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
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.
10 #include "regcharclass.h"
12 /* Convert branch sequences to more efficient trie ops? */
13 #define PERL_ENABLE_TRIE_OPTIMISATION 1
15 /* Be really aggressive about optimising patterns with trie sequences? */
16 #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
18 /* Should the optimiser take positive assertions into account? */
19 #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
21 /* Not for production use: */
22 #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
24 /* Activate offsets code - set to if 1 to enable */
26 #define RE_TRACK_PATTERN_OFFSETS
29 /* Unless the next line is uncommented it is illegal to combine lazy
30 matching with possessive matching. Frankly it doesn't make much sense
31 to allow it as X*?+ matches nothing, X+?+ matches a single char only,
32 and X{min,max}?+ matches min times only.
34 /* #define REG_ALLOW_MINMOD_SUSPEND */
37 * The "internal use only" fields in regexp.h are present to pass info from
38 * compile to execute that permits the execute phase to run lots faster on
39 * simple cases. They are:
41 * regstart sv that must begin a match; NULL if none obvious
42 * reganch is the match anchored (at beginning-of-line only)?
43 * regmust string (pointer into program) that match must include, or NULL
44 * [regmust changed to SV* for bminstr()--law]
45 * regmlen length of regmust string
46 * [regmlen not used currently]
48 * Regstart and reganch permit very fast decisions on suitable starting points
49 * for a match, cutting down the work a lot. Regmust permits fast rejection
50 * of lines that cannot possibly match. The regmust tests are costly enough
51 * that pregcomp() supplies a regmust only if the r.e. contains something
52 * potentially expensive (at present, the only such thing detected is * or +
53 * at the start of the r.e., which can involve a lot of backup). Regmlen is
54 * supplied because the test in pregexec() needs it and pregcomp() is computing
56 * [regmust is now supplied always. The tests that use regmust have a
57 * heuristic that disables the test if it usually matches.]
59 * [In fact, we now use regmust in many cases to locate where the search
60 * starts in the string, so if regback is >= 0, the regmust search is never
61 * wasted effort. The regback variable says how many characters back from
62 * where regmust matched is the earliest possible start of the match.
63 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
67 * Structure for regexp "program". This is essentially a linear encoding
68 * of a nondeterministic finite-state machine (aka syntax charts or
69 * "railroad normal form" in parsing technology). Each node is an opcode
70 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
71 * all nodes except BRANCH implement concatenation; a "next" pointer with
72 * a BRANCH on both ends of it is connecting two alternatives. (Here we
73 * have one of the subtle syntax dependencies: an individual BRANCH (as
74 * opposed to a collection of them) is never concatenated with anything
75 * because of operator precedence.) The operand of some types of node is
76 * a literal string; for others, it is a node leading into a sub-FSM. In
77 * particular, the operand of a BRANCH node is the first node of the branch.
78 * (NB this is *not* a tree structure: the tail of the branch connects
79 * to the thing following the set of BRANCHes.) The opcodes are defined
80 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
84 * A node is one char of opcode followed by two chars of "next" pointer.
85 * "Next" pointers are stored as two 8-bit pieces, high order first. The
86 * value is a positive offset from the opcode of the node containing it.
87 * An operand, if any, simply follows the node. (Note that much of the
88 * code generation knows about this implicit relationship.)
90 * Using two bytes for the "next" pointer is vast overkill for most things,
91 * but allows patterns to get big without disasters.
93 * [The "next" pointer is always aligned on an even
94 * boundary, and reads the offset directly as a short. Also, there is no
95 * special test to reverse the sign of BACK pointers since the offset is
99 /* This is the stuff that used to live in regexp.h that was truly
100 private to the engine itself. It now lives here. */
104 typedef struct regexp_internal {
105 int name_list_idx; /* Optional data index of an array of paren names */
107 U32 *offsets; /* offset annotations 20001228 MJD
108 data about mapping the program to the
110 offsets[0] is proglen when this is used
115 regnode *regstclass; /* Optional startclass as identified or constructed
117 struct reg_data *data; /* Additional miscellaneous data used by the program.
118 Used to make it easier to clone and free arbitrary
119 data that the regops need. Often the ARG field of
120 a regop is an index into this structure */
121 struct reg_code_block *code_blocks;/* positions of literal (?{}) */
122 int num_code_blocks; /* size of code_blocks[] */
123 regnode program[1]; /* Unwarranted chumminess with compiler. */
126 #define RXi_SET(x,y) (x)->pprivate = (void*)(y)
127 #define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
128 #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
130 * Flags stored in regexp->intflags
131 * These are used only internally to the regexp engine
133 * See regexp.h for flags used externally to the regexp engine
135 #define PREGf_SKIP 0x00000001
136 #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
137 #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
138 #define PREGf_VERBARG_SEEN 0x00000008
139 #define PREGf_CUTGROUP_SEEN 0x00000010
140 #define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
143 /* this is where the old regcomp.h started */
145 struct regnode_string {
152 /* Argument bearing node - workhorse,
153 arg1 is often for the data field */
161 /* Similar to a regnode_1 but with an extra signed argument */
170 /* 'Two field' -- Two 16 bit unsigned args */
180 #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
182 /* also used by trie */
183 struct regnode_charclass {
187 U32 arg1; /* used as ptr in S_regclass */
188 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
191 /* has runtime (locale) \d, \w, ..., [:posix:] classes */
192 struct regnode_charclass_class {
193 U8 flags; /* ANYOF_CLASS bit must go here */
196 U32 arg1; /* used as ptr in S_regclass */
197 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
198 U32 classflags; /* and run-time */
201 /* XXX fix this description.
202 Impose a limit of REG_INFTY on various pattern matching operations
203 to limit stack growth and to avoid "infinite" recursions.
205 /* The default size for REG_INFTY is I16_MAX, which is the same as
206 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
207 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
208 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
209 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
211 --Andy Dougherty 11 June 1998
215 # define REG_INFTY ((1<<15)-1)
220 # define REG_INFTY I16_MAX
223 #define ARG_VALUE(arg) (arg)
224 #define ARG__SET(arg,val) ((arg) = (val))
230 #define ARG(p) ARG_VALUE(ARG_LOC(p))
231 #define ARG1(p) ARG_VALUE(ARG1_LOC(p))
232 #define ARG2(p) ARG_VALUE(ARG2_LOC(p))
233 #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
235 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
236 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
237 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
238 #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
243 #define NEXT_OFF(p) ((p)->next_off)
244 #define NODE_ALIGN(node)
245 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
247 #define SIZE_ALIGN NODE_ALIGN
254 #define OP(p) ((p)->type)
255 #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
256 regnode types. For some, it's the \
257 character set of the regnode */
258 #define OPERAND(p) (((struct regnode_string *)p)->string)
259 #define MASK(p) ((char*)OPERAND(p))
260 #define STR_LEN(p) (((struct regnode_string *)p)->str_len)
261 #define STRING(p) (((struct regnode_string *)p)->string)
262 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
263 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
270 #define NODE_ALIGN(node)
271 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
272 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
273 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
274 #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
276 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
277 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
279 #define NODE_STEP_B 4
281 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
282 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
284 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
285 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
286 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
287 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
289 #define REG_MAGIC 0234
291 #define SIZE_ONLY (RExC_emit == &PL_regdummy)
293 /* If the bitmap doesn't fully represent what this ANYOF node can match, the
294 * ARG is set to this special value (since 0, 1, ... are legal, but will never
295 * reach this high). */
296 #define ANYOF_NONBITMAP_EMPTY ((U32) -1)
298 /* The information used to be stored as as combination of the ANYOF_UTF8 and
299 * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there
300 * to free up a bit for other uses. This tries to hide the change from
301 * existing code as much as possible. Now, the data structure that goes in ARG
302 * is not allocated unless it is needed, and that is what is used to determine
303 * if there is something outside the bitmap. The code now assumes that if
304 * that structure exists, that any UTF-8 encoded string should be tried against
305 * it, but a non-UTF8-encoded string will be tried only if the
306 * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */
307 #define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY)
309 /* Flags for node->flags of ANYOF. These are in short supply, so some games
310 * are done to share them, as described below. Already, the ANYOF_LOCALE and
311 * ANYOF_CLASS bits are shared, making a space penalty for all locale nodes.
312 * An option would be to push them into new nodes. E.g. there could be an
313 * ANYOF_LOCALE node that would be in place of the flag of the same name. But
314 * there are better options. The UNICODE_ALL bit could be freed up by
315 * resorting to creating a swash containing everything above 255. This
316 * introduces a performance penalty. Better would be to split it off into a
317 * separate node, which actually would improve performance by allowing adding a
318 * case statement to regexec.c use the bit map for code points under 256, and
319 * to match everything above. If flags need to be added that are applicable to
320 * the synthetic start class only, with some work, they could be put in the
321 * next-node field, or in an unused bit of the classflags field. This could be
322 * done with the current EOS flag, and a new node type created that is just for
323 * the scc, freeing up that bit */
325 #define ANYOF_LOCALE 0x01 /* /l modifier */
327 /* The fold is calculated and stored in the bitmap where possible at compile
328 * time. However under locale, the actual folding varies depending on
329 * what the locale is at the time of execution, so it has to be deferred until
331 #define ANYOF_LOC_FOLD 0x02
333 #define ANYOF_INVERT 0x04
335 /* Set if this is a struct regnode_charclass_class vs a regnode_charclass. This
336 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
337 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
338 * at compile-time. Now shared with ANYOF_LOCALE, forcing all locale nodes to
340 #define ANYOF_CLASS ANYOF_LOCALE
341 #define ANYOF_LARGE ANYOF_CLASS /* Same; name retained for back compat */
343 /* Should this character class warn if matched against a character above
345 #define ANYOF_WARN_SUPER 0x08
347 /* EOS, meaning that it can match an empty string too, is used for the
348 * synthetic start class only. */
349 #define ANYOF_EOS 0x10
351 /* ? Is this node the synthetic start class (ssc). This bit is shared with
352 * ANYOF_EOS, as the latter is used only for the ssc, and then not used by
353 * regexec.c. And, the code is structured so that if it is set, the ssc is
354 * not used, so it is guaranteed to be 0 for the ssc by the time regexec.c
355 * gets executed, and 0 for a non-ssc ANYOF node, as it only ever gets set for
356 * a potential ssc candidate. Thus setting it to 1 after it has been
357 * determined that the ssc will be used is not ambiguous */
358 #define ANYOF_IS_SYNTHETIC ANYOF_EOS
360 /* Can match something outside the bitmap that isn't in utf8 */
361 #define ANYOF_NONBITMAP_NON_UTF8 0x20
363 /* Matches every code point 0x100 and above*/
364 #define ANYOF_UNICODE_ALL 0x40
366 /* Match all Latin1 characters that aren't ASCII when the target string is not
368 #define ANYOF_NON_UTF8_LATIN1_ALL 0x80
370 #define ANYOF_FLAGS_ALL 0xff
372 /* These are the flags that ANYOF_INVERT being set or not doesn't affect
373 * whether they are operative or not. e.g., the node still has LOCALE
374 * regardless of being inverted; whereas ANYOF_UNICODE_ALL means something
375 * different if inverted */
376 #define INVERSION_UNAFFECTED_FLAGS (ANYOF_LOCALE \
380 |ANYOF_NONBITMAP_NON_UTF8)
382 /* Character classes for node->classflags of ANYOF */
383 /* Should be synchronized with a table in regprop() */
384 /* 2n should be the normal one, paired with its complement at 2n+1 */
386 #define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
387 #define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
388 #define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
389 #define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
390 #define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
391 #define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
392 #define ANYOF_ALNUMC ((_CC_ALNUMC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum, ALNUMC */
393 #define ANYOF_NALNUMC ((ANYOF_ALNUMC) + 1)
394 #define ANYOF_ALPHA ((_CC_ALPHA) * 2)
395 #define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
396 #define ANYOF_ASCII ((_CC_ASCII) * 2)
397 #define ANYOF_NASCII ((ANYOF_ASCII) + 1)
398 #define ANYOF_CNTRL ((_CC_CNTRL) * 2)
399 #define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
400 #define ANYOF_GRAPH ((_CC_GRAPH) * 2)
401 #define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
402 #define ANYOF_LOWER ((_CC_LOWER) * 2)
403 #define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
404 #define ANYOF_PRINT ((_CC_PRINT) * 2)
405 #define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
406 #define ANYOF_PUNCT ((_CC_PUNCT) * 2)
407 #define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
408 #define ANYOF_UPPER ((_CC_UPPER) * 2)
409 #define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
410 #define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
411 #define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
412 #define ANYOF_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */
413 #define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1)
414 #define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
415 #define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
417 /* pseudo classes below this, not stored in the class bitmap, but used as flags
418 during compilation of char classes */
420 #define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
421 #define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
423 /* It is best if this is the last one, as all above it are stored as bits in a
424 * bitmap, and it isn't part of that bitmap */
425 #if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
426 # error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
429 #define ANYOF_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
431 * Hence doesn't include VERTWS, as that
432 * is a pseudo class */
433 #if (ANYOF_MAX > 32) /* Must fit in 32-bit word */
434 # error Problem with handy.h _CC_foo #defines
437 #define ANYOF_HORIZWS ((ANYOF_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
438 #define ANYOF_NHORIZWS ((ANYOF_MAX)+3)
440 #define ANYOF_UNIPROP ((ANYOF_MAX)+4) /* Used to indicate a Unicode
441 property: \p{} or \P{} */
443 /* Backward source code compatibility. */
445 #define ANYOF_ALNUML ANYOF_ALNUM
446 #define ANYOF_NALNUML ANYOF_NALNUM
447 #define ANYOF_SPACEL ANYOF_SPACE
448 #define ANYOF_NSPACEL ANYOF_NSPACE
449 #define ANYOF_ALNUM ANYOF_WORDCHAR
450 #define ANYOF_NALNUM ANYOF_NWORDCHAR
452 /* Utility macros for the bitmap and classes of ANYOF */
454 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
455 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
457 #define ANYOF_FLAGS(p) ((p)->flags)
459 #define ANYOF_BIT(c) (1 << ((c) & 7))
461 #define ANYOF_CLASS_SET(p, c) (((struct regnode_charclass_class*) (p))->classflags |= (1U << (c)))
462 #define ANYOF_CLASS_CLEAR(p, c) (((struct regnode_charclass_class*) (p))->classflags &= ~ (1U <<(c)))
463 #define ANYOF_CLASS_TEST(p, c) (((struct regnode_charclass_class*) (p))->classflags & (1U << (c)))
465 #define ANYOF_CLASS_ZERO(ret) STMT_START { ((struct regnode_charclass_class*) (ret))->classflags = 0; } STMT_END
467 /* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
468 #define ANYOF_CLASS_SETALL(ret) STMT_START { ((struct regnode_charclass_class*) (ret))->classflags = ((1U << ((ANYOF_MAX) - 1))) - 1; } STMT_END
470 #define ANYOF_CLASS_OR(source, dest) STMT_START { (dest)->classflags |= source->classflags ; } STMT_END
472 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
473 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
474 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
475 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
476 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
477 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
479 #define ANYOF_BITMAP_SETALL(p) \
480 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
481 #define ANYOF_BITMAP_CLEARALL(p) \
482 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
483 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */
484 #define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
485 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)
487 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
488 #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
490 #define ANYOF_CLASS_TEST_ANY_SET(p) \
491 ((ANYOF_FLAGS(p) & (ANYOF_CLASS|ANYOF_IS_SYNTHETIC)) \
492 && (((struct regnode_charclass_class*)(p))->classflags))
493 /*#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
498 * Utility definitions.
501 # define UCHARAT(p) ((int)*(const U8*)(p))
503 # define UCHARAT(p) ((int)*(p)&CHARMASK)
506 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
508 #define REG_SEEN_ZERO_LEN 0x00000001
509 #define REG_SEEN_LOOKBEHIND 0x00000002
510 #define REG_SEEN_GPOS 0x00000004
512 #define REG_SEEN_CANY 0x00000010
513 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
514 #define REG_SEEN_RECURSE 0x00000020
515 #define REG_TOP_LEVEL_BRANCHES 0x00000040
516 #define REG_SEEN_VERBARG 0x00000080
517 #define REG_SEEN_CUTGROUP 0x00000100
518 #define REG_SEEN_RUN_ON_COMMENT 0x00000200
519 #define REG_SEEN_EXACTF_SHARP_S 0x00000400
523 #ifdef PLUGGABLE_RE_EXTENSION
524 #include "re_nodes.h"
526 #include "regnodes.h"
529 #ifndef PLUGGABLE_RE_EXTENSION
531 EXTCONST regexp_engine PL_core_reg_engine;
533 EXTCONST regexp_engine PL_core_reg_engine = {
536 Perl_re_intuit_start,
537 Perl_re_intuit_string,
538 Perl_regfree_internal,
539 Perl_reg_numbered_buff_fetch,
540 Perl_reg_numbered_buff_store,
541 Perl_reg_numbered_buff_length,
543 Perl_reg_named_buff_iter,
545 #if defined(USE_ITHREADS)
546 Perl_regdupe_internal,
551 #endif /* PLUGGABLE_RE_EXTENSION */
557 /* .what is a character array with one character for each member of .data
558 * The character describes the function of the corresponding .data item:
559 * a - AV for paren_name_list under DEBUGGING
560 * f - start-class data for regstclass optimization
561 * l - start op for literal (?{EVAL}) item
562 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
563 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
564 * s - swash for Unicode-style character class, and the multicharacter
565 * strings resulting from casefolding the single-character entries
566 * in the character class
568 * u - trie struct's widecharmap (a HV, so can't share, must dup)
569 * also used for revcharmap and words under DEBUGGING
570 * T - aho-trie struct
571 * S - sv for named capture lookup
572 * 20010712 mjd@plover.com
573 * (Remember to update re_dup() and pregfree() if you add any items.)
581 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
582 anchored* and float* via array indexes 0 and 1. */
583 #define anchored_substr substrs->data[0].substr
584 #define anchored_utf8 substrs->data[0].utf8_substr
585 #define anchored_offset substrs->data[0].min_offset
586 #define anchored_end_shift substrs->data[0].end_shift
588 #define float_substr substrs->data[1].substr
589 #define float_utf8 substrs->data[1].utf8_substr
590 #define float_min_offset substrs->data[1].min_offset
591 #define float_max_offset substrs->data[1].max_offset
592 #define float_end_shift substrs->data[1].end_shift
594 #define check_substr substrs->data[2].substr
595 #define check_utf8 substrs->data[2].utf8_substr
596 #define check_offset_min substrs->data[2].min_offset
597 #define check_offset_max substrs->data[2].max_offset
598 #define check_end_shift substrs->data[2].end_shift
600 #define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
601 #define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
602 #define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
603 #define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
605 /* trie related stuff */
607 /* a transition record for the state machine. the
608 check field determines which state "owns" the
609 transition. the char the transition is for is
610 determined by offset from the owning states base
611 field. the next field determines which state
612 is to be transitioned to if any.
614 struct _reg_trie_trans {
619 /* a transition list element for the list based representation */
620 struct _reg_trie_trans_list_elem {
624 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
626 /* a state for compressed nodes. base is an offset
627 into an array of reg_trie_trans array. If wordnum is
628 nonzero the state is accepting. if base is zero then
629 the state has no children (and will be accepting)
631 struct _reg_trie_state {
635 reg_trie_trans_le* list;
639 /* info per word; indexed by wordnum */
641 U16 prev; /* previous word in acceptance chain; eg in
642 * zzz|abc|ab/ after matching the chars abc, the
643 * accepted word is #2, and the previous accepted
645 U32 len; /* how many chars long is this word? */
646 U32 accept; /* accept state for this word */
650 typedef struct _reg_trie_state reg_trie_state;
651 typedef struct _reg_trie_trans reg_trie_trans;
654 /* anything in here that needs to be freed later
655 should be dealt with in pregfree.
656 refcount is first in both this and _reg_ac_data to allow a space
657 optimisation in Perl_regdupe. */
658 struct _reg_trie_data {
659 U32 refcount; /* number of times this trie is referenced */
660 U32 lasttrans; /* last valid transition element */
661 U16 *charmap; /* byte to charid lookup array */
662 reg_trie_state *states; /* state data */
663 reg_trie_trans *trans; /* array of transition elements */
664 char *bitmap; /* stclass bitmap */
665 U16 *jump; /* optional 1 indexed array of offsets before tail
666 for the node following a given word. */
667 reg_trie_wordinfo *wordinfo; /* array of info per word */
668 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
669 U32 startstate; /* initial state - used for common prefix optimisation */
670 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
671 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
672 U32 prefixlen; /* #chars in common prefix */
673 U32 statecount; /* Build only - number of states in the states array
674 (including the unused zero state) */
675 U32 wordcount; /* Build only */
677 STRLEN charcount; /* Build only */
680 /* There is one (3 under DEBUGGING) pointers that logically belong in this
681 structure, but are held outside as they need duplication on thread cloning,
682 whereas the rest of the structure can be read only:
683 HV *widecharmap; code points > 255 to charid
685 AV *words; Array of words contained in trie, for dumping
686 AV *revcharmap; Map of each charid back to its character representation
690 #define TRIE_WORDS_OFFSET 2
692 typedef struct _reg_trie_data reg_trie_data;
694 /* refcount is first in both this and _reg_trie_data to allow a space
695 optimisation in Perl_regdupe. */
696 struct _reg_ac_data {
700 reg_trie_state *states;
702 typedef struct _reg_ac_data reg_ac_data;
704 /* ANY_BIT doesn't use the structure, so we can borrow it here.
705 This is simpler than refactoring all of it as wed end up with
706 three different sets... */
708 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
709 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
710 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
711 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
712 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
714 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
715 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
718 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
719 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
721 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
722 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
723 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
724 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
727 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
729 #define TRIE_CHARCOUNT(trie) (trie_charcount)
732 #define RE_TRIE_MAXBUF_INIT 65536
733 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
734 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
738 RE_DEBUG_FLAGS is used to control what debug output is emitted
739 its divided into three groups of options, some of which interact.
740 The three groups are: Compile, Execute, Extra. There is room for a
741 further group, as currently only the low three bytes are used.
762 If you modify any of these make sure you make corresponding changes to
763 re.pm, especially to the documentation.
769 #define RE_DEBUG_COMPILE_MASK 0x0000FF
770 #define RE_DEBUG_COMPILE_PARSE 0x000001
771 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
772 #define RE_DEBUG_COMPILE_TRIE 0x000004
773 #define RE_DEBUG_COMPILE_DUMP 0x000008
774 #define RE_DEBUG_COMPILE_FLAGS 0x000010
777 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
778 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
779 #define RE_DEBUG_EXECUTE_MATCH 0x000200
780 #define RE_DEBUG_EXECUTE_TRIE 0x000400
783 #define RE_DEBUG_EXTRA_MASK 0xFF0000
784 #define RE_DEBUG_EXTRA_TRIE 0x010000
785 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
786 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
787 #define RE_DEBUG_EXTRA_STATE 0x080000
788 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
789 #define RE_DEBUG_EXTRA_BUFFERS 0x400000
790 #define RE_DEBUG_EXTRA_GPOS 0x800000
792 #define RE_DEBUG_EXTRA_STACK 0x280000
794 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
796 #define DEBUG_COMPILE_r(x) DEBUG_r( \
797 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
798 #define DEBUG_PARSE_r(x) DEBUG_r( \
799 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
800 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
801 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
802 #define DEBUG_PARSE_r(x) DEBUG_r( \
803 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
804 #define DEBUG_DUMP_r(x) DEBUG_r( \
805 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
806 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
807 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
808 #define DEBUG_FLAGS_r(x) DEBUG_r( \
809 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
811 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
812 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
813 #define DEBUG_INTUIT_r(x) DEBUG_r( \
814 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
815 #define DEBUG_MATCH_r(x) DEBUG_r( \
816 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
817 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
818 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
821 #define DEBUG_EXTRA_r(x) DEBUG_r( \
822 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
823 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
824 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
825 #define DEBUG_STATE_r(x) DEBUG_r( \
826 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
827 #define DEBUG_STACK_r(x) DEBUG_r( \
828 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
829 #define DEBUG_BUFFERS_r(x) DEBUG_r( \
830 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
832 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
833 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
834 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
835 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
836 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
837 Perl_warn_nocontext x )
838 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
839 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
840 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
841 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
843 #define DEBUG_TRIE_r(x) DEBUG_r( \
844 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
845 | RE_DEBUG_EXECUTE_TRIE )) x )
846 #define DEBUG_GPOS_r(x) DEBUG_r( \
847 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
850 /* get_sv() can return NULL during global destruction. re_debug_flags can get
851 * clobbered by a longjmp, so must be initialized */
852 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
853 SV * re_debug_flags_sv = NULL; \
854 re_debug_flags = 0; \
855 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
856 if (re_debug_flags_sv) { \
857 if (!SvIOK(re_debug_flags_sv)) \
858 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
859 re_debug_flags=SvIV(re_debug_flags_sv); \
865 #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
866 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
868 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
869 const char * const rpv = \
870 pv_pretty((dsv), (pv), (l), (m), \
871 PL_colors[(c1)],PL_colors[(c2)], \
872 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
873 const int rlen = SvCUR(dsv)
875 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
876 const char * const rpv = \
877 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
878 PL_colors[(c1)],PL_colors[(c2)], \
879 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
881 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
882 const char * const rpv = \
883 pv_pretty((dsv), (pv), (l), (m), \
884 PL_colors[0], PL_colors[1], \
885 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
886 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
889 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
890 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
892 #else /* if not DEBUGGING */
894 #define GET_RE_DEBUG_FLAGS_DECL
895 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
896 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
897 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
898 #define RE_SV_DUMPLEN(ItEm)
899 #define RE_SV_TAIL(ItEm)
901 #endif /* DEBUG RELATED DEFINES */
905 * c-indentation-style: bsd
907 * indent-tabs-mode: nil
910 * ex: set ts=8 sts=4 sw=4 et: