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
30 * The "internal use only" fields in regexp.h are present to pass info from
31 * compile to execute that permits the execute phase to run lots faster on
32 * simple cases. They are:
34 * regstart sv that must begin a match; NULL if none obvious
35 * reganch is the match anchored (at beginning-of-line only)?
36 * regmust string (pointer into program) that match must include, or NULL
37 * [regmust changed to SV* for bminstr()--law]
38 * regmlen length of regmust string
39 * [regmlen not used currently]
41 * Regstart and reganch permit very fast decisions on suitable starting points
42 * for a match, cutting down the work a lot. Regmust permits fast rejection
43 * of lines that cannot possibly match. The regmust tests are costly enough
44 * that pregcomp() supplies a regmust only if the r.e. contains something
45 * potentially expensive (at present, the only such thing detected is * or +
46 * at the start of the r.e., which can involve a lot of backup). Regmlen is
47 * supplied because the test in pregexec() needs it and pregcomp() is computing
49 * [regmust is now supplied always. The tests that use regmust have a
50 * heuristic that disables the test if it usually matches.]
52 * [In fact, we now use regmust in many cases to locate where the search
53 * starts in the string, so if regback is >= 0, the regmust search is never
54 * wasted effort. The regback variable says how many characters back from
55 * where regmust matched is the earliest possible start of the match.
56 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
60 * Structure for regexp "program". This is essentially a linear encoding
61 * of a nondeterministic finite-state machine (aka syntax charts or
62 * "railroad normal form" in parsing technology). Each node is an opcode
63 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
64 * all nodes except BRANCH implement concatenation; a "next" pointer with
65 * a BRANCH on both ends of it is connecting two alternatives. (Here we
66 * have one of the subtle syntax dependencies: an individual BRANCH (as
67 * opposed to a collection of them) is never concatenated with anything
68 * because of operator precedence.) The operand of some types of node is
69 * a literal string; for others, it is a node leading into a sub-FSM. In
70 * particular, the operand of a BRANCH node is the first node of the branch.
71 * (NB this is *not* a tree structure: the tail of the branch connects
72 * to the thing following the set of BRANCHes.) The opcodes are defined
73 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
77 * A node is one char of opcode followed by two chars of "next" pointer.
78 * "Next" pointers are stored as two 8-bit pieces, high order first. The
79 * value is a positive offset from the opcode of the node containing it.
80 * An operand, if any, simply follows the node. (Note that much of the
81 * code generation knows about this implicit relationship.)
83 * Using two bytes for the "next" pointer is vast overkill for most things,
84 * but allows patterns to get big without disasters.
86 * [The "next" pointer is always aligned on an even
87 * boundary, and reads the offset directly as a short. Also, there is no
88 * special test to reverse the sign of BACK pointers since the offset is
92 /* This is the stuff that used to live in regexp.h that was truly
93 private to the engine itself. It now lives here. */
97 typedef struct regexp_internal {
98 int name_list_idx; /* Optional data index of an array of paren names */
100 U32 *offsets; /* offset annotations 20001228 MJD
101 data about mapping the program to the
103 offsets[0] is proglen when this is used
108 regnode *regstclass; /* Optional startclass as identified or constructed
110 struct reg_data *data; /* Additional miscellaneous data used by the program.
111 Used to make it easier to clone and free arbitrary
112 data that the regops need. Often the ARG field of
113 a regop is an index into this structure */
114 struct reg_code_block *code_blocks;/* positions of literal (?{}) */
115 int num_code_blocks; /* size of code_blocks[] */
116 regnode program[1]; /* Unwarranted chumminess with compiler. */
119 #define RXi_SET(x,y) (x)->pprivate = (void*)(y)
120 #define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
121 #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
123 * Flags stored in regexp->intflags
124 * These are used only internally to the regexp engine
126 * See regexp.h for flags used externally to the regexp engine
128 #define PREGf_SKIP 0x00000001
129 #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
130 #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
131 #define PREGf_VERBARG_SEEN 0x00000008
132 #define PREGf_CUTGROUP_SEEN 0x00000010
133 #define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
136 /* this is where the old regcomp.h started */
138 struct regnode_string {
145 /* Argument bearing node - workhorse,
146 arg1 is often for the data field */
154 /* Similar to a regnode_1 but with an extra signed argument */
163 /* 'Two field' -- Two 16 bit unsigned args */
173 #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
175 /* also used by trie */
176 struct regnode_charclass {
180 U32 arg1; /* used as ptr in S_regclass */
181 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
184 /* has runtime (locale) \d, \w, ..., [:posix:] classes */
185 struct regnode_charclass_class {
186 U8 flags; /* ANYOF_POSIXL bit must go here */
189 U32 arg1; /* used as ptr in S_regclass */
190 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
191 U32 classflags; /* and run-time */
194 /* Synthetic start class; is a regnode_charclass_class plus an SV*. Note that
195 * the 'next_off' field is unused, as the SSC stands alone, so there is never a
198 U8 flags; /* ANYOF_POSIXL bit must go here */
201 U32 arg1; /* used as ptr in S_regclass */
202 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
203 U32 classflags; /* and run-time */
204 SV* invlist; /* list of code points matched */
207 /* XXX fix this description.
208 Impose a limit of REG_INFTY on various pattern matching operations
209 to limit stack growth and to avoid "infinite" recursions.
211 /* The default size for REG_INFTY is I16_MAX, which is the same as
212 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
213 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
214 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
215 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
217 --Andy Dougherty 11 June 1998
221 # define REG_INFTY ((1<<15)-1)
226 # define REG_INFTY I16_MAX
229 #define ARG_VALUE(arg) (arg)
230 #define ARG__SET(arg,val) ((arg) = (val))
236 #define ARG(p) ARG_VALUE(ARG_LOC(p))
237 #define ARG1(p) ARG_VALUE(ARG1_LOC(p))
238 #define ARG2(p) ARG_VALUE(ARG2_LOC(p))
239 #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
241 #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
242 #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
243 #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
244 #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
249 #define NEXT_OFF(p) ((p)->next_off)
250 #define NODE_ALIGN(node)
251 #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
253 #define SIZE_ALIGN NODE_ALIGN
260 #define OP(p) ((p)->type)
261 #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
262 regnode types. For some, it's the \
263 character set of the regnode */
264 #define OPERAND(p) (((struct regnode_string *)p)->string)
265 #define MASK(p) ((char*)OPERAND(p))
266 #define STR_LEN(p) (((struct regnode_string *)p)->str_len)
267 #define STRING(p) (((struct regnode_string *)p)->string)
268 #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
269 #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
276 #define NODE_ALIGN(node)
277 #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
278 #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
279 #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
280 #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
282 #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
283 #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
285 #define NODE_STEP_B 4
287 #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
288 #define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
290 #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
291 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
292 #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
293 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
295 #define REG_MAGIC 0234
297 #define SIZE_ONLY (RExC_emit == (regnode *) & RExC_emit_dummy)
298 #define PASS1 SIZE_ONLY
299 #define PASS2 (! SIZE_ONLY)
301 /* If the bitmap doesn't fully represent what this ANYOF node can match, the
302 * ARG is set to this special value (since 0, 1, ... are legal, but will never
303 * reach this high). */
304 #define ANYOF_NONBITMAP_EMPTY ((U32) -1)
306 /* The information used to be stored as as combination of the ANYOF_UTF8 and
307 * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there
308 * to free up a bit for other uses. This tries to hide the change from
309 * existing code as much as possible. Now, the data structure that goes in ARG
310 * is not allocated unless it is needed, and that is what is used to determine
311 * if there is something outside the bitmap. The code now assumes that if
312 * that structure exists, that any UTF-8 encoded string should be tried against
313 * it, but a non-UTF8-encoded string will be tried only if the
314 * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */
315 #define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY)
317 /* Flags for node->flags of ANYOF. These are in short supply, with none
318 * currently available. If more are needed, the ANYOF_LOCALE and
319 * ANYOF_POSIXL bits could be shared, making a space penalty for all locale
320 * nodes. Also, the ABOVE_LATIN1_ALL bit could be freed up by resorting to
321 * creating a swash containing everything above 255. This introduces a
322 * performance penalty. Better would be to split it off into a separate node,
323 * which actually would improve performance a bit by allowing regexec.c to test
324 * for a UTF-8 character being above 255 without having to call a function nor
325 * calculate its code point value. However, this solution might need to have a
326 * second node type, ANYOF_SYNTHETIC_ABOVE_LATIN1_ALL. Several flags are not
327 * used in synthetic start class (SSC) nodes, so could be shared should new
328 * flags be needed for SSCs. */
330 /* regexec.c is expecting this to be in the low bit */
331 #define ANYOF_INVERT 0x01
333 /* For the SSC node only, which cannot be inverted, so is shared with that bit.
