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 typedef OP OP_4tree; /* Will be redefined later. */
15 /* Convert branch sequences to more efficient trie ops? */
16 #define PERL_ENABLE_TRIE_OPTIMISATION 1
18 /* Be really aggressive about optimising patterns with trie sequences? */
19 #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
21 /* Use old style unicode mappings for perl and posix character classes
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 * consistent regardless of internal string type, and make character class inversions
30 * consistent. The tests that fail in the regex engine are basically broken tests.
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
38 #define PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS 1
40 /* Should the optimiser take positive assertions into account? */
41 #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
43 /* Not for production use: */
44 #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
46 /* Activate offsets code - set to if 1 to enable */
48 #define RE_TRACK_PATTERN_OFFSETS
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.
56 /* #define REG_ALLOW_MINMOD_SUSPEND */
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:
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]
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
78 * [regmust is now supplied always. The tests that use regmust have a
79 * heuristic that disables the test if it usually matches.]
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.]
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.
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.)
112 * Using two bytes for the "next" pointer is vast overkill for most things,
113 * but allows patterns to get big without disasters.
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
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. */
126 typedef struct regexp_internal {
127 int name_list_idx; /* Optional data index of an array of paren names */
129 U32 *offsets; /* offset annotations 20001228 MJD
130 data about mapping the program to the
132 offsets[0] is proglen when this is used
137 regnode *regstclass; /* Optional startclass as identified or constructed
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. */
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)
150 * Flags stored in regexp->intflags
151 * These are used only internally to the regexp engine
153 * See regexp.h for flags used externally to the regexp engine
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
162 /* this is where the old regcomp.h started */
164 struct regnode_string {
171 /* Argument bearing node - workhorse,
172 arg1 is often for the data field */
180 /* Similar to a regnode_1 but with an extra signed argument */
189 /* 'Two field' -- Two 16 bit unsigned args */
199 #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
200 #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
202 /* also used by trie */
203 struct regnode_charclass {
207 U32 arg1; /* used as ptr in S_regclass */
208 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
211 /* has runtime (locale) \d, \w, ..., [:posix:] classes */
212 struct regnode_charclass_class {
213 U8 flags; /* ANYOF_CLASS bit must go here */
216 U32 arg1; /* used as ptr in S_regclass */
217 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
218 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
221 /* XXX fix this description.
222 Impose a limit of REG_INFTY on various pattern matching operations
223 to limit stack growth and to avoid "infinite" recursions.
225 /* The default size for REG_INFTY is I16_MAX, which is the same as
226 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
227 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
228 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
229 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
231 --Andy Dougherty 11 June 1998
235 # define REG_INFTY ((1<<15)-1)
240 # define REG_INFTY I16_MAX
243 #define ARG_VALUE(arg) (arg)
244 #define ARG__SET(arg,val) ((arg) = (val))
250 #define ARG(p) ARG_VALUE(ARG_LOC(p))
251 #define ARG1(p) ARG_VALUE(ARG1_LOC(p))
252 #define ARG2(p) ARG_VALUE(ARG2_LOC(p))
253 #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
255 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
256 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
257 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
258 #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
263 #define NEXT_OFF(p) ((p)->next_off)
264 #define NODE_ALIGN(node)
265 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
267 #define SIZE_ALIGN NODE_ALIGN
274 #define OP(p) ((p)->type)
275 #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
276 regnode types. For some, it's the \
277 character set of the regnode */
278 #define OPERAND(p) (((struct regnode_string *)p)->string)
279 #define MASK(p) ((char*)OPERAND(p))
280 #define STR_LEN(p) (((struct regnode_string *)p)->str_len)
281 #define STRING(p) (((struct regnode_string *)p)->string)
282 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
283 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
290 #define NODE_ALIGN(node)
291 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
292 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
293 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
294 #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
296 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
297 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
299 #define NODE_STEP_B 4
301 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
302 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
304 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
305 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
306 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
307 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
309 #define REG_MAGIC 0234
311 #define SIZE_ONLY (RExC_emit == &PL_regdummy)
313 /* Flags for node->flags of several of the node types */
316 /* Flags for node->flags of ANYOF. These are in short supply, so some games
317 * are done to share them, as described below. If necessary, the ANYOF_LOCALE
318 * and ANYOF_CLASS bits could be shared with a space penalty for locale nodes
319 * (and the code at the time this comment was written, is written so that all
320 * that is necessary to make the change would be to redefine the ANYOF_CLASS
321 * define). Once the planned change to compile all the above-latin1 code points
322 * is done, then the UNICODE_ALL bit can be freed up. If flags need to be
323 * added that are applicable to the synthetic start class only, with some work,
324 * they could be put in the next-node field, or in an unused bit of the
325 * classflags field. */
327 #define ANYOF_LOCALE 0x01
329 /* The fold is calculated and stored in the bitmap where possible at compile
330 * time. However there are two cases where it isn't possible. These share
331 * this bit: 1) under locale, where the actual folding varies depending on
332 * what the locale is at the time of execution; and 2) where the folding is
333 * specified in a swash, not the bitmap, such as characters which aren't
334 * specified in the bitmap, or properties that aren't looked at at compile time
336 #define ANYOF_LOC_NONBITMAP_FOLD 0x02
338 #define ANYOF_INVERT 0x04
340 /* EOS, meaning that it can match an empty string too, is used for the
341 * synthetic start class (ssc) only. It can share the INVERT bit, as the ssc
342 * is never inverted. The bit just needs to be turned off before regexec.c
343 * gets a hold of it so that regexec.c doesn't think it's inverted, but this
344 * happens automatically, as if the ssc can match an EOS, the ssc is discarded,
345 * and never passed to regexec.c */
346 #define ANYOF_EOS ANYOF_INVERT
348 /* CLASS is never set unless LOCALE is too: has runtime \d, \w, [:posix:], ...
349 * The non-locale ones are resolved at compile-time */
350 #define ANYOF_CLASS 0x08
351 #define ANYOF_LARGE ANYOF_CLASS /* Same; name retained for back compat */
353 /* Can match something outside the bitmap that is expressible only in utf8 */
354 #define ANYOF_UTF8 0x10
356 /* Can match something outside the bitmap that isn't in utf8 */
357 #define ANYOF_NONBITMAP_NON_UTF8 0x20
359 /* Set if the bitmap doesn't fully represent what this node can match */
360 #define ANYOF_NONBITMAP (ANYOF_UTF8|ANYOF_NONBITMAP_NON_UTF8)
361 #define ANYOF_UNICODE ANYOF_NONBITMAP /* old name, for back compat */
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 /* Character classes for node->classflags of ANYOF */
373 /* Should be synchronized with a table in regprop() */
374 /* 2n should pair with 2n+1 */
376 #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
377 #define ANYOF_NALNUM 1
378 #define ANYOF_SPACE 2 /* \s */
379 #define ANYOF_NSPACE 3
380 #define ANYOF_DIGIT 4 /* \d */
381 #define ANYOF_NDIGIT 5
382 #define ANYOF_ALNUMC 6 /* [[:alnum:]] isalnum(3), utf8::IsAlnum, ALNUMC */
383 #define ANYOF_NALNUMC 7
384 #define ANYOF_ALPHA 8
385 #define ANYOF_NALPHA 9
386 #define ANYOF_ASCII 10
387 #define ANYOF_NASCII 11
388 #define ANYOF_CNTRL 12
389 #define ANYOF_NCNTRL 13
390 #define ANYOF_GRAPH 14
391 #define ANYOF_NGRAPH 15
392 #define ANYOF_LOWER 16
393 #define ANYOF_NLOWER 17
394 #define ANYOF_PRINT 18
395 #define ANYOF_NPRINT 19
396 #define ANYOF_PUNCT 20
397 #define ANYOF_NPUNCT 21
398 #define ANYOF_UPPER 22
399 #define ANYOF_NUPPER 23
400 #define ANYOF_XDIGIT 24
401 #define ANYOF_NXDIGIT 25
402 #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
403 #define ANYOF_NPSXSPC 27
404 #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
405 #define ANYOF_NBLANK 29
409 /* pseudo classes, not stored in the class bitmap, but used as flags
410 during compilation of char classes */
412 #define ANYOF_VERTWS (ANYOF_MAX+1)
413 #define ANYOF_NVERTWS (ANYOF_MAX+2)
414 #define ANYOF_HORIZWS (ANYOF_MAX+3)
415 #define ANYOF_NHORIZWS (ANYOF_MAX+4)
417 /* Backward source code compatibility. */
419 #define ANYOF_ALNUML ANYOF_ALNUM
420 #define ANYOF_NALNUML ANYOF_NALNUM
421 #define ANYOF_SPACEL ANYOF_SPACE
422 #define ANYOF_NSPACEL ANYOF_NSPACE
424 /* Utility macros for the bitmap and classes of ANYOF */
426 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
427 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
429 #define ANYOF_FLAGS(p) ((p)->flags)
431 #define ANYOF_BIT(c) (1 << ((c) & 7))
433 #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
434 #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
435 #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
436 #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
438 #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
439 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
441 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
442 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
443 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
444 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
445 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
447 #define ANYOF_BITMAP_SETALL(p) \
448 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
449 #define ANYOF_BITMAP_CLEARALL(p) \
450 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
451 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */
452 #define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
453 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)
455 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
456 #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
458 /* The class bit can be set to the locale one if necessary to save bits at the
459 * expense of having locale ANYOF nodes always have a class bit map, and hence
460 * take up extra space. This allows convenient changing it as development
461 * proceeds on this */
462 #if ANYOF_CLASS == ANYOF_LOCALE
463 # undef ANYOF_CLASS_ADD_SKIP
464 # define ANYOF_ADD_LOC_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
466 /* Quicker way to see if there are actually any tests. This is because
467 * currently the set of tests can be empty even when the class bitmap is
469 # if ANYOF_CLASSBITMAP_SIZE != 4
470 # error ANYOF_CLASSBITMAP_SIZE is expected to be 4
472 # define ANYOF_CLASS_TEST_ANY_SET(p) /* assumes sizeof(p) = 4 */ \
473 memNE (((struct regnode_charclass_class*)(p))->classflags, \
474 "\0\0\0\0", ANYOF_CLASSBITMAP_SIZE)
476 # define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
477 # undef ANYOF_ADD_LOC_SKIP
478 # define ANYOF_CLASS_TEST_ANY_SET(p) (ANYOF_FLAGS(p) & ANYOF_CLASS)
483 * Utility definitions.
486 # define UCHARAT(p) ((int)*(const U8*)(p))
488 # define UCHARAT(p) ((int)*(p)&CHARMASK)
491 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
493 #define REG_SEEN_ZERO_LEN 0x00000001
494 #define REG_SEEN_LOOKBEHIND 0x00000002
495 #define REG_SEEN_GPOS 0x00000004
496 #define REG_SEEN_EVAL 0x00000008
497 #define REG_SEEN_CANY 0x00000010
498 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
499 #define REG_SEEN_RECURSE 0x00000020
500 #define REG_TOP_LEVEL_BRANCHES 0x00000040
501 #define REG_SEEN_VERBARG 0x00000080
502 #define REG_SEEN_CUTGROUP 0x00000100
503 #define REG_SEEN_RUN_ON_COMMENT 0x00000200
507 #ifdef PLUGGABLE_RE_EXTENSION
508 #include "re_nodes.h"
510 #include "regnodes.h"
513 #ifndef PLUGGABLE_RE_EXTENSION
515 EXTCONST regexp_engine PL_core_reg_engine;
517 EXTCONST regexp_engine PL_core_reg_engine = {
520 Perl_re_intuit_start,
521 Perl_re_intuit_string,
522 Perl_regfree_internal,
523 Perl_reg_numbered_buff_fetch,
524 Perl_reg_numbered_buff_store,
525 Perl_reg_numbered_buff_length,
527 Perl_reg_named_buff_iter,
529 #if defined(USE_ITHREADS)
530 Perl_regdupe_internal
534 #endif /* PLUGGABLE_RE_EXTENSION */
540 /* .what is a character array with one character for each member of .data
541 * The character describes the function of the corresponding .data item:
542 * a - AV for paren_name_list under DEBUGGING
543 * f - start-class data for regstclass optimization
544 * n - Root of op tree for (?{EVAL}) item
545 * o - Start op for (?{EVAL}) item
546 * p - Pad for (?{EVAL}) item
547 * s - swash for Unicode-style character class, and the multicharacter
548 * strings resulting from casefolding the single-character entries
549 * in the character class
551 * u - trie struct's widecharmap (a HV, so can't share, must dup)
552 * also used for revcharmap and words under DEBUGGING
553 * T - aho-trie struct
554 * S - sv for named capture lookup
555 * 20010712 mjd@plover.com
556 * (Remember to update re_dup() and pregfree() if you add any items.)
564 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
565 anchored* and float* via array indexes 0 and 1. */
566 #define anchored_substr substrs->data[0].substr
567 #define anchored_utf8 substrs->data[0].utf8_substr
568 #define anchored_offset substrs->data[0].min_offset
569 #define anchored_end_shift substrs->data[0].end_shift
571 #define float_substr substrs->data[1].substr
572 #define float_utf8 substrs->data[1].utf8_substr
573 #define float_min_offset substrs->data[1].min_offset
574 #define float_max_offset substrs->data[1].max_offset
575 #define float_end_shift substrs->data[1].end_shift
577 #define check_substr substrs->data[2].substr
578 #define check_utf8 substrs->data[2].utf8_substr
579 #define check_offset_min substrs->data[2].min_offset
580 #define check_offset_max substrs->data[2].max_offset
581 #define check_end_shift substrs->data[2].end_shift
583 #define RX_ANCHORED_SUBSTR(rx) (((struct regexp *)SvANY(rx))->anchored_substr)
584 #define RX_ANCHORED_UTF8(rx) (((struct regexp *)SvANY(rx))->anchored_utf8)
585 #define RX_FLOAT_SUBSTR(rx) (((struct regexp *)SvANY(rx))->float_substr)
586 #define RX_FLOAT_UTF8(rx) (((struct regexp *)SvANY(rx))->float_utf8)
588 /* trie related stuff */
590 /* a transition record for the state machine. the
591 check field determines which state "owns" the
592 transition. the char the transition is for is
593 determined by offset from the owning states base
594 field. the next field determines which state
595 is to be transitioned to if any.
597 struct _reg_trie_trans {
602 /* a transition list element for the list based representation */
603 struct _reg_trie_trans_list_elem {
607 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
609 /* a state for compressed nodes. base is an offset
610 into an array of reg_trie_trans array. If wordnum is
611 nonzero the state is accepting. if base is zero then
612 the state has no children (and will be accepting)
614 struct _reg_trie_state {
618 reg_trie_trans_le* list;
622 /* info per word; indexed by wordnum */
624 U16 prev; /* previous word in acceptance chain; eg in
625 * zzz|abc|ab/ after matching the chars abc, the
626 * accepted word is #2, and the previous accepted
628 U32 len; /* how many chars long is this word? */
629 U32 accept; /* accept state for this word */
633 typedef struct _reg_trie_state reg_trie_state;
634 typedef struct _reg_trie_trans reg_trie_trans;
637 /* anything in here that needs to be freed later
638 should be dealt with in pregfree.
639 refcount is first in both this and _reg_ac_data to allow a space
640 optimisation in Perl_regdupe. */
641 struct _reg_trie_data {
642 U32 refcount; /* number of times this trie is referenced */
643 U32 lasttrans; /* last valid transition element */
644 U16 *charmap; /* byte to charid lookup array */
645 reg_trie_state *states; /* state data */
646 reg_trie_trans *trans; /* array of transition elements */
647 char *bitmap; /* stclass bitmap */
648 U16 *jump; /* optional 1 indexed array of offsets before tail
649 for the node following a given word. */
650 reg_trie_wordinfo *wordinfo; /* array of info per word */
651 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
652 U32 startstate; /* initial state - used for common prefix optimisation */
653 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
654 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
655 U32 prefixlen; /* #chars in common prefix */
656 U32 statecount; /* Build only - number of states in the states array
657 (including the unused zero state) */
658 U32 wordcount; /* Build only */
660 STRLEN charcount; /* Build only */
663 /* There is one (3 under DEBUGGING) pointers that logically belong in this
664 structure, but are held outside as they need duplication on thread cloning,
665 whereas the rest of the structure can be read only:
666 HV *widecharmap; code points > 255 to charid
668 AV *words; Array of words contained in trie, for dumping
669 AV *revcharmap; Map of each charid back to its character representation
673 #define TRIE_WORDS_OFFSET 2
675 typedef struct _reg_trie_data reg_trie_data;
677 /* refcount is first in both this and _reg_trie_data to allow a space
678 optimisation in Perl_regdupe. */
679 struct _reg_ac_data {
683 reg_trie_state *states;
685 typedef struct _reg_ac_data reg_ac_data;
687 /* ANY_BIT doesn't use the structure, so we can borrow it here.
688 This is simpler than refactoring all of it as wed end up with
689 three different sets... */
691 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
692 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
693 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
694 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
695 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
697 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
698 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
701 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
702 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
704 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
705 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
706 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
707 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
710 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
712 #define TRIE_CHARCOUNT(trie) (trie_charcount)
715 #define RE_TRIE_MAXBUF_INIT 65536
716 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
717 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
721 RE_DEBUG_FLAGS is used to control what debug output is emitted
722 its divided into three groups of options, some of which interact.
723 The three groups are: Compile, Execute, Extra. There is room for a
724 further group, as currently only the low three bytes are used.
745 If you modify any of these make sure you make corresponding changes to
746 re.pm, especially to the documentation.
752 #define RE_DEBUG_COMPILE_MASK 0x0000FF
753 #define RE_DEBUG_COMPILE_PARSE 0x000001
754 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
755 #define RE_DEBUG_COMPILE_TRIE 0x000004
756 #define RE_DEBUG_COMPILE_DUMP 0x000008
757 #define RE_DEBUG_COMPILE_FLAGS 0x000010
760 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
761 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
762 #define RE_DEBUG_EXECUTE_MATCH 0x000200
763 #define RE_DEBUG_EXECUTE_TRIE 0x000400
766 #define RE_DEBUG_EXTRA_MASK 0xFF0000
767 #define RE_DEBUG_EXTRA_TRIE 0x010000
768 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
769 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
770 #define RE_DEBUG_EXTRA_STATE 0x080000
771 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
772 #define RE_DEBUG_EXTRA_BUFFERS 0x400000
773 #define RE_DEBUG_EXTRA_GPOS 0x800000
775 #define RE_DEBUG_EXTRA_STACK 0x280000
777 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
779 #define DEBUG_COMPILE_r(x) DEBUG_r( \
780 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
781 #define DEBUG_PARSE_r(x) DEBUG_r( \
782 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
783 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
784 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
785 #define DEBUG_PARSE_r(x) DEBUG_r( \
786 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
787 #define DEBUG_DUMP_r(x) DEBUG_r( \
788 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
789 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
790 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
791 #define DEBUG_FLAGS_r(x) DEBUG_r( \
792 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
794 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
795 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
796 #define DEBUG_INTUIT_r(x) DEBUG_r( \
797 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
798 #define DEBUG_MATCH_r(x) DEBUG_r( \
799 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
800 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
801 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
804 #define DEBUG_EXTRA_r(x) DEBUG_r( \
805 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
806 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
807 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
808 #define DEBUG_STATE_r(x) DEBUG_r( \
809 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
810 #define DEBUG_STACK_r(x) DEBUG_r( \
811 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
812 #define DEBUG_BUFFERS_r(x) DEBUG_r( \
813 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
815 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
816 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
817 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
818 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
819 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
820 Perl_warn_nocontext x )
821 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
822 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
823 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
824 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
826 #define DEBUG_TRIE_r(x) DEBUG_r( \
827 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
828 | RE_DEBUG_EXECUTE_TRIE )) x )
829 #define DEBUG_GPOS_r(x) DEBUG_r( \
830 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
833 /* get_sv() can return NULL during global destruction. re_debug_flags can get
834 * clobbered by a longjmp, so must be initialized */
835 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
836 SV * re_debug_flags_sv = NULL; \
837 re_debug_flags = 0; \
838 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
839 if (re_debug_flags_sv) { \
840 if (!SvIOK(re_debug_flags_sv)) \
841 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
842 re_debug_flags=SvIV(re_debug_flags_sv); \
848 #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
850 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
851 const char * const rpv = \
852 pv_pretty((dsv), (pv), (l), (m), \
853 PL_colors[(c1)],PL_colors[(c2)], \
854 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
855 const int rlen = SvCUR(dsv)
857 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
858 const char * const rpv = \
859 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
860 PL_colors[(c1)],PL_colors[(c2)], \
861 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
863 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
864 const char * const rpv = \
865 pv_pretty((dsv), (pv), (l), (m), \
866 PL_colors[0], PL_colors[1], \
867 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
868 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
871 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
872 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
874 #else /* if not DEBUGGING */
876 #define GET_RE_DEBUG_FLAGS_DECL
877 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
878 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
879 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
880 #define RE_SV_DUMPLEN(ItEm)
881 #define RE_SV_TAIL(ItEm)
883 #endif /* DEBUG RELATED DEFINES */
887 * c-indentation-style: bsd
889 * indent-tabs-mode: t
892 * ex: set ts=8 sts=4 sw=4 noet: