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 /* Should the optimiser take positive assertions into account? */
22 #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
24 /* Not for production use: */
25 #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
27 /* Activate offsets code - set to if 1 to enable */
29 #define RE_TRACK_PATTERN_OFFSETS
32 /* Unless the next line is uncommented it is illegal to combine lazy
33 matching with possessive matching. Frankly it doesn't make much sense
34 to allow it as X*?+ matches nothing, X+?+ matches a single char only,
35 and X{min,max}?+ matches min times only.
37 /* #define REG_ALLOW_MINMOD_SUSPEND */
40 * The "internal use only" fields in regexp.h are present to pass info from
41 * compile to execute that permits the execute phase to run lots faster on
42 * simple cases. They are:
44 * regstart sv that must begin a match; NULL if none obvious
45 * reganch is the match anchored (at beginning-of-line only)?
46 * regmust string (pointer into program) that match must include, or NULL
47 * [regmust changed to SV* for bminstr()--law]
48 * regmlen length of regmust string
49 * [regmlen not used currently]
51 * Regstart and reganch permit very fast decisions on suitable starting points
52 * for a match, cutting down the work a lot. Regmust permits fast rejection
53 * of lines that cannot possibly match. The regmust tests are costly enough
54 * that pregcomp() supplies a regmust only if the r.e. contains something
55 * potentially expensive (at present, the only such thing detected is * or +
56 * at the start of the r.e., which can involve a lot of backup). Regmlen is
57 * supplied because the test in pregexec() needs it and pregcomp() is computing
59 * [regmust is now supplied always. The tests that use regmust have a
60 * heuristic that disables the test if it usually matches.]
62 * [In fact, we now use regmust in many cases to locate where the search
63 * starts in the string, so if regback is >= 0, the regmust search is never
64 * wasted effort. The regback variable says how many characters back from
65 * where regmust matched is the earliest possible start of the match.
66 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
70 * Structure for regexp "program". This is essentially a linear encoding
71 * of a nondeterministic finite-state machine (aka syntax charts or
72 * "railroad normal form" in parsing technology). Each node is an opcode
73 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
74 * all nodes except BRANCH implement concatenation; a "next" pointer with
75 * a BRANCH on both ends of it is connecting two alternatives. (Here we
76 * have one of the subtle syntax dependencies: an individual BRANCH (as
77 * opposed to a collection of them) is never concatenated with anything
78 * because of operator precedence.) The operand of some types of node is
79 * a literal string; for others, it is a node leading into a sub-FSM. In
80 * particular, the operand of a BRANCH node is the first node of the branch.
81 * (NB this is *not* a tree structure: the tail of the branch connects
82 * to the thing following the set of BRANCHes.) The opcodes are defined
83 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
87 * A node is one char of opcode followed by two chars of "next" pointer.
88 * "Next" pointers are stored as two 8-bit pieces, high order first. The
89 * value is a positive offset from the opcode of the node containing it.
90 * An operand, if any, simply follows the node. (Note that much of the
91 * code generation knows about this implicit relationship.)
93 * Using two bytes for the "next" pointer is vast overkill for most things,
94 * but allows patterns to get big without disasters.
96 * [The "next" pointer is always aligned on an even
97 * boundary, and reads the offset directly as a short. Also, there is no
98 * special test to reverse the sign of BACK pointers since the offset is
102 /* This is the stuff that used to live in regexp.h that was truly
103 private to the engine itself. It now lives here. */
107 typedef struct regexp_internal {
108 int name_list_idx; /* Optional data index of an array of paren names */
110 U32 *offsets; /* offset annotations 20001228 MJD
111 data about mapping the program to the
113 offsets[0] is proglen when this is used
118 regnode *regstclass; /* Optional startclass as identified or constructed
120 struct reg_data *data; /* Additional miscellaneous data used by the program.
