5264 | * pop to the state marked by yes_state and continue from there */ | |
77cb431f | 5265 | assert(st != yes_state); |
5bc10b2c DM |
5266 | #ifdef DEBUGGING |
5267 | while (st != yes_state) { | |
5268 | st--; | |
5269 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
5270 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5271 | st = SLAB_LAST(PL_regmatch_slab); | |
5272 | } | |
e2e6a0f1 | 5273 | DEBUG_STATE_r({ |
54612592 YO |
5274 | if (no_final) { |
5275 | DEBUG_STATE_pp("pop (no final)"); | |
5276 | } else { | |
5277 | DEBUG_STATE_pp("pop (yes)"); | |
5278 | } | |
e2e6a0f1 | 5279 | }); |
5bc10b2c DM |
5280 | depth--; |
5281 | } | |
5282 | #else | |
77cb431f DM |
5283 | while (yes_state < SLAB_FIRST(PL_regmatch_slab) |
5284 | || yes_state > SLAB_LAST(PL_regmatch_slab)) | |
5285 | { | |
5286 | /* not in this slab, pop slab */ | |
5287 | depth -= (st - SLAB_FIRST(PL_regmatch_slab) + 1); | |
5288 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5289 | st = SLAB_LAST(PL_regmatch_slab); | |
5290 | } | |
5291 | depth -= (st - yes_state); | |
5bc10b2c | 5292 | #endif |
77cb431f DM |
5293 | st = yes_state; |
5294 | yes_state = st->u.yes.prev_yes_state; | |
5295 | PL_regmatch_state = st; | |
24b23f37 | 5296 | |
5d458dd8 YO |
5297 | if (no_final) { |
5298 | locinput= st->locinput; | |
5299 | nextchr = UCHARAT(locinput); | |
5300 | } | |
54612592 | 5301 | state_num = st->resume_state + no_final; |
24d3c4a9 | 5302 | goto reenter_switch; |
77cb431f DM |
5303 | } |
5304 | ||
a3621e74 | 5305 | DEBUG_EXECUTE_r(PerlIO_printf(Perl_debug_log, "%sMatch successful!%s\n", |
e4584336 | 5306 | PL_colors[4], PL_colors[5])); |
02db2b7b | 5307 | |
19b95bf0 DM |
5308 | if (PL_reg_eval_set) { |
5309 | /* each successfully executed (?{...}) block does the equivalent of | |
5310 | * local $^R = do {...} | |
5311 | * When popping the save stack, all these locals would be undone; | |
5312 | * bypass this by setting the outermost saved $^R to the latest | |
5313 | * value */ | |
5314 | if (oreplsv != GvSV(PL_replgv)) | |
5315 | sv_setsv(oreplsv, GvSV(PL_replgv)); | |
5316 | } | |
95b24440 | 5317 | result = 1; |
aa283a38 | 5318 | goto final_exit; |
4633a7c4 LW |
5319 | |
5320 | no: | |
a3621e74 | 5321 | DEBUG_EXECUTE_r( |
7821416a | 5322 | PerlIO_printf(Perl_debug_log, |
786e8c11 | 5323 | "%*s %sfailed...%s\n", |
5bc10b2c | 5324 | REPORT_CODE_OFF+depth*2, "", |
786e8c11 | 5325 | PL_colors[4], PL_colors[5]) |
7821416a | 5326 | ); |
aa283a38 | 5327 | |
262b90c4 | 5328 | no_silent: |
54612592 YO |
5329 | if (no_final) { |
5330 | if (yes_state) { | |
5331 | goto yes; | |
5332 | } else { | |
5333 | goto final_exit; | |
5334 | } | |
5335 | } | |
aa283a38 DM |
5336 | if (depth) { |
5337 | /* there's a previous state to backtrack to */ | |
40a82448 DM |
5338 | st--; |
5339 | if (st < SLAB_FIRST(PL_regmatch_slab)) { | |
5340 | PL_regmatch_slab = PL_regmatch_slab->prev; | |
5341 | st = SLAB_LAST(PL_regmatch_slab); | |
5342 | } | |
5343 | PL_regmatch_state = st; | |
40a82448 DM |
5344 | locinput= st->locinput; |
5345 | nextchr = UCHARAT(locinput); | |
5346 | ||
5bc10b2c DM |
5347 | DEBUG_STATE_pp("pop"); |
5348 | depth--; | |
262b90c4 DM |
5349 | if (yes_state == st) |
5350 | yes_state = st->u.yes.prev_yes_state; | |
5bc10b2c | 5351 | |
24d3c4a9 DM |
5352 | state_num = st->resume_state + 1; /* failure = success + 1 */ |
5353 | goto reenter_switch; | |
95b24440 | 5354 | } |
24d3c4a9 | 5355 | result = 0; |
aa283a38 | 5356 | |
262b90c4 | 5357 | final_exit: |
bbe252da | 5358 | if (rex->intflags & PREGf_VERBARG_SEEN) { |
5d458dd8 YO |
5359 | SV *sv_err = get_sv("REGERROR", 1); |
5360 | SV *sv_mrk = get_sv("REGMARK", 1); | |
5361 | if (result) { | |
e2e6a0f1 | 5362 | sv_commit = &PL_sv_no; |
5d458dd8 YO |
5363 | if (!sv_yes_mark) |
5364 | sv_yes_mark = &PL_sv_yes; | |
5365 | } else { | |
5366 | if (!sv_commit) | |
5367 | sv_commit = &PL_sv_yes; | |
5368 | sv_yes_mark = &PL_sv_no; | |
5369 | } | |
5370 | sv_setsv(sv_err, sv_commit); | |
5371 | sv_setsv(sv_mrk, sv_yes_mark); | |
e2e6a0f1 | 5372 | } |
19b95bf0 | 5373 | |
2f554ef7 DM |
5374 | /* clean up; in particular, free all slabs above current one */ |
5375 | LEAVE_SCOPE(oldsave); | |
5d9a96ca | 5376 | |
95b24440 | 5377 | return result; |
a687059c LW |
5378 | } |
5379 | ||
5380 | /* | |
5381 | - regrepeat - repeatedly match something simple, report how many | |
5382 | */ | |
5383 | /* | |
5384 | * [This routine now assumes that it will only match on things of length 1. | |
5385 | * That was true before, but now we assume scan - reginput is the count, | |
a0ed51b3 | 5386 | * rather than incrementing count on every character. [Er, except utf8.]] |
a687059c | 5387 | */ |
76e3520e | 5388 | STATIC I32 |
e2e6a0f1 | 5389 | S_regrepeat(pTHX_ const regexp *prog, const regnode *p, I32 max, int depth) |
a687059c | 5390 | { |
27da23d5 | 5391 | dVAR; |
a0d0e21e | 5392 | register char *scan; |
a0d0e21e | 5393 | register I32 c; |
3280af22 | 5394 | register char *loceol = PL_regeol; |
a0ed51b3 | 5395 | register I32 hardcount = 0; |
53c4c00c | 5396 | register bool do_utf8 = PL_reg_match_utf8; |
4f55667c SP |
5397 | #ifndef DEBUGGING |
5398 | PERL_UNUSED_ARG(depth); | |
5399 | #endif | |
a0d0e21e | 5400 | |
7918f24d NC |
5401 | PERL_ARGS_ASSERT_REGREPEAT; |
5402 | ||
3280af22 | 5403 | scan = PL_reginput; |
faf11cac HS |
5404 | if (max == REG_INFTY) |
5405 | max = I32_MAX; | |
5406 | else if (max < loceol - scan) | |
7f596f4c | 5407 | loceol = scan + max; |
a0d0e21e | 5408 | switch (OP(p)) { |
22c35a8c | 5409 | case REG_ANY: |
1aa99e6b | 5410 | if (do_utf8) { |
ffc61ed2 | 5411 | loceol = PL_regeol; |
1aa99e6b | 5412 | while (scan < loceol && hardcount < max && *scan != '\n') { |
ffc61ed2 JH |
5413 | scan += UTF8SKIP(scan); |
5414 | hardcount++; | |
5415 | } | |
5416 | } else { | |
5417 | while (scan < loceol && *scan != '\n') | |
5418 | scan++; | |
a0ed51b3 LW |
5419 | } |
5420 | break; | |
ffc61ed2 | 5421 | case SANY: |
def8e4ea JH |
5422 | if (do_utf8) { |
5423 | loceol = PL_regeol; | |
a0804c9e | 5424 | while (scan < loceol && hardcount < max) { |
def8e4ea JH |
5425 | scan += UTF8SKIP(scan); |
5426 | hardcount++; | |
5427 | } | |
5428 | } | |
5429 | else | |
5430 | scan = loceol; | |
a0ed51b3 | 5431 | break; |
f33976b4 DB |
5432 | case CANY: |
5433 | scan = loceol; | |
5434 | break; | |
090f7165 JH |
5435 | case EXACT: /* length of string is 1 */ |
5436 | c = (U8)*STRING(p); | |
5437 | while (scan < loceol && UCHARAT(scan) == c) | |
5438 | scan++; | |
bbce6d69 | 5439 | break; |
5440 | case EXACTF: /* length of string is 1 */ | |
cd439c50 | 5441 | c = (U8)*STRING(p); |
bbce6d69 | 5442 | while (scan < loceol && |
22c35a8c | 5443 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold[c])) |
bbce6d69 | 5444 | scan++; |
5445 | break; | |
5446 | case EXACTFL: /* length of string is 1 */ | |
3280af22 | 5447 | PL_reg_flags |= RF_tainted; |
cd439c50 | 5448 | c = (U8)*STRING(p); |
bbce6d69 | 5449 | while (scan < loceol && |
22c35a8c | 5450 | (UCHARAT(scan) == c || UCHARAT(scan) == PL_fold_locale[c])) |
a0d0e21e LW |
5451 | scan++; |
5452 | break; | |
5453 | case ANYOF: | |
ffc61ed2 JH |
5454 | if (do_utf8) { |
5455 | loceol = PL_regeol; | |
cfc92286 | 5456 | while (hardcount < max && scan < loceol && |
32fc9b6a | 5457 | reginclass(prog, p, (U8*)scan, 0, do_utf8)) { |
ffc61ed2 JH |
5458 | scan += UTF8SKIP(scan); |
5459 | hardcount++; | |
5460 | } | |
5461 | } else { | |
32fc9b6a | 5462 | while (scan < loceol && REGINCLASS(prog, p, (U8*)scan)) |
ffc61ed2 JH |
5463 | scan++; |
5464 | } | |
a0d0e21e LW |
5465 | break; |
5466 | case ALNUM: | |
1aa99e6b | 5467 | if (do_utf8) { |
ffc61ed2 | 5468 | loceol = PL_regeol; |
1a4fad37 | 5469 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5470 | while (hardcount < max && scan < loceol && |
3568d838 | 5471 | swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5472 | scan += UTF8SKIP(scan); |
5473 | hardcount++; | |
5474 | } | |
5475 | } else { | |
5476 | while (scan < loceol && isALNUM(*scan)) | |
5477 | scan++; | |
a0ed51b3 LW |
5478 | } |
5479 | break; | |
bbce6d69 | 5480 | case ALNUML: |
3280af22 | 5481 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5482 | if (do_utf8) { |
ffc61ed2 | 5483 | loceol = PL_regeol; |
1aa99e6b IH |
5484 | while (hardcount < max && scan < loceol && |
5485 | isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5486 | scan += UTF8SKIP(scan); |
5487 | hardcount++; | |
5488 | } | |
5489 | } else { | |
5490 | while (scan < loceol && isALNUM_LC(*scan)) | |
5491 | scan++; | |
a0ed51b3 LW |
5492 | } |
5493 | break; | |
a0d0e21e | 5494 | case NALNUM: |
1aa99e6b | 5495 | if (do_utf8) { |
ffc61ed2 | 5496 | loceol = PL_regeol; |
1a4fad37 | 5497 | LOAD_UTF8_CHARCLASS_ALNUM(); |
1aa99e6b | 5498 | while (hardcount < max && scan < loceol && |
3568d838 | 5499 | !swash_fetch(PL_utf8_alnum, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5500 | scan += UTF8SKIP(scan); |
5501 | hardcount++; | |
5502 | } | |
5503 | } else { | |
5504 | while (scan < loceol && !isALNUM(*scan)) | |
5505 | scan++; | |
a0ed51b3 LW |
5506 | } |
5507 | break; | |
bbce6d69 | 5508 | case NALNUML: |
3280af22 | 5509 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5510 | if (do_utf8) { |
ffc61ed2 | 5511 | loceol = PL_regeol; |
1aa99e6b IH |
5512 | while (hardcount < max && scan < loceol && |
5513 | !isALNUM_LC_utf8((U8*)scan)) { | |
ffc61ed2 JH |
5514 | scan += UTF8SKIP(scan); |
5515 | hardcount++; | |
5516 | } | |
5517 | } else { | |
5518 | while (scan < loceol && !isALNUM_LC(*scan)) | |
5519 | scan++; | |
a0ed51b3 LW |
5520 | } |
5521 | break; | |
a0d0e21e | 5522 | case SPACE: |
1aa99e6b | 5523 | if (do_utf8) { |
ffc61ed2 | 5524 | loceol = PL_regeol; |
1a4fad37 | 5525 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5526 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5527 | (*scan == ' ' || |
5528 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5529 | scan += UTF8SKIP(scan); |
5530 | hardcount++; | |
5531 | } | |
5532 | } else { | |
5533 | while (scan < loceol && isSPACE(*scan)) | |
5534 | scan++; | |
a0ed51b3 LW |
5535 | } |
5536 | break; | |
bbce6d69 | 5537 | case SPACEL: |
3280af22 | 5538 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5539 | if (do_utf8) { |
ffc61ed2 | 5540 | loceol = PL_regeol; |
1aa99e6b | 5541 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5542 | (*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5543 | scan += UTF8SKIP(scan); | |
5544 | hardcount++; | |
5545 | } | |
5546 | } else { | |
5547 | while (scan < loceol && isSPACE_LC(*scan)) | |
5548 | scan++; | |
a0ed51b3 LW |
5549 | } |
5550 | break; | |
a0d0e21e | 5551 | case NSPACE: |
1aa99e6b | 5552 | if (do_utf8) { |
ffc61ed2 | 5553 | loceol = PL_regeol; |
1a4fad37 | 5554 | LOAD_UTF8_CHARCLASS_SPACE(); |
1aa99e6b | 5555 | while (hardcount < max && scan < loceol && |
3568d838 JH |
5556 | !(*scan == ' ' || |
5557 | swash_fetch(PL_utf8_space,(U8*)scan, do_utf8))) { | |
ffc61ed2 JH |
5558 | scan += UTF8SKIP(scan); |
5559 | hardcount++; | |
5560 | } | |
5561 | } else { | |
5562 | while (scan < loceol && !isSPACE(*scan)) | |
5563 | scan++; | |
a0ed51b3 | 5564 | } |
0008a298 | 5565 | break; |
bbce6d69 | 5566 | case NSPACEL: |
3280af22 | 5567 | PL_reg_flags |= RF_tainted; |
1aa99e6b | 5568 | if (do_utf8) { |
ffc61ed2 | 5569 | loceol = PL_regeol; |
1aa99e6b | 5570 | while (hardcount < max && scan < loceol && |
ffc61ed2 JH |
5571 | !(*scan == ' ' || isSPACE_LC_utf8((U8*)scan))) { |
5572 | scan += UTF8SKIP(scan); | |
5573 | hardcount++; | |
5574 | } | |
5575 | } else { | |
5576 | while (scan < loceol && !isSPACE_LC(*scan)) | |
5577 | scan++; | |
a0ed51b3 LW |
5578 | } |
5579 | break; | |
a0d0e21e | 5580 | case DIGIT: |
1aa99e6b | 5581 | if (do_utf8) { |
ffc61ed2 | 5582 | loceol = PL_regeol; |
1a4fad37 | 5583 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5584 | while (hardcount < max && scan < loceol && |
3568d838 | 5585 | swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5586 | scan += UTF8SKIP(scan); |
5587 | hardcount++; | |
5588 | } | |
5589 | } else { | |
5590 | while (scan < loceol && isDIGIT(*scan)) | |
5591 | scan++; | |
a0ed51b3 LW |
5592 | } |
5593 | break; | |
a0d0e21e | 5594 | case NDIGIT: |
1aa99e6b | 5595 | if (do_utf8) { |
ffc61ed2 | 5596 | loceol = PL_regeol; |
1a4fad37 | 5597 | LOAD_UTF8_CHARCLASS_DIGIT(); |
1aa99e6b | 5598 | while (hardcount < max && scan < loceol && |
3568d838 | 5599 | !swash_fetch(PL_utf8_digit, (U8*)scan, do_utf8)) { |
ffc61ed2 JH |
5600 | scan += UTF8SKIP(scan); |
5601 | hardcount++; | |
5602 | } | |
5603 | } else { | |
5604 | while (scan < loceol && !isDIGIT(*scan)) | |
5605 | scan++; | |
a0ed51b3 | 5606 | } |
e1d1eefb YO |
5607 | case LNBREAK: |
5608 | if (do_utf8) { | |
5609 | loceol = PL_regeol; | |
5610 | while (hardcount < max && scan < loceol && (c=is_LNBREAK_utf8(scan))) { | |
5611 | scan += c; | |
5612 | hardcount++; | |
5613 | } | |
5614 | } else { | |
5615 | /* | |
5616 | LNBREAK can match two latin chars, which is ok, | |
5617 | because we have a null terminated string, but we | |
5618 | have to use hardcount in this situation | |
5619 | */ | |
5620 | while (scan < loceol && (c=is_LNBREAK_latin1(scan))) { | |
5621 | scan+=c; | |
5622 | hardcount++; | |
5623 | } | |
5624 | } | |
5625 | break; | |
5626 | case HORIZWS: | |
5627 | if (do_utf8) { | |
5628 | loceol = PL_regeol; | |
5629 | while (hardcount < max && scan < loceol && (c=is_HORIZWS_utf8(scan))) { | |
5630 | scan += c; | |
5631 | hardcount++; | |
5632 | } | |
5633 | } else { | |
5634 | while (scan < loceol && is_HORIZWS_latin1(scan)) | |
5635 | scan++; | |
5636 | } | |
a0ed51b3 | 5637 | break; |
e1d1eefb YO |
5638 | case NHORIZWS: |
5639 | if (do_utf8) { | |
5640 | loceol = PL_regeol; | |
5641 | while (hardcount < max && scan < loceol && !is_HORIZWS_utf8(scan)) { | |
5642 | scan += UTF8SKIP(scan); | |
5643 | hardcount++; | |
5644 | } | |
5645 | } else { | |
5646 | while (scan < loceol && !is_HORIZWS_latin1(scan)) | |
5647 | scan++; | |
5648 | ||
5649 | } | |
5650 | break; | |
5651 | case VERTWS: | |
5652 | if (do_utf8) { | |
5653 | loceol = PL_regeol; | |
5654 | while (hardcount < max && scan < loceol && (c=is_VERTWS_utf8(scan))) { | |
5655 | scan += c; | |
5656 | hardcount++; | |
5657 | } | |
5658 | } else { | |
5659 | while (scan < loceol && is_VERTWS_latin1(scan)) | |
5660 | scan++; | |
5661 | ||
5662 | } | |
5663 | break; | |
5664 | case NVERTWS: | |
5665 | if (do_utf8) { | |
5666 | loceol = PL_regeol; | |
5667 | while (hardcount < max && scan < loceol && !is_VERTWS_utf8(scan)) { | |
5668 | scan += UTF8SKIP(scan); | |
5669 | hardcount++; | |
5670 | } | |
5671 | } else { | |
5672 | while (scan < loceol && !is_VERTWS_latin1(scan)) | |
5673 | scan++; | |
5674 | ||
5675 | } | |
5676 | break; | |
5677 | ||
a0d0e21e LW |
5678 | default: /* Called on something of 0 width. */ |
5679 | break; /* So match right here or not at all. */ | |
5680 | } | |
a687059c | 5681 | |
a0ed51b3 LW |
5682 | if (hardcount) |
5683 | c = hardcount; | |
5684 | else | |
5685 | c = scan - PL_reginput; | |
3280af22 | 5686 | PL_reginput = scan; |
a687059c | 5687 | |
a3621e74 | 5688 | DEBUG_r({ |
e68ec53f | 5689 | GET_RE_DEBUG_FLAGS_DECL; |
be8e71aa | 5690 | DEBUG_EXECUTE_r({ |
e68ec53f YO |
5691 | SV * const prop = sv_newmortal(); |
5692 | regprop(prog, prop, p); | |
5693 | PerlIO_printf(Perl_debug_log, | |
be8e71aa | 5694 | "%*s %s can match %"IVdf" times out of %"IVdf"...\n", |
e2e6a0f1 | 5695 | REPORT_CODE_OFF + depth*2, "", SvPVX_const(prop),(IV)c,(IV)max); |
a3621e74 | 5696 | }); |
be8e71aa | 5697 | }); |
9041c2e3 | 5698 | |
a0d0e21e | 5699 | return(c); |
a687059c LW |
5700 | } |
5701 | ||
c277df42 | 5702 | |
be8e71aa | 5703 | #if !defined(PERL_IN_XSUB_RE) || defined(PLUGGABLE_RE_EXTENSION) |
c277df42 | 5704 | /* |
ffc61ed2 JH |
5705 | - regclass_swash - prepare the utf8 swash |
5706 | */ | |
5707 | ||
5708 | SV * | |
32fc9b6a | 5709 | Perl_regclass_swash(pTHX_ const regexp *prog, register const regnode* node, bool doinit, SV** listsvp, SV **altsvp) |
ffc61ed2 | 5710 | { |
97aff369 | 5711 | dVAR; |
9e55ce06 JH |
5712 | SV *sw = NULL; |
5713 | SV *si = NULL; | |
5714 | SV *alt = NULL; | |
f8fc2ecf YO |
5715 | RXi_GET_DECL(prog,progi); |
5716 | const struct reg_data * const data = prog ? progi->data : NULL; | |
ffc61ed2 | 5717 | |
7918f24d NC |
5718 | PERL_ARGS_ASSERT_REGCLASS_SWASH; |
5719 | ||
4f639d21 | 5720 | if (data && data->count) { |
a3b680e6 | 5721 | const U32 n = ARG(node); |
ffc61ed2 | 5722 | |
4f639d21 | 5723 | if (data->what[n] == 's') { |
ad64d0ec NC |
5724 | SV * const rv = MUTABLE_SV(data->data[n]); |
5725 | AV * const av = MUTABLE_AV(SvRV(rv)); | |
2d03de9c | 5726 | SV **const ary = AvARRAY(av); |
9e55ce06 | 5727 | SV **a, **b; |
9041c2e3 | 5728 | |
711a919c | 5729 | /* See the end of regcomp.c:S_regclass() for |
9e55ce06 JH |
5730 | * documentation of these array elements. */ |
5731 | ||
b11f357e | 5732 | si = *ary; |
fe5bfecd JH |
5733 | a = SvROK(ary[1]) ? &ary[1] : NULL; |
5734 | b = SvTYPE(ary[2]) == SVt_PVAV ? &ary[2] : NULL; | |
b11f357e | 5735 | |
ffc61ed2 JH |
5736 | if (a) |
5737 | sw = *a; | |
5738 | else if (si && doinit) { | |
5739 | sw = swash_init("utf8", "", si, 1, 0); | |
5740 | (void)av_store(av, 1, sw); | |
5741 | } | |
9e55ce06 JH |
5742 | if (b) |
5743 | alt = *b; | |
ffc61ed2 JH |
5744 | } |
5745 | } | |
5746 | ||
9e55ce06 JH |
5747 | if (listsvp) |
5748 | *listsvp = si; | |
5749 | if (altsvp) | |
5750 | *altsvp = alt; | |
ffc61ed2 JH |
5751 | |
5752 | return sw; | |
5753 | } | |
76234dfb | 5754 | #endif |
ffc61ed2 JH |
5755 | |
5756 | /* | |
ba7b4546 | 5757 | - reginclass - determine if a character falls into a character class |
832705d4 JH |
5758 | |
5759 | The n is the ANYOF regnode, the p is the target string, lenp | |
5760 | is pointer to the maximum length of how far to go in the p | |
5761 | (if the lenp is zero, UTF8SKIP(p) is used), | |
5762 | do_utf8 tells whether the target string is in UTF-8. | |
5763 | ||
bbce6d69 | 5764 | */ |
5765 | ||
76e3520e | 5766 | STATIC bool |
32fc9b6a | 5767 | S_reginclass(pTHX_ const regexp *prog, register const regnode *n, register const U8* p, STRLEN* lenp, register bool do_utf8) |
bbce6d69 | 5768 | { |
27da23d5 | 5769 | dVAR; |
a3b680e6 | 5770 | const char flags = ANYOF_FLAGS(n); |
bbce6d69 | 5771 | bool match = FALSE; |
cc07378b | 5772 | UV c = *p; |
ae9ddab8 | 5773 | STRLEN len = 0; |
9e55ce06 | 5774 | STRLEN plen; |
1aa99e6b | 5775 | |
7918f24d NC |
5776 | PERL_ARGS_ASSERT_REGINCLASS; |
5777 | ||
19f67299 TS |
5778 | if (do_utf8 && !UTF8_IS_INVARIANT(c)) { |
5779 | c = utf8n_to_uvchr(p, UTF8_MAXBYTES, &len, | |
4ad0818d DM |
5780 | (UTF8_ALLOW_DEFAULT & UTF8_ALLOW_ANYUV) | UTF8_CHECK_ONLY); |
5781 | /* see [perl #37836] for UTF8_ALLOW_ANYUV */ | |
e8a70c6f SP |
5782 | if (len == (STRLEN)-1) |
5783 | Perl_croak(aTHX_ "Malformed UTF-8 character (fatal)"); | |
19f67299 | 5784 | } |
bbce6d69 | 5785 | |
0f0076b4 | 5786 | plen = lenp ? *lenp : UNISKIP(NATIVE_TO_UNI(c)); |
ffc61ed2 | 5787 | if (do_utf8 || (flags & ANYOF_UNICODE)) { |
9e55ce06 JH |
5788 | if (lenp) |
5789 | *lenp = 0; | |
ffc61ed2 | 5790 | if (do_utf8 && !ANYOF_RUNTIME(n)) { |
ffc61ed2 JH |
5791 | if (len != (STRLEN)-1 && c < 256 && ANYOF_BITMAP_TEST(n, c)) |
5792 | match = TRUE; | |
bbce6d69 | 5793 | } |
3568d838 | 5794 | if (!match && do_utf8 && (flags & ANYOF_UNICODE_ALL) && c >= 256) |
1aa99e6b | 5795 | match = TRUE; |
ffc61ed2 | 5796 | if (!match) { |
9e55ce06 | 5797 | AV *av; |
32fc9b6a | 5798 | SV * const sw = regclass_swash(prog, n, TRUE, 0, (SV**)&av); |
ffc61ed2 JH |
5799 | |
5800 | if (sw) { | |
3f0c5693 KW |
5801 | U8 * utf8_p; |
5802 | if (do_utf8) { | |
5803 | utf8_p = (U8 *) p; | |
5804 | } else { | |
5805 | STRLEN len = 1; | |
5806 | utf8_p = bytes_to_utf8(p, &len); | |
5807 | } | |
5808 | if (swash_fetch(sw, utf8_p, 1)) | |
ffc61ed2 JH |
5809 | match = TRUE; |
5810 | else if (flags & ANYOF_FOLD) { | |
9e55ce06 JH |
5811 | if (!match && lenp && av) { |
5812 | I32 i; | |
9e55ce06 | 5813 | for (i = 0; i <= av_len(av); i++) { |
890ce7af | 5814 | SV* const sv = *av_fetch(av, i, FALSE); |
9e55ce06 | 5815 | STRLEN len; |
890ce7af | 5816 | const char * const s = SvPV_const(sv, len); |
3f0c5693 | 5817 | if (len <= plen && memEQ(s, (char*)utf8_p, len)) { |
9e55ce06 JH |
5818 | *lenp = len; |
5819 | match = TRUE; | |
5820 | break; | |
5821 | } | |
5822 | } | |
5823 | } | |
5824 | if (!match) { | |
89ebb4a3 | 5825 | U8 tmpbuf[UTF8_MAXBYTES_CASE+1]; |
4a623e43 | 5826 | |
3f0c5693 KW |
5827 | STRLEN tmplen; |
5828 | to_utf8_fold(utf8_p, tmpbuf, &tmplen); | |
5829 | if (swash_fetch(sw, tmpbuf, 1)) | |
9e55ce06 JH |
5830 | match = TRUE; |
5831 | } | |
ffc61ed2 | 5832 | } |
b3a04dd3 KW |
5833 | |
5834 | /* If we allocated a string above, free it */ | |
5835 | if (! do_utf8) Safefree(utf8_p); | |
ffc61ed2 | 5836 | } |
bbce6d69 | 5837 | } |
9e55ce06 | 5838 | if (match && lenp && *lenp == 0) |
0f0076b4 | 5839 | *lenp = UNISKIP(NATIVE_TO_UNI(c)); |
bbce6d69 | 5840 | } |
1aa99e6b | 5841 | if (!match && c < 256) { |
ffc61ed2 JH |
5842 | if (ANYOF_BITMAP_TEST(n, c)) |
5843 | match = TRUE; | |
5844 | else if (flags & ANYOF_FOLD) { | |
eb160463 | 5845 | U8 f; |
a0ed51b3 | 5846 | |
ffc61ed2 JH |
5847 | if (flags & ANYOF_LOCALE) { |
5848 | PL_reg_flags |= RF_tainted; | |
5849 | f = PL_fold_locale[c]; | |
5850 | } | |
5851 | else | |
5852 | f = PL_fold[c]; | |
5853 | if (f != c && ANYOF_BITMAP_TEST(n, f)) | |
5854 | match = TRUE; | |
5855 | } | |
5856 | ||
5857 | if (!match && (flags & ANYOF_CLASS)) { | |
a0ed51b3 | 5858 | PL_reg_flags |= RF_tainted; |
ffc61ed2 JH |
5859 | if ( |
5860 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUM) && isALNUM_LC(c)) || | |
5861 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUM) && !isALNUM_LC(c)) || | |
5862 | (ANYOF_CLASS_TEST(n, ANYOF_SPACE) && isSPACE_LC(c)) || | |
5863 | (ANYOF_CLASS_TEST(n, ANYOF_NSPACE) && !isSPACE_LC(c)) || | |
5864 | (ANYOF_CLASS_TEST(n, ANYOF_DIGIT) && isDIGIT_LC(c)) || | |
5865 | (ANYOF_CLASS_TEST(n, ANYOF_NDIGIT) && !isDIGIT_LC(c)) || | |
5866 | (ANYOF_CLASS_TEST(n, ANYOF_ALNUMC) && isALNUMC_LC(c)) || | |
5867 | (ANYOF_CLASS_TEST(n, ANYOF_NALNUMC) && !isALNUMC_LC(c)) || | |
5868 | (ANYOF_CLASS_TEST(n, ANYOF_ALPHA) && isALPHA_LC(c)) || | |
5869 | (ANYOF_CLASS_TEST(n, ANYOF_NALPHA) && !isALPHA_LC(c)) || | |
5870 | (ANYOF_CLASS_TEST(n, ANYOF_ASCII) && isASCII(c)) || | |
5871 | (ANYOF_CLASS_TEST(n, ANYOF_NASCII) && !isASCII(c)) || | |
5872 | (ANYOF_CLASS_TEST(n, ANYOF_CNTRL) && isCNTRL_LC(c)) || | |
5873 | (ANYOF_CLASS_TEST(n, ANYOF_NCNTRL) && !isCNTRL_LC(c)) || | |
5874 | (ANYOF_CLASS_TEST(n, ANYOF_GRAPH) && isGRAPH_LC(c)) || | |
5875 | (ANYOF_CLASS_TEST(n, ANYOF_NGRAPH) && !isGRAPH_LC(c)) || | |
5876 | (ANYOF_CLASS_TEST(n, ANYOF_LOWER) && isLOWER_LC(c)) || | |
5877 | (ANYOF_CLASS_TEST(n, ANYOF_NLOWER) && !isLOWER_LC(c)) || | |
5878 | (ANYOF_CLASS_TEST(n, ANYOF_PRINT) && isPRINT_LC(c)) || | |
5879 | (ANYOF_CLASS_TEST(n, ANYOF_NPRINT) && !isPRINT_LC(c)) || | |
5880 | (ANYOF_CLASS_TEST(n, ANYOF_PUNCT) && isPUNCT_LC(c)) || | |
5881 | (ANYOF_CLASS_TEST(n, ANYOF_NPUNCT) && !isPUNCT_LC(c)) || | |
5882 | (ANYOF_CLASS_TEST(n, ANYOF_UPPER) && isUPPER_LC(c)) || | |
5883 | (ANYOF_CLASS_TEST(n, ANYOF_NUPPER) && !isUPPER_LC(c)) || | |
5884 | (ANYOF_CLASS_TEST(n, ANYOF_XDIGIT) && isXDIGIT(c)) || | |
5885 | (ANYOF_CLASS_TEST(n, ANYOF_NXDIGIT) && !isXDIGIT(c)) || | |
5886 | (ANYOF_CLASS_TEST(n, ANYOF_PSXSPC) && isPSXSPC(c)) || | |
5887 | (ANYOF_CLASS_TEST(n, ANYOF_NPSXSPC) && !isPSXSPC(c)) || | |
5888 | (ANYOF_CLASS_TEST(n, ANYOF_BLANK) && isBLANK(c)) || | |
5889 | (ANYOF_CLASS_TEST(n, ANYOF_NBLANK) && !isBLANK(c)) | |
5890 | ) /* How's that for a conditional? */ | |
5891 | { | |
5892 | match = TRUE; | |
5893 | } | |
a0ed51b3 | 5894 | } |
a0ed51b3 LW |
5895 | } |
5896 | ||
a0ed51b3 LW |
5897 | return (flags & ANYOF_INVERT) ? !match : match; |
5898 | } | |
161b471a | 5899 | |
dfe13c55 | 5900 | STATIC U8 * |
0ce71af7 | 5901 | S_reghop3(U8 *s, I32 off, const U8* lim) |
9041c2e3 | 5902 | { |
97aff369 | 5903 | dVAR; |
7918f24d NC |
5904 | |
5905 | PERL_ARGS_ASSERT_REGHOP3; | |
5906 | ||
a0ed51b3 | 5907 | if (off >= 0) { |
1aa99e6b | 5908 | while (off-- && s < lim) { |
ffc61ed2 | 5909 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5910 | s += UTF8SKIP(s); |
ffc61ed2 | 5911 | } |
a0ed51b3 LW |
5912 | } |
5913 | else { | |
1de06328 YO |
5914 | while (off++ && s > lim) { |
5915 | s--; | |
5916 | if (UTF8_IS_CONTINUED(*s)) { | |
5917 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5918 | s--; | |
a0ed51b3 | 5919 | } |
1de06328 | 5920 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5921 | } |
5922 | } | |
5923 | return s; | |
5924 | } | |
161b471a | 5925 | |
f9f4320a YO |
5926 | #ifdef XXX_dmq |
5927 | /* there are a bunch of places where we use two reghop3's that should | |
5928 | be replaced with this routine. but since thats not done yet | |
5929 | we ifdef it out - dmq | |
5930 | */ | |
dfe13c55 | 5931 | STATIC U8 * |
1de06328 YO |
5932 | S_reghop4(U8 *s, I32 off, const U8* llim, const U8* rlim) |
5933 | { | |
5934 | dVAR; | |
7918f24d NC |
5935 | |
5936 | PERL_ARGS_ASSERT_REGHOP4; | |
5937 | ||
1de06328 YO |
5938 | if (off >= 0) { |
5939 | while (off-- && s < rlim) { | |
5940 | /* XXX could check well-formedness here */ | |
5941 | s += UTF8SKIP(s); | |
5942 | } | |
5943 | } | |
5944 | else { | |
5945 | while (off++ && s > llim) { | |
5946 | s--; | |
5947 | if (UTF8_IS_CONTINUED(*s)) { | |
5948 | while (s > llim && UTF8_IS_CONTINUATION(*s)) | |
5949 | s--; | |
5950 | } | |
5951 | /* XXX could check well-formedness here */ | |
5952 | } | |
5953 | } | |
5954 | return s; | |
5955 | } | |
f9f4320a | 5956 | #endif |
1de06328 YO |
5957 | |
5958 | STATIC U8 * | |
0ce71af7 | 5959 | S_reghopmaybe3(U8* s, I32 off, const U8* lim) |
a0ed51b3 | 5960 | { |
97aff369 | 5961 | dVAR; |
7918f24d NC |
5962 | |
5963 | PERL_ARGS_ASSERT_REGHOPMAYBE3; | |
5964 | ||
a0ed51b3 | 5965 | if (off >= 0) { |
1aa99e6b | 5966 | while (off-- && s < lim) { |
ffc61ed2 | 5967 | /* XXX could check well-formedness here */ |
a0ed51b3 | 5968 | s += UTF8SKIP(s); |
ffc61ed2 | 5969 | } |
a0ed51b3 | 5970 | if (off >= 0) |
3dab1dad | 5971 | return NULL; |
a0ed51b3 LW |
5972 | } |
5973 | else { | |
1de06328 YO |
5974 | while (off++ && s > lim) { |
5975 | s--; | |
5976 | if (UTF8_IS_CONTINUED(*s)) { | |
5977 | while (s > lim && UTF8_IS_CONTINUATION(*s)) | |
5978 | s--; | |
a0ed51b3 | 5979 | } |
1de06328 | 5980 | /* XXX could check well-formedness here */ |
a0ed51b3 LW |
5981 | } |
5982 | if (off <= 0) | |
3dab1dad | 5983 | return NULL; |
a0ed51b3 LW |
5984 | } |
5985 | return s; | |
5986 | } | |
51371543 | 5987 | |
51371543 | 5988 | static void |
acfe0abc | 5989 | restore_pos(pTHX_ void *arg) |
51371543 | 5990 | { |
97aff369 | 5991 | dVAR; |
097eb12c | 5992 | regexp * const rex = (regexp *)arg; |
51371543 GS |
5993 | if (PL_reg_eval_set) { |
5994 | if (PL_reg_oldsaved) { | |
4f639d21 DM |
5995 | rex->subbeg = PL_reg_oldsaved; |
5996 | rex->sublen = PL_reg_oldsavedlen; | |
f8c7b90f | 5997 | #ifdef PERL_OLD_COPY_ON_WRITE |
4f639d21 | 5998 | rex->saved_copy = PL_nrs; |
ed252734 | 5999 | #endif |
07bc277f | 6000 | RXp_MATCH_COPIED_on(rex); |
51371543 GS |
6001 | } |
6002 | PL_reg_magic->mg_len = PL_reg_oldpos; | |
6003 | PL_reg_eval_set = 0; | |
6004 | PL_curpm = PL_reg_oldcurpm; | |
6005 | } | |
6006 | } | |
33b8afdf JH |
6007 | |
6008 | STATIC void | |
6009 | S_to_utf8_substr(pTHX_ register regexp *prog) | |
6010 | { | |
a1cac82e | 6011 | int i = 1; |
7918f24d NC |
6012 | |
6013 | PERL_ARGS_ASSERT_TO_UTF8_SUBSTR; | |
6014 | ||
a1cac82e NC |
6015 | do { |
6016 | if (prog->substrs->data[i].substr | |
6017 | && !prog->substrs->data[i].utf8_substr) { | |
6018 | SV* const sv = newSVsv(prog->substrs->data[i].substr); | |
6019 | prog->substrs->data[i].utf8_substr = sv; | |
6020 | sv_utf8_upgrade(sv); | |
610460f9 NC |
6021 | if (SvVALID(prog->substrs->data[i].substr)) { |
6022 | const U8 flags = BmFLAGS(prog->substrs->data[i].substr); | |
6023 | if (flags & FBMcf_TAIL) { | |
6024 | /* Trim the trailing \n that fbm_compile added last | |
6025 | time. */ | |
6026 | SvCUR_set(sv, SvCUR(sv) - 1); | |
6027 | /* Whilst this makes the SV technically "invalid" (as its | |
6028 | buffer is no longer followed by "\0") when fbm_compile() | |
6029 | adds the "\n" back, a "\0" is restored. */ | |
6030 | } | |
6031 | fbm_compile(sv, flags); | |
6032 | } | |
a1cac82e NC |
6033 | if (prog->substrs->data[i].substr == prog->check_substr) |
6034 | prog->check_utf8 = sv; | |
6035 | } | |
6036 | } while (i--); | |
33b8afdf JH |
6037 | } |
6038 | ||
6039 | STATIC void | |
6040 | S_to_byte_substr(pTHX_ register regexp *prog) | |
6041 | { | |
97aff369 | 6042 | dVAR; |
a1cac82e | 6043 | int i = 1; |
7918f24d NC |
6044 | |
6045 | PERL_ARGS_ASSERT_TO_BYTE_SUBSTR; | |
6046 | ||
a1cac82e NC |
6047 | do { |
6048 | if (prog->substrs->data[i].utf8_substr | |
6049 | && !prog->substrs->data[i].substr) { | |
6050 | SV* sv = newSVsv(prog->substrs->data[i].utf8_substr); | |
6051 | if (sv_utf8_downgrade(sv, TRUE)) { | |
610460f9 NC |
6052 | if (SvVALID(prog->substrs->data[i].utf8_substr)) { |
6053 | const U8 flags | |
6054 | = BmFLAGS(prog->substrs->data[i].utf8_substr); | |
6055 | if (flags & FBMcf_TAIL) { | |
6056 | /* Trim the trailing \n that fbm_compile added last | |
6057 | time. */ | |
6058 | SvCUR_set(sv, SvCUR(sv) - 1); | |
6059 | } | |
6060 | fbm_compile(sv, flags); | |
6061 | } | |
a1cac82e NC |
6062 | } else { |
6063 | SvREFCNT_dec(sv); | |
6064 | sv = &PL_sv_undef; | |
6065 | } | |
6066 | prog->substrs->data[i].substr = sv; | |
6067 | if (prog->substrs->data[i].utf8_substr == prog->check_utf8) | |
6068 | prog->check_substr = sv; | |
33b8afdf | 6069 | } |
a1cac82e | 6070 | } while (i--); |
33b8afdf | 6071 | } |
66610fdd RGS |
6072 | |
6073 | /* | |
6074 | * Local variables: | |
6075 | * c-indentation-style: bsd | |
6076 | * c-basic-offset: 4 | |
6077 | * indent-tabs-mode: t | |
6078 | * End: | |
6079 | * | |
37442d52 RGS |
6080 | * ex: set ts=8 sts=4 sw=4 noet: |
6081 | */ |