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 agressive 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 * 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.
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 {
208 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
211 struct regnode_charclass_class { /* has [[:blah:]] classes */
212 U8 flags; /* should have ANYOF_CLASS here */
216 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
217 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
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.
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
230 --Andy Dougherty 11 June 1998
234 # define REG_INFTY ((1<<15)-1)
239 # define REG_INFTY I16_MAX
242 #define ARG_VALUE(arg) (arg)
243 #define ARG__SET(arg,val) ((arg) = (val))
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))
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))
262 #define NEXT_OFF(p) ((p)->next_off)
263 #define NODE_ALIGN(node)
264 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
266 #define SIZE_ALIGN NODE_ALIGN
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)
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)
292 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
293 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
295 #define NODE_STEP_B 4
297 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
298 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
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
305 #define REG_MAGIC 0234
307 #define SIZE_ONLY (RExC_emit == &PL_regdummy)
309 /* Flags for node->flags of ANYOF */
311 #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
312 #define ANYOF_INVERT 0x04
313 #define ANYOF_FOLD 0x02
314 #define ANYOF_LOCALE 0x01
316 /* Used for regstclass only */
317 #define ANYOF_EOS 0x10 /* Can match an empty string too */
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 */
323 /* size of node is large (includes class pointer) */
324 #define ANYOF_LARGE 0x80
326 /* Are there any runtime flags on in this node? */
327 #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
329 #define ANYOF_FLAGS_ALL 0xff
331 /* Character classes for node->classflags of ANYOF */
332 /* Should be synchronized with a table in regprop() */
333 /* 2n should pair with 2n+1 */
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
368 /* pseudo classes, not stored in the class bitmap, but used as flags
369 during compilation of char classes */
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)
376 /* Backward source code compatibility. */
378 #define ANYOF_ALNUML ANYOF_ALNUM
379 #define ANYOF_NALNUML ANYOF_NALNUM
380 #define ANYOF_SPACEL ANYOF_SPACE
381 #define ANYOF_NSPACEL ANYOF_NSPACE
383 /* Utility macros for the bitmap and classes of ANYOF */
385 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
386 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
388 #define ANYOF_FLAGS(p) ((p)->flags)
390 #define ANYOF_BIT(c) (1 << ((c) & 7))
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))
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)
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))
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)
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)
420 * Utility definitions.
423 # define UCHARAT(p) ((int)*(const U8*)(p))
425 # define UCHARAT(p) ((int)*(p)&CHARMASK)
428 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
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
444 #ifdef PLUGGABLE_RE_EXTENSION
445 #include "re_nodes.h"
447 #include "regnodes.h"
450 #ifndef PLUGGABLE_RE_EXTENSION
452 EXTCONST regexp_engine PL_core_reg_engine;
454 EXTCONST regexp_engine PL_core_reg_engine = {
457 Perl_re_intuit_start,
458 Perl_re_intuit_string,
459 Perl_regfree_internal,
460 Perl_reg_numbered_buff_fetch,
461 Perl_reg_numbered_buff_store,
462 Perl_reg_numbered_buff_length,
464 Perl_reg_named_buff_iter,
466 #if defined(USE_ITHREADS)
467 Perl_regdupe_internal
471 #endif /* PLUGGABLE_RE_EXTENSION */
477 /* .what is a character array with one character for each member of .data
478 * The character describes the function of the corresponding .data item:
479 * a - AV for paren_name_list under DEBUGGING
480 * f - start-class data for regstclass optimization
481 * n - Root of op tree for (?{EVAL}) item
482 * o - Start op for (?{EVAL}) item
483 * p - Pad for (?{EVAL}) item
484 * s - swash for Unicode-style character class, and the multicharacter
485 * strings resulting from casefolding the single-character entries
486 * in the character class
488 * u - trie struct's widecharmap (a HV, so can't share, must dup)
489 * also used for revcharmap and words under DEBUGGING
490 * T - aho-trie struct
491 * S - sv for named capture lookup
492 * 20010712 mjd@plover.com
493 * (Remember to update re_dup() and pregfree() if you add any items.)
501 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
502 anchored* and float* via array indexes 0 and 1. */
503 #define anchored_substr substrs->data[0].substr
504 #define anchored_utf8 substrs->data[0].utf8_substr
505 #define anchored_offset substrs->data[0].min_offset
506 #define anchored_end_shift substrs->data[0].end_shift
508 #define float_substr substrs->data[1].substr
509 #define float_utf8 substrs->data[1].utf8_substr
510 #define float_min_offset substrs->data[1].min_offset
511 #define float_max_offset substrs->data[1].max_offset
512 #define float_end_shift substrs->data[1].end_shift
514 #define check_substr substrs->data[2].substr
515 #define check_utf8 substrs->data[2].utf8_substr
516 #define check_offset_min substrs->data[2].min_offset
517 #define check_offset_max substrs->data[2].max_offset
518 #define check_end_shift substrs->data[2].end_shift
520 #define RX_ANCHORED_SUBSTR(rx) (((struct regexp *)SvANY(rx))->anchored_substr)
521 #define RX_ANCHORED_UTF8(rx) (((struct regexp *)SvANY(rx))->anchored_utf8)
522 #define RX_FLOAT_SUBSTR(rx) (((struct regexp *)SvANY(rx))->float_substr)
523 #define RX_FLOAT_UTF8(rx) (((struct regexp *)SvANY(rx))->float_utf8)
525 /* trie related stuff */
527 /* a transition record for the state machine. the
528 check field determines which state "owns" the
529 transition. the char the transition is for is
530 determined by offset from the owning states base
531 field. the next field determines which state
532 is to be transitioned to if any.
534 struct _reg_trie_trans {
539 /* a transition list element for the list based representation */
540 struct _reg_trie_trans_list_elem {
544 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
546 /* a state for compressed nodes. base is an offset
547 into an array of reg_trie_trans array. If wordnum is
548 nonzero the state is accepting. if base is zero then
549 the state has no children (and will be accepting)
551 struct _reg_trie_state {
555 reg_trie_trans_le* list;
559 /* info per word; indexed by wordnum */
561 U16 prev; /* previous word in acceptance chain; eg in
562 * zzz|abc|ab/ after matching the chars abc, the
563 * accepted word is #2, and the previous accepted
565 U32 len; /* how many chars long is this word? */
566 U32 accept; /* accept state for this word */
570 typedef struct _reg_trie_state reg_trie_state;
571 typedef struct _reg_trie_trans reg_trie_trans;
574 /* anything in here that needs to be freed later
575 should be dealt with in pregfree.
576 refcount is first in both this and _reg_ac_data to allow a space
577 optimisation in Perl_regdupe. */
578 struct _reg_trie_data {
579 U32 refcount; /* number of times this trie is referenced */
580 U32 lasttrans; /* last valid transition element */
581 U16 *charmap; /* byte to charid lookup array */
582 reg_trie_state *states; /* state data */
583 reg_trie_trans *trans; /* array of transition elements */
584 char *bitmap; /* stclass bitmap */
585 U16 *jump; /* optional 1 indexed array of offsets before tail
586 for the node following a given word. */
587 reg_trie_wordinfo *wordinfo; /* array of info per word */
588 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
589 U32 startstate; /* initial state - used for common prefix optimisation */
590 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
591 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
592 U32 prefixlen; /* #chars in common prefix */
593 U32 statecount; /* Build only - number of states in the states array
594 (including the unused zero state) */
595 U32 wordcount; /* Build only */
597 STRLEN charcount; /* Build only */
600 /* There is one (3 under DEBUGGING) pointers that logically belong in this
601 structure, but are held outside as they need duplication on thread cloning,
602 whereas the rest of the structure can be read only:
603 HV *widecharmap; code points > 255 to charid
605 AV *words; Array of words contained in trie, for dumping
606 AV *revcharmap; Map of each charid back to its character representation
610 #define TRIE_WORDS_OFFSET 2
612 typedef struct _reg_trie_data reg_trie_data;
614 /* refcount is first in both this and _reg_trie_data to allow a space
615 optimisation in Perl_regdupe. */
616 struct _reg_ac_data {
620 reg_trie_state *states;
622 typedef struct _reg_ac_data reg_ac_data;
624 /* ANY_BIT doesnt use the structure, so we can borrow it here.
625 This is simpler than refactoring all of it as wed end up with
626 three different sets... */
628 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
629 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
630 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
631 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
632 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
634 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
635 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
638 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
639 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
641 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
642 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
643 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
644 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
647 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
649 #define TRIE_CHARCOUNT(trie) (trie_charcount)
652 #define RE_TRIE_MAXBUF_INIT 65536
653 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
654 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
658 RE_DEBUG_FLAGS is used to control what debug output is emitted
659 its divided into three groups of options, some of which interact.
660 The three groups are: Compile, Execute, Extra. There is room for a
661 further group, as currently only the low three bytes are used.
682 If you modify any of these make sure you make corresponding changes to
683 re.pm, especially to the documentation.
689 #define RE_DEBUG_COMPILE_MASK 0x0000FF
690 #define RE_DEBUG_COMPILE_PARSE 0x000001
691 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
692 #define RE_DEBUG_COMPILE_TRIE 0x000004
693 #define RE_DEBUG_COMPILE_DUMP 0x000008
694 #define RE_DEBUG_COMPILE_FLAGS 0x000010
697 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
698 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
699 #define RE_DEBUG_EXECUTE_MATCH 0x000200
700 #define RE_DEBUG_EXECUTE_TRIE 0x000400
703 #define RE_DEBUG_EXTRA_MASK 0xFF0000
704 #define RE_DEBUG_EXTRA_TRIE 0x010000
705 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
706 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
707 #define RE_DEBUG_EXTRA_STATE 0x080000
708 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
709 #define RE_DEBUG_EXTRA_BUFFERS 0x400000
710 #define RE_DEBUG_EXTRA_GPOS 0x800000
712 #define RE_DEBUG_EXTRA_STACK 0x280000
714 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
716 #define DEBUG_COMPILE_r(x) DEBUG_r( \
717 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
718 #define DEBUG_PARSE_r(x) DEBUG_r( \
719 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
720 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
721 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
722 #define DEBUG_PARSE_r(x) DEBUG_r( \
723 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
724 #define DEBUG_DUMP_r(x) DEBUG_r( \
725 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
726 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
727 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
728 #define DEBUG_FLAGS_r(x) DEBUG_r( \
729 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
731 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
732 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
733 #define DEBUG_INTUIT_r(x) DEBUG_r( \
734 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
735 #define DEBUG_MATCH_r(x) DEBUG_r( \
736 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
737 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
738 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
741 #define DEBUG_EXTRA_r(x) DEBUG_r( \
742 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
743 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
744 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
745 #define DEBUG_STATE_r(x) DEBUG_r( \
746 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
747 #define DEBUG_STACK_r(x) DEBUG_r( \
748 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
749 #define DEBUG_BUFFERS_r(x) DEBUG_r( \
750 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
752 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
753 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
754 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
755 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
756 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
757 Perl_warn_nocontext x )
758 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
759 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
760 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
761 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
763 #define DEBUG_TRIE_r(x) DEBUG_r( \
764 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
765 | RE_DEBUG_EXECUTE_TRIE )) x )
766 #define DEBUG_GPOS_r(x) DEBUG_r( \
767 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
770 /* get_sv() can return NULL during global destruction. */
771 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
772 SV * re_debug_flags_sv = NULL; \
773 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
774 if (re_debug_flags_sv) { \
775 if (!SvIOK(re_debug_flags_sv)) \
776 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
777 re_debug_flags=SvIV(re_debug_flags_sv); \
783 #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
785 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
786 const char * const rpv = \
787 pv_pretty((dsv), (pv), (l), (m), \
788 PL_colors[(c1)],PL_colors[(c2)], \
789 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
790 const int rlen = SvCUR(dsv)
792 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
793 const char * const rpv = \
794 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
795 PL_colors[(c1)],PL_colors[(c2)], \
796 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
798 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
799 const char * const rpv = \
800 pv_pretty((dsv), (pv), (l), (m), \
801 PL_colors[0], PL_colors[1], \
802 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELLIPSES | \
803 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
806 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
807 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
809 #else /* if not DEBUGGING */
811 #define GET_RE_DEBUG_FLAGS_DECL
812 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
813 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
814 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
815 #define RE_SV_DUMPLEN(ItEm)
816 #define RE_SV_TAIL(ItEm)
818 #endif /* DEBUG RELATED DEFINES */
822 * c-indentation-style: bsd
824 * indent-tabs-mode: t
827 * ex: set ts=8 sts=4 sw=4 noet: