| 1 | /* regexp.h |
| 2 | * |
| 3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
| 4 | * by Larry Wall and others |
| 5 | * |
| 6 | * You may distribute under the terms of either the GNU General Public |
| 7 | * License or the Artistic License, as specified in the README file. |
| 8 | * |
| 9 | */ |
| 10 | |
| 11 | /* |
| 12 | * Definitions etc. for regexp(3) routines. |
| 13 | * |
| 14 | * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], |
| 15 | * not the System V one. |
| 16 | */ |
| 17 | #ifndef PLUGGABLE_RE_EXTENSION |
| 18 | /* we don't want to include this stuff if we are inside Nicholas' |
| 19 | * pluggable regex engine code */ |
| 20 | |
| 21 | struct regnode { |
| 22 | U8 flags; |
| 23 | U8 type; |
| 24 | U16 next_off; |
| 25 | }; |
| 26 | |
| 27 | typedef struct regnode regnode; |
| 28 | |
| 29 | struct reg_substr_data; |
| 30 | |
| 31 | struct reg_data; |
| 32 | |
| 33 | struct regexp_engine; |
| 34 | |
| 35 | typedef struct regexp { |
| 36 | I32 *startp; |
| 37 | I32 *endp; |
| 38 | regnode *regstclass; |
| 39 | struct reg_substr_data *substrs; |
| 40 | char *precomp; /* pre-compilation regular expression */ |
| 41 | struct reg_data *data; /* Additional data. */ |
| 42 | char *subbeg; /* saved or original string |
| 43 | so \digit works forever. */ |
| 44 | #ifdef PERL_OLD_COPY_ON_WRITE |
| 45 | SV *saved_copy; /* If non-NULL, SV which is COW from original */ |
| 46 | #endif |
| 47 | U32 *offsets; /* offset annotations 20001228 MJD */ |
| 48 | I32 sublen; /* Length of string pointed by subbeg */ |
| 49 | I32 refcnt; |
| 50 | I32 minlen; /* mininum possible length of string to match */ |
| 51 | I32 minlenret; /* mininum possible length of $& */ |
| 52 | I32 prelen; /* length of precomp */ |
| 53 | U32 nparens; /* number of parentheses */ |
| 54 | U32 lastparen; /* last paren matched */ |
| 55 | U32 lastcloseparen; /* last paren matched */ |
| 56 | U32 reganch; /* Internal use only + |
| 57 | Tainted information used by regexec? */ |
| 58 | HV *paren_names; /* Paren names */ |
| 59 | const struct regexp_engine* engine; |
| 60 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
| 61 | } regexp; |
| 62 | |
| 63 | |
| 64 | typedef struct re_scream_pos_data_s |
| 65 | { |
| 66 | char **scream_olds; /* match pos */ |
| 67 | I32 *scream_pos; /* Internal iterator of scream. */ |
| 68 | } re_scream_pos_data; |
| 69 | |
| 70 | typedef struct regexp_engine { |
| 71 | regexp* (*comp) (pTHX_ char* exp, char* xend, PMOP* pm); |
| 72 | I32 (*exec) (pTHX_ regexp* prog, char* stringarg, char* strend, |
| 73 | char* strbeg, I32 minend, SV* screamer, |
| 74 | void* data, U32 flags); |
| 75 | char* (*intuit) (pTHX_ regexp *prog, SV *sv, char *strpos, |
| 76 | char *strend, U32 flags, |
| 77 | struct re_scream_pos_data_s *data); |
| 78 | SV* (*checkstr) (pTHX_ regexp *prog); |
| 79 | void (*free) (pTHX_ struct regexp* r); |
| 80 | char* (*as_str) (pTHX_ MAGIC *mg, STRLEN *lp, U32 *flags, I32 *haseval); |
| 81 | #ifdef USE_ITHREADS |
| 82 | regexp* (*dupe) (pTHX_ const regexp *r, CLONE_PARAMS *param); |
| 83 | #endif |
| 84 | } regexp_engine; |
| 85 | |
| 86 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
| 87 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) |
| 88 | #define ROPT_ANCH_BOL 0x00000001 |
| 89 | #define ROPT_ANCH_MBOL 0x00000002 |
| 90 | #define ROPT_ANCH_SBOL 0x00000004 |
| 91 | #define ROPT_ANCH_GPOS 0x00000008 |
| 92 | #define ROPT_SKIP 0x00000010 |
| 93 | #define ROPT_IMPLICIT 0x00000020 /* Converted .* to ^.* */ |
| 94 | #define ROPT_NOSCAN 0x00000040 /* Check-string always at start. */ |
| 95 | #define ROPT_GPOS_SEEN 0x00000080 |
| 96 | #define ROPT_CHECK_ALL 0x00000100 |
| 97 | #define ROPT_LOOKBEHIND_SEEN 0x00000200 |
| 98 | #define ROPT_EVAL_SEEN 0x00000400 |
| 99 | #define ROPT_CANY_SEEN 0x00000800 |
| 100 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ |
| 101 | #define ROPT_GPOS_CHECK (ROPT_GPOS_SEEN|ROPT_ANCH_GPOS) |
| 102 | |
| 103 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
| 104 | |
| 105 | #define ROPT_UTF8 0x00010000 |
| 106 | #define ROPT_NAUGHTY 0x00020000 /* how exponential is this pattern? */ |
| 107 | #define ROPT_COPY_DONE 0x00040000 /* subbeg is a copy of the string */ |
| 108 | #define ROPT_TAINTED_SEEN 0x00080000 |
| 109 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
| 110 | #define ROPT_RECURSE_SEEN 0x20000000 |
| 111 | #define ROPT_VERBARG_SEEN 0x40000000 |
| 112 | |
| 113 | #define RE_USE_INTUIT_NOML 0x00100000 /* Best to intuit before matching */ |
| 114 | #define RE_USE_INTUIT_ML 0x00200000 |
| 115 | #define REINT_AUTORITATIVE_NOML 0x00400000 /* Can trust a positive answer */ |
| 116 | #define REINT_AUTORITATIVE_ML 0x00800000 |
| 117 | #define REINT_ONCE_NOML 0x01000000 /* Intuit can succed once only. */ |
| 118 | #define REINT_ONCE_ML 0x02000000 |
| 119 | #define RE_INTUIT_ONECHAR 0x04000000 |
| 120 | #define RE_INTUIT_TAIL 0x08000000 |
| 121 | |
| 122 | |
| 123 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) |
| 124 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) |
| 125 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) |
| 126 | |
| 127 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
| 128 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
| 129 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) |
| 130 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
| 131 | ? RX_MATCH_TAINTED_on(prog) \ |
| 132 | : RX_MATCH_TAINTED_off(prog)) |
| 133 | |
| 134 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
| 135 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) |
| 136 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) |
| 137 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
| 138 | ? RX_MATCH_COPIED_on(prog) \ |
| 139 | : RX_MATCH_COPIED_off(prog)) |
| 140 | |
| 141 | #endif /* PLUGGABLE_RE_EXTENSION */ |
| 142 | |
| 143 | /* Stuff that needs to be included in the plugable extension goes below here */ |
| 144 | |
| 145 | #define RE_DEBUG_BIT 0x20000000 |
| 146 | #define RX_DEBUG(prog) ((prog)->reganch & RE_DEBUG_BIT) |
| 147 | #define RX_DEBUG_on(prog) ((prog)->reganch |= RE_DEBUG_BIT) |
| 148 | |
| 149 | #ifdef PERL_OLD_COPY_ON_WRITE |
| 150 | #define RX_MATCH_COPY_FREE(rx) \ |
| 151 | STMT_START {if (rx->saved_copy) { \ |
| 152 | SV_CHECK_THINKFIRST_COW_DROP(rx->saved_copy); \ |
| 153 | } \ |
| 154 | if (RX_MATCH_COPIED(rx)) { \ |
| 155 | Safefree(rx->subbeg); \ |
| 156 | RX_MATCH_COPIED_off(rx); \ |
| 157 | }} STMT_END |
| 158 | #else |
| 159 | #define RX_MATCH_COPY_FREE(rx) \ |
| 160 | STMT_START {if (RX_MATCH_COPIED(rx)) { \ |
| 161 | Safefree(rx->subbeg); \ |
| 162 | RX_MATCH_COPIED_off(rx); \ |
| 163 | }} STMT_END |
| 164 | #endif |
| 165 | |
| 166 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
| 167 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) |
| 168 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) |
| 169 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
| 170 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
| 171 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
| 172 | |
| 173 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
| 174 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
| 175 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
| 176 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
| 177 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
| 178 | |
| 179 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
| 180 | #define ReREFCNT_dec(re) CALLREGFREE(re) |
| 181 | |
| 182 | #define FBMcf_TAIL_DOLLAR 1 |
| 183 | #define FBMcf_TAIL_DOLLARM 2 |
| 184 | #define FBMcf_TAIL_Z 4 |
| 185 | #define FBMcf_TAIL_z 8 |
| 186 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
| 187 | |
| 188 | #define FBMrf_MULTILINE 1 |
| 189 | |
| 190 | struct re_scream_pos_data_s; |
| 191 | |
| 192 | /* an accepting state/position*/ |
| 193 | struct _reg_trie_accepted { |
| 194 | U8 *endpos; |
| 195 | U16 wordnum; |
| 196 | }; |
| 197 | typedef struct _reg_trie_accepted reg_trie_accepted; |
| 198 | |
| 199 | /* some basic information about the current match that is created by |
| 200 | * Perl_regexec_flags and then passed to regtry(), regmatch() etc */ |
| 201 | |
| 202 | typedef struct { |
| 203 | regexp *prog; |
| 204 | char *bol; |
| 205 | char *till; |
| 206 | SV *sv; |
| 207 | char *ganch; |
| 208 | char *cutpoint; |
| 209 | } regmatch_info; |
| 210 | |
| 211 | |
| 212 | /* structures for holding and saving the state maintained by regmatch() */ |
| 213 | |
| 214 | #define MAX_RECURSE_EVAL_NOCHANGE_DEPTH 50 |
| 215 | |
| 216 | typedef I32 CHECKPOINT; |
| 217 | |
| 218 | typedef struct regmatch_state { |
| 219 | int resume_state; /* where to jump to on return */ |
| 220 | char *locinput; /* where to backtrack in string on failure */ |
| 221 | |
| 222 | union { |
| 223 | |
| 224 | /* this is a fake union member that matches the first element |
| 225 | * of each member that needs to store positive backtrack |
| 226 | * information */ |
| 227 | struct { |
| 228 | struct regmatch_state *prev_yes_state; |
| 229 | } yes; |
| 230 | |
| 231 | struct { |
| 232 | reg_trie_accepted *accept_buff; |
| 233 | U32 accepted; /* how many accepting states we have seen */ |
| 234 | U16 *jump; /* positive offsets from me */ |
| 235 | regnode *B; /* node following the trie */ |
| 236 | regnode *me; /* Which node am I - needed for jump tries*/ |
| 237 | } trie; |
| 238 | |
| 239 | struct { |
| 240 | /* this first element must match u.yes */ |
| 241 | struct regmatch_state *prev_yes_state; |
| 242 | struct regmatch_state *prev_eval; |
| 243 | struct regmatch_state *prev_curlyx; |
| 244 | regexp *prev_rex; |
| 245 | U32 toggle_reg_flags; /* what bits in PL_reg_flags to |
| 246 | flip when transitioning between |
| 247 | inner and outer rexen */ |
| 248 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 249 | CHECKPOINT lastcp; |
| 250 | regnode *B; /* the node following us */ |
| 251 | U32 close_paren; /* which close bracket is our end */ |
| 252 | } eval; |
| 253 | |
| 254 | struct { |
| 255 | /* this first element must match u.yes */ |
| 256 | struct regmatch_state *prev_yes_state; |
| 257 | struct regmatch_state *prev_curlyx; /* previous cur_curlyx */ |
| 258 | CHECKPOINT cp; /* remember current savestack index */ |
| 259 | bool minmod; |
| 260 | int parenfloor;/* how far back to strip paren data */ |
| 261 | int min; /* the minimal number of A's to match */ |
| 262 | int max; /* the maximal number of A's to match */ |
| 263 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
| 264 | |
| 265 | /* these two are modified by WHILEM */ |
| 266 | int count; /* how many instances of A we've matched */ |
| 267 | char *lastloc;/* where previous A matched (0-len detect) */ |
| 268 | } curlyx; |
| 269 | |
| 270 | struct { |
| 271 | /* this first element must match u.yes */ |
| 272 | struct regmatch_state *prev_yes_state; |
| 273 | struct regmatch_state *save_curlyx; |
| 274 | CHECKPOINT cp; /* remember current savestack indexes */ |
| 275 | CHECKPOINT lastcp; |
| 276 | char *save_lastloc; /* previous curlyx.lastloc */ |
| 277 | I32 cache_offset; |
| 278 | I32 cache_mask; |
| 279 | } whilem; |
| 280 | |
| 281 | struct { |
| 282 | U32 lastparen; |
| 283 | regnode *next_branch; /* next branch node */ |
| 284 | CHECKPOINT cp; |
| 285 | } branch; |
| 286 | |
| 287 | struct { |
| 288 | /* this first element must match u.yes */ |
| 289 | struct regmatch_state *prev_yes_state; |
| 290 | I32 c1, c2; /* case fold search */ |
| 291 | CHECKPOINT cp; |
| 292 | I32 alen; /* length of first-matched A string */ |
| 293 | I32 count; |
| 294 | bool minmod; |
| 295 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
| 296 | regnode *me; /* the curlym node */ |
| 297 | } curlym; |
| 298 | |
| 299 | struct { |
| 300 | U32 paren; |
| 301 | CHECKPOINT cp; |
| 302 | I32 c1, c2; /* case fold search */ |
| 303 | char *maxpos; /* highest possible point in string to match */ |
| 304 | char *oldloc; /* the previous locinput */ |
| 305 | int count; |
| 306 | int min, max; /* {m,n} */ |
| 307 | regnode *A, *B; /* the nodes corresponding to /A*B/ */ |
| 308 | } curly; /* and CURLYN/PLUS/STAR */ |
| 309 | |
| 310 | struct { |
| 311 | /* this first element must match u.yes */ |
| 312 | struct regmatch_state *prev_yes_state; |
| 313 | I32 wanted; |
| 314 | I32 logical; /* saved copy of 'logical' var */ |
| 315 | regnode *me; /* the IFMATCH/SUSPEND/UNLESSM node */ |
| 316 | } ifmatch; /* and SUSPEND/UNLESSM */ |
| 317 | |
| 318 | struct { |
| 319 | /* this first element must match u.yes */ |
| 320 | struct regmatch_state *prev_yes_state; |
| 321 | struct regmatch_state *prev_mark; |
| 322 | SV* mark_name; |
| 323 | char *mark_loc; |
| 324 | } mark; |
| 325 | } u; |
| 326 | } regmatch_state; |
| 327 | |
| 328 | /* how many regmatch_state structs to allocate as a single slab. |
| 329 | * We do it in 4K blocks for efficiency. The "3" is 2 for the next/prev |
| 330 | * pointers, plus 1 for any mythical malloc overhead. */ |
| 331 | |
| 332 | #define PERL_REGMATCH_SLAB_SLOTS \ |
| 333 | ((4096 - 3 * sizeof (void*)) / sizeof(regmatch_state)) |
| 334 | |
| 335 | typedef struct regmatch_slab { |
| 336 | regmatch_state states[PERL_REGMATCH_SLAB_SLOTS]; |
| 337 | struct regmatch_slab *prev, *next; |
| 338 | } regmatch_slab; |
| 339 | |
| 340 | #define PL_reg_flags PL_reg_state.re_state_reg_flags |
| 341 | #define PL_bostr PL_reg_state.re_state_bostr |
| 342 | #define PL_reginput PL_reg_state.re_state_reginput |
| 343 | #define PL_regeol PL_reg_state.re_state_regeol |
| 344 | #define PL_regstartp PL_reg_state.re_state_regstartp |
| 345 | #define PL_regendp PL_reg_state.re_state_regendp |
| 346 | #define PL_reglastparen PL_reg_state.re_state_reglastparen |
| 347 | #define PL_reglastcloseparen PL_reg_state.re_state_reglastcloseparen |
| 348 | #define PL_reg_start_tmp PL_reg_state.re_state_reg_start_tmp |
| 349 | #define PL_reg_start_tmpl PL_reg_state.re_state_reg_start_tmpl |
| 350 | #define PL_reg_eval_set PL_reg_state.re_state_reg_eval_set |
| 351 | #define PL_reg_match_utf8 PL_reg_state.re_state_reg_match_utf8 |
| 352 | #define PL_reg_magic PL_reg_state.re_state_reg_magic |
| 353 | #define PL_reg_oldpos PL_reg_state.re_state_reg_oldpos |
| 354 | #define PL_reg_oldcurpm PL_reg_state.re_state_reg_oldcurpm |
| 355 | #define PL_reg_curpm PL_reg_state.re_state_reg_curpm |
| 356 | #define PL_reg_oldsaved PL_reg_state.re_state_reg_oldsaved |
| 357 | #define PL_reg_oldsavedlen PL_reg_state.re_state_reg_oldsavedlen |
| 358 | #define PL_reg_maxiter PL_reg_state.re_state_reg_maxiter |
| 359 | #define PL_reg_leftiter PL_reg_state.re_state_reg_leftiter |
| 360 | #define PL_reg_poscache PL_reg_state.re_state_reg_poscache |
| 361 | #define PL_reg_poscache_size PL_reg_state.re_state_reg_poscache_size |
| 362 | #define PL_regsize PL_reg_state.re_state_regsize |
| 363 | #define PL_reg_starttry PL_reg_state.re_state_reg_starttry |
| 364 | #define PL_nrs PL_reg_state.re_state_nrs |
| 365 | |
| 366 | struct re_save_state { |
| 367 | U32 re_state_reg_flags; /* from regexec.c */ |
| 368 | char *re_state_bostr; |
| 369 | char *re_state_reginput; /* String-input pointer. */ |
| 370 | char *re_state_regeol; /* End of input, for $ check. */ |
| 371 | I32 *re_state_regstartp; /* Pointer to startp array. */ |
| 372 | I32 *re_state_regendp; /* Ditto for endp. */ |
| 373 | U32 *re_state_reglastparen; /* Similarly for lastparen. */ |
| 374 | U32 *re_state_reglastcloseparen; /* Similarly for lastcloseparen. */ |
| 375 | char **re_state_reg_start_tmp; /* from regexec.c */ |
| 376 | U32 re_state_reg_start_tmpl; /* from regexec.c */ |
| 377 | I32 re_state_reg_eval_set; /* from regexec.c */ |
| 378 | bool re_state_reg_match_utf8; /* from regexec.c */ |
| 379 | MAGIC *re_state_reg_magic; /* from regexec.c */ |
| 380 | I32 re_state_reg_oldpos; /* from regexec.c */ |
| 381 | PMOP *re_state_reg_oldcurpm; /* from regexec.c */ |
| 382 | PMOP *re_state_reg_curpm; /* from regexec.c */ |
| 383 | char *re_state_reg_oldsaved; /* old saved substr during match */ |
| 384 | STRLEN re_state_reg_oldsavedlen; /* old length of saved substr during match */ |
| 385 | I32 re_state_reg_maxiter; /* max wait until caching pos */ |
| 386 | I32 re_state_reg_leftiter; /* wait until caching pos */ |
| 387 | char *re_state_reg_poscache; /* cache of pos of WHILEM */ |
| 388 | STRLEN re_state_reg_poscache_size; /* size of pos cache of WHILEM */ |
| 389 | U32 re_state_regsize; /* from regexec.c */ |
| 390 | char *re_state_reg_starttry; /* from regexec.c */ |
| 391 | #ifdef PERL_OLD_COPY_ON_WRITE |
| 392 | SV *re_state_nrs; /* was placeholder: unused since 5.8.0 (5.7.2 patch #12027 for bug ID 20010815.012). Used to save rx->saved_copy */ |
| 393 | #endif |
| 394 | }; |
| 395 | |
| 396 | #define SAVESTACK_ALLOC_FOR_RE_SAVE_STATE \ |
| 397 | (1 + ((sizeof(struct re_save_state) - 1) / sizeof(*PL_savestack))) |
| 398 | /* |
| 399 | * Local variables: |
| 400 | * c-indentation-style: bsd |
| 401 | * c-basic-offset: 4 |
| 402 | * indent-tabs-mode: t |
| 403 | * End: |
| 404 | * |
| 405 | * ex: set ts=8 sts=4 sw=4 noet: |
| 406 | */ |