1 /* regcomp.h 2 * 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 4 * 2000, 2001, 2002, 2003, 2005, 2006, 2007, 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 #include "regcharclass.h" 11 12 typedef OP OP_4tree; /* Will be redefined later. */ 13 14 15 /* Convert branch sequences to more efficient trie ops? */ 16 #define PERL_ENABLE_TRIE_OPTIMISATION 1 17 18 /* Be really agressive about optimising patterns with trie sequences? */ 19 #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1 20 21 /* Use old style unicode mappings for perl and posix character classes 22 * 23 * NOTE: Enabling this essentially breaks character class matching against unicode 24 * strings, so that POSIX char classes match when they shouldn't, and \d matches 25 * way more than 10 characters, and sometimes a charclass and its complement either 26 * both match or neither match. 27 * NOTE: Disabling this will cause various backwards compatibility issues to rear 28 * their head, and tests to fail. However it will make the charclass behaviour 29 * consistant regardless of internal string type, and make character class inversions 30 * consistant. The tests that fail in the regex engine are basically broken tests. 31 * 32 * Personally I think 5.12 should disable this for sure. Its a bit more debatable for 33 * 5.10, so for now im leaving it enabled. 34 * XXX: It is now enabled for 5.11/5.12 35 * 36 * -demerphq 37 */ 38 #define PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS 1 39 40 /* Should the optimiser take positive assertions into account? */ 41 #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0 42 43 /* Not for production use: */ 44 #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 45 46 /* Activate offsets code - set to if 1 to enable */ 47 #ifdef DEBUGGING 48 #define RE_TRACK_PATTERN_OFFSETS 49 #endif 50 51 /* Unless the next line is uncommented it is illegal to combine lazy 52 matching with possessive matching. Frankly it doesn't make much sense 53 to allow it as X*?+ matches nothing, X+?+ matches a single char only, 54 and X{min,max}?+ matches min times only. 55 */ 56 /* #define REG_ALLOW_MINMOD_SUSPEND */ 57 58 /* 59 * The "internal use only" fields in regexp.h are present to pass info from 60 * compile to execute that permits the execute phase to run lots faster on 61 * simple cases. They are: 62 * 63 * regstart sv that must begin a match; NULL if none obvious 64 * reganch is the match anchored (at beginning-of-line only)? 65 * regmust string (pointer into program) that match must include, or NULL 66 * [regmust changed to SV* for bminstr()--law] 67 * regmlen length of regmust string 68 * [regmlen not used currently] 69 * 70 * Regstart and reganch permit very fast decisions on suitable starting points 71 * for a match, cutting down the work a lot. Regmust permits fast rejection 72 * of lines that cannot possibly match. The regmust tests are costly enough 73 * that pregcomp() supplies a regmust only if the r.e. contains something 74 * potentially expensive (at present, the only such thing detected is * or + 75 * at the start of the r.e., which can involve a lot of backup). Regmlen is 76 * supplied because the test in pregexec() needs it and pregcomp() is computing 77 * it anyway. 78 * [regmust is now supplied always. The tests that use regmust have a 79 * heuristic that disables the test if it usually matches.] 80 * 81 * [In fact, we now use regmust in many cases to locate where the search 82 * starts in the string, so if regback is >= 0, the regmust search is never 83 * wasted effort. The regback variable says how many characters back from 84 * where regmust matched is the earliest possible start of the match. 85 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] 86 */ 87 88 /* 89 * Structure for regexp "program". This is essentially a linear encoding 90 * of a nondeterministic finite-state machine (aka syntax charts or 91 * "railroad normal form" in parsing technology). Each node is an opcode 92 * plus a "next" pointer, possibly plus an operand. "Next" pointers of 93 * all nodes except BRANCH implement concatenation; a "next" pointer with 94 * a BRANCH on both ends of it is connecting two alternatives. (Here we 95 * have one of the subtle syntax dependencies: an individual BRANCH (as 96 * opposed to a collection of them) is never concatenated with anything 97 * because of operator precedence.) The operand of some types of node is 98 * a literal string; for others, it is a node leading into a sub-FSM. In 99 * particular, the operand of a BRANCH node is the first node of the branch. 100 * (NB this is *not* a tree structure: the tail of the branch connects 101 * to the thing following the set of BRANCHes.) The opcodes are defined 102 * in regnodes.h which is generated from regcomp.sym by regcomp.pl. 103 */ 104 105 /* 106 * A node is one char of opcode followed by two chars of "next" pointer. 107 * "Next" pointers are stored as two 8-bit pieces, high order first. The 108 * value is a positive offset from the opcode of the node containing it. 109 * An operand, if any, simply follows the node. (Note that much of the 110 * code generation knows about this implicit relationship.) 111 * 112 * Using two bytes for the "next" pointer is vast overkill for most things, 113 * but allows patterns to get big without disasters. 114 * 115 * [The "next" pointer is always aligned on an even 116 * boundary, and reads the offset directly as a short. Also, there is no 117 * special test to reverse the sign of BACK pointers since the offset is 118 * stored negative.] 119 */ 120 121 /* This is the stuff that used to live in regexp.h that was truly 122 private to the engine itself. It now lives here. */ 123 124 125 126 typedef struct regexp_internal { 127 int name_list_idx; /* Optional data index of an array of paren names */ 128 union { 129 U32 *offsets; /* offset annotations 20001228 MJD 130 data about mapping the program to the 131 string - 132 offsets[0] is proglen when this is used 133 */ 134 U32 proglen; 135 } u; 136 137 regnode *regstclass; /* Optional startclass as identified or constructed 138 by the optimiser */ 139 struct reg_data *data; /* Additional miscellaneous data used by the program. 140 Used to make it easier to clone and free arbitrary 141 data that the regops need. Often the ARG field of 142 a regop is an index into this structure */ 143 regnode program[1]; /* Unwarranted chumminess with compiler. */ 144 } regexp_internal; 145 146 #define RXi_SET(x,y) (x)->pprivate = (void*)(y) 147 #define RXi_GET(x) ((regexp_internal *)((x)->pprivate)) 148 #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r) 149 /* 150 * Flags stored in regexp->intflags 151 * These are used only internally to the regexp engine 152 * 153 * See regexp.h for flags used externally to the regexp engine 154 */ 155 #define PREGf_SKIP 0x00000001 156 #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */ 157 #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */ 158 #define PREGf_VERBARG_SEEN 0x00000008 159 #define PREGf_CUTGROUP_SEEN 0x00000010 160 161 162 /* this is where the old regcomp.h started */ 163 164 struct regnode_string { 165 U8 str_len; 166 U8 type; 167 U16 next_off; 168 char string[1]; 169 }; 170 171 /* Argument bearing node - workhorse, 172 arg1 is often for the data field */ 173 struct regnode_1 { 174 U8 flags; 175 U8 type; 176 U16 next_off; 177 U32 arg1; 178 }; 179 180 /* Similar to a regnode_1 but with an extra signed argument */ 181 struct regnode_2L { 182 U8 flags; 183 U8 type; 184 U16 next_off; 185 U32 arg1; 186 I32 arg2; 187 }; 188 189 /* 'Two field' -- Two 16 bit unsigned args */ 190 struct regnode_2 { 191 U8 flags; 192 U8 type; 193 U16 next_off; 194 U16 arg1; 195 U16 arg2; 196 }; 197 198 199 #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ 200 #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ 201 202 /* also used by trie */ 203 struct regnode_charclass { 204 U8 flags; 205 U8 type; 206 U16 next_off; 207 U32 arg1; 208 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ 209 }; 210 211 struct regnode_charclass_class { /* has [[:blah:]] classes */ 212 U8 flags; /* should have ANYOF_CLASS here */ 213 U8 type; 214 U16 next_off; 215 U32 arg1; 216 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ 217 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ 218 }; 219 220 /* XXX fix this description. 221 Impose a limit of REG_INFTY on various pattern matching operations 222 to limit stack growth and to avoid "infinite" recursions. 223 */ 224 /* The default size for REG_INFTY is I16_MAX, which is the same as 225 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits 226 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is 227 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is 228 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a 229 smaller default. 230 --Andy Dougherty 11 June 1998 231 */ 232 #if SHORTSIZE > 2 233 # ifndef REG_INFTY 234 # define REG_INFTY ((1<<15)-1) 235 # endif 236 #endif 237 238 #ifndef REG_INFTY 239 # define REG_INFTY I16_MAX 240 #endif 241 242 #define ARG_VALUE(arg) (arg) 243 #define ARG__SET(arg,val) ((arg) = (val)) 244 245 #undef ARG 246 #undef ARG1 247 #undef ARG2 248 249 #define ARG(p) ARG_VALUE(ARG_LOC(p)) 250 #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) 251 #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) 252 #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) 253 254 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) 255 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) 256 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) 257 #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) 258 259 #undef NEXT_OFF 260 #undef NODE_ALIGN 261 262 #define NEXT_OFF(p) ((p)->next_off) 263 #define NODE_ALIGN(node) 264 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ 265 266 #define SIZE_ALIGN NODE_ALIGN 267 268 #undef OP 269 #undef OPERAND 270 #undef MASK 271 #undef STRING 272 273 #define OP(p) ((p)->type) 274 #define OPERAND(p) (((struct regnode_string *)p)->string) 275 #define MASK(p) ((char*)OPERAND(p)) 276 #define STR_LEN(p) (((struct regnode_string *)p)->str_len) 277 #define STRING(p) (((struct regnode_string *)p)->string) 278 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) 279 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) 280 281 #undef NODE_ALIGN 282 #undef ARG_LOC 283 #undef NEXTOPER 284 #undef PREVOPER 285 286 #define NODE_ALIGN(node) 287 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) 288 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) 289 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) 290 #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) 291 292 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ 293 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) 294 295 #define NODE_STEP_B 4 296 297 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) 298 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) 299 300 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ 301 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END 302 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ 303 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END 304 305 #define REG_MAGIC 0234 306 307 #define SIZE_ONLY (RExC_emit == &PL_regdummy) 308 309 /* Flags for node->flags of ANYOF */ 310 311 #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ 312 #define ANYOF_INVERT 0x04 313 #define ANYOF_FOLD 0x02 314 #define ANYOF_LOCALE 0x01 315 316 /* Used for regstclass only */ 317 #define ANYOF_EOS 0x10 /* Can match an empty string too */ 318 319 /* There is a character or a range past 0xff */ 320 #define ANYOF_UNICODE 0x20 321 #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ 322 323 /* size of node is large (includes class pointer) */ 324 #define ANYOF_LARGE 0x80 325 326 /* Are there any runtime flags on in this node? */ 327 #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) 328 329 #define ANYOF_FLAGS_ALL 0xff 330 331 /* Character classes for node->classflags of ANYOF */ 332 /* Should be synchronized with a table in regprop() */ 333 /* 2n should pair with 2n+1 */ 334 335 #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ 336 #define ANYOF_NALNUM 1 337 #define ANYOF_SPACE 2 /* \s */ 338 #define ANYOF_NSPACE 3 339 #define ANYOF_DIGIT 4 /* \d */ 340 #define ANYOF_NDIGIT 5 341 #define ANYOF_ALNUMC 6 /* [[:alnum:]] isalnum(3), utf8::IsAlnum, ALNUMC */ 342 #define ANYOF_NALNUMC 7 343 #define ANYOF_ALPHA 8 344 #define ANYOF_NALPHA 9 345 #define ANYOF_ASCII 10 346 #define ANYOF_NASCII 11 347 #define ANYOF_CNTRL 12 348 #define ANYOF_NCNTRL 13 349 #define ANYOF_GRAPH 14 350 #define ANYOF_NGRAPH 15 351 #define ANYOF_LOWER 16 352 #define ANYOF_NLOWER 17 353 #define ANYOF_PRINT 18 354 #define ANYOF_NPRINT 19 355 #define ANYOF_PUNCT 20 356 #define ANYOF_NPUNCT 21 357 #define ANYOF_UPPER 22 358 #define ANYOF_NUPPER 23 359 #define ANYOF_XDIGIT 24 360 #define ANYOF_NXDIGIT 25 361 #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ 362 #define ANYOF_NPSXSPC 27 363 #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ 364 #define ANYOF_NBLANK 29 365 366 #define ANYOF_MAX 32 367 368 /* pseudo classes, not stored in the class bitmap, but used as flags 369 during compilation of char classes */ 370 371 #define ANYOF_VERTWS (ANYOF_MAX+1) 372 #define ANYOF_NVERTWS (ANYOF_MAX+2) 373 #define ANYOF_HORIZWS (ANYOF_MAX+3) 374 #define ANYOF_NHORIZWS (ANYOF_MAX+4) 375 376 /* Backward source code compatibility. */ 377 378 #define ANYOF_ALNUML ANYOF_ALNUM 379 #define ANYOF_NALNUML ANYOF_NALNUM 380 #define ANYOF_SPACEL ANYOF_SPACE 381 #define ANYOF_NSPACEL ANYOF_NSPACE 382 383 /* Utility macros for the bitmap and classes of ANYOF */ 384 385 #define ANYOF_SIZE (sizeof(struct regnode_charclass)) 386 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) 387 388 #define ANYOF_FLAGS(p) ((p)->flags) 389 390 #define ANYOF_BIT(c) (1 << ((c) & 7)) 391 392 #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) 393 #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) 394 #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) 395 #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) 396 397 #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) 398 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) 399 400 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) 401 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) 402 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) 403 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) 404 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) 405 406 #define ANYOF_BITMAP_SETALL(p) \ 407 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) 408 #define ANYOF_BITMAP_CLEARALL(p) \ 409 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) 410 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ 411 #define ANYOF_BITMAP_TESTALLSET(p) \ 412 memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE) 413 414 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) 415 #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) 416 #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) 417 418 419 /* 420 * Utility definitions. 421 */ 422 #ifndef CHARMASK 423 # define UCHARAT(p) ((int)*(const U8*)(p)) 424 #else 425 # define UCHARAT(p) ((int)*(p)&CHARMASK) 426 #endif 427 428 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) 429 430 #define REG_SEEN_ZERO_LEN 0x00000001 431 #define REG_SEEN_LOOKBEHIND 0x00000002 432 #define REG_SEEN_GPOS 0x00000004 433 #define REG_SEEN_EVAL 0x00000008 434 #define REG_SEEN_CANY 0x00000010 435 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ 436 #define REG_SEEN_RECURSE 0x00000020 437 #define REG_TOP_LEVEL_BRANCHES 0x00000040 438 #define REG_SEEN_VERBARG 0x00000080 439 #define REG_SEEN_CUTGROUP 0x00000100 440 #define REG_SEEN_RUN_ON_COMMENT 0x00000200 441 442 START_EXTERN_C 443 444 #ifdef PLUGGABLE_RE_EXTENSION 445 #include "re_nodes.h" 446 #else 447 #include "regnodes.h" 448 #endif 449 450 /* The following have no fixed length. U8 so we can do strchr() on it. */ 451 #ifndef DOINIT 452 EXTCONST U8 PL_varies[]; 453 #else 454 EXTCONST U8 PL_varies[] = { 455 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, 456 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 457 NREF, NREFF, NREFFL, 458 0 459 }; 460 #endif 461 462 /* The following always have a length of 1. U8 we can do strchr() on it. */ 463 /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ 464 #ifndef DOINIT 465 EXTCONST U8 PL_simple[]; 466 #else 467 EXTCONST U8 PL_simple[] = { 468 REG_ANY, SANY, CANY, 469 ANYOF, 470 ALNUM, ALNUML, 471 NALNUM, NALNUML, 472 SPACE, SPACEL, 473 NSPACE, NSPACEL, 474 DIGIT, NDIGIT, 475 VERTWS, NVERTWS, 476 HORIZWS, NHORIZWS, 477 0 478 }; 479 #endif 480 481 #ifndef PLUGGABLE_RE_EXTENSION 482 #ifndef DOINIT 483 EXTCONST regexp_engine PL_core_reg_engine; 484 #else /* DOINIT */ 485 EXTCONST regexp_engine PL_core_reg_engine = { 486 Perl_re_compile, 487 Perl_regexec_flags, 488 Perl_re_intuit_start, 489 Perl_re_intuit_string, 490 Perl_regfree_internal, 491 Perl_reg_numbered_buff_fetch, 492 Perl_reg_numbered_buff_store, 493 Perl_reg_numbered_buff_length, 494 Perl_reg_named_buff, 495 Perl_reg_named_buff_iter, 496 Perl_reg_qr_package, 497 #if defined(USE_ITHREADS) 498 Perl_regdupe_internal 499 #endif 500 }; 501 #endif /* DOINIT */ 502 #endif /* PLUGGABLE_RE_EXTENSION */ 503 504 505 END_EXTERN_C 506 507 508 /* .what is a character array with one character for each member of .data 509 * The character describes the function of the corresponding .data item: 510 * f - start-class data for regstclass optimization 511 * n - Root of op tree for (?{EVAL}) item 512 * o - Start op for (?{EVAL}) item 513 * p - Pad for (?{EVAL}) item 514 * s - swash for Unicode-style character class, and the multicharacter 515 * strings resulting from casefolding the single-character entries 516 * in the character class 517 * t - trie struct 518 * u - trie struct's widecharmap (a HV, so can't share, must dup) 519 * also used for revcharmap and words under DEBUGGING 520 * T - aho-trie struct 521 * S - sv for named capture lookup 522 * 20010712 mjd@plover.com 523 * (Remember to update re_dup() and pregfree() if you add any items.) 524 */ 525 struct reg_data { 526 U32 count; 527 U8 *what; 528 void* data[1]; 529 }; 530 531 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses 532 anchored* and float* via array indexes 0 and 1. */ 533 #define anchored_substr substrs->data[0].substr 534 #define anchored_utf8 substrs->data[0].utf8_substr 535 #define anchored_offset substrs->data[0].min_offset 536 #define anchored_end_shift substrs->data[0].end_shift 537 538 #define float_substr substrs->data[1].substr 539 #define float_utf8 substrs->data[1].utf8_substr 540 #define float_min_offset substrs->data[1].min_offset 541 #define float_max_offset substrs->data[1].max_offset 542 #define float_end_shift substrs->data[1].end_shift 543 544 #define check_substr substrs->data[2].substr 545 #define check_utf8 substrs->data[2].utf8_substr 546 #define check_offset_min substrs->data[2].min_offset 547 #define check_offset_max substrs->data[2].max_offset 548 #define check_end_shift substrs->data[2].end_shift 549 550 #define RX_ANCHORED_SUBSTR(rx) (((struct regexp *)SvANY(rx))->anchored_substr) 551 #define RX_ANCHORED_UTF8(rx) (((struct regexp *)SvANY(rx))->anchored_utf8) 552 #define RX_FLOAT_SUBSTR(rx) (((struct regexp *)SvANY(rx))->float_substr) 553 #define RX_FLOAT_UTF8(rx) (((struct regexp *)SvANY(rx))->float_utf8) 554 555 /* trie related stuff */ 556 557 /* a transition record for the state machine. the 558 check field determines which state "owns" the 559 transition. the char the transition is for is 560 determined by offset from the owning states base 561 field. the next field determines which state 562 is to be transitioned to if any. 563 */ 564 struct _reg_trie_trans { 565 U32 next; 566 U32 check; 567 }; 568 569 /* a transition list element for the list based representation */ 570 struct _reg_trie_trans_list_elem { 571 U16 forid; 572 U32 newstate; 573 }; 574 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; 575 576 /* a state for compressed nodes. base is an offset 577 into an array of reg_trie_trans array. If wordnum is 578 nonzero the state is accepting. if base is zero then 579 the state has no children (and will be accepting) 580 */ 581 struct _reg_trie_state { 582 U16 wordnum; 583 union { 584 U32 base; 585 reg_trie_trans_le* list; 586 } trans; 587 }; 588 589 590 591 typedef struct _reg_trie_state reg_trie_state; 592 typedef struct _reg_trie_trans reg_trie_trans; 593 594 595 /* anything in here that needs to be freed later 596 should be dealt with in pregfree. 597 refcount is first in both this and _reg_ac_data to allow a space 598 optimisation in Perl_regdupe. */ 599 struct _reg_trie_data { 600 U32 refcount; /* number of times this trie is referenced */ 601 U32 lasttrans; /* last valid transition element */ 602 U16 *charmap; /* byte to charid lookup array */ 603 reg_trie_state *states; /* state data */ 604 reg_trie_trans *trans; /* array of transition elements */ 605 char *bitmap; /* stclass bitmap */ 606 U32 *wordlen; /* array of lengths of words */ 607 U16 *jump; /* optional 1 indexed array of offsets before tail 608 for the node following a given word. */ 609 U16 *nextword; /* optional 1 indexed array to support linked list 610 of duplicate wordnums */ 611 U16 uniquecharcount; /* unique chars in trie (width of trans table) */ 612 U32 startstate; /* initial state - used for common prefix optimisation */ 613 STRLEN minlen; /* minimum length of words in trie - build/opt only? */ 614 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ 615 U32 statecount; /* Build only - number of states in the states array 616 (including the unused zero state) */ 617 U32 wordcount; /* Build only */ 618 #ifdef DEBUGGING 619 STRLEN charcount; /* Build only */ 620 #endif 621 }; 622 /* There is one (3 under DEBUGGING) pointers that logically belong in this 623 structure, but are held outside as they need duplication on thread cloning, 624 whereas the rest of the structure can be read only: 625 HV *widecharmap; code points > 255 to charid 626 #ifdef DEBUGGING 627 AV *words; Array of words contained in trie, for dumping 628 AV *revcharmap; Map of each charid back to its character representation 629 #endif 630 */ 631 632 #define TRIE_WORDS_OFFSET 2 633 634 typedef struct _reg_trie_data reg_trie_data; 635 636 /* refcount is first in both this and _reg_trie_data to allow a space 637 optimisation in Perl_regdupe. */ 638 struct _reg_ac_data { 639 U32 refcount; 640 U32 trie; 641 U32 *fail; 642 reg_trie_state *states; 643 }; 644 typedef struct _reg_ac_data reg_ac_data; 645 646 /* ANY_BIT doesnt use the structure, so we can borrow it here. 647 This is simpler than refactoring all of it as wed end up with 648 three different sets... */ 649 650 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) 651 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) 652 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) 653 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) 654 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) 655 656 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) 657 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) 658 659 660 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) 661 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) 662 663 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ 664 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) 665 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) 666 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) 667 668 #ifdef DEBUGGING 669 #define TRIE_CHARCOUNT(trie) ((trie)->charcount) 670 #else 671 #define TRIE_CHARCOUNT(trie) (trie_charcount) 672 #endif 673 674 #define RE_TRIE_MAXBUF_INIT 65536 675 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" 676 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" 677 678 /* 679 680 RE_DEBUG_FLAGS is used to control what debug output is emitted 681 its divided into three groups of options, some of which interact. 682 The three groups are: Compile, Execute, Extra. There is room for a 683 further group, as currently only the low three bytes are used. 684 685 Compile Options: 686 687 PARSE 688 PEEP 689 TRIE 690 PROGRAM 691 OFFSETS 692 693 Execute Options: 694 695 INTUIT 696 MATCH 697 TRIE 698 699 Extra Options 700 701 TRIE 702 OFFSETS 703 704 If you modify any of these make sure you make corresponding changes to 705 re.pm, especially to the documentation. 706 707 */ 708 709 710 /* Compile */ 711 #define RE_DEBUG_COMPILE_MASK 0x0000FF 712 #define RE_DEBUG_COMPILE_PARSE 0x000001 713 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 714 #define RE_DEBUG_COMPILE_TRIE 0x000004 715 #define RE_DEBUG_COMPILE_DUMP 0x000008 716 #define RE_DEBUG_COMPILE_FLAGS 0x000010 717 718 /* Execute */ 719 #define RE_DEBUG_EXECUTE_MASK 0x00FF00 720 #define RE_DEBUG_EXECUTE_INTUIT 0x000100 721 #define RE_DEBUG_EXECUTE_MATCH 0x000200 722 #define RE_DEBUG_EXECUTE_TRIE 0x000400 723 724 /* Extra */ 725 #define RE_DEBUG_EXTRA_MASK 0xFF0000 726 #define RE_DEBUG_EXTRA_TRIE 0x010000 727 #define RE_DEBUG_EXTRA_OFFSETS 0x020000 728 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 729 #define RE_DEBUG_EXTRA_STATE 0x080000 730 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 731 #define RE_DEBUG_EXTRA_BUFFERS 0x400000 732 #define RE_DEBUG_EXTRA_GPOS 0x800000 733 /* combined */ 734 #define RE_DEBUG_EXTRA_STACK 0x280000 735 736 #define RE_DEBUG_FLAG(x) (re_debug_flags & x) 737 /* Compile */ 738 #define DEBUG_COMPILE_r(x) DEBUG_r( \ 739 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) 740 #define DEBUG_PARSE_r(x) DEBUG_r( \ 741 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) 742 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ 743 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) 744 #define DEBUG_PARSE_r(x) DEBUG_r( \ 745 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) 746 #define DEBUG_DUMP_r(x) DEBUG_r( \ 747 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) 748 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ 749 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) 750 #define DEBUG_FLAGS_r(x) DEBUG_r( \ 751 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x ) 752 /* Execute */ 753 #define DEBUG_EXECUTE_r(x) DEBUG_r( \ 754 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) 755 #define DEBUG_INTUIT_r(x) DEBUG_r( \ 756 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) 757 #define DEBUG_MATCH_r(x) DEBUG_r( \ 758 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) 759 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ 760 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) 761 762 /* Extra */ 763 #define DEBUG_EXTRA_r(x) DEBUG_r( \ 764 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) 765 #define DEBUG_OFFSETS_r(x) DEBUG_r( \ 766 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) 767 #define DEBUG_STATE_r(x) DEBUG_r( \ 768 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) 769 #define DEBUG_STACK_r(x) DEBUG_r( \ 770 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x ) 771 #define DEBUG_BUFFERS_r(x) DEBUG_r( \ 772 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x ) 773 774 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ 775 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ 776 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) 777 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ 778 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ 779 Perl_warn_nocontext x ) 780 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ 781 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) 782 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ 783 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) 784 785 #define DEBUG_TRIE_r(x) DEBUG_r( \ 786 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ 787 | RE_DEBUG_EXECUTE_TRIE )) x ) 788 #define DEBUG_GPOS_r(x) DEBUG_r( \ 789 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x ) 790 791 /* initialization */ 792 /* get_sv() can return NULL during global destruction. */ 793 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ 794 SV * re_debug_flags_sv = NULL; \ 795 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ 796 if (re_debug_flags_sv) { \ 797 if (!SvIOK(re_debug_flags_sv)) \ 798 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ 799 re_debug_flags=SvIV(re_debug_flags_sv); \ 800 }\ 801 }) 802 803 #ifdef DEBUGGING 804 805 #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; 806 807 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ 808 const char * const rpv = \ 809 pv_pretty((dsv), (pv), (l), (m), \ 810 PL_colors[(c1)],PL_colors[(c2)], \ 811 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ 812 const int rlen = SvCUR(dsv) 813 814 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ 815 const char * const rpv = \ 816 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ 817 PL_colors[(c1)],PL_colors[(c2)], \ 818 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) 819 820 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ 821 const char * const rpv = \ 822 pv_pretty((dsv), (pv), (l), (m), \ 823 PL_colors[0], PL_colors[1], \ 824 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELLIPSES | \ 825 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ 826 ) 827 828 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) 829 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") 830 831 #else /* if not DEBUGGING */ 832 833 #define GET_RE_DEBUG_FLAGS_DECL 834 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) 835 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) 836 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) 837 #define RE_SV_DUMPLEN(ItEm) 838 #define RE_SV_TAIL(ItEm) 839 840 #endif /* DEBUG RELATED DEFINES */ 841 842 /* 843 * Local variables: 844 * c-indentation-style: bsd 845 * c-basic-offset: 4 846 * indent-tabs-mode: t 847 * End: 848 * 849 * ex: set ts=8 sts=4 sw=4 noet: 850 */ 851