This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Further tweaks to make it easier to create regexp engine plug ins.
[perl5.git] / regcomp.h
CommitLineData
a0d0e21e 1/* regcomp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
1d325971 4 * 2000, 2001, 2002, 2003, 2005 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
LW
9 */
10
c277df42
IZ
11typedef OP OP_4tree; /* Will be redefined later. */
12
07be1b83 13
58e23c8d 14/* Convert branch sequences to more efficient trie ops? */
07be1b83 15#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d
YO
16
17/* Be really agressive about optimising patterns with trie sequences? */
07be1b83 18#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d
YO
19
20/* Should the optimiser take positive assertions into account? */
1de06328 21#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 1
58e23c8d
YO
22
23/* Not for production use: */
07be1b83 24#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 25
b9b4dddf
YO
26/* Unless the next line is uncommented it is illegal to combine lazy
27 matching with possessive matching. Frankly it doesn't make much sense
28 to allow it as X*?+ matches nothing, X+?+ matches a single char only,
29 and X{min,max}?+ matches min times only.
30 */
31/* #define REG_ALLOW_MINMOD_SUSPEND */
07be1b83 32
a687059c
LW
33/*
34 * The "internal use only" fields in regexp.h are present to pass info from
35 * compile to execute that permits the execute phase to run lots faster on
36 * simple cases. They are:
37 *
bd61b366 38 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
39 * reganch is the match anchored (at beginning-of-line only)?
40 * regmust string (pointer into program) that match must include, or NULL
79072805 41 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
42 * regmlen length of regmust string
43 * [regmlen not used currently]
44 *
45 * Regstart and reganch permit very fast decisions on suitable starting points
46 * for a match, cutting down the work a lot. Regmust permits fast rejection
47 * of lines that cannot possibly match. The regmust tests are costly enough
e50aee73 48 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
49 * potentially expensive (at present, the only such thing detected is * or +
50 * at the start of the r.e., which can involve a lot of backup). Regmlen is
e50aee73 51 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
52 * it anyway.
53 * [regmust is now supplied always. The tests that use regmust have a
54 * heuristic that disables the test if it usually matches.]
55 *
56 * [In fact, we now use regmust in many cases to locate where the search
57 * starts in the string, so if regback is >= 0, the regmust search is never
58 * wasted effort. The regback variable says how many characters back from
59 * where regmust matched is the earliest possible start of the match.
60 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
61 */
62
63/*
64 * Structure for regexp "program". This is essentially a linear encoding
65 * of a nondeterministic finite-state machine (aka syntax charts or
66 * "railroad normal form" in parsing technology). Each node is an opcode
67 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
68 * all nodes except BRANCH implement concatenation; a "next" pointer with
69 * a BRANCH on both ends of it is connecting two alternatives. (Here we
70 * have one of the subtle syntax dependencies: an individual BRANCH (as
71 * opposed to a collection of them) is never concatenated with anything
72 * because of operator precedence.) The operand of some types of node is
73 * a literal string; for others, it is a node leading into a sub-FSM. In
74 * particular, the operand of a BRANCH node is the first node of the branch.
75 * (NB this is *not* a tree structure: the tail of the branch connects
a3621e74
YO
76 * to the thing following the set of BRANCHes.) The opcodes are defined
77 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
a687059c
LW
78 */
79
a687059c
LW
80/*
81 * A node is one char of opcode followed by two chars of "next" pointer.
82 * "Next" pointers are stored as two 8-bit pieces, high order first. The
83 * value is a positive offset from the opcode of the node containing it.
84 * An operand, if any, simply follows the node. (Note that much of the
85 * code generation knows about this implicit relationship.)
86 *
87 * Using two bytes for the "next" pointer is vast overkill for most things,
88 * but allows patterns to get big without disasters.
89 *
e91177ed 90 * [The "next" pointer is always aligned on an even
a687059c
LW
91 * boundary, and reads the offset directly as a short. Also, there is no
92 * special test to reverse the sign of BACK pointers since the offset is
93 * stored negative.]
94 */
785a26d5
YO
95
96/* This is the stuff that used to live in regexp.h that was truly
97 private to the engine itself. It now lives here. */
98
99/* swap buffer for paren structs */
100typedef struct regexp_paren_ofs {
101 I32 *startp;
102 I32 *endp;
103} regexp_paren_ofs;
104
105 typedef struct regexp_internal {
6d5c990f
RGS
106 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
107 U32 *offsets; /* offset annotations 20001228 MJD
108 data about mapping the program to the
109 string*/
110 regnode *regstclass; /* Optional startclass as identified or constructed
111 by the optimiser */
112 struct reg_data *data; /* Additional miscellaneous data used by the program.
113 Used to make it easier to clone and free arbitrary
114 data that the regops need. Often the ARG field of
115 a regop is an index into this structure */
116 regnode program[1]; /* Unwarranted chumminess with compiler. */
117} regexp_internal;
118
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)
785a26d5
YO
122/*
123 * Flags stored in regexp->intflags
124 * These are used only internally to the regexp engine
125 *
126 * See regexp.h for flags used externally to the regexp engine
127 */
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
134
135/* this is where the old regcomp.h started */
a687059c 136
c277df42 137struct regnode_string {
cd439c50 138 U8 str_len;
c277df42
IZ
139 U8 type;
140 U16 next_off;
cd439c50 141 char string[1];
c277df42 142};
a687059c 143
6bda09f9
YO
144/* Argument bearing node - workhorse,
145 arg1 is often for the data field */
c277df42
IZ
146struct regnode_1 {
147 U8 flags;
148 U8 type;
149 U16 next_off;
150 U32 arg1;
151};
152
6bda09f9
YO
153/* Similar to a regnode_1 but with an extra signed argument */
154struct regnode_2L {
155 U8 flags;
156 U8 type;
157 U16 next_off;
158 U32 arg1;
159 I32 arg2;
160};
161
162/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
163struct regnode_2 {
164 U8 flags;
165 U8 type;
166 U16 next_off;
167 U16 arg1;
168 U16 arg2;
169};
170
6bda09f9 171
936ed897 172#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
ffc61ed2 173#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 174
786e8c11 175/* also used by trie */
936ed897
IZ
176struct regnode_charclass {
177 U8 flags;
178 U8 type;
179 U16 next_off;
ffc61ed2 180 U32 arg1;
19103b91 181 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
182};
183
19103b91
JH
184struct regnode_charclass_class { /* has [[:blah:]] classes */
185 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
186 U8 type;
187 U16 next_off;
ffc61ed2 188 U32 arg1;
19103b91
JH
189 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
190 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
191};
192
9c7e81e8
AD
193/* XXX fix this description.
194 Impose a limit of REG_INFTY on various pattern matching operations
195 to limit stack growth and to avoid "infinite" recursions.
196*/
197/* The default size for REG_INFTY is I16_MAX, which is the same as
198 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
199 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
200 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
201 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
202 smaller default.
203 --Andy Dougherty 11 June 1998
204*/
205#if SHORTSIZE > 2
206# ifndef REG_INFTY
207# define REG_INFTY ((1<<15)-1)
208# endif
209#endif
210
211#ifndef REG_INFTY
83cfe48c
IZ
212# define REG_INFTY I16_MAX
213#endif
a687059c 214
e91177ed
IZ
215#define ARG_VALUE(arg) (arg)
216#define ARG__SET(arg,val) ((arg) = (val))
c277df42 217
e8e3b0fb
JH
218#undef ARG
219#undef ARG1
220#undef ARG2
221
c277df42
IZ
222#define ARG(p) ARG_VALUE(ARG_LOC(p))
223#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
224#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 225#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 226
c277df42
IZ
227#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
228#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
229#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 230#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 231
e8e3b0fb
JH
232#undef NEXT_OFF
233#undef NODE_ALIGN
234
075abff3
AL
235#define NEXT_OFF(p) ((p)->next_off)
236#define NODE_ALIGN(node)
237#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 238
c277df42
IZ
239#define SIZE_ALIGN NODE_ALIGN
240
e8e3b0fb
JH
241#undef OP
242#undef OPERAND
243#undef MASK
244#undef STRING
245
e91177ed
IZ
246#define OP(p) ((p)->type)
247#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
248#define MASK(p) ((char*)OPERAND(p))
249#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
250#define STRING(p) (((struct regnode_string *)p)->string)
251#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
252#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
253
e8e3b0fb
JH
254#undef NODE_ALIGN
255#undef ARG_LOC
256#undef NEXTOPER
257#undef PREVOPER
258
e91177ed
IZ
259#define NODE_ALIGN(node)
260#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
261#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
262#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 263#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 264
e91177ed
IZ
265#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
266#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
267
268#define NODE_STEP_B 4
c277df42
IZ
269
270#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
271#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
272
e91177ed 273#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 274 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 275#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 276 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 277
22c35a8c 278#define REG_MAGIC 0234
a687059c 279
830247a4 280#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 281
936ed897 282/* Flags for node->flags of ANYOF */
b8c5462f 283
19103b91 284#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
285#define ANYOF_INVERT 0x04
286#define ANYOF_FOLD 0x02
287#define ANYOF_LOCALE 0x01
b8c5462f 288
653099ff 289/* Used for regstclass only */
ffc61ed2
JH
290#define ANYOF_EOS 0x10 /* Can match an empty string too */
291
292/* There is a character or a range past 0xff */
293#define ANYOF_UNICODE 0x20
1aa99e6b 294#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 295
4f66b38d
HS
296/* size of node is large (includes class pointer) */
297#define ANYOF_LARGE 0x80
298
ffc61ed2
JH
299/* Are there any runtime flags on in this node? */
300#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
301
302#define ANYOF_FLAGS_ALL 0xff
653099ff 303
936ed897 304/* Character classes for node->classflags of ANYOF */
653099ff
GS
305/* Should be synchronized with a table in regprop() */
306/* 2n should pair with 2n+1 */
b8c5462f 307
289d4f09 308#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 309#define ANYOF_NALNUM 1
aaa51d5e 310#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
311#define ANYOF_NSPACE 3
312#define ANYOF_DIGIT 4
313#define ANYOF_NDIGIT 5
289d4f09 314#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
315#define ANYOF_NALNUMC 7
316#define ANYOF_ALPHA 8
317#define ANYOF_NALPHA 9
318#define ANYOF_ASCII 10
319#define ANYOF_NASCII 11
320#define ANYOF_CNTRL 12
321#define ANYOF_NCNTRL 13
322#define ANYOF_GRAPH 14
323#define ANYOF_NGRAPH 15
324#define ANYOF_LOWER 16
325#define ANYOF_NLOWER 17
326#define ANYOF_PRINT 18
327#define ANYOF_NPRINT 19
328#define ANYOF_PUNCT 20
329#define ANYOF_NPUNCT 21
330#define ANYOF_UPPER 22
331#define ANYOF_NUPPER 23
332#define ANYOF_XDIGIT 24
333#define ANYOF_NXDIGIT 25
aaa51d5e
JF
334#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
335#define ANYOF_NPSXSPC 27
ffc61ed2 336#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 337#define ANYOF_NBLANK 29
b8c5462f 338
aaa51d5e 339#define ANYOF_MAX 32
b8c5462f
JH
340
341/* Backward source code compatibility. */
342
343#define ANYOF_ALNUML ANYOF_ALNUM
344#define ANYOF_NALNUML ANYOF_NALNUM
345#define ANYOF_SPACEL ANYOF_SPACE
346#define ANYOF_NSPACEL ANYOF_NSPACE
347
348/* Utility macros for the bitmap and classes of ANYOF */
349
936ed897
IZ
350#define ANYOF_SIZE (sizeof(struct regnode_charclass))
351#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 352
936ed897 353#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
354
355#define ANYOF_BIT(c) (1 << ((c) & 7))
356
936ed897 357#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
358#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
359#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
360#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
361
936ed897
IZ
362#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
363#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
364
365#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 366#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
367#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
368#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
369#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
370
f8bef550
NC
371#define ANYOF_BITMAP_SETALL(p) \
372 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
373#define ANYOF_BITMAP_CLEARALL(p) \
374 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
375/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
376#define ANYOF_BITMAP_TESTALLSET(p) \
377 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)
378
936ed897
IZ
379#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
380#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
381#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 382
786e8c11 383
a687059c
LW
384/*
385 * Utility definitions.
386 */
2304df62 387#ifndef CHARMASK
075abff3 388# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 389#else
075abff3 390# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 391#endif
a687059c 392
c277df42
IZ
393#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
394
3dab1dad
YO
395#define REG_SEEN_ZERO_LEN 0x00000001
396#define REG_SEEN_LOOKBEHIND 0x00000002
397#define REG_SEEN_GPOS 0x00000004
398#define REG_SEEN_EVAL 0x00000008
399#define REG_SEEN_CANY 0x00000010
cce850e4 400#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 401#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 402#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 403#define REG_SEEN_VERBARG 0x00000080
5d458dd8 404#define REG_SEEN_CUTGROUP 0x00000100
d09b2d29 405
73c4f7a1
GS
406START_EXTERN_C
407
be8e71aa
YO
408#ifdef PLUGGABLE_RE_EXTENSION
409#include "re_nodes.h"
410#else
d09b2d29 411#include "regnodes.h"
be8e71aa 412#endif
d09b2d29 413
4d61ec05 414/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 415#ifndef DOINIT
4d61ec05 416EXTCONST U8 PL_varies[];
d09b2d29 417#else
4d61ec05 418EXTCONST U8 PL_varies[] = {
0111c4fd 419 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
99ed40ff
YO
420 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
421 NREF, NREFF, NREFFL,
422 0
c277df42 423};
d09b2d29 424#endif
c277df42 425
4d61ec05
GS
426/* The following always have a length of 1. U8 we can do strchr() on it. */
427/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 428#ifndef DOINIT
4d61ec05 429EXTCONST U8 PL_simple[];
d09b2d29 430#else
4d61ec05 431EXTCONST U8 PL_simple[] = {
f33976b4 432 REG_ANY, SANY, CANY,
ffc61ed2
JH
433 ANYOF,
434 ALNUM, ALNUML,
435 NALNUM, NALNUML,
436 SPACE, SPACEL,
437 NSPACE, NSPACEL,
438 DIGIT, NDIGIT,
439 0
c277df42
IZ
440};
441#endif
442
f9f4320a
YO
443#ifndef PLUGGABLE_RE_EXTENSION
444#ifndef DOINIT
445EXTCONST regexp_engine PL_core_reg_engine;
446#else /* DOINIT */
447EXTCONST regexp_engine PL_core_reg_engine = {
3c6f95d1 448 Perl_re_compile,
f9f4320a
YO
449 Perl_regexec_flags,
450 Perl_re_intuit_start,
451 Perl_re_intuit_string,
f8149455 452 Perl_regfree_internal,
f9f4320a 453#if defined(USE_ITHREADS)
f8149455 454 Perl_regdupe_internal
f9f4320a
YO
455#endif
456};
457#endif /* DOINIT */
458#endif /* PLUGGABLE_RE_EXTENSION */
459
460
73c4f7a1 461END_EXTERN_C
cad2e5aa 462
cad2e5aa 463
261faec3
MJD
464/* .what is a character array with one character for each member of .data
465 * The character describes the function of the corresponding .data item:
0111c4fd 466 * f - start-class data for regstclass optimization
261faec3
MJD
467 * n - Root of op tree for (?{EVAL}) item
468 * o - Start op for (?{EVAL}) item
3dab1dad 469 * p - Pad for (?{EVAL}) item
9e55ce06
JH
470 * s - swash for unicode-style character class, and the multicharacter
471 * strings resulting from casefolding the single-character entries
472 * in the character class
a3621e74 473 * t - trie struct
55eed653 474 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 475 * also used for revcharmap and words under DEBUGGING
07be1b83 476 * T - aho-trie struct
81714fb9 477 * S - sv for named capture lookup
261faec3 478 * 20010712 mjd@plover.com
c3613ed7 479 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 480 */
cad2e5aa
JH
481struct reg_data {
482 U32 count;
483 U8 *what;
484 void* data[1];
485};
486
cad2e5aa 487#define anchored_substr substrs->data[0].substr
33b8afdf 488#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 489#define anchored_offset substrs->data[0].min_offset
1de06328
YO
490#define anchored_end_shift substrs->data[0].end_shift
491
cad2e5aa 492#define float_substr substrs->data[1].substr
33b8afdf 493#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
494#define float_min_offset substrs->data[1].min_offset
495#define float_max_offset substrs->data[1].max_offset
1de06328
YO
496#define float_end_shift substrs->data[1].end_shift
497
cad2e5aa 498#define check_substr substrs->data[2].substr
33b8afdf 499#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
500#define check_offset_min substrs->data[2].min_offset
501#define check_offset_max substrs->data[2].max_offset
1de06328 502#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
503
504
505
506/* trie related stuff */
5d9a96ca 507
a3621e74
YO
508/* a transition record for the state machine. the
509 check field determines which state "owns" the
510 transition. the char the transition is for is
511 determined by offset from the owning states base
512 field. the next field determines which state
513 is to be transitioned to if any.
514*/
515struct _reg_trie_trans {
516 U32 next;
517 U32 check;
518};
519
520/* a transition list element for the list based representation */
521struct _reg_trie_trans_list_elem {
522 U16 forid;
523 U32 newstate;
524};
525typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
526
527/* a state for compressed nodes. base is an offset
528 into an array of reg_trie_trans array. If wordnum is
529 nonzero the state is accepting. if base is zero then
530 the state has no children (and will be accepting)
531*/
532struct _reg_trie_state {
533 U16 wordnum;
534 union {
535 U32 base;
536 reg_trie_trans_le* list;
537 } trans;
538};
539
540
541
a3621e74
YO
542typedef struct _reg_trie_state reg_trie_state;
543typedef struct _reg_trie_trans reg_trie_trans;
544
545
546/* anything in here that needs to be freed later
23eab42c
NC
547 should be dealt with in pregfree.
548 refcount is first in both this and _reg_ac_data to allow a space
549 optimisation in Perl_regdupe. */
a3621e74 550struct _reg_trie_data {
23eab42c 551 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
552 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
553 U32 lasttrans; /* last valid transition element */
554 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
555 reg_trie_state *states; /* state data */
556 reg_trie_trans *trans; /* array of transition elements */
557 char *bitmap; /* stclass bitmap */
786e8c11
YO
558 U32 startstate; /* initial state - used for common prefix optimisation */
559 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
560 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
561 U32 *wordlen; /* array of lengths of words */
562 U16 *jump; /* optional 1 indexed array of offsets before tail
563 for the node following a given word. */
564 U16 *nextword; /* optional 1 indexed array to support linked list
565 of duplicate wordnums */
1e2e3d02
YO
566 U32 statecount; /* Build only - number of states in the states array
567 (including the unused zero state) */
786e8c11 568 U32 wordcount; /* Build only */
a3621e74 569#ifdef DEBUGGING
786e8c11 570 STRLEN charcount; /* Build only */
a3621e74
YO
571#endif
572};
2b8b4781
NC
573/* There is one (3 under DEBUGGING) pointers that logically belong in this
574 structure, but are held outside as they need duplication on thread cloning,
575 whereas the rest of the structure can be read only:
576 HV *widecharmap; code points > 255 to charid
577#ifdef DEBUGGING
578 AV *words; Array of words contained in trie, for dumping
579 AV *revcharmap; Map of each charid back to its character representation
580#endif
581*/
582
583#define TRIE_WORDS_OFFSET 2
584
a3621e74
YO
585typedef struct _reg_trie_data reg_trie_data;
586
23eab42c
NC
587/* refcount is first in both this and _reg_trie_data to allow a space
588 optimisation in Perl_regdupe. */
07be1b83 589struct _reg_ac_data {
23eab42c 590 U32 refcount;
07be1b83
YO
591 U32 *fail;
592 reg_trie_state *states;
3251b653 593 U32 trie;
07be1b83
YO
594};
595typedef struct _reg_ac_data reg_ac_data;
596
3dab1dad
YO
597/* ANY_BIT doesnt use the structure, so we can borrow it here.
598 This is simpler than refactoring all of it as wed end up with
599 three different sets... */
600
601#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 602#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
603#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
604#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
605#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 606
1de06328
YO
607#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
608#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
609
610
786e8c11
YO
611#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
612#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 613
a3621e74
YO
614/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
615#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
616#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
617#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
618
3dab1dad 619#ifdef DEBUGGING
3dab1dad 620#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 621#else
3dab1dad 622#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 623#endif
a3621e74 624
0111c4fd
RGS
625#define RE_TRIE_MAXBUF_INIT 65536
626#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
627#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
628
be8e71aa
YO
629/*
630
631RE_DEBUG_FLAGS is used to control what debug output is emitted
632its divided into three groups of options, some of which interact.
633The three groups are: Compile, Execute, Extra. There is room for a
634further group, as currently only the low three bytes are used.
635
636 Compile Options:
07be1b83 637
be8e71aa
YO
638 PARSE
639 PEEP
640 TRIE
641 PROGRAM
642 OFFSETS
a3621e74 643
be8e71aa 644 Execute Options:
a3621e74 645
be8e71aa
YO
646 INTUIT
647 MATCH
648 TRIE
a3621e74 649
be8e71aa 650 Extra Options
a3621e74 651
be8e71aa
YO
652 TRIE
653 OFFSETS
654
655If you modify any of these make sure you make corresponding changes to
656re.pm, especially to the documentation.
a3621e74 657
be8e71aa
YO
658*/
659
660
661/* Compile */
662#define RE_DEBUG_COMPILE_MASK 0x0000FF
663#define RE_DEBUG_COMPILE_PARSE 0x000001
664#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
665#define RE_DEBUG_COMPILE_TRIE 0x000004
666#define RE_DEBUG_COMPILE_DUMP 0x000008
be8e71aa
YO
667
668/* Execute */
669#define RE_DEBUG_EXECUTE_MASK 0x00FF00
670#define RE_DEBUG_EXECUTE_INTUIT 0x000100
671#define RE_DEBUG_EXECUTE_MATCH 0x000200
672#define RE_DEBUG_EXECUTE_TRIE 0x000400
673
674/* Extra */
675#define RE_DEBUG_EXTRA_MASK 0xFF0000
676#define RE_DEBUG_EXTRA_TRIE 0x010000
677#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
678#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
679#define RE_DEBUG_EXTRA_STATE 0x080000
680#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
24b23f37
YO
681/* combined */
682#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 683
e68ec53f 684#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
685/* Compile */
686#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 687 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 688#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 689 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 690#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 691 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 692#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 693 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 694#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 695 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 696#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 697 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
698
699/* Execute */
700#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 701 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 702#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 703 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 704#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 705 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 706#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 707 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
708
709/* Extra */
710#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 711 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
712#define DEBUG_OFFSETS_r(x) DEBUG_r( \
713 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
714#define DEBUG_STATE_r(x) DEBUG_r( \
715 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
716#define DEBUG_STACK_r(x) DEBUG_r( \
717 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
a5ca303d
YO
718#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
719 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
720 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 721#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 722 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
723 Perl_warn_nocontext x )
724#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 725 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 726#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 727 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
728
729#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 730 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
731 | RE_DEBUG_EXECUTE_TRIE )) x )
732
733/* initialization */
734/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
735#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
736 SV * re_debug_flags_sv = NULL; \
737 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
738 if (re_debug_flags_sv) { \
8e9a8a48 739 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 740 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 741 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
742 }\
743})
a3621e74
YO
744
745#ifdef DEBUGGING
ab3bbdeb 746
8e9a8a48 747#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
748
749#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
750 const char * const rpv = \
ddc5bc0f 751 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
752 PL_colors[(c1)],PL_colors[(c2)], \
753 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 754 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
755
756#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
757 const char * const rpv = \
ddc5bc0f 758 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb
YO
759 PL_colors[(c1)],PL_colors[(c2)], \
760 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
761
762#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
763 const char * const rpv = \
ddc5bc0f 764 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
765 PL_colors[0], PL_colors[1], \
766 ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \
767 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
768 )
769
770#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
771#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
772
773#else /* if not DEBUGGING */
774
a3621e74 775#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
776#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
777#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
778#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
779#define RE_SV_DUMPLEN(ItEm)
780#define RE_SV_TAIL(ItEm)
781
782#endif /* DEBUG RELATED DEFINES */
a3621e74 783
99ed40ff 784