121 Used to make it easier to clone and free arbitrary
122 data that the regops need. Often the ARG field of
123 a regop is an index into this structure */
124 regnode program[1]; /* Unwarranted chumminess with compiler. */
127 #define RXi_SET(x,y) (x)->pprivate = (void*)(y)
128 #define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
129 #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
131 * Flags stored in regexp->intflags
132 * These are used only internally to the regexp engine
134 * See regexp.h for flags used externally to the regexp engine
136 #define PREGf_SKIP 0x00000001
137 #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
138 #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
139 #define PREGf_VERBARG_SEEN 0x00000008
140 #define PREGf_CUTGROUP_SEEN 0x00000010
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) */
181 #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
183 /* also used by trie */
184 struct regnode_charclass {
188 U32 arg1; /* used as ptr in S_regclass */
189 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
192 /* has runtime (locale) \d, \w, ..., [:posix:] classes */
193 struct regnode_charclass_class {
194 U8 flags; /* ANYOF_CLASS bit must go here */
197 U32 arg1; /* used as ptr in S_regclass */
198 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
199 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
202 /* XXX fix this description.
203 Impose a limit of REG_INFTY on various pattern matching operations
204 to limit stack growth and to avoid "infinite" recursions.
206 /* The default size for REG_INFTY is I16_MAX, which is the same as
207 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
208 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
209 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
210 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
212 --Andy Dougherty 11 June 1998
216 # define REG_INFTY ((1<<15)-1)
221 # define REG_INFTY I16_MAX
224 #define ARG_VALUE(arg) (arg)
225 #define ARG__SET(arg,val) ((arg) = (val))
231 #define ARG(p) ARG_VALUE(ARG_LOC(p))
232 #define ARG1(p) ARG_VALUE(ARG1_LOC(p))
233 #define ARG2(p) ARG_VALUE(ARG2_LOC(p))
234 #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
236 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
237 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
238 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
239 #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
244 #define NEXT_OFF(p) ((p)->next_off)
245 #define NODE_ALIGN(node)
246 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
248 #define SIZE_ALIGN NODE_ALIGN
255 #define OP(p) ((p)->type)
256 #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
257 regnode types. For some, it's the \
258 character set of the regnode */
259 #define OPERAND(p) (((struct regnode_string *)p)->string)
260 #define MASK(p) ((char*)OPERAND(p))
261 #define STR_LEN(p) (((struct regnode_string *)p)->str_len)
262 #define STRING(p) (((struct regnode_string *)p)->string)
263 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
264 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
271 #define NODE_ALIGN(node)
272 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
273 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
274 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
275 #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
277 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
278 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
280 #define NODE_STEP_B 4
282 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
283 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
285 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
286 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
287 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
288 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
290 #define REG_MAGIC 0234
292 #define SIZE_ONLY (RExC_emit == &PL_regdummy)
294 /* If the bitmap doesn't fully represent what this ANYOF node can match, the
295 * ARG is set to this special value (since 0, 1, ... are legal, but will never
296 * reach this high). */
297 #define ANYOF_NONBITMAP_EMPTY ((U32) -1)
299 /* The information used to be stored as as combination of the ANYOF_UTF8 and
300 * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there
301 * to free up a bit for other uses. This tries to hide the change from
302 * existing code as much as possible. Now, the data structure that goes in ARG
303 * is not allocated unless it is needed, and that is what is used to determine
304 * if there is something outside the bitmap. The code now assumes that if
305 * that structure exists, that any UTF-8 encoded string should be tried against
306 * it, but a non-UTF8-encoded string will be tried only if the
307 * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */
308 #define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY)
310 /* Flags for node->flags of ANYOF. These are in short supply, so some games
311 * are done to share them, as described below. If necessary, the ANYOF_LOCALE
312 * and ANYOF_CLASS bits could be shared with a space penalty for locale nodes
313 * (and the code at the time this comment was written, is written so that all
314 * that is necessary to make the change would be to redefine the ANYOF_CLASS
315 * define). Once the planned change to compile all the above-latin1 code points
316 * is done, then the UNICODE_ALL bit can be freed up. If flags need to be
317 * added that are applicable to the synthetic start class only, with some work,
318 * they could be put in the next-node field, or in an unused bit of the
319 * classflags field. */
321 #define ANYOF_LOCALE 0x01
323 /* The fold is calculated and stored in the bitmap where possible at compile
324 * time. However there are two cases where it isn't possible. These share
325 * this bit: 1) under locale, where the actual folding varies depending on
326 * what the locale is at the time of execution; and 2) where the folding is
327 * specified in a swash, not the bitmap, such as characters which aren't
328 * specified in the bitmap, or properties that aren't looked at at compile time
330 #define ANYOF_LOC_NONBITMAP_FOLD 0x02
332 #define ANYOF_INVERT 0x04
334 /* EOS, meaning that it can match an empty string too, is used for the
335 * synthetic start class (ssc) only. It looks like it could share the INVERT
336 * bit, as the ssc is never inverted. But doing that caused this reges to
338 * 'foo/file.fob' =~ m,^(?=[^\.])[^/]* /(?=[^\.])[^/]*\.fo[^/]$,;
339 * (except the space between the * and the / above shouldn't be there; it was
340 * inserted to make this comment continue on.)
341 * Rather than try to figure out what was going on in the optimizer, I (khw)
342 * found a way to save a different bit. But my original line of reasoning was
343 * "The bit just needs to be turned off before regexec.c gets a hold of it so
344 * that regexec.c doesn't think it's inverted, but this happens automatically,
345 * as if the ssc can match an EOS, the ssc is discarded, and never passed to
347 #define ANYOF_EOS 0x10
349 /* CLASS is never set unless LOCALE is too: has runtime \d, \w, [:posix:], ...
350 * The non-locale ones are resolved at compile-time */
351 #define ANYOF_CLASS 0x08
352 #define ANYOF_LARGE ANYOF_CLASS /* Same; name retained for back compat */
354 /* Can match something outside the bitmap that isn't in utf8 */
355 #define ANYOF_NONBITMAP_NON_UTF8 0x20
357 /* Matches every code point 0x100 and above*/
358 #define ANYOF_UNICODE_ALL 0x40
360 /* Match all Latin1 characters that aren't ASCII when the target string is not
362 #define ANYOF_NON_UTF8_LATIN1_ALL 0x80
364 #define ANYOF_FLAGS_ALL 0xff
366 /* Character classes for node->classflags of ANYOF */
367 /* Should be synchronized with a table in regprop() */
368 /* 2n should pair with 2n+1 */
370 #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
371 #define ANYOF_NALNUM 1
372 #define ANYOF_SPACE 2 /* \s */
373 #define ANYOF_NSPACE 3
374 #define ANYOF_DIGIT 4 /* \d */
375 #define ANYOF_NDIGIT 5
376 #define ANYOF_ALNUMC 6 /* [[:alnum:]] isalnum(3), utf8::IsAlnum, ALNUMC */
377 #define ANYOF_NALNUMC 7
378 #define ANYOF_ALPHA 8
379 #define ANYOF_NALPHA 9
380 #define ANYOF_ASCII 10
381 #define ANYOF_NASCII 11
382 #define ANYOF_CNTRL 12
383 #define ANYOF_NCNTRL 13
384 #define ANYOF_GRAPH 14
385 #define ANYOF_NGRAPH 15
386 #define ANYOF_LOWER 16
387 #define ANYOF_NLOWER 17
388 #define ANYOF_PRINT 18
389 #define ANYOF_NPRINT 19
390 #define ANYOF_PUNCT 20
391 #define ANYOF_NPUNCT 21
392 #define ANYOF_UPPER 22
393 #define ANYOF_NUPPER 23
394 #define ANYOF_XDIGIT 24
395 #define ANYOF_NXDIGIT 25
396 #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
397 #define ANYOF_NPSXSPC 27
398 #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
399 #define ANYOF_NBLANK 29
403 /* pseudo classes, not stored in the class bitmap, but used as flags
404 during compilation of char classes */
406 #define ANYOF_VERTWS (ANYOF_MAX+1)
407 #define ANYOF_NVERTWS (ANYOF_MAX+2)
408 #define ANYOF_HORIZWS (ANYOF_MAX+3)
409 #define ANYOF_NHORIZWS (ANYOF_MAX+4)
411 /* Backward source code compatibility. */
413 #define ANYOF_ALNUML ANYOF_ALNUM
414 #define ANYOF_NALNUML ANYOF_NALNUM
415 #define ANYOF_SPACEL ANYOF_SPACE
416 #define ANYOF_NSPACEL ANYOF_NSPACE
418 /* Utility macros for the bitmap and classes of ANYOF */
420 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
421 #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
423 #define ANYOF_FLAGS(p) ((p)->flags)
425 #define ANYOF_BIT(c) (1 << ((c) & 7))
427 #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
428 #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
429 #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
430 #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
432 #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
433 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
435 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
436 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
437 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
438 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
439 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
441 #define ANYOF_BITMAP_SETALL(p) \
442 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
443 #define ANYOF_BITMAP_CLEARALL(p) \
444 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
445 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */
446 #define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
447 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)
449 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
450 #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
452 /* The class bit can be set to the locale one if necessary to save bits at the
453 * expense of having locale ANYOF nodes always have a class bit map, and hence
454 * take up extra space. This allows convenient changing it as development
455 * proceeds on this */
456 #if ANYOF_CLASS == ANYOF_LOCALE
457 # undef ANYOF_CLASS_ADD_SKIP
458 # define ANYOF_ADD_LOC_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
460 /* Quicker way to see if there are actually any tests. This is because
461 * currently the set of tests can be empty even when the class bitmap is
463 # if ANYOF_CLASSBITMAP_SIZE != 4
464 # error ANYOF_CLASSBITMAP_SIZE is expected to be 4
466 # define ANYOF_CLASS_TEST_ANY_SET(p) /* assumes sizeof(p) = 4 */ \
467 memNE (((struct regnode_charclass_class*)(p))->classflags, \
468 "\0\0\0\0", ANYOF_CLASSBITMAP_SIZE)
470 # define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
471 # undef ANYOF_ADD_LOC_SKIP
472 # define ANYOF_CLASS_TEST_ANY_SET(p) (ANYOF_FLAGS(p) & ANYOF_CLASS)
477 * Utility definitions.
480 # define UCHARAT(p) ((int)*(const U8*)(p))
482 # define UCHARAT(p) ((int)*(p)&CHARMASK)
485 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
487 #define REG_SEEN_ZERO_LEN 0x00000001
488 #define REG_SEEN_LOOKBEHIND 0x00000002
489 #define REG_SEEN_GPOS 0x00000004
490 #define REG_SEEN_EVAL 0x00000008
491 #define REG_SEEN_CANY 0x00000010
492 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
493 #define REG_SEEN_RECURSE 0x00000020
494 #define REG_TOP_LEVEL_BRANCHES 0x00000040
495 #define REG_SEEN_VERBARG 0x00000080
496 #define REG_SEEN_CUTGROUP 0x00000100
497 #define REG_SEEN_RUN_ON_COMMENT 0x00000200
501 #ifdef PLUGGABLE_RE_EXTENSION
502 #include "re_nodes.h"
504 #include "regnodes.h"
507 #ifndef PLUGGABLE_RE_EXTENSION
509 EXTCONST regexp_engine PL_core_reg_engine;
511 EXTCONST regexp_engine PL_core_reg_engine = {
514 Perl_re_intuit_start,
515 Perl_re_intuit_string,
516 Perl_regfree_internal,
517 Perl_reg_numbered_buff_fetch,
518 Perl_reg_numbered_buff_store,
519 Perl_reg_numbered_buff_length,
521 Perl_reg_named_buff_iter,
523 #if defined(USE_ITHREADS)
524 Perl_regdupe_internal
528 #endif /* PLUGGABLE_RE_EXTENSION */
534 /* .what is a character array with one character for each member of .data
535 * The character describes the function of the corresponding .data item:
536 * a - AV for paren_name_list under DEBUGGING
537 * f - start-class data for regstclass optimization
538 * n - Root of op tree for (?{EVAL}) item
539 * o - Start op for (?{EVAL}) item
540 * p - Pad for (?{EVAL}) item
541 * s - swash for Unicode-style character class, and the multicharacter
542 * strings resulting from casefolding the single-character entries
543 * in the character class
545 * u - trie struct's widecharmap (a HV, so can't share, must dup)
546 * also used for revcharmap and words under DEBUGGING
547 * T - aho-trie struct
548 * S - sv for named capture lookup
549 * 20010712 mjd@plover.com
550 * (Remember to update re_dup() and pregfree() if you add any items.)
