This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: [PATCH] Callbacks for named captures (%+ and %-)
[perl5.git] / regcomp.h
CommitLineData
a0d0e21e 1/* regcomp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
699a97de 4 * 2000, 2001, 2002, 2003, 2005, 2006 by Larry Wall and others
d6376244
JH
5 *
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.
8 *
a687059c 9 */
e1d1eefb 10#include "regcharclass.h"
a687059c 11
c277df42
IZ
12typedef OP OP_4tree; /* Will be redefined later. */
13
07be1b83 14
58e23c8d 15/* Convert branch sequences to more efficient trie ops? */
07be1b83 16#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d
YO
17
18/* Be really agressive about optimising patterns with trie sequences? */
07be1b83 19#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d
YO
20
21/* Should the optimiser take positive assertions into account? */
68ba3a3f 22#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
58e23c8d
YO
23
24/* Not for production use: */
07be1b83 25#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 26
7122b237
YO
27/* Activate offsets code - set to if 1 to enable */
28#ifdef DEBUGGING
29#define RE_TRACK_PATTERN_OFFSETS
30#endif
31
b9b4dddf
YO
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.
36 */
37/* #define REG_ALLOW_MINMOD_SUSPEND */
07be1b83 38
a687059c
LW
39/*
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:
43 *
bd61b366 44 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
45 * reganch is the match anchored (at beginning-of-line only)?
46 * regmust string (pointer into program) that match must include, or NULL
79072805 47 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
48 * regmlen length of regmust string
49 * [regmlen not used currently]
50 *
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
e50aee73 54 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
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
e50aee73 57 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
58 * it anyway.
59 * [regmust is now supplied always. The tests that use regmust have a
60 * heuristic that disables the test if it usually matches.]
61 *
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.]
67 */
68
69/*
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
a3621e74
YO
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.
a687059c
LW
84 */
85
a687059c
LW
86/*
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.)
92 *
93 * Using two bytes for the "next" pointer is vast overkill for most things,
94 * but allows patterns to get big without disasters.
95 *
e91177ed 96 * [The "next" pointer is always aligned on an even
a687059c
LW
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
99 * stored negative.]
100 */
785a26d5
YO
101
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. */
104
28d8d7f4 105
785a26d5
YO
106
107 typedef struct regexp_internal {
1f1031fe 108 int name_list_idx; /* Optional data index of an array of paren names */
7122b237
YO
109 union {
110 U32 *offsets; /* offset annotations 20001228 MJD
111 data about mapping the program to the
112 string -
113 offsets[0] is proglen when this is used
114 */
115 U32 proglen;
116 } u;
117
6d5c990f
RGS
118 regnode *regstclass; /* Optional startclass as identified or constructed
119 by the optimiser */
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. */
125} regexp_internal;
126
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)
785a26d5
YO
130/*
131 * Flags stored in regexp->intflags
132 * These are used only internally to the regexp engine
133 *
134 * See regexp.h for flags used externally to the regexp engine
135 */
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
141
142
143/* this is where the old regcomp.h started */
a687059c 144
c277df42 145struct regnode_string {
cd439c50 146 U8 str_len;
c277df42
IZ
147 U8 type;
148 U16 next_off;
cd439c50 149 char string[1];
c277df42 150};
a687059c 151
6bda09f9
YO
152/* Argument bearing node - workhorse,
153 arg1 is often for the data field */
c277df42
IZ
154struct regnode_1 {
155 U8 flags;
156 U8 type;
157 U16 next_off;
158 U32 arg1;
159};
160
6bda09f9
YO
161/* Similar to a regnode_1 but with an extra signed argument */
162struct regnode_2L {
163 U8 flags;
164 U8 type;
165 U16 next_off;
166 U32 arg1;
167 I32 arg2;
168};
169
170/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
171struct regnode_2 {
172 U8 flags;
173 U8 type;
174 U16 next_off;
175 U16 arg1;
176 U16 arg2;
177};
178
6bda09f9 179
936ed897 180#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
e1d1eefb 181#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 40 (8*5) named classes */
936ed897 182
786e8c11 183/* also used by trie */
936ed897
IZ
184struct regnode_charclass {
185 U8 flags;
186 U8 type;
187 U16 next_off;
ffc61ed2 188 U32 arg1;
19103b91 189 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
190};
191
19103b91
JH
192struct regnode_charclass_class { /* has [[:blah:]] classes */
193 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
194 U8 type;
195 U16 next_off;
ffc61ed2 196 U32 arg1;
19103b91
JH
197 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
198 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
199};
200
9c7e81e8
AD
201/* XXX fix this description.
202 Impose a limit of REG_INFTY on various pattern matching operations
203 to limit stack growth and to avoid "infinite" recursions.
204*/
205/* The default size for REG_INFTY is I16_MAX, which is the same as
206 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
207 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
208 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
209 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
210 smaller default.
211 --Andy Dougherty 11 June 1998
212*/
213#if SHORTSIZE > 2
214# ifndef REG_INFTY
215# define REG_INFTY ((1<<15)-1)
216# endif
217#endif
218
219#ifndef REG_INFTY
83cfe48c
IZ
220# define REG_INFTY I16_MAX
221#endif
a687059c 222
e91177ed
IZ
223#define ARG_VALUE(arg) (arg)
224#define ARG__SET(arg,val) ((arg) = (val))
c277df42 225
e8e3b0fb
JH
226#undef ARG
227#undef ARG1
228#undef ARG2
229
c277df42
IZ
230#define ARG(p) ARG_VALUE(ARG_LOC(p))
231#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
232#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 233#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 234
c277df42
IZ
235#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
236#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
237#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 238#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 239
e8e3b0fb
JH
240#undef NEXT_OFF
241#undef NODE_ALIGN
242
075abff3
AL
243#define NEXT_OFF(p) ((p)->next_off)
244#define NODE_ALIGN(node)
245#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 246
c277df42
IZ
247#define SIZE_ALIGN NODE_ALIGN
248
e8e3b0fb
JH
249#undef OP
250#undef OPERAND
251#undef MASK
252#undef STRING
253
e91177ed
IZ
254#define OP(p) ((p)->type)
255#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
256#define MASK(p) ((char*)OPERAND(p))
257#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
258#define STRING(p) (((struct regnode_string *)p)->string)
259#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
260#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
261
e8e3b0fb
JH
262#undef NODE_ALIGN
263#undef ARG_LOC
264#undef NEXTOPER
265#undef PREVOPER
266
e91177ed
IZ
267#define NODE_ALIGN(node)
268#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
269#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
270#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 271#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 272
e91177ed
IZ
273#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
274#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
275
276#define NODE_STEP_B 4
c277df42
IZ
277
278#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
279#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
280
e91177ed 281#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 282 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 283#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 284 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 285
22c35a8c 286#define REG_MAGIC 0234
a687059c 287
830247a4 288#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 289
936ed897 290/* Flags for node->flags of ANYOF */
b8c5462f 291
19103b91 292#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
293#define ANYOF_INVERT 0x04
294#define ANYOF_FOLD 0x02
295#define ANYOF_LOCALE 0x01
b8c5462f 296
653099ff 297/* Used for regstclass only */
ffc61ed2
JH
298#define ANYOF_EOS 0x10 /* Can match an empty string too */
299
300/* There is a character or a range past 0xff */
301#define ANYOF_UNICODE 0x20
1aa99e6b 302#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 303
4f66b38d
HS
304/* size of node is large (includes class pointer) */
305#define ANYOF_LARGE 0x80
306
ffc61ed2
JH
307/* Are there any runtime flags on in this node? */
308#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
309
310#define ANYOF_FLAGS_ALL 0xff
653099ff 311
936ed897 312/* Character classes for node->classflags of ANYOF */
653099ff
GS
313/* Should be synchronized with a table in regprop() */
314/* 2n should pair with 2n+1 */
b8c5462f 315
289d4f09 316#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 317#define ANYOF_NALNUM 1
aaa51d5e 318#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
319#define ANYOF_NSPACE 3
320#define ANYOF_DIGIT 4
321#define ANYOF_NDIGIT 5
289d4f09 322#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
323#define ANYOF_NALNUMC 7
324#define ANYOF_ALPHA 8
325#define ANYOF_NALPHA 9
326#define ANYOF_ASCII 10
327#define ANYOF_NASCII 11
328#define ANYOF_CNTRL 12
329#define ANYOF_NCNTRL 13
330#define ANYOF_GRAPH 14
331#define ANYOF_NGRAPH 15
332#define ANYOF_LOWER 16
333#define ANYOF_NLOWER 17
334#define ANYOF_PRINT 18
335#define ANYOF_NPRINT 19
336#define ANYOF_PUNCT 20
337#define ANYOF_NPUNCT 21
338#define ANYOF_UPPER 22
339#define ANYOF_NUPPER 23
340#define ANYOF_XDIGIT 24
341#define ANYOF_NXDIGIT 25
aaa51d5e
JF
342#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
343#define ANYOF_NPSXSPC 27
ffc61ed2 344#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 345#define ANYOF_NBLANK 29
b8c5462f 346
aaa51d5e 347#define ANYOF_MAX 32
b8c5462f 348
e1d1eefb
YO
349/* pseudo classes, not stored in the class bitmap, but used as flags
350 during compilation of char classes */
351
352#define ANYOF_VERTWS (ANYOF_MAX+1)
353#define ANYOF_NVERTWS (ANYOF_MAX+2)
354#define ANYOF_HORIZWS (ANYOF_MAX+3)
355#define ANYOF_NHORIZWS (ANYOF_MAX+4)
356
b8c5462f
JH
357/* Backward source code compatibility. */
358
359#define ANYOF_ALNUML ANYOF_ALNUM
360#define ANYOF_NALNUML ANYOF_NALNUM
361#define ANYOF_SPACEL ANYOF_SPACE
362#define ANYOF_NSPACEL ANYOF_NSPACE
363
364/* Utility macros for the bitmap and classes of ANYOF */
365
936ed897
IZ
366#define ANYOF_SIZE (sizeof(struct regnode_charclass))
367#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 368
936ed897 369#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
370
371#define ANYOF_BIT(c) (1 << ((c) & 7))
372
936ed897 373#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
374#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
375#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
376#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
377
936ed897
IZ
378#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
379#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
380
381#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 382#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
383#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
384#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
385#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
386
f8bef550
NC
387#define ANYOF_BITMAP_SETALL(p) \
388 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
389#define ANYOF_BITMAP_CLEARALL(p) \
390 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
391/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
392#define ANYOF_BITMAP_TESTALLSET(p) \
393 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)
394
936ed897
IZ
395#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
396#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
397#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 398
786e8c11 399
a687059c
LW
400/*
401 * Utility definitions.
402 */
2304df62 403#ifndef CHARMASK
075abff3 404# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 405#else
075abff3 406# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 407#endif
a687059c 408
c277df42
IZ
409#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
410
3dab1dad
YO
411#define REG_SEEN_ZERO_LEN 0x00000001
412#define REG_SEEN_LOOKBEHIND 0x00000002
413#define REG_SEEN_GPOS 0x00000004
414#define REG_SEEN_EVAL 0x00000008
415#define REG_SEEN_CANY 0x00000010
cce850e4 416#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 417#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 418#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 419#define REG_SEEN_VERBARG 0x00000080
5d458dd8 420#define REG_SEEN_CUTGROUP 0x00000100
bcdf7404 421#define REG_SEEN_RUN_ON_COMMENT 0x00000200
d09b2d29 422
73c4f7a1
GS
423START_EXTERN_C
424
be8e71aa
YO
425#ifdef PLUGGABLE_RE_EXTENSION
426#include "re_nodes.h"
427#else
d09b2d29 428#include "regnodes.h"
be8e71aa 429#endif
d09b2d29 430
4d61ec05 431/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 432#ifndef DOINIT
4d61ec05 433EXTCONST U8 PL_varies[];
d09b2d29 434#else
4d61ec05 435EXTCONST U8 PL_varies[] = {
0111c4fd 436 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
99ed40ff
YO
437 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
438 NREF, NREFF, NREFFL,
439 0
c277df42 440};
d09b2d29 441#endif
c277df42 442
4d61ec05
GS
443/* The following always have a length of 1. U8 we can do strchr() on it. */
444/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 445#ifndef DOINIT
4d61ec05 446EXTCONST U8 PL_simple[];
d09b2d29 447#else
4d61ec05 448EXTCONST U8 PL_simple[] = {
f33976b4 449 REG_ANY, SANY, CANY,
ffc61ed2
JH
450 ANYOF,
451 ALNUM, ALNUML,
452 NALNUM, NALNUML,
453 SPACE, SPACEL,
454 NSPACE, NSPACEL,
455 DIGIT, NDIGIT,
e1d1eefb
YO
456 VERTWS, NVERTWS,
457 HORIZWS, NHORIZWS,
ffc61ed2 458 0
c277df42
IZ
459};
460#endif
461
f9f4320a
YO
462#ifndef PLUGGABLE_RE_EXTENSION
463#ifndef DOINIT
464EXTCONST regexp_engine PL_core_reg_engine;
465#else /* DOINIT */
466EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 467 Perl_re_compile,
2fdbfb4d 468 Perl_regexec_flags,
f9f4320a
YO
469 Perl_re_intuit_start,
470 Perl_re_intuit_string,
2fdbfb4d
AB
471 Perl_regfree_internal,
472 Perl_reg_numbered_buff_fetch,
473 Perl_reg_numbered_buff_store,
474 Perl_reg_numbered_buff_length,
192b9cd1
AB
475 Perl_reg_named_buff,
476 Perl_reg_named_buff_iter,
49d7dfbc 477 Perl_reg_qr_package,
f9f4320a 478#if defined(USE_ITHREADS)
f8149455 479 Perl_regdupe_internal
f9f4320a
YO
480#endif
481};
482#endif /* DOINIT */
483#endif /* PLUGGABLE_RE_EXTENSION */
484
485
73c4f7a1 486END_EXTERN_C
cad2e5aa 487
cad2e5aa 488
261faec3
MJD
489/* .what is a character array with one character for each member of .data
490 * The character describes the function of the corresponding .data item:
0111c4fd 491 * f - start-class data for regstclass optimization
261faec3
MJD
492 * n - Root of op tree for (?{EVAL}) item
493 * o - Start op for (?{EVAL}) item
3dab1dad 494 * p - Pad for (?{EVAL}) item
9e55ce06
JH
495 * s - swash for unicode-style character class, and the multicharacter
496 * strings resulting from casefolding the single-character entries
497 * in the character class
a3621e74 498 * t - trie struct
55eed653 499 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 500 * also used for revcharmap and words under DEBUGGING
07be1b83 501 * T - aho-trie struct
81714fb9 502 * S - sv for named capture lookup
261faec3 503 * 20010712 mjd@plover.com
c3613ed7 504 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 505 */
cad2e5aa
JH
506struct reg_data {
507 U32 count;
508 U8 *what;
509 void* data[1];
510};
511
a1cac82e
NC
512/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
513 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 514#define anchored_substr substrs->data[0].substr
33b8afdf 515#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 516#define anchored_offset substrs->data[0].min_offset
1de06328
YO
517#define anchored_end_shift substrs->data[0].end_shift
518
cad2e5aa 519#define float_substr substrs->data[1].substr
33b8afdf 520#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
521#define float_min_offset substrs->data[1].min_offset
522#define float_max_offset substrs->data[1].max_offset
1de06328
YO
523#define float_end_shift substrs->data[1].end_shift
524
cad2e5aa 525#define check_substr substrs->data[2].substr
33b8afdf 526#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
527#define check_offset_min substrs->data[2].min_offset
528#define check_offset_max substrs->data[2].max_offset
1de06328 529#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
530
531
532
533/* trie related stuff */
5d9a96ca 534
a3621e74
YO
535/* a transition record for the state machine. the
536 check field determines which state "owns" the
537 transition. the char the transition is for is
538 determined by offset from the owning states base
539 field. the next field determines which state
540 is to be transitioned to if any.
541*/
542struct _reg_trie_trans {
543 U32 next;
544 U32 check;
545};
546
547/* a transition list element for the list based representation */
548struct _reg_trie_trans_list_elem {
549 U16 forid;
550 U32 newstate;
551};
552typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
553
554/* a state for compressed nodes. base is an offset
555 into an array of reg_trie_trans array. If wordnum is
556 nonzero the state is accepting. if base is zero then
557 the state has no children (and will be accepting)
558*/
559struct _reg_trie_state {
560 U16 wordnum;
561 union {
562 U32 base;
563 reg_trie_trans_le* list;
564 } trans;
565};
566
567
568
a3621e74
YO
569typedef struct _reg_trie_state reg_trie_state;
570typedef struct _reg_trie_trans reg_trie_trans;
571
572
573/* anything in here that needs to be freed later
23eab42c
NC
574 should be dealt with in pregfree.
575 refcount is first in both this and _reg_ac_data to allow a space
576 optimisation in Perl_regdupe. */
a3621e74 577struct _reg_trie_data {
23eab42c 578 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
579 U32 lasttrans; /* last valid transition element */
580 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
581 reg_trie_state *states; /* state data */
582 reg_trie_trans *trans; /* array of transition elements */
583 char *bitmap; /* stclass bitmap */
786e8c11
YO
584 U32 *wordlen; /* array of lengths of words */
585 U16 *jump; /* optional 1 indexed array of offsets before tail
586 for the node following a given word. */
587 U16 *nextword; /* optional 1 indexed array to support linked list
588 of duplicate wordnums */
1b5c067c
NC
589 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
590 U32 startstate; /* initial state - used for common prefix optimisation */
591 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
592 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
1e2e3d02
YO
593 U32 statecount; /* Build only - number of states in the states array
594 (including the unused zero state) */
786e8c11 595 U32 wordcount; /* Build only */
a3621e74 596#ifdef DEBUGGING
786e8c11 597 STRLEN charcount; /* Build only */
a3621e74
YO
598#endif
599};
2b8b4781
NC
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
604#ifdef DEBUGGING
605 AV *words; Array of words contained in trie, for dumping
606 AV *revcharmap; Map of each charid back to its character representation
607#endif
608*/
609
610#define TRIE_WORDS_OFFSET 2
611
a3621e74
YO
612typedef struct _reg_trie_data reg_trie_data;
613
23eab42c
NC
614/* refcount is first in both this and _reg_trie_data to allow a space
615 optimisation in Perl_regdupe. */
07be1b83 616struct _reg_ac_data {
23eab42c 617 U32 refcount;
1b5c067c 618 U32 trie;
07be1b83
YO
619 U32 *fail;
620 reg_trie_state *states;
07be1b83
YO
621};
622typedef struct _reg_ac_data reg_ac_data;
623
3dab1dad
YO
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... */
627
628#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 629#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
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))
3dab1dad 633
1de06328
YO
634#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
635#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
636
637
786e8c11
YO
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))
3dab1dad 640
a3621e74
YO
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))
645
3dab1dad 646#ifdef DEBUGGING
3dab1dad 647#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 648#else
3dab1dad 649#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 650#endif
a3621e74 651
0111c4fd
RGS
652#define RE_TRIE_MAXBUF_INIT 65536
653#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
654#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
655
be8e71aa
YO
656/*
657
658RE_DEBUG_FLAGS is used to control what debug output is emitted
659its divided into three groups of options, some of which interact.
660The three groups are: Compile, Execute, Extra. There is room for a
661further group, as currently only the low three bytes are used.
662
663 Compile Options:
07be1b83 664
be8e71aa
YO
665 PARSE
666 PEEP
667 TRIE
668 PROGRAM
669 OFFSETS
a3621e74 670
be8e71aa 671 Execute Options:
a3621e74 672
be8e71aa
YO
673 INTUIT
674 MATCH
675 TRIE
a3621e74 676
be8e71aa 677 Extra Options
a3621e74 678
be8e71aa
YO
679 TRIE
680 OFFSETS
681
682If you modify any of these make sure you make corresponding changes to
683re.pm, especially to the documentation.
a3621e74 684
be8e71aa
YO
685*/
686
687
688/* Compile */
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
be8e71aa
YO
694
695/* Execute */
696#define RE_DEBUG_EXECUTE_MASK 0x00FF00
697#define RE_DEBUG_EXECUTE_INTUIT 0x000100
698#define RE_DEBUG_EXECUTE_MATCH 0x000200
699#define RE_DEBUG_EXECUTE_TRIE 0x000400
700
701/* Extra */
702#define RE_DEBUG_EXTRA_MASK 0xFF0000
703#define RE_DEBUG_EXTRA_TRIE 0x010000
704#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
705#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
706#define RE_DEBUG_EXTRA_STATE 0x080000
707#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 708#define RE_DEBUG_EXTRA_BUFFERS 0x400000
24b23f37
YO
709/* combined */
710#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 711
e68ec53f 712#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
713/* Compile */
714#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 715 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 716#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 717 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 718#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 719 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 720#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 721 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 722#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 723 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 724#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 725 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
726
727/* Execute */
728#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 729 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 730#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 731 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 732#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 733 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 734#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 735 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
736
737/* Extra */
738#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 739 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
740#define DEBUG_OFFSETS_r(x) DEBUG_r( \
741 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
742#define DEBUG_STATE_r(x) DEBUG_r( \
743 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
744#define DEBUG_STACK_r(x) DEBUG_r( \
745 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
746#define DEBUG_BUFFERS_r(x) DEBUG_r( \
747 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
748
a5ca303d
YO
749#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
750 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
751 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 752#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 753 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
754 Perl_warn_nocontext x )
755#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 756 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 757#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 758 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
759
760#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 761 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
762 | RE_DEBUG_EXECUTE_TRIE )) x )
763
764/* initialization */
765/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
766#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
767 SV * re_debug_flags_sv = NULL; \
768 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
769 if (re_debug_flags_sv) { \
8e9a8a48 770 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 771 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 772 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
773 }\
774})
a3621e74
YO
775
776#ifdef DEBUGGING
ab3bbdeb 777
8e9a8a48 778#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
779
780#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
781 const char * const rpv = \
ddc5bc0f 782 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 783 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 784 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 785 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
786
787#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
788 const char * const rpv = \
ddc5bc0f 789 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 790 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 791 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
792
793#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
794 const char * const rpv = \
ddc5bc0f 795 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 796 PL_colors[0], PL_colors[1], \
0a0e9afc 797 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELIPSES | \
ab3bbdeb 798 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 799 )
ab3bbdeb
YO
800
801#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
802#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
803
804#else /* if not DEBUGGING */
805
a3621e74 806#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
807#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
808#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
809#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
810#define RE_SV_DUMPLEN(ItEm)
811#define RE_SV_TAIL(ItEm)
812
813#endif /* DEBUG RELATED DEFINES */
a3621e74 814
99ed40ff 815
e1d1eefb 816