This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
The new regexp compilation function must be added to the engine structure.
[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 */
95
c277df42 96struct regnode_string {
cd439c50 97 U8 str_len;
c277df42
IZ
98 U8 type;
99 U16 next_off;
cd439c50 100 char string[1];
c277df42 101};
a687059c 102
6bda09f9
YO
103/* Argument bearing node - workhorse,
104 arg1 is often for the data field */
c277df42
IZ
105struct regnode_1 {
106 U8 flags;
107 U8 type;
108 U16 next_off;
109 U32 arg1;
110};
111
6bda09f9
YO
112/* Similar to a regnode_1 but with an extra signed argument */
113struct regnode_2L {
114 U8 flags;
115 U8 type;
116 U16 next_off;
117 U32 arg1;
118 I32 arg2;
119};
120
121/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
122struct regnode_2 {
123 U8 flags;
124 U8 type;
125 U16 next_off;
126 U16 arg1;
127 U16 arg2;
128};
129
6bda09f9 130
936ed897 131#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
ffc61ed2 132#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 133
786e8c11 134/* also used by trie */
936ed897
IZ
135struct regnode_charclass {
136 U8 flags;
137 U8 type;
138 U16 next_off;
ffc61ed2 139 U32 arg1;
19103b91 140 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
141};
142
19103b91
JH
143struct regnode_charclass_class { /* has [[:blah:]] classes */
144 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
145 U8 type;
146 U16 next_off;
ffc61ed2 147 U32 arg1;
19103b91
JH
148 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
149 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
150};
151
9c7e81e8
AD
152/* XXX fix this description.
153 Impose a limit of REG_INFTY on various pattern matching operations
154 to limit stack growth and to avoid "infinite" recursions.
155*/
156/* The default size for REG_INFTY is I16_MAX, which is the same as
157 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
158 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
159 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
160 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
161 smaller default.
162 --Andy Dougherty 11 June 1998
163*/
164#if SHORTSIZE > 2
165# ifndef REG_INFTY
166# define REG_INFTY ((1<<15)-1)
167# endif
168#endif
169
170#ifndef REG_INFTY
83cfe48c
IZ
171# define REG_INFTY I16_MAX
172#endif
a687059c 173
e91177ed
IZ
174#define ARG_VALUE(arg) (arg)
175#define ARG__SET(arg,val) ((arg) = (val))
c277df42 176
e8e3b0fb
JH
177#undef ARG
178#undef ARG1
179#undef ARG2
180
c277df42
IZ
181#define ARG(p) ARG_VALUE(ARG_LOC(p))
182#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
183#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 184#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 185
c277df42
IZ
186#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
187#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
188#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 189#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 190
e8e3b0fb
JH
191#undef NEXT_OFF
192#undef NODE_ALIGN
193
075abff3
AL
194#define NEXT_OFF(p) ((p)->next_off)
195#define NODE_ALIGN(node)
196#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 197
c277df42
IZ
198#define SIZE_ALIGN NODE_ALIGN
199
e8e3b0fb
JH
200#undef OP
201#undef OPERAND
202#undef MASK
203#undef STRING
204
e91177ed
IZ
205#define OP(p) ((p)->type)
206#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
207#define MASK(p) ((char*)OPERAND(p))
208#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
209#define STRING(p) (((struct regnode_string *)p)->string)
210#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
211#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
212
e8e3b0fb
JH
213#undef NODE_ALIGN
214#undef ARG_LOC
215#undef NEXTOPER
216#undef PREVOPER
217
e91177ed
IZ
218#define NODE_ALIGN(node)
219#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
220#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
221#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 222#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 223
e91177ed
IZ
224#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
225#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
226
227#define NODE_STEP_B 4
c277df42
IZ
228
229#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
230#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
231
e91177ed 232#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 233 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 234#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 235 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 236
22c35a8c 237#define REG_MAGIC 0234
a687059c 238
830247a4 239#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 240
936ed897 241/* Flags for node->flags of ANYOF */
b8c5462f 242
19103b91 243#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
244#define ANYOF_INVERT 0x04
245#define ANYOF_FOLD 0x02
246#define ANYOF_LOCALE 0x01
b8c5462f 247
653099ff 248/* Used for regstclass only */
ffc61ed2
JH
249#define ANYOF_EOS 0x10 /* Can match an empty string too */
250
251/* There is a character or a range past 0xff */
252#define ANYOF_UNICODE 0x20
1aa99e6b 253#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 254
4f66b38d
HS
255/* size of node is large (includes class pointer) */
256#define ANYOF_LARGE 0x80
257
ffc61ed2
JH
258/* Are there any runtime flags on in this node? */
259#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
260
261#define ANYOF_FLAGS_ALL 0xff
653099ff 262
936ed897 263/* Character classes for node->classflags of ANYOF */
653099ff
GS
264/* Should be synchronized with a table in regprop() */
265/* 2n should pair with 2n+1 */
b8c5462f 266
289d4f09 267#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 268#define ANYOF_NALNUM 1
aaa51d5e 269#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
270#define ANYOF_NSPACE 3
271#define ANYOF_DIGIT 4
272#define ANYOF_NDIGIT 5
289d4f09 273#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
274#define ANYOF_NALNUMC 7
275#define ANYOF_ALPHA 8
276#define ANYOF_NALPHA 9
277#define ANYOF_ASCII 10
278#define ANYOF_NASCII 11
279#define ANYOF_CNTRL 12
280#define ANYOF_NCNTRL 13
281#define ANYOF_GRAPH 14
282#define ANYOF_NGRAPH 15
283#define ANYOF_LOWER 16
284#define ANYOF_NLOWER 17
285#define ANYOF_PRINT 18
286#define ANYOF_NPRINT 19
287#define ANYOF_PUNCT 20
288#define ANYOF_NPUNCT 21
289#define ANYOF_UPPER 22
290#define ANYOF_NUPPER 23
291#define ANYOF_XDIGIT 24
292#define ANYOF_NXDIGIT 25
aaa51d5e
JF
293#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
294#define ANYOF_NPSXSPC 27
ffc61ed2 295#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 296#define ANYOF_NBLANK 29
b8c5462f 297
aaa51d5e 298#define ANYOF_MAX 32
b8c5462f
JH
299
300/* Backward source code compatibility. */
301
302#define ANYOF_ALNUML ANYOF_ALNUM
303#define ANYOF_NALNUML ANYOF_NALNUM
304#define ANYOF_SPACEL ANYOF_SPACE
305#define ANYOF_NSPACEL ANYOF_NSPACE
306
307/* Utility macros for the bitmap and classes of ANYOF */
308
936ed897
IZ
309#define ANYOF_SIZE (sizeof(struct regnode_charclass))
310#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 311
936ed897 312#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
313
314#define ANYOF_BIT(c) (1 << ((c) & 7))
315
936ed897 316#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
317#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
318#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
319#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
320
936ed897
IZ
321#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
322#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
323
324#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 325#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
326#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
327#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
328#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
329
f8bef550
NC
330#define ANYOF_BITMAP_SETALL(p) \
331 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
332#define ANYOF_BITMAP_CLEARALL(p) \
333 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
334/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
335#define ANYOF_BITMAP_TESTALLSET(p) \
336 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)
337
936ed897
IZ
338#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
339#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
340#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 341
786e8c11 342
a687059c
LW
343/*
344 * Utility definitions.
345 */
2304df62 346#ifndef CHARMASK
075abff3 347# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 348#else
075abff3 349# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 350#endif
a687059c 351
c277df42
IZ
352#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
353
3dab1dad
YO
354#define REG_SEEN_ZERO_LEN 0x00000001
355#define REG_SEEN_LOOKBEHIND 0x00000002
356#define REG_SEEN_GPOS 0x00000004
357#define REG_SEEN_EVAL 0x00000008
358#define REG_SEEN_CANY 0x00000010
cce850e4 359#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 360#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 361#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 362#define REG_SEEN_VERBARG 0x00000080
5d458dd8 363#define REG_SEEN_CUTGROUP 0x00000100
d09b2d29 364
73c4f7a1
GS
365START_EXTERN_C
366
be8e71aa
YO
367#ifdef PLUGGABLE_RE_EXTENSION
368#include "re_nodes.h"
369#else
d09b2d29 370#include "regnodes.h"
be8e71aa 371#endif
d09b2d29 372
4d61ec05 373/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 374#ifndef DOINIT
4d61ec05 375EXTCONST U8 PL_varies[];
d09b2d29 376#else
4d61ec05 377EXTCONST U8 PL_varies[] = {
0111c4fd 378 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
99ed40ff
YO
379 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
380 NREF, NREFF, NREFFL,
381 0
c277df42 382};
d09b2d29 383#endif
c277df42 384
4d61ec05
GS
385/* The following always have a length of 1. U8 we can do strchr() on it. */
386/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 387#ifndef DOINIT
4d61ec05 388EXTCONST U8 PL_simple[];
d09b2d29 389#else
4d61ec05 390EXTCONST U8 PL_simple[] = {
f33976b4 391 REG_ANY, SANY, CANY,
ffc61ed2
JH
392 ANYOF,
393 ALNUM, ALNUML,
394 NALNUM, NALNUML,
395 SPACE, SPACEL,
396 NSPACE, NSPACEL,
397 DIGIT, NDIGIT,
398 0
c277df42
IZ
399};
400#endif
401
f9f4320a
YO
402#ifndef PLUGGABLE_RE_EXTENSION
403#ifndef DOINIT
404EXTCONST regexp_engine PL_core_reg_engine;
405#else /* DOINIT */
406EXTCONST regexp_engine PL_core_reg_engine = {
407 Perl_pregcomp,
3c6f95d1 408 Perl_re_compile,
f9f4320a
YO
409 Perl_regexec_flags,
410 Perl_re_intuit_start,
411 Perl_re_intuit_string,
f8149455 412 Perl_regfree_internal,
f9f4320a 413#if defined(USE_ITHREADS)
f8149455 414 Perl_regdupe_internal
f9f4320a
YO
415#endif
416};
417#endif /* DOINIT */
418#endif /* PLUGGABLE_RE_EXTENSION */
419
420
73c4f7a1 421END_EXTERN_C
cad2e5aa 422
cad2e5aa 423
261faec3
MJD
424/* .what is a character array with one character for each member of .data
425 * The character describes the function of the corresponding .data item:
0111c4fd 426 * f - start-class data for regstclass optimization
261faec3
MJD
427 * n - Root of op tree for (?{EVAL}) item
428 * o - Start op for (?{EVAL}) item
3dab1dad 429 * p - Pad for (?{EVAL}) item
9e55ce06
JH
430 * s - swash for unicode-style character class, and the multicharacter
431 * strings resulting from casefolding the single-character entries
432 * in the character class
a3621e74 433 * t - trie struct
55eed653 434 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 435 * also used for revcharmap and words under DEBUGGING
07be1b83 436 * T - aho-trie struct
81714fb9 437 * S - sv for named capture lookup
261faec3 438 * 20010712 mjd@plover.com
c3613ed7 439 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 440 */
cad2e5aa
JH
441struct reg_data {
442 U32 count;
443 U8 *what;
444 void* data[1];
445};
446
447struct reg_substr_datum {
448 I32 min_offset;
449 I32 max_offset;
33b8afdf
JH
450 SV *substr; /* non-utf8 variant */
451 SV *utf8_substr; /* utf8 variant */
1de06328 452 I32 end_shift;
cad2e5aa
JH
453};
454
455struct reg_substr_data {
456 struct reg_substr_datum data[3]; /* Actual array */
457};
458
459#define anchored_substr substrs->data[0].substr
33b8afdf 460#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 461#define anchored_offset substrs->data[0].min_offset
1de06328
YO
462#define anchored_end_shift substrs->data[0].end_shift
463
cad2e5aa 464#define float_substr substrs->data[1].substr
33b8afdf 465#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
466#define float_min_offset substrs->data[1].min_offset
467#define float_max_offset substrs->data[1].max_offset
1de06328
YO
468#define float_end_shift substrs->data[1].end_shift
469
cad2e5aa 470#define check_substr substrs->data[2].substr
33b8afdf 471#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
472#define check_offset_min substrs->data[2].min_offset
473#define check_offset_max substrs->data[2].max_offset
1de06328 474#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
475
476
477
478/* trie related stuff */
5d9a96ca 479
a3621e74
YO
480/* a transition record for the state machine. the
481 check field determines which state "owns" the
482 transition. the char the transition is for is
483 determined by offset from the owning states base
484 field. the next field determines which state
485 is to be transitioned to if any.
486*/
487struct _reg_trie_trans {
488 U32 next;
489 U32 check;
490};
491
492/* a transition list element for the list based representation */
493struct _reg_trie_trans_list_elem {
494 U16 forid;
495 U32 newstate;
496};
497typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
498
499/* a state for compressed nodes. base is an offset
500 into an array of reg_trie_trans array. If wordnum is
501 nonzero the state is accepting. if base is zero then
502 the state has no children (and will be accepting)
503*/
504struct _reg_trie_state {
505 U16 wordnum;
506 union {
507 U32 base;
508 reg_trie_trans_le* list;
509 } trans;
510};
511
512
513
a3621e74
YO
514typedef struct _reg_trie_state reg_trie_state;
515typedef struct _reg_trie_trans reg_trie_trans;
516
517
518/* anything in here that needs to be freed later
23eab42c
NC
519 should be dealt with in pregfree.
520 refcount is first in both this and _reg_ac_data to allow a space
521 optimisation in Perl_regdupe. */
a3621e74 522struct _reg_trie_data {
23eab42c 523 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
524 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
525 U32 lasttrans; /* last valid transition element */
526 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
527 reg_trie_state *states; /* state data */
528 reg_trie_trans *trans; /* array of transition elements */
529 char *bitmap; /* stclass bitmap */
786e8c11
YO
530 U32 startstate; /* initial state - used for common prefix optimisation */
531 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
532 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
533 U32 *wordlen; /* array of lengths of words */
534 U16 *jump; /* optional 1 indexed array of offsets before tail
535 for the node following a given word. */
536 U16 *nextword; /* optional 1 indexed array to support linked list
537 of duplicate wordnums */
1e2e3d02
YO
538 U32 statecount; /* Build only - number of states in the states array
539 (including the unused zero state) */
786e8c11 540 U32 wordcount; /* Build only */
a3621e74 541#ifdef DEBUGGING
786e8c11 542 STRLEN charcount; /* Build only */
a3621e74
YO
543#endif
544};
2b8b4781
NC
545/* There is one (3 under DEBUGGING) pointers that logically belong in this
546 structure, but are held outside as they need duplication on thread cloning,
547 whereas the rest of the structure can be read only:
548 HV *widecharmap; code points > 255 to charid
549#ifdef DEBUGGING
550 AV *words; Array of words contained in trie, for dumping
551 AV *revcharmap; Map of each charid back to its character representation
552#endif
553*/
554
555#define TRIE_WORDS_OFFSET 2
556
a3621e74
YO
557typedef struct _reg_trie_data reg_trie_data;
558
23eab42c
NC
559/* refcount is first in both this and _reg_trie_data to allow a space
560 optimisation in Perl_regdupe. */
07be1b83 561struct _reg_ac_data {
23eab42c 562 U32 refcount;
07be1b83
YO
563 U32 *fail;
564 reg_trie_state *states;
3251b653 565 U32 trie;
07be1b83
YO
566};
567typedef struct _reg_ac_data reg_ac_data;
568
3dab1dad
YO
569/* ANY_BIT doesnt use the structure, so we can borrow it here.
570 This is simpler than refactoring all of it as wed end up with
571 three different sets... */
572
573#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 574#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
575#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
576#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
577#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 578
1de06328
YO
579#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
580#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
581
582
786e8c11
YO
583#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
584#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 585
a3621e74
YO
586/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
587#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
588#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
589#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
590
3dab1dad 591#ifdef DEBUGGING
3dab1dad 592#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 593#else
3dab1dad 594#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 595#endif
a3621e74 596
0111c4fd
RGS
597#define RE_TRIE_MAXBUF_INIT 65536
598#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
599#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
600
be8e71aa
YO
601/*
602
603RE_DEBUG_FLAGS is used to control what debug output is emitted
604its divided into three groups of options, some of which interact.
605The three groups are: Compile, Execute, Extra. There is room for a
606further group, as currently only the low three bytes are used.
607
608 Compile Options:
07be1b83 609
be8e71aa
YO
610 PARSE
611 PEEP
612 TRIE
613 PROGRAM
614 OFFSETS
a3621e74 615
be8e71aa 616 Execute Options:
a3621e74 617
be8e71aa
YO
618 INTUIT
619 MATCH
620 TRIE
a3621e74 621
be8e71aa 622 Extra Options
a3621e74 623
be8e71aa
YO
624 TRIE
625 OFFSETS
626
627If you modify any of these make sure you make corresponding changes to
628re.pm, especially to the documentation.
a3621e74 629
be8e71aa
YO
630*/
631
632
633/* Compile */
634#define RE_DEBUG_COMPILE_MASK 0x0000FF
635#define RE_DEBUG_COMPILE_PARSE 0x000001
636#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
637#define RE_DEBUG_COMPILE_TRIE 0x000004
638#define RE_DEBUG_COMPILE_DUMP 0x000008
be8e71aa
YO
639
640/* Execute */
641#define RE_DEBUG_EXECUTE_MASK 0x00FF00
642#define RE_DEBUG_EXECUTE_INTUIT 0x000100
643#define RE_DEBUG_EXECUTE_MATCH 0x000200
644#define RE_DEBUG_EXECUTE_TRIE 0x000400
645
646/* Extra */
647#define RE_DEBUG_EXTRA_MASK 0xFF0000
648#define RE_DEBUG_EXTRA_TRIE 0x010000
649#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
650#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
651#define RE_DEBUG_EXTRA_STATE 0x080000
652#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
24b23f37
YO
653/* combined */
654#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 655
e68ec53f 656#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
657/* Compile */
658#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 659 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 660#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 661 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 662#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 663 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 664#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 665 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 666#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 667 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 668#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 669 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
670
671/* Execute */
672#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 673 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 674#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 675 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 676#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 677 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 678#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 679 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
680
681/* Extra */
682#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 683 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
684#define DEBUG_OFFSETS_r(x) DEBUG_r( \
685 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
686#define DEBUG_STATE_r(x) DEBUG_r( \
687 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
688#define DEBUG_STACK_r(x) DEBUG_r( \
689 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
a5ca303d
YO
690#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
691 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
692 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 693#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 694 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
695 Perl_warn_nocontext x )
696#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 697 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 698#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 699 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
700
701#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 702 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
703 | RE_DEBUG_EXECUTE_TRIE )) x )
704
705/* initialization */
706/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
707#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
708 SV * re_debug_flags_sv = NULL; \
709 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
710 if (re_debug_flags_sv) { \
8e9a8a48 711 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 712 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 713 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
714 }\
715})
a3621e74
YO
716
717#ifdef DEBUGGING
ab3bbdeb 718
8e9a8a48 719#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
720
721#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
722 const char * const rpv = \
ddc5bc0f 723 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
724 PL_colors[(c1)],PL_colors[(c2)], \
725 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 726 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
727
728#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
729 const char * const rpv = \
ddc5bc0f 730 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb
YO
731 PL_colors[(c1)],PL_colors[(c2)], \
732 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
733
734#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
735 const char * const rpv = \
ddc5bc0f 736 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
737 PL_colors[0], PL_colors[1], \
738 ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \
739 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
740 )
741
742#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
743#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
744
745#else /* if not DEBUGGING */
746
a3621e74 747#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
748#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
749#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
750#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
751#define RE_SV_DUMPLEN(ItEm)
752#define RE_SV_TAIL(ItEm)
753
754#endif /* DEBUG RELATED DEFINES */
a3621e74 755
99ed40ff 756