334 * This means "Does this SSC match an empty string?" This is used only during
335 * regex compilation. */
336 #define ANYOF_EMPTY_STRING ANYOF_INVERT
338 #define ANYOF_LOCALE 0x02 /* /l modifier */
340 /* The fold is calculated and stored in the bitmap where possible at compile
341 * time. However under locale, the actual folding varies depending on
342 * what the locale is at the time of execution, so it has to be deferred until
344 #define ANYOF_LOC_FOLD 0x04
346 /* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This
347 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
348 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
349 * at compile-time. Could be shared with ANYOF_LOCALE, forcing all locale
350 * nodes to be large */
351 #define ANYOF_POSIXL 0x08
352 #define ANYOF_CLASS ANYOF_POSIXL
353 #define ANYOF_LARGE ANYOF_POSIXL
355 /* Should we raise a warning if matching against an above-Unicode code point?
357 #define ANYOF_WARN_SUPER 0x10
359 /* Can match something outside the bitmap that isn't in utf8 */
360 #define ANYOF_NONBITMAP_NON_UTF8 0x20
362 /* Matches every code point 0x100 and above*/
363 #define ANYOF_ABOVE_LATIN1_ALL 0x40
364 #define ANYOF_UNICODE_ALL ANYOF_ABOVE_LATIN1_ALL
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 #define ANYOF_LOCALE_FLAGS (ANYOF_LOCALE \
376 /* These are the flags that apply to both regular ANYOF nodes and synthetic
377 * start class nodes during construction of the SSC. During finalization of
378 * the SSC, other of the flags could be added to it */
379 #define ANYOF_COMMON_FLAGS (ANYOF_LOCALE_FLAGS | ANYOF_WARN_SUPER)
381 /* Character classes for node->classflags of ANYOF */
382 /* Should be synchronized with a table in regprop() */
383 /* 2n should be the normal one, paired with its complement at 2n+1 */
385 #define ANYOF_ALPHA ((_CC_ALPHA) * 2)
386 #define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
387 #define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
388 #define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
389 #define ANYOF_ASCII ((_CC_ASCII) * 2)
390 #define ANYOF_NASCII ((ANYOF_ASCII) + 1)
391 #define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
392 #define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
393 #define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
394 [:upper:] under /i */
395 #define ANYOF_NCASED ((ANYOF_CASED) + 1)
396 #define ANYOF_CNTRL ((_CC_CNTRL) * 2)
397 #define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
398 #define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
399 #define ANYOF_NDIGIT ((ANYOF_DIGIT) + 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_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */
407 #define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1)
408 #define ANYOF_PUNCT ((_CC_PUNCT) * 2)
409 #define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
410 #define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
411 #define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
412 #define ANYOF_UPPER ((_CC_UPPER) * 2)
413 #define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
414 #define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
415 #define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
416 #define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
417 #define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
419 /* pseudo classes below this, not stored in the class bitmap, but used as flags
420 during compilation of char classes */
422 #define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
423 #define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
425 /* It is best if this is the last one, as all above it are stored as bits in a
426 * bitmap, and it isn't part of that bitmap */
427 #if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
428 # error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
431 #define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
433 * Hence doesn't include VERTWS, as that
434 * is a pseudo class */
435 #define ANYOF_MAX ANYOF_POSIXL_MAX
437 #if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */
438 # error Problem with handy.h _CC_foo #defines
441 #define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
442 #define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3)
444 #define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode
445 property: \p{} or \P{} */
447 /* Backward source code compatibility. */
449 #define ANYOF_ALNUML ANYOF_ALNUM
450 #define ANYOF_NALNUML ANYOF_NALNUM
451 #define ANYOF_SPACEL ANYOF_SPACE
452 #define ANYOF_NSPACEL ANYOF_NSPACE
453 #define ANYOF_ALNUM ANYOF_WORDCHAR
454 #define ANYOF_NALNUM ANYOF_NWORDCHAR
456 /* Utility macros for the bitmap and classes of ANYOF */
458 #define ANYOF_SIZE (sizeof(struct regnode_charclass))
459 #define ANYOF_POSIXL_SIZE (sizeof(regnode_charclass_posixl))
460 #define ANYOF_CLASS_SIZE ANYOF_POSIXL_SIZE
462 #define ANYOF_FLAGS(p) ((p)->flags)
464 #define ANYOF_BIT(c) (1 << ((c) & 7))
466 #define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c)))
467 #define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c))
469 #define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c)))
470 #define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c))
472 #define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c)))
473 #define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c))
475 #define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END
476 #define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret)
478 /* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
479 #define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END
480 #define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret)
482 #define ANYOF_POSIXL_TEST_ANY_SET(p) \
483 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
484 && (((regnode_charclass_posixl*)(p))->classflags))
485 #define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p)
487 #define ANYOF_POSIXL_TEST_ALL_SET(p) \
488 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
489 && ((regnode_charclass_posixl*) (p))->classflags == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
491 #define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END
492 #define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest))
494 #define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END
496 #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
497 #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
498 #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
499 #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
500 #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
501 #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
503 #define ANYOF_BITMAP_SETALL(p) \
504 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
505 #define ANYOF_BITMAP_CLEARALL(p) \
506 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
507 /* Check that all 256 bits are all set. Used in S_cl_is_anything() */
508 #define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
509 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)
511 #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
512 #define ANYOF_POSIXL_SKIP ((ANYOF_POSIXL_SIZE - 1)/sizeof(regnode))
513 #define ANYOF_CLASS_SKIP ANYOF_POSIXL_SKIP
516 * Utility definitions.
519 # define UCHARAT(p) ((int)*(const U8*)(p))
521 # define UCHARAT(p) ((int)*(p)&CHARMASK)
524 #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
526 #define REG_SEEN_ZERO_LEN 0x00000001
527 #define REG_SEEN_LOOKBEHIND 0x00000002
528 #define REG_SEEN_GPOS 0x00000004
530 #define REG_SEEN_CANY 0x00000010
531 #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
532 #define REG_SEEN_RECURSE 0x00000020
533 #define REG_TOP_LEVEL_BRANCHES 0x00000040
534 #define REG_SEEN_VERBARG 0x00000080
535 #define REG_SEEN_CUTGROUP 0x00000100
536 #define REG_SEEN_RUN_ON_COMMENT 0x00000200
537 #define REG_SEEN_EXACTF_SHARP_S 0x00000400
538 #define REG_SEEN_GOSTART 0x00000800
542 #ifdef PLUGGABLE_RE_EXTENSION
543 #include "re_nodes.h"
545 #include "regnodes.h"
548 #ifndef PLUGGABLE_RE_EXTENSION
550 EXTCONST regexp_engine PL_core_reg_engine;
552 EXTCONST regexp_engine PL_core_reg_engine = {
555 Perl_re_intuit_start,
556 Perl_re_intuit_string,
557 Perl_regfree_internal,
558 Perl_reg_numbered_buff_fetch,
559 Perl_reg_numbered_buff_store,
560 Perl_reg_numbered_buff_length,
562 Perl_reg_named_buff_iter,
564 #if defined(USE_ITHREADS)
565 Perl_regdupe_internal,
570 #endif /* PLUGGABLE_RE_EXTENSION */
576 /* .what is a character array with one character for each member of .data
577 * The character describes the function of the corresponding .data item:
578 * a - AV for paren_name_list under DEBUGGING
579 * f - start-class data for regstclass optimization
580 * l - start op for literal (?{EVAL}) item
581 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
582 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
583 * s - swash for Unicode-style character class, and the multicharacter
584 * strings resulting from casefolding the single-character entries
585 * in the character class
587 * u - trie struct's widecharmap (a HV, so can't share, must dup)
588 * also used for revcharmap and words under DEBUGGING
589 * T - aho-trie struct
590 * S - sv for named capture lookup
591 * 20010712 mjd@plover.com
592 * (Remember to update re_dup() and pregfree() if you add any items.)
600 /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
601 anchored* and float* via array indexes 0 and 1. */
602 #define anchored_substr substrs->data[0].substr
603 #define anchored_utf8 substrs->data[0].utf8_substr
604 #define anchored_offset substrs->data[0].min_offset
605 #define anchored_end_shift substrs->data[0].end_shift
607 #define float_substr substrs->data[1].substr
608 #define float_utf8 substrs->data[1].utf8_substr
609 #define float_min_offset substrs->data[1].min_offset
610 #define float_max_offset substrs->data[1].max_offset
611 #define float_end_shift substrs->data[1].end_shift
613 #define check_substr substrs->data[2].substr
614 #define check_utf8 substrs->data[2].utf8_substr
615 #define check_offset_min substrs->data[2].min_offset
616 #define check_offset_max substrs->data[2].max_offset
617 #define check_end_shift substrs->data[2].end_shift
619 #define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
620 #define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
621 #define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
622 #define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
624 /* trie related stuff */
626 /* a transition record for the state machine. the
627 check field determines which state "owns" the
628 transition. the char the transition is for is
629 determined by offset from the owning states base
630 field. the next field determines which state
631 is to be transitioned to if any.
633 struct _reg_trie_trans {
638 /* a transition list element for the list based representation */
639 struct _reg_trie_trans_list_elem {
643 typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
645 /* a state for compressed nodes. base is an offset
646 into an array of reg_trie_trans array. If wordnum is
647 nonzero the state is accepting. if base is zero then
648 the state has no children (and will be accepting)
650 struct _reg_trie_state {
654 reg_trie_trans_le* list;
658 /* info per word; indexed by wordnum */
660 U16 prev; /* previous word in acceptance chain; eg in
661 * zzz|abc|ab/ after matching the chars abc, the
662 * accepted word is #2, and the previous accepted
664 U32 len; /* how many chars long is this word? */
665 U32 accept; /* accept state for this word */
669 typedef struct _reg_trie_state reg_trie_state;
670 typedef struct _reg_trie_trans reg_trie_trans;
673 /* anything in here that needs to be freed later
674 should be dealt with in pregfree.
675 refcount is first in both this and _reg_ac_data to allow a space
676 optimisation in Perl_regdupe. */
677 struct _reg_trie_data {
678 U32 refcount; /* number of times this trie is referenced */
679 U32 lasttrans; /* last valid transition element */
680 U16 *charmap; /* byte to charid lookup array */
681 reg_trie_state *states; /* state data */
682 reg_trie_trans *trans; /* array of transition elements */
683 char *bitmap; /* stclass bitmap */
684 U16 *jump; /* optional 1 indexed array of offsets before tail
685 for the node following a given word. */
686 reg_trie_wordinfo *wordinfo; /* array of info per word */
687 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
688 U32 startstate; /* initial state - used for common prefix optimisation */
689 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
690 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
691 U32 prefixlen; /* #chars in common prefix */
692 U32 statecount; /* Build only - number of states in the states array
693 (including the unused zero state) */
694 U32 wordcount; /* Build only */
696 STRLEN charcount; /* Build only */
699 /* There is one (3 under DEBUGGING) pointers that logically belong in this
700 structure, but are held outside as they need duplication on thread cloning,
701 whereas the rest of the structure can be read only:
702 HV *widecharmap; code points > 255 to charid
704 AV *words; Array of words contained in trie, for dumping
705 AV *revcharmap; Map of each charid back to its character representation
709 #define TRIE_WORDS_OFFSET 2
711 typedef struct _reg_trie_data reg_trie_data;
713 /* refcount is first in both this and _reg_trie_data to allow a space
714 optimisation in Perl_regdupe. */
715 struct _reg_ac_data {
719 reg_trie_state *states;
721 typedef struct _reg_ac_data reg_ac_data;
723 /* ANY_BIT doesn't use the structure, so we can borrow it here.
724 This is simpler than refactoring all of it as wed end up with
725 three different sets... */
727 #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
728 #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
729 #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
730 #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
731 #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
733 #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
734 #define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
737 #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
738 #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
740 /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
741 #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
742 #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
743 #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
746 #define TRIE_CHARCOUNT(trie) ((trie)->charcount)
748 #define TRIE_CHARCOUNT(trie) (trie_charcount)
751 #define RE_TRIE_MAXBUF_INIT 65536
752 #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
753 #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
757 RE_DEBUG_FLAGS is used to control what debug output is emitted
758 its divided into three groups of options, some of which interact.
759 The three groups are: Compile, Execute, Extra. There is room for a
760 further group, as currently only the low three bytes are used.
781 If you modify any of these make sure you make corresponding changes to
782 re.pm, especially to the documentation.
788 #define RE_DEBUG_COMPILE_MASK 0x0000FF
789 #define RE_DEBUG_COMPILE_PARSE 0x000001
790 #define RE_DEBUG_COMPILE_OPTIMISE 0x000002
791 #define RE_DEBUG_COMPILE_TRIE 0x000004
792 #define RE_DEBUG_COMPILE_DUMP 0x000008
793 #define RE_DEBUG_COMPILE_FLAGS 0x000010
796 #define RE_DEBUG_EXECUTE_MASK 0x00FF00
797 #define RE_DEBUG_EXECUTE_INTUIT 0x000100
798 #define RE_DEBUG_EXECUTE_MATCH 0x000200
799 #define RE_DEBUG_EXECUTE_TRIE 0x000400
802 #define RE_DEBUG_EXTRA_MASK 0xFF0000
803 #define RE_DEBUG_EXTRA_TRIE 0x010000
804 #define RE_DEBUG_EXTRA_OFFSETS 0x020000
805 #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
806 #define RE_DEBUG_EXTRA_STATE 0x080000
807 #define RE_DEBUG_EXTRA_OPTIMISE 0x100000
808 #define RE_DEBUG_EXTRA_BUFFERS 0x400000
809 #define RE_DEBUG_EXTRA_GPOS 0x800000
811 #define RE_DEBUG_EXTRA_STACK 0x280000
813 #define RE_DEBUG_FLAG(x) (re_debug_flags & x)
815 #define DEBUG_COMPILE_r(x) DEBUG_r( \
816 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
817 #define DEBUG_PARSE_r(x) DEBUG_r( \
818 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
819 #define DEBUG_OPTIMISE_r(x) DEBUG_r( \
820 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
821 #define DEBUG_PARSE_r(x) DEBUG_r( \
822 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
823 #define DEBUG_DUMP_r(x) DEBUG_r( \
824 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
825 #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
826 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
827 #define DEBUG_FLAGS_r(x) DEBUG_r( \
828 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
830 #define DEBUG_EXECUTE_r(x) DEBUG_r( \
831 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
832 #define DEBUG_INTUIT_r(x) DEBUG_r( \
833 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
834 #define DEBUG_MATCH_r(x) DEBUG_r( \
835 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
836 #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
837 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
840 #define DEBUG_EXTRA_r(x) DEBUG_r( \
841 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
842 #define DEBUG_OFFSETS_r(x) DEBUG_r( \
843 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
844 #define DEBUG_STATE_r(x) DEBUG_r( \
845 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
846 #define DEBUG_STACK_r(x) DEBUG_r( \
847 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
848 #define DEBUG_BUFFERS_r(x) DEBUG_r( \
849 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
851 #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
852 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
853 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
854 #define MJD_OFFSET_DEBUG(x) DEBUG_r( \
855 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
856 Perl_warn_nocontext x )
857 #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
858 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
859 #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
860 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
862 #define DEBUG_TRIE_r(x) DEBUG_r( \
863 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
864 | RE_DEBUG_EXECUTE_TRIE )) x )
865 #define DEBUG_GPOS_r(x) DEBUG_r( \
866 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
869 /* get_sv() can return NULL during global destruction. */
870 #define GET_RE_DEBUG_FLAGS DEBUG_r({ \
871 SV * re_debug_flags_sv = NULL; \
872 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
873 if (re_debug_flags_sv) { \
874 if (!SvIOK(re_debug_flags_sv)) \
875 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
876 re_debug_flags=SvIV(re_debug_flags_sv); \
882 #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
883 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
885 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
886 const char * const rpv = \
887 pv_pretty((dsv), (pv), (l), (m), \
888 PL_colors[(c1)],PL_colors[(c2)], \
889 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
890 const int rlen = SvCUR(dsv)
892 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
893 const char * const rpv = \
894 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
895 PL_colors[(c1)],PL_colors[(c2)], \
896 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
898 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
899 const char * const rpv = \
900 pv_pretty((dsv), (pv), (l), (m), \
901 PL_colors[0], PL_colors[1], \
902 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
903 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
906 #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
907 #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
909 #else /* if not DEBUGGING */
911 #define GET_RE_DEBUG_FLAGS_DECL
912 #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
913 #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
914 #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
915 #define RE_SV_DUMPLEN(ItEm)
916 #define RE_SV_TAIL(ItEm)
918 #endif /* DEBUG RELATED DEFINES */
922 * c-indentation-style: bsd
924 * indent-tabs-mode: nil
927 * ex: set ts=8 sts=4 sw=4 et: