This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Typo fix, by Robin Barker
[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
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 {
1f1031fe 106 int name_list_idx; /* Optional data index of an array of paren names */
6d5c990f
RGS
107 U32 *offsets; /* offset annotations 20001228 MJD
108 data about mapping the program to the
109 string*/
1f1031fe 110 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
6d5c990f
RGS
111 regnode *regstclass; /* Optional startclass as identified or constructed
112 by the optimiser */
113 struct reg_data *data; /* Additional miscellaneous data used by the program.
114 Used to make it easier to clone and free arbitrary
115 data that the regops need. Often the ARG field of
116 a regop is an index into this structure */
117 regnode program[1]; /* Unwarranted chumminess with compiler. */
118} regexp_internal;
119
120#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
121#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
122#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
123/*
124 * Flags stored in regexp->intflags
125 * These are used only internally to the regexp engine
126 *
127 * See regexp.h for flags used externally to the regexp engine
128 */
129#define PREGf_SKIP 0x00000001
130#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
131#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
132#define PREGf_VERBARG_SEEN 0x00000008
133#define PREGf_CUTGROUP_SEEN 0x00000010
134
135
136/* this is where the old regcomp.h started */
a687059c 137
c277df42 138struct regnode_string {
cd439c50 139 U8 str_len;
c277df42
IZ
140 U8 type;
141 U16 next_off;
cd439c50 142 char string[1];
c277df42 143};
a687059c 144
6bda09f9
YO
145/* Argument bearing node - workhorse,
146 arg1 is often for the data field */
c277df42
IZ
147struct regnode_1 {
148 U8 flags;
149 U8 type;
150 U16 next_off;
151 U32 arg1;
152};
153
6bda09f9
YO
154/* Similar to a regnode_1 but with an extra signed argument */
155struct regnode_2L {
156 U8 flags;
157 U8 type;
158 U16 next_off;
159 U32 arg1;
160 I32 arg2;
161};
162
163/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
164struct regnode_2 {
165 U8 flags;
166 U8 type;
167 U16 next_off;
168 U16 arg1;
169 U16 arg2;
170};
171
6bda09f9 172
936ed897 173#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
ffc61ed2 174#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 175
786e8c11 176/* also used by trie */
936ed897
IZ
177struct regnode_charclass {
178 U8 flags;
179 U8 type;
180 U16 next_off;
ffc61ed2 181 U32 arg1;
19103b91 182 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
183};
184
19103b91
JH
185struct regnode_charclass_class { /* has [[:blah:]] classes */
186 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
187 U8 type;
188 U16 next_off;
ffc61ed2 189 U32 arg1;
19103b91
JH
190 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
191 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
192};
193
9c7e81e8
AD
194/* XXX fix this description.
195 Impose a limit of REG_INFTY on various pattern matching operations
196 to limit stack growth and to avoid "infinite" recursions.
197*/
198/* The default size for REG_INFTY is I16_MAX, which is the same as
199 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
200 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
201 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
202 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
203 smaller default.
204 --Andy Dougherty 11 June 1998
205*/
206#if SHORTSIZE > 2
207# ifndef REG_INFTY
208# define REG_INFTY ((1<<15)-1)
209# endif
210#endif
211
212#ifndef REG_INFTY
83cfe48c
IZ
213# define REG_INFTY I16_MAX
214#endif
a687059c 215
e91177ed
IZ
216#define ARG_VALUE(arg) (arg)
217#define ARG__SET(arg,val) ((arg) = (val))
c277df42 218
e8e3b0fb
JH
219#undef ARG
220#undef ARG1
221#undef ARG2
222
c277df42
IZ
223#define ARG(p) ARG_VALUE(ARG_LOC(p))
224#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
225#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 226#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 227
c277df42
IZ
228#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
229#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
230#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 231#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 232
e8e3b0fb
JH
233#undef NEXT_OFF
234#undef NODE_ALIGN
235
075abff3
AL
236#define NEXT_OFF(p) ((p)->next_off)
237#define NODE_ALIGN(node)
238#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 239
c277df42
IZ
240#define SIZE_ALIGN NODE_ALIGN
241
e8e3b0fb
JH
242#undef OP
243#undef OPERAND
244#undef MASK
245#undef STRING
246
e91177ed
IZ
247#define OP(p) ((p)->type)
248#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
249#define MASK(p) ((char*)OPERAND(p))
250#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
251#define STRING(p) (((struct regnode_string *)p)->string)
252#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
253#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
254
e8e3b0fb
JH
255#undef NODE_ALIGN
256#undef ARG_LOC
257#undef NEXTOPER
258#undef PREVOPER
259
e91177ed
IZ
260#define NODE_ALIGN(node)
261#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
262#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
263#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 264#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 265
e91177ed
IZ
266#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
267#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
268
269#define NODE_STEP_B 4
c277df42
IZ
270
271#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
272#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
273
e91177ed 274#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 275 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 276#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 277 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 278
22c35a8c 279#define REG_MAGIC 0234
a687059c 280
830247a4 281#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 282
936ed897 283/* Flags for node->flags of ANYOF */
b8c5462f 284
19103b91 285#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
286#define ANYOF_INVERT 0x04
287#define ANYOF_FOLD 0x02
288#define ANYOF_LOCALE 0x01
b8c5462f 289
653099ff 290/* Used for regstclass only */
ffc61ed2
JH
291#define ANYOF_EOS 0x10 /* Can match an empty string too */
292
293/* There is a character or a range past 0xff */
294#define ANYOF_UNICODE 0x20
1aa99e6b 295#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 296
4f66b38d
HS
297/* size of node is large (includes class pointer) */
298#define ANYOF_LARGE 0x80
299
ffc61ed2
JH
300/* Are there any runtime flags on in this node? */
301#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
302
303#define ANYOF_FLAGS_ALL 0xff
653099ff 304
936ed897 305/* Character classes for node->classflags of ANYOF */
653099ff
GS
306/* Should be synchronized with a table in regprop() */
307/* 2n should pair with 2n+1 */
b8c5462f 308
289d4f09 309#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 310#define ANYOF_NALNUM 1
aaa51d5e 311#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
312#define ANYOF_NSPACE 3
313#define ANYOF_DIGIT 4
314#define ANYOF_NDIGIT 5
289d4f09 315#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
316#define ANYOF_NALNUMC 7
317#define ANYOF_ALPHA 8
318#define ANYOF_NALPHA 9
319#define ANYOF_ASCII 10
320#define ANYOF_NASCII 11
321#define ANYOF_CNTRL 12
322#define ANYOF_NCNTRL 13
323#define ANYOF_GRAPH 14
324#define ANYOF_NGRAPH 15
325#define ANYOF_LOWER 16
326#define ANYOF_NLOWER 17
327#define ANYOF_PRINT 18
328#define ANYOF_NPRINT 19
329#define ANYOF_PUNCT 20
330#define ANYOF_NPUNCT 21
331#define ANYOF_UPPER 22
332#define ANYOF_NUPPER 23
333#define ANYOF_XDIGIT 24
334#define ANYOF_NXDIGIT 25
aaa51d5e
JF
335#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
336#define ANYOF_NPSXSPC 27
ffc61ed2 337#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 338#define ANYOF_NBLANK 29
b8c5462f 339
aaa51d5e 340#define ANYOF_MAX 32
b8c5462f
JH
341
342/* Backward source code compatibility. */
343
344#define ANYOF_ALNUML ANYOF_ALNUM
345#define ANYOF_NALNUML ANYOF_NALNUM
346#define ANYOF_SPACEL ANYOF_SPACE
347#define ANYOF_NSPACEL ANYOF_NSPACE
348
349/* Utility macros for the bitmap and classes of ANYOF */
350
936ed897
IZ
351#define ANYOF_SIZE (sizeof(struct regnode_charclass))
352#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 353
936ed897 354#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
355
356#define ANYOF_BIT(c) (1 << ((c) & 7))
357
936ed897 358#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
359#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
360#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
361#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
362
936ed897
IZ
363#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
364#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
365
366#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 367#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
368#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
369#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
370#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
371
f8bef550
NC
372#define ANYOF_BITMAP_SETALL(p) \
373 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
374#define ANYOF_BITMAP_CLEARALL(p) \
375 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
376/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
377#define ANYOF_BITMAP_TESTALLSET(p) \
378 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)
379
936ed897
IZ
380#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
381#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
382#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 383
786e8c11 384
a687059c
LW
385/*
386 * Utility definitions.
387 */
2304df62 388#ifndef CHARMASK
075abff3 389# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 390#else
075abff3 391# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 392#endif
a687059c 393
c277df42
IZ
394#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
395
3dab1dad
YO
396#define REG_SEEN_ZERO_LEN 0x00000001
397#define REG_SEEN_LOOKBEHIND 0x00000002
398#define REG_SEEN_GPOS 0x00000004
399#define REG_SEEN_EVAL 0x00000008
400#define REG_SEEN_CANY 0x00000010
cce850e4 401#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 402#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 403#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 404#define REG_SEEN_VERBARG 0x00000080
5d458dd8 405#define REG_SEEN_CUTGROUP 0x00000100
d09b2d29 406
73c4f7a1
GS
407START_EXTERN_C
408
be8e71aa
YO
409#ifdef PLUGGABLE_RE_EXTENSION
410#include "re_nodes.h"
411#else
d09b2d29 412#include "regnodes.h"
be8e71aa 413#endif
d09b2d29 414
4d61ec05 415/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 416#ifndef DOINIT
4d61ec05 417EXTCONST U8 PL_varies[];
d09b2d29 418#else
4d61ec05 419EXTCONST U8 PL_varies[] = {
0111c4fd 420 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
99ed40ff
YO
421 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
422 NREF, NREFF, NREFFL,
423 0
c277df42 424};
d09b2d29 425#endif
c277df42 426
4d61ec05
GS
427/* The following always have a length of 1. U8 we can do strchr() on it. */
428/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 429#ifndef DOINIT
4d61ec05 430EXTCONST U8 PL_simple[];
d09b2d29 431#else
4d61ec05 432EXTCONST U8 PL_simple[] = {
f33976b4 433 REG_ANY, SANY, CANY,
ffc61ed2
JH
434 ANYOF,
435 ALNUM, ALNUML,
436 NALNUM, NALNUML,
437 SPACE, SPACEL,
438 NSPACE, NSPACEL,
439 DIGIT, NDIGIT,
440 0
c277df42
IZ
441};
442#endif
443
f9f4320a
YO
444#ifndef PLUGGABLE_RE_EXTENSION
445#ifndef DOINIT
446EXTCONST regexp_engine PL_core_reg_engine;
447#else /* DOINIT */
448EXTCONST regexp_engine PL_core_reg_engine = {
3c6f95d1 449 Perl_re_compile,
f9f4320a
YO
450 Perl_regexec_flags,
451 Perl_re_intuit_start,
452 Perl_re_intuit_string,
f8149455 453 Perl_regfree_internal,
f9f4320a 454#if defined(USE_ITHREADS)
f8149455 455 Perl_regdupe_internal
f9f4320a
YO
456#endif
457};
458#endif /* DOINIT */
459#endif /* PLUGGABLE_RE_EXTENSION */
460
461
73c4f7a1 462END_EXTERN_C
cad2e5aa 463
cad2e5aa 464
261faec3
MJD
465/* .what is a character array with one character for each member of .data
466 * The character describes the function of the corresponding .data item:
0111c4fd 467 * f - start-class data for regstclass optimization
261faec3
MJD
468 * n - Root of op tree for (?{EVAL}) item
469 * o - Start op for (?{EVAL}) item
3dab1dad 470 * p - Pad for (?{EVAL}) item
9e55ce06
JH
471 * s - swash for unicode-style character class, and the multicharacter
472 * strings resulting from casefolding the single-character entries
473 * in the character class
a3621e74 474 * t - trie struct
55eed653 475 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 476 * also used for revcharmap and words under DEBUGGING
07be1b83 477 * T - aho-trie struct
81714fb9 478 * S - sv for named capture lookup
261faec3 479 * 20010712 mjd@plover.com
c3613ed7 480 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 481 */
cad2e5aa
JH
482struct reg_data {
483 U32 count;
484 U8 *what;
485 void* data[1];
486};
487
a1cac82e
NC
488/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
489 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 490#define anchored_substr substrs->data[0].substr
33b8afdf 491#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 492#define anchored_offset substrs->data[0].min_offset
1de06328
YO
493#define anchored_end_shift substrs->data[0].end_shift
494
cad2e5aa 495#define float_substr substrs->data[1].substr
33b8afdf 496#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
497#define float_min_offset substrs->data[1].min_offset
498#define float_max_offset substrs->data[1].max_offset
1de06328
YO
499#define float_end_shift substrs->data[1].end_shift
500
cad2e5aa 501#define check_substr substrs->data[2].substr
33b8afdf 502#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
503#define check_offset_min substrs->data[2].min_offset
504#define check_offset_max substrs->data[2].max_offset
1de06328 505#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
506
507
508
509/* trie related stuff */
5d9a96ca 510
a3621e74
YO
511/* a transition record for the state machine. the
512 check field determines which state "owns" the
513 transition. the char the transition is for is
514 determined by offset from the owning states base
515 field. the next field determines which state
516 is to be transitioned to if any.
517*/
518struct _reg_trie_trans {
519 U32 next;
520 U32 check;
521};
522
523/* a transition list element for the list based representation */
524struct _reg_trie_trans_list_elem {
525 U16 forid;
526 U32 newstate;
527};
528typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
529
530/* a state for compressed nodes. base is an offset
531 into an array of reg_trie_trans array. If wordnum is
532 nonzero the state is accepting. if base is zero then
533 the state has no children (and will be accepting)
534*/
535struct _reg_trie_state {
536 U16 wordnum;
537 union {
538 U32 base;
539 reg_trie_trans_le* list;
540 } trans;
541};
542
543
544
a3621e74
YO
545typedef struct _reg_trie_state reg_trie_state;
546typedef struct _reg_trie_trans reg_trie_trans;
547
548
549/* anything in here that needs to be freed later
23eab42c
NC
550 should be dealt with in pregfree.
551 refcount is first in both this and _reg_ac_data to allow a space
552 optimisation in Perl_regdupe. */
a3621e74 553struct _reg_trie_data {
23eab42c 554 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
555 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
556 U32 lasttrans; /* last valid transition element */
557 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
558 reg_trie_state *states; /* state data */
559 reg_trie_trans *trans; /* array of transition elements */
560 char *bitmap; /* stclass bitmap */
786e8c11
YO
561 U32 startstate; /* initial state - used for common prefix optimisation */
562 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
563 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
564 U32 *wordlen; /* array of lengths of words */
565 U16 *jump; /* optional 1 indexed array of offsets before tail
566 for the node following a given word. */
567 U16 *nextword; /* optional 1 indexed array to support linked list
568 of duplicate wordnums */
1e2e3d02
YO
569 U32 statecount; /* Build only - number of states in the states array
570 (including the unused zero state) */
786e8c11 571 U32 wordcount; /* Build only */
a3621e74 572#ifdef DEBUGGING
786e8c11 573 STRLEN charcount; /* Build only */
a3621e74
YO
574#endif
575};
2b8b4781
NC
576/* There is one (3 under DEBUGGING) pointers that logically belong in this
577 structure, but are held outside as they need duplication on thread cloning,
578 whereas the rest of the structure can be read only:
579 HV *widecharmap; code points > 255 to charid
580#ifdef DEBUGGING
581 AV *words; Array of words contained in trie, for dumping
582 AV *revcharmap; Map of each charid back to its character representation
583#endif
584*/
585
586#define TRIE_WORDS_OFFSET 2
587
a3621e74
YO
588typedef struct _reg_trie_data reg_trie_data;
589
23eab42c
NC
590/* refcount is first in both this and _reg_trie_data to allow a space
591 optimisation in Perl_regdupe. */
07be1b83 592struct _reg_ac_data {
23eab42c 593 U32 refcount;
07be1b83
YO
594 U32 *fail;
595 reg_trie_state *states;
3251b653 596 U32 trie;
07be1b83
YO
597};
598typedef struct _reg_ac_data reg_ac_data;
599
3dab1dad
YO
600/* ANY_BIT doesnt use the structure, so we can borrow it here.
601 This is simpler than refactoring all of it as wed end up with
602 three different sets... */
603
604#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 605#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
606#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
607#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
608#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 609
1de06328
YO
610#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
611#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
612
613
786e8c11
YO
614#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
615#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 616
a3621e74
YO
617/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
618#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
619#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
620#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
621
3dab1dad 622#ifdef DEBUGGING
3dab1dad 623#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 624#else
3dab1dad 625#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 626#endif
a3621e74 627
0111c4fd
RGS
628#define RE_TRIE_MAXBUF_INIT 65536
629#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
630#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
631
be8e71aa
YO
632/*
633
634RE_DEBUG_FLAGS is used to control what debug output is emitted
635its divided into three groups of options, some of which interact.
636The three groups are: Compile, Execute, Extra. There is room for a
637further group, as currently only the low three bytes are used.
638
639 Compile Options:
07be1b83 640
be8e71aa
YO
641 PARSE
642 PEEP
643 TRIE
644 PROGRAM
645 OFFSETS
a3621e74 646
be8e71aa 647 Execute Options:
a3621e74 648
be8e71aa
YO
649 INTUIT
650 MATCH
651 TRIE
a3621e74 652
be8e71aa 653 Extra Options
a3621e74 654
be8e71aa
YO
655 TRIE
656 OFFSETS
657
658If you modify any of these make sure you make corresponding changes to
659re.pm, especially to the documentation.
a3621e74 660
be8e71aa
YO
661*/
662
663
664/* Compile */
665#define RE_DEBUG_COMPILE_MASK 0x0000FF
666#define RE_DEBUG_COMPILE_PARSE 0x000001
667#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
668#define RE_DEBUG_COMPILE_TRIE 0x000004
669#define RE_DEBUG_COMPILE_DUMP 0x000008
be8e71aa
YO
670
671/* Execute */
672#define RE_DEBUG_EXECUTE_MASK 0x00FF00
673#define RE_DEBUG_EXECUTE_INTUIT 0x000100
674#define RE_DEBUG_EXECUTE_MATCH 0x000200
675#define RE_DEBUG_EXECUTE_TRIE 0x000400
676
677/* Extra */
678#define RE_DEBUG_EXTRA_MASK 0xFF0000
679#define RE_DEBUG_EXTRA_TRIE 0x010000
680#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
681#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
682#define RE_DEBUG_EXTRA_STATE 0x080000
683#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
24b23f37
YO
684/* combined */
685#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 686
e68ec53f 687#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
688/* Compile */
689#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 690 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 691#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 692 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 693#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 694 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 695#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 696 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 697#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 698 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 699#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 700 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
701
702/* Execute */
703#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 704 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 705#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 706 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 707#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 708 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 709#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 710 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
711
712/* Extra */
713#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 714 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
715#define DEBUG_OFFSETS_r(x) DEBUG_r( \
716 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
717#define DEBUG_STATE_r(x) DEBUG_r( \
718 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
719#define DEBUG_STACK_r(x) DEBUG_r( \
720 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
a5ca303d
YO
721#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
722 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
723 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 724#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 725 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
726 Perl_warn_nocontext x )
727#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 728 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 729#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 730 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
731
732#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 733 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
734 | RE_DEBUG_EXECUTE_TRIE )) x )
735
736/* initialization */
737/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
738#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
739 SV * re_debug_flags_sv = NULL; \
740 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
741 if (re_debug_flags_sv) { \
8e9a8a48 742 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 743 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 744 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
745 }\
746})
a3621e74
YO
747
748#ifdef DEBUGGING
ab3bbdeb 749
8e9a8a48 750#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
751
752#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
753 const char * const rpv = \
ddc5bc0f 754 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
755 PL_colors[(c1)],PL_colors[(c2)], \
756 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 757 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
758
759#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
760 const char * const rpv = \
ddc5bc0f 761 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb
YO
762 PL_colors[(c1)],PL_colors[(c2)], \
763 ((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
764
765#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
766 const char * const rpv = \
ddc5bc0f 767 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb
YO
768 PL_colors[0], PL_colors[1], \
769 ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \
770 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
771 )
772
773#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
774#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
775
776#else /* if not DEBUGGING */
777
a3621e74 778#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
779#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
780#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
781#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
782#define RE_SV_DUMPLEN(ItEm)
783#define RE_SV_TAIL(ItEm)
784
785#endif /* DEBUG RELATED DEFINES */
a3621e74 786
99ed40ff 787