558 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
559 anchored* and float* via array indexes 0 and 1. */
560 #define anchored_substr substrs->data[0].substr
561 #define anchored_utf8 substrs->data[0].utf8_substr
562 #define anchored_offset substrs->data[0].min_offset
563 #define anchored_end_shift substrs->data[0].end_shift
565 #define float_substr substrs->data[1].substr
566 #define float_utf8 substrs->data[1].utf8_substr
567 #define float_min_offset substrs->data[1].min_offset
568 #define float_max_offset substrs->data[1].max_offset
569 #define float_end_shift substrs->data[1].end_shift
571 #define check_substr substrs->data[2].substr
572 #define check_utf8 substrs->data[2].utf8_substr
573 #define check_offset_min substrs->data[2].min_offset
574 #define check_offset_max substrs->data[2].max_offset
575 #define check_end_shift substrs->data[2].end_shift
577 #define RX_ANCHORED_SUBSTR(rx) (((struct regexp *)SvANY(rx))->anchored_substr)
578 #define RX_ANCHORED_UTF8(rx) (((struct regexp *)SvANY(rx))->anchored_utf8)
579 #define RX_FLOAT_SUBSTR(rx) (((struct regexp *)SvANY(rx))->float_substr)
580 #define RX_FLOAT_UTF8(rx) (((struct regexp *)SvANY(rx))->float_utf8)
582 /* trie related stuff */
584 /* a transition record for the state machine. the
585 check field determines which state "owns" the
586 transition. the char the transition is for is
587 determined by offset from the owning states base
588 field. the next field determines which state
589 is to be transitioned to if any.
591 struct _reg_trie_trans {
596 /* a transition list element for the list based representation */
597 struct _reg_trie_trans_list_elem {
601 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
603 /* a state for compressed nodes. base is an offset
604 into an array of reg_trie_trans array. If wordnum is
605 nonzero the state is accepting. if base is zero then
606 the state has no children (and will be accepting)
608 struct _reg_trie_state {
612 reg_trie_trans_le* list;
616 /* info per word; indexed by wordnum */
618 U16 prev; /* previous word in acceptance chain; eg in
619 * zzz|abc|ab/ after matching the chars abc, the
620 * accepted word is #2, and the previous accepted
622 U32 len; /* how many chars long is this word? */
623 U32 accept; /* accept state for this word */
627 typedef struct _reg_trie_state reg_trie_state;
628 typedef struct _reg_trie_trans reg_trie_trans;
631 /* anything in here that needs to be freed later
632 should be dealt with in pregfree.
633 refcount is first in both this and _reg_ac_data to allow a space
634 optimisation in Perl_regdupe. */
635 struct _reg_trie_data {
636 U32 refcount; /* number of times this trie is referenced */
637 U32 lasttrans; /* last valid transition element */
638 U16 *charmap; /* byte to charid lookup array */
639 reg_trie_state *states; /* state data */
640 reg_trie_trans *trans; /* array of transition elements */
641 char *bitmap; /* stclass bitmap */
642 U16 *jump; /* optional 1 indexed array of offsets before tail
643 for the node following a given word. */
644 reg_trie_wordinfo *wordinfo; /* array of info per word */
645 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
646 U32 startstate; /* initial state - used for common prefix optimisation */
647 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
648 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
649 U32 prefixlen; /* #chars in common prefix */
650 U32 statecount; /* Build only - number of states in the states array
651 (including the unused zero state) */
652 U32 wordcount; /* Build only */
654 STRLEN charcount; /* Build only */
657 /* There is one (3 under DEBUGGING) pointers that logically belong in this
658 structure, but are held outside as they need duplication on thread cloning,
659 whereas the rest of the structure can be read only:
660 HV *widecharmap; code points > 255 to charid
662 AV *words; Array of words contained in trie, for dumping
663 AV *revcharmap; Map of each charid back to its character representation
667 #define TRIE_WORDS_OFFSET 2
669 typedef struct _reg_trie_data reg_trie_data;
671 /* refcount is first in both this and _reg_trie_data to allow a space
672 optimisation in Perl_regdupe. */
673 struct _reg_ac_data {
677 reg_trie_state *states;
679 typedef struct _reg_ac_data reg_ac_data;
681 /* ANY_BIT doesn't use the structure, so we can borrow it here.
682 This is simpler than refactoring all of it as wed end up with
683 three different sets... */
685 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
686 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
687 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
688 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
689 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
691 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
692 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
695 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
696 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
698 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
699 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
700 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
701 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
704 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
706 #define TRIE_CHARCOUNT(trie) (trie_charcount)
709 #define RE_TRIE_MAXBUF_INIT 65536
710 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
711 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
715 RE_DEBUG_FLAGS is used to control what debug output is emitted
716 its divided into three groups of options, some of which interact.
717 The three groups are: Compile, Execute, Extra. There is room for a
718 further group, as currently only the low three bytes are used.
739 If you modify any of these make sure you make corresponding changes to
740 re.pm, especially to the documentation.
746 #define RE_DEBUG_COMPILE_MASK 0x0000FF
747 #define RE_DEBUG_COMPILE_PARSE 0x000001
748 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
749 #define RE_DEBUG_COMPILE_TRIE 0x000004
750 #define RE_DEBUG_COMPILE_DUMP 0x000008
751 #define RE_DEBUG_COMPILE_FLAGS 0x000010
754 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
755 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
756 #define RE_DEBUG_EXECUTE_MATCH 0x000200
757 #define RE_DEBUG_EXECUTE_TRIE 0x000400
760 #define RE_DEBUG_EXTRA_MASK 0xFF0000
761 #define RE_DEBUG_EXTRA_TRIE 0x010000
762 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
763 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
764 #define RE_DEBUG_EXTRA_STATE 0x080000
765 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
766 #define RE_DEBUG_EXTRA_BUFFERS 0x400000
767 #define RE_DEBUG_EXTRA_GPOS 0x800000
769 #define RE_DEBUG_EXTRA_STACK 0x280000
771 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
773 #define DEBUG_COMPILE_r(x) DEBUG_r( \
774 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
775 #define DEBUG_PARSE_r(x) DEBUG_r( \
776 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
777 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
778 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
779 #define DEBUG_PARSE_r(x) DEBUG_r( \
780 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
781 #define DEBUG_DUMP_r(x) DEBUG_r( \
782 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
783 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
784 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
785 #define DEBUG_FLAGS_r(x) DEBUG_r( \
786 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
788 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
789 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
790 #define DEBUG_INTUIT_r(x) DEBUG_r( \
791 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
792 #define DEBUG_MATCH_r(x) DEBUG_r( \
793 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
794 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
795 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
798 #define DEBUG_EXTRA_r(x) DEBUG_r( \
799 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
800 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
801 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
802 #define DEBUG_STATE_r(x) DEBUG_r( \
803 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
804 #define DEBUG_STACK_r(x) DEBUG_r( \
805 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
806 #define DEBUG_BUFFERS_r(x) DEBUG_r( \
807 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
809 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
810 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
811 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
812 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
813 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
814 Perl_warn_nocontext x )
815 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
816 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
817 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
818 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
820 #define DEBUG_TRIE_r(x) DEBUG_r( \
821 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
822 | RE_DEBUG_EXECUTE_TRIE )) x )
823 #define DEBUG_GPOS_r(x) DEBUG_r( \
824 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
827 /* get_sv() can return NULL during global destruction. re_debug_flags can get
828 * clobbered by a longjmp, so must be initialized */
829 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
830 SV * re_debug_flags_sv = NULL; \
831 re_debug_flags = 0; \
832 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
833 if (re_debug_flags_sv) { \
834 if (!SvIOK(re_debug_flags_sv)) \
835 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
836 re_debug_flags=SvIV(re_debug_flags_sv); \
842 #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
844 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
845 const char * const rpv = \
846 pv_pretty((dsv), (pv), (l), (m), \
847 PL_colors[(c1)],PL_colors[(c2)], \
848 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
849 const int rlen = SvCUR(dsv)
851 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
852 const char * const rpv = \
853 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
854 PL_colors[(c1)],PL_colors[(c2)], \
855 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
857 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
858 const char * const rpv = \
859 pv_pretty((dsv), (pv), (l), (m), \
860 PL_colors[0], PL_colors[1], \
861 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
862 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
865 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
866 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
868 #else /* if not DEBUGGING */
870 #define GET_RE_DEBUG_FLAGS_DECL
871 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
872 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
873 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
874 #define RE_SV_DUMPLEN(ItEm)
875 #define RE_SV_TAIL(ItEm)
877 #endif /* DEBUG RELATED DEFINES */
881 * c-indentation-style: bsd
883 * indent-tabs-mode: t
886 * ex: set ts=8 sts=4 sw=4 noet: