This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add ANYOFD regex node
[perl5.git] / regcomp.h
CommitLineData
a0d0e21e 1/* regcomp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
663f364b 4 * 2000, 2001, 2002, 2003, 2005, 2006, 2007, by Larry Wall and others
d6376244
JH
5 *
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
8 *
a687059c 9 */
e1d1eefb 10#include "regcharclass.h"
a687059c 11
58e23c8d 12/* Convert branch sequences to more efficient trie ops? */
07be1b83 13#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d 14
486ec47a 15/* Be really aggressive about optimising patterns with trie sequences? */
07be1b83 16#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d
YO
17
18/* Should the optimiser take positive assertions into account? */
68ba3a3f 19#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
58e23c8d
YO
20
21/* Not for production use: */
07be1b83 22#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 23
7122b237
YO
24/* Activate offsets code - set to if 1 to enable */
25#ifdef DEBUGGING
26#define RE_TRACK_PATTERN_OFFSETS
27#endif
28
a687059c
LW
29/*
30 * The "internal use only" fields in regexp.h are present to pass info from
31 * compile to execute that permits the execute phase to run lots faster on
32 * simple cases. They are:
33 *
bd61b366 34 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
35 * reganch is the match anchored (at beginning-of-line only)?
36 * regmust string (pointer into program) that match must include, or NULL
79072805 37 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
38 * regmlen length of regmust string
39 * [regmlen not used currently]
40 *
41 * Regstart and reganch permit very fast decisions on suitable starting points
42 * for a match, cutting down the work a lot. Regmust permits fast rejection
43 * of lines that cannot possibly match. The regmust tests are costly enough
e50aee73 44 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
45 * potentially expensive (at present, the only such thing detected is * or +
46 * at the start of the r.e., which can involve a lot of backup). Regmlen is
e50aee73 47 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
48 * it anyway.
49 * [regmust is now supplied always. The tests that use regmust have a
50 * heuristic that disables the test if it usually matches.]
51 *
52 * [In fact, we now use regmust in many cases to locate where the search
53 * starts in the string, so if regback is >= 0, the regmust search is never
54 * wasted effort. The regback variable says how many characters back from
55 * where regmust matched is the earliest possible start of the match.
56 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
57 */
58
59/*
60 * Structure for regexp "program". This is essentially a linear encoding
61 * of a nondeterministic finite-state machine (aka syntax charts or
62 * "railroad normal form" in parsing technology). Each node is an opcode
63 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
64 * all nodes except BRANCH implement concatenation; a "next" pointer with
65 * a BRANCH on both ends of it is connecting two alternatives. (Here we
66 * have one of the subtle syntax dependencies: an individual BRANCH (as
67 * opposed to a collection of them) is never concatenated with anything
68 * because of operator precedence.) The operand of some types of node is
69 * a literal string; for others, it is a node leading into a sub-FSM. In
70 * particular, the operand of a BRANCH node is the first node of the branch.
71 * (NB this is *not* a tree structure: the tail of the branch connects
a3621e74
YO
72 * to the thing following the set of BRANCHes.) The opcodes are defined
73 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
a687059c
LW
74 */
75
a687059c
LW
76/*
77 * A node is one char of opcode followed by two chars of "next" pointer.
78 * "Next" pointers are stored as two 8-bit pieces, high order first. The
79 * value is a positive offset from the opcode of the node containing it.
80 * An operand, if any, simply follows the node. (Note that much of the
81 * code generation knows about this implicit relationship.)
82 *
83 * Using two bytes for the "next" pointer is vast overkill for most things,
84 * but allows patterns to get big without disasters.
85 *
e91177ed 86 * [The "next" pointer is always aligned on an even
62e6ef33 87 * boundary, and reads the offset directly as a short.]
a687059c 88 */
785a26d5
YO
89
90/* This is the stuff that used to live in regexp.h that was truly
91 private to the engine itself. It now lives here. */
92
28d8d7f4 93
785a26d5
YO
94
95 typedef struct regexp_internal {
1f1031fe 96 int name_list_idx; /* Optional data index of an array of paren names */
7122b237
YO
97 union {
98 U32 *offsets; /* offset annotations 20001228 MJD
99 data about mapping the program to the
100 string -
101 offsets[0] is proglen when this is used
102 */
103 U32 proglen;
104 } u;
105
6d5c990f
RGS
106 regnode *regstclass; /* Optional startclass as identified or constructed
107 by the optimiser */
108 struct reg_data *data; /* Additional miscellaneous data used by the program.
109 Used to make it easier to clone and free arbitrary
110 data that the regops need. Often the ARG field of
111 a regop is an index into this structure */
3d2bd50a
DM
112 struct reg_code_block *code_blocks;/* positions of literal (?{}) */
113 int num_code_blocks; /* size of code_blocks[] */
6d5c990f
RGS
114 regnode program[1]; /* Unwarranted chumminess with compiler. */
115} regexp_internal;
116
117#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
118#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
119#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
120/*
121 * Flags stored in regexp->intflags
122 * These are used only internally to the regexp engine
123 *
124 * See regexp.h for flags used externally to the regexp engine
125 */
e3e400ec
YO
126#define RXp_INTFLAGS(rx) ((rx)->intflags)
127#define RX_INTFLAGS(prog) RXp_INTFLAGS(ReANY(prog))
128
785a26d5
YO
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
732caac7 134#define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
e3e400ec
YO
135/* these used to be extflags, but are now intflags */
136#define PREGf_NOSCAN 0x00000040
33c28ab2 137 /* spare */
8e1490ee
YO
138#define PREGf_GPOS_SEEN 0x00000100
139#define PREGf_GPOS_FLOAT 0x00000200
785a26d5 140
d3d47aac
YO
141#define PREGf_ANCH_MBOL 0x00000400
142#define PREGf_ANCH_SBOL 0x00000800
143#define PREGf_ANCH_GPOS 0x00001000
8e1490ee 144
d3d47aac
YO
145#define PREGf_ANCH \
146 ( PREGf_ANCH_SBOL | PREGf_ANCH_GPOS | PREGf_ANCH_MBOL )
785a26d5
YO
147
148/* this is where the old regcomp.h started */
a687059c 149
c277df42 150struct regnode_string {
cd439c50 151 U8 str_len;
c277df42
IZ
152 U8 type;
153 U16 next_off;
cd439c50 154 char string[1];
c277df42 155};
a687059c 156
6bda09f9
YO
157/* Argument bearing node - workhorse,
158 arg1 is often for the data field */
c277df42
IZ
159struct regnode_1 {
160 U8 flags;
161 U8 type;
162 U16 next_off;
163 U32 arg1;
164};
165
6bda09f9
YO
166/* Similar to a regnode_1 but with an extra signed argument */
167struct regnode_2L {
168 U8 flags;
169 U8 type;
170 U16 next_off;
171 U32 arg1;
172 I32 arg2;
173};
174
175/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
176struct regnode_2 {
177 U8 flags;
178 U8 type;
179 U16 next_off;
180 U16 arg1;
181 U16 arg2;
182};
183
e0a1ff7a
KW
184/* This give the number of code points that can be in the bitmap of an ANYOF
185 * node. The shift number must currently be one of: 8..12. It can't be less
186 * than 8 (256) because some code relies on it being at least that. Above 12
187 * (4096), and you start running into warnings that some data structure widths
188 * have been exceeded, though the test suite as of this writing still passes
189 * for up through 16, which is as high as anyone would ever want to go,
190 * encompassing all of the Unicode BMP, and thus including all the economically
191 * important world scripts. At 12 most of them are: including Arabic,
192 * Cyrillic, Greek, Hebrew, Indian subcontinent, Latin, and Thai; but not Han,
193 * Japanese, nor Korean. (The regarglen structure in regnodes.h is a U8, and
194 * the trie types TRIEC and AHOCORASICKC are larger than U8 for shift values
195 * below above 12.) Be sure to benchmark before changing, as larger sizes do
196 * significantly slow down the test suite */
197#define NUM_ANYOF_CODE_POINTS (1 << 8)
6bda09f9 198
dcb20b36 199#define ANYOF_BITMAP_SIZE (NUM_ANYOF_CODE_POINTS / 8) /* 8 bits/Byte */
936ed897 200
1ee208c4
KW
201/* Note that these form structs which are supersets of the next smaller one, by
202 * appending fields. Alignment problems can occur if one of those optional
203 * fields requires stricter alignment than the base struct. And formal
204 * parameters that can really be two or more of the structs should be
205 * declared as the smallest one it could be. See commit message for
206 * 7dcac5f6a5195002b55c935ee1d67f67e1df280b. Regnode allocation is done
207 * without regard to alignment, and changing it to would also require changing
208 * the code that inserts and deletes regnodes. The basic single-argument
209 * regnode has a U32, which is what reganode() allocates as a unit. Therefore
210 * no field can require stricter alignment than U32. */
211
786e8c11 212/* also used by trie */
936ed897
IZ
213struct regnode_charclass {
214 U8 flags;
215 U8 type;
216 U16 next_off;
9da99c5e 217 U32 arg1; /* set by set_ANYOF_arg() */
19103b91 218 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
219};
220
bf3f2d85
KW
221/* has runtime (locale) \d, \w, ..., [:posix:] classes */
222struct regnode_charclass_class {
93e92956 223 U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */
936ed897
IZ
224 U8 type;
225 U16 next_off;
7e817188 226 U32 arg1;
03fa83ba 227 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */
9cf78fa0 228 U32 classflags; /* and run-time */
936ed897
IZ
229};
230
85c8e306
KW
231/* A synthetic start class (SSC); is a regnode_charclass_posixl_fold, plus an
232 * extra SV*, used only during its construction and which is not used by
233 * regexec.c. Note that the 'next_off' field is unused, as the SSC stands
234 * alone, so there is never a next node. Also, there is no alignment issue,
b8fda935 235 * because these are declared or allocated as a complete unit so the compiler
85c8e306
KW
236 * takes care of alignment. This is unlike the other regnodes which are
237 * allocated in terms of multiples of a single-argument regnode. SSC nodes can
238 * have a pointer field because there is no alignment issue, and because it is
239 * set to NULL after construction, before any cloning of the pattern */
b8f7bb16 240struct regnode_ssc {
93e92956 241 U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */
b8f7bb16
KW
242 U8 type;
243 U16 next_off;
7e817188 244 U32 arg1;
03fa83ba 245 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */
631e9161 246 U32 classflags; /* ... and run-time */
85c8e306
KW
247
248 /* Auxiliary, only used during construction; NULL afterwards: list of code
249 * points matched */
250 SV* invlist;
b8f7bb16
KW
251};
252
71068078
KW
253/* We take advantage of 'next_off' not otherwise being used in the SSC by
254 * actually using it: by setting it to 1. This allows us to test and
255 * distinguish between an SSC and other ANYOF node types, as 'next_off' cannot
256 * otherwise be 1, because it is the offset to the next regnode expressed in
257 * units of regnodes. Since an ANYOF node contains extra fields, it adds up
258 * to 12 regnode units on 32-bit systems, (hence the minimum this can be (if
259 * not 0) is 11 there. Even if things get tightly packed on a 64-bit system,
260 * it still would be more than 1. */
261#define set_ANYOF_SYNTHETIC(n) STMT_START{ OP(n) = ANYOF; \
262 NEXT_OFF(n) = 1; \
263 } STMT_END
264#define is_ANYOF_SYNTHETIC(n) (OP(n) == ANYOF && NEXT_OFF(n) == 1)
265
9c7e81e8
AD
266/* XXX fix this description.
267 Impose a limit of REG_INFTY on various pattern matching operations
268 to limit stack growth and to avoid "infinite" recursions.
269*/
270/* The default size for REG_INFTY is I16_MAX, which is the same as
271 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
272 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
273 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
274 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
275 smaller default.
276 --Andy Dougherty 11 June 1998
277*/
278#if SHORTSIZE > 2
279# ifndef REG_INFTY
280# define REG_INFTY ((1<<15)-1)
281# endif
282#endif
283
284#ifndef REG_INFTY
83cfe48c
IZ
285# define REG_INFTY I16_MAX
286#endif
a687059c 287
e91177ed
IZ
288#define ARG_VALUE(arg) (arg)
289#define ARG__SET(arg,val) ((arg) = (val))
c277df42 290
e8e3b0fb
JH
291#undef ARG
292#undef ARG1
293#undef ARG2
294
c277df42
IZ
295#define ARG(p) ARG_VALUE(ARG_LOC(p))
296#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
297#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 298#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 299
c277df42
IZ
300#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
301#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
302#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 303#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 304
e8e3b0fb
JH
305#undef NEXT_OFF
306#undef NODE_ALIGN
307
075abff3
AL
308#define NEXT_OFF(p) ((p)->next_off)
309#define NODE_ALIGN(node)
aa48e906
YO
310/* the following define was set to 0xde in 075abff3
311 * as part of some linting logic. I have set it to 0
312 * as otherwise in every place where we /might/ set flags
313 * we have to set it 0 explicitly, which duplicates
314 * assignments and IMO adds an unacceptable level of
315 * surprise to working in the regex engine. If this
316 * is changed from 0 then at the very least make sure
317 * that SBOL for /^/ sets the flags to 0 explicitly.
318 * -- Yves */
319#define NODE_ALIGN_FILL(node) ((node)->flags = 0)
a687059c 320
c277df42
IZ
321#define SIZE_ALIGN NODE_ALIGN
322
e8e3b0fb
JH
323#undef OP
324#undef OPERAND
325#undef MASK
326#undef STRING
327
e91177ed 328#define OP(p) ((p)->type)
b988e673
KW
329#define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
330 regnode types. For some, it's the \
331 character set of the regnode */
e91177ed 332#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
333#define MASK(p) ((char*)OPERAND(p))
334#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
335#define STRING(p) (((struct regnode_string *)p)->string)
336#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
337#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
338
e8e3b0fb
JH
339#undef NODE_ALIGN
340#undef ARG_LOC
341#undef NEXTOPER
342#undef PREVOPER
343
e91177ed
IZ
344#define NODE_ALIGN(node)
345#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
346#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
347#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 348#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 349
e91177ed
IZ
350#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
351#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
352
353#define NODE_STEP_B 4
c277df42
IZ
354
355#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
356#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
357
e91177ed 358#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 359 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 360#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 361 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
9990d5ab
KW
362#define FILL_ADVANCE_NODE_2L_ARG(ptr, op, arg1, arg2) \
363 STMT_START { \
364 ARG_SET(ptr, arg1); \
365 ARG2L_SET(ptr, arg2); \
366 FILL_ADVANCE_NODE(ptr, op); \
367 (ptr) += 2; \
368 } STMT_END
a687059c 369
22c35a8c 370#define REG_MAGIC 0234
a687059c 371
9a81a976 372#define SIZE_ONLY (RExC_emit == (regnode *) & RExC_emit_dummy)
6d15ebf3
KW
373#define PASS1 SIZE_ONLY
374#define PASS2 (! SIZE_ONLY)
c277df42 375
6f16c8da 376/* If the bitmap fully represents what this ANYOF node can match, the
137165a6
KW
377 * ARG is set to this special value (since 0, 1, ... are legal, but will never
378 * reach this high). */
93e92956 379#define ANYOF_ONLY_HAS_BITMAP ((U32) -1)
137165a6 380
3b04b210 381/* Flags for node->flags of ANYOF. These are in short supply, with none
93e92956 382 * currently available. The ABOVE_BITMAP_ALL bit could be freed up
7bc66b18
KW
383 * by resorting to creating a swash containing everything above 255. This
384 * introduces a performance penalty. An option that wouldn't slow things down
8f536e26 385 * would be to split one of the two LOC flags out into a separate
7bc66b18
KW
386 * node, like what was done with ANYOF_NON_UTF8_NON_ASCII_ALL in commit
387 * 34fdef848b1687b91892ba55e9e0c3430e0770f6 (but which was reverted because it
388 * wasn't the best option available at the time), and using a LOC flag is
389 * probably better than that commit anyway. But it could be reinstated if we
390 * need a bit. The LOC flags are only for /l nodes; the reverted commit was
8f536e26 391 * only for /d, so there are no combinatorial issues. The LOC flag to use is
a4525e78
KW
392 * probably the POSIXL one. Now that there is an ANYOFL (locale) node, another
393 * option would be to make all of those include the POSIXL data structure,
394 * which would get rid of needing a separate POSIXL flag. But it would
395 * increase the size of all such nodes, so it's probably not as atractive as
396 * having an ANYOF_POSIXL node type. But if we did do it, note that not all 32
397 * bits of that extra space are used, one bit of that could be set aside for
398 * the LOC_FOLD flag, yielding yet another bit. This would require extra code
399 * for masking, so again not the most attractive solution.
400 *
34fdef84 401 * Several flags are not used in synthetic start class (SSC) nodes, so could be
93e92956
KW
402 * shared should new flags be needed for SSCs, like SSC_MATCHES_EMPTY_STRING
403 * now. */
b8c5462f 404
5dbb0c08 405/* regexec.c is expecting this to be in the low bit */
93e92956 406#define ANYOF_INVERT 0x01
1984f6a0 407
a0dd4231 408/* For the SSC node only, which cannot be inverted, so is shared with that bit.
93e92956
KW
409 * This is used only during regex compilation. */
410#define SSC_MATCHES_EMPTY_STRING ANYOF_INVERT
a0dd4231 411
93e92956
KW
412/* Are there things outside the bitmap that will match only if the target
413 * string is encoded in UTF-8? (This is not set if ANYOF_ABOVE_BITMAP_ALL is
414 * set) */
415#define ANYOF_HAS_UTF8_NONBITMAP_MATCHES 0x02
b63f73da
KW
416
417/* The fold is calculated and stored in the bitmap where possible at compile
418 * time. However under locale, the actual folding varies depending on
419 * what the locale is at the time of execution, so it has to be deferred until
420 * then */
93e92956 421#define ANYOF_LOC_FOLD 0x04
b63f73da 422
d1d040e5 423/* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This
dd58aee1
KW
424 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
425 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
8f536e26 426 * at compile-time */
93e92956 427#define ANYOF_MATCHES_POSIXL 0x08
c8d3cd88 428
ae986089
KW
429/* Should we raise a warning if matching against an above-Unicode code point?
430 * */
93e92956 431#define ANYOF_WARN_SUPER 0x10
f4244008 432
ef87b810 433/* Can match something outside the bitmap that isn't in utf8 */
93e92956 434#define ANYOF_HAS_NONBITMAP_NON_UTF8_MATCHES 0x20
ef87b810 435
93e92956
KW
436/* Matches every code point NUM_ANYOF_CODE_POINTS and above*/
437#define ANYOF_MATCHES_ALL_ABOVE_BITMAP 0x40
ffc61ed2 438
7bc66b18
KW
439/* Match all Latin1 characters that aren't ASCII when the target string is not
440 * in utf8. */
93e92956 441#define ANYOF_MATCHES_ALL_NON_UTF8_NON_ASCII 0x80
7bc66b18 442
3b04b210 443#define ANYOF_FLAGS_ALL (0xff)
653099ff 444
93e92956 445#define ANYOF_LOCALE_FLAGS (ANYOF_LOC_FOLD | ANYOF_MATCHES_POSIXL)
ace6b0e4 446
eff8b7dc
KW
447/* These are the flags that apply to both regular ANYOF nodes and synthetic
448 * start class nodes during construction of the SSC. During finalization of
449 * the SSC, other of the flags could be added to it */
93e92956 450#define ANYOF_COMMON_FLAGS (ANYOF_WARN_SUPER|ANYOF_HAS_UTF8_NONBITMAP_MATCHES)
eff8b7dc 451
936ed897 452/* Character classes for node->classflags of ANYOF */
653099ff 453/* Should be synchronized with a table in regprop() */
575a8a29 454/* 2n should be the normal one, paired with its complement at 2n+1 */
b8c5462f 455
08ada29e
KW
456#define ANYOF_ALPHA ((_CC_ALPHA) * 2)
457#define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
e8d596e0
KW
458#define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
459#define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
08ada29e
KW
460#define ANYOF_ASCII ((_CC_ASCII) * 2)
461#define ANYOF_NASCII ((ANYOF_ASCII) + 1)
e8d596e0
KW
462#define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
463#define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
b0d691b2
KW
464#define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
465 [:upper:] under /i */
466#define ANYOF_NCASED ((ANYOF_CASED) + 1)
08ada29e
KW
467#define ANYOF_CNTRL ((_CC_CNTRL) * 2)
468#define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
e8d596e0
KW
469#define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
470#define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
08ada29e
KW
471#define ANYOF_GRAPH ((_CC_GRAPH) * 2)
472#define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
473#define ANYOF_LOWER ((_CC_LOWER) * 2)
474#define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
475#define ANYOF_PRINT ((_CC_PRINT) * 2)
476#define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
477#define ANYOF_PUNCT ((_CC_PUNCT) * 2)
478#define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
e8d596e0
KW
479#define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
480#define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
08ada29e
KW
481#define ANYOF_UPPER ((_CC_UPPER) * 2)
482#define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
e8d596e0
KW
483#define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
484#define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
08ada29e
KW
485#define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
486#define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
b8c5462f 487
87cad4c0 488/* pseudo classes below this, not stored in the class bitmap, but used as flags
e1d1eefb
YO
489 during compilation of char classes */
490
b2ef2e34
KW
491#define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
492#define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
e1d1eefb 493
b2ef2e34
KW
494/* It is best if this is the last one, as all above it are stored as bits in a
495 * bitmap, and it isn't part of that bitmap */
496#if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
497# error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
498#endif
499
8efd3f97
KW
500#define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
501 * '< ANYOF_MAX'
502 * Hence doesn't include VERTWS, as that
503 * is a pseudo class */
504#define ANYOF_MAX ANYOF_POSIXL_MAX
505
506#if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */
b2ef2e34 507# error Problem with handy.h _CC_foo #defines
a0947d7b
KW
508#endif
509
8efd3f97
KW
510#define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
511#define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3)
87cad4c0 512
8efd3f97
KW
513#define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode
514 property: \p{} or \P{} */
f39a4dc2 515
b8c5462f
JH
516/* Backward source code compatibility. */
517
518#define ANYOF_ALNUML ANYOF_ALNUM
519#define ANYOF_NALNUML ANYOF_NALNUM
520#define ANYOF_SPACEL ANYOF_SPACE
521#define ANYOF_NSPACEL ANYOF_NSPACE
8ebfc29a
KW
522#define ANYOF_ALNUM ANYOF_WORDCHAR
523#define ANYOF_NALNUM ANYOF_NWORDCHAR
b8c5462f
JH
524
525/* Utility macros for the bitmap and classes of ANYOF */
526
936ed897 527#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f 528
bc51fd78 529#define ANYOF_BIT(c) (1U << ((c) & 7))
b8c5462f 530
d1d040e5 531#define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c)))
8efd3f97
KW
532#define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c))
533
d1d040e5 534#define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c)))
8efd3f97
KW
535#define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c))
536
d1d040e5 537#define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c)))
8efd3f97 538#define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c))
b8c5462f 539
d1d040e5 540#define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END
8efd3f97 541#define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret)
9cf78fa0
KW
542
543/* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
d1d040e5 544#define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END
8efd3f97 545#define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret)
9cf78fa0 546
8efd3f97 547#define ANYOF_POSIXL_TEST_ANY_SET(p) \
93e92956 548 ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \
d1d040e5 549 && (((regnode_charclass_posixl*)(p))->classflags))
8efd3f97 550#define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p)
e867ce45 551
e0e1be5f
KW
552/* Since an SSC always has this field, we don't have to test for that; nor do
553 * we want to because the bit isn't set for SSC during its construction */
554#define ANYOF_POSIXL_SSC_TEST_ANY_SET(p) \
555 cBOOL(((regnode_ssc*)(p))->classflags)
556#define ANYOF_POSIXL_SSC_TEST_ALL_SET(p) /* Are all bits set? */ \
557 (((regnode_ssc*) (p))->classflags \
558 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
559
560#define ANYOF_POSIXL_TEST_ALL_SET(p) \
93e92956 561 ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \
e0e1be5f
KW
562 && ((regnode_charclass_posixl*) (p))->classflags \
563 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
fb38762f 564
b27fd97a 565#define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END
8efd3f97 566#define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest))
936ed897 567
cdd87c1d
KW
568#define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END
569
9cf78fa0 570#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
936ed897 571#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 572#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
573#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
574#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
bc51fd78 575#define ANYOF_BITMAP_TEST(p, c) cBOOL(ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
b8c5462f 576
f8bef550
NC
577#define ANYOF_BITMAP_SETALL(p) \
578 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
579#define ANYOF_BITMAP_CLEARALL(p) \
580 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
f8bef550 581
888ee784
KW
582#define ANYOF_SKIP EXTRA_SIZE(struct regnode_charclass)
583#define ANYOF_POSIXL_SKIP EXTRA_SIZE(regnode_charclass_posixl)
3a15e693 584
a687059c
LW
585/*
586 * Utility definitions.
587 */
2304df62 588#ifndef CHARMASK
075abff3 589# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 590#else
075abff3 591# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 592#endif
a687059c 593
c277df42
IZ
594#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
595
e384d5c1
YO
596#define REG_ZERO_LEN_SEEN 0x00000001
597#define REG_LOOKBEHIND_SEEN 0x00000002
598#define REG_GPOS_SEEN 0x00000004
471f5387 599/* spare */
e384d5c1
YO
600#define REG_RECURSE_SEEN 0x00000020
601#define REG_TOP_LEVEL_BRANCHES_SEEN 0x00000040
602#define REG_VERBARG_SEEN 0x00000080
603#define REG_CUTGROUP_SEEN 0x00000100
604#define REG_RUN_ON_COMMENT_SEEN 0x00000200
605#define REG_UNFOLDED_MULTI_SEEN 0x00000400
606#define REG_GOSTART_SEEN 0x00000800
ee273784
YO
607#define REG_UNBOUNDED_QUANTIFIER_SEEN 0x00001000
608
d09b2d29 609
73c4f7a1
GS
610START_EXTERN_C
611
be8e71aa
YO
612#ifdef PLUGGABLE_RE_EXTENSION
613#include "re_nodes.h"
614#else
d09b2d29 615#include "regnodes.h"
be8e71aa 616#endif
d09b2d29 617
f9f4320a
YO
618#ifndef PLUGGABLE_RE_EXTENSION
619#ifndef DOINIT
620EXTCONST regexp_engine PL_core_reg_engine;
621#else /* DOINIT */
622EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 623 Perl_re_compile,
2fdbfb4d 624 Perl_regexec_flags,
f9f4320a
YO
625 Perl_re_intuit_start,
626 Perl_re_intuit_string,
2fdbfb4d
AB
627 Perl_regfree_internal,
628 Perl_reg_numbered_buff_fetch,
629 Perl_reg_numbered_buff_store,
630 Perl_reg_numbered_buff_length,
192b9cd1
AB
631 Perl_reg_named_buff,
632 Perl_reg_named_buff_iter,
49d7dfbc 633 Perl_reg_qr_package,
f9f4320a 634#if defined(USE_ITHREADS)
3c13cae6 635 Perl_regdupe_internal,
f9f4320a 636#endif
3c13cae6 637 Perl_re_op_compile
f9f4320a
YO
638};
639#endif /* DOINIT */
640#endif /* PLUGGABLE_RE_EXTENSION */
641
642
73c4f7a1 643END_EXTERN_C
cad2e5aa 644
cad2e5aa 645
261faec3
MJD
646/* .what is a character array with one character for each member of .data
647 * The character describes the function of the corresponding .data item:
af534a04 648 * a - AV for paren_name_list under DEBUGGING
0111c4fd 649 * f - start-class data for regstclass optimization
68e2671b 650 * l - start op for literal (?{EVAL}) item
d63c20f2 651 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
b30fcab9 652 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
38a44b82 653 * s - swash for Unicode-style character class, and the multicharacter
9e55ce06
JH
654 * strings resulting from casefolding the single-character entries
655 * in the character class
a3621e74 656 * t - trie struct
55eed653 657 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 658 * also used for revcharmap and words under DEBUGGING
07be1b83 659 * T - aho-trie struct
81714fb9 660 * S - sv for named capture lookup
261faec3 661 * 20010712 mjd@plover.com
c3613ed7 662 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 663 */
cad2e5aa
JH
664struct reg_data {
665 U32 count;
666 U8 *what;
667 void* data[1];
668};
669
a1cac82e
NC
670/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
671 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 672#define anchored_substr substrs->data[0].substr
33b8afdf 673#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 674#define anchored_offset substrs->data[0].min_offset
1de06328
YO
675#define anchored_end_shift substrs->data[0].end_shift
676
cad2e5aa 677#define float_substr substrs->data[1].substr
33b8afdf 678#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
679#define float_min_offset substrs->data[1].min_offset
680#define float_max_offset substrs->data[1].max_offset
1de06328
YO
681#define float_end_shift substrs->data[1].end_shift
682
cad2e5aa 683#define check_substr substrs->data[2].substr
33b8afdf 684#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
685#define check_offset_min substrs->data[2].min_offset
686#define check_offset_max substrs->data[2].max_offset
1de06328 687#define check_end_shift substrs->data[2].end_shift
a3621e74 688
8d919b0a
FC
689#define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
690#define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
691#define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
692#define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
a3621e74
YO
693
694/* trie related stuff */
5d9a96ca 695
a3621e74
YO
696/* a transition record for the state machine. the
697 check field determines which state "owns" the
698 transition. the char the transition is for is
699 determined by offset from the owning states base
700 field. the next field determines which state
701 is to be transitioned to if any.
702*/
703struct _reg_trie_trans {
704 U32 next;
705 U32 check;
706};
707
708/* a transition list element for the list based representation */
709struct _reg_trie_trans_list_elem {
710 U16 forid;
711 U32 newstate;
712};
713typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
714
715/* a state for compressed nodes. base is an offset
716 into an array of reg_trie_trans array. If wordnum is
717 nonzero the state is accepting. if base is zero then
718 the state has no children (and will be accepting)
719*/
720struct _reg_trie_state {
721 U16 wordnum;
722 union {
723 U32 base;
724 reg_trie_trans_le* list;
725 } trans;
726};
727
2e64971a
DM
728/* info per word; indexed by wordnum */
729typedef struct {
730 U16 prev; /* previous word in acceptance chain; eg in
731 * zzz|abc|ab/ after matching the chars abc, the
732 * accepted word is #2, and the previous accepted
733 * word is #3 */
734 U32 len; /* how many chars long is this word? */
735 U32 accept; /* accept state for this word */
736} reg_trie_wordinfo;
a3621e74
YO
737
738
a3621e74
YO
739typedef struct _reg_trie_state reg_trie_state;
740typedef struct _reg_trie_trans reg_trie_trans;
741
742
743/* anything in here that needs to be freed later
23eab42c
NC
744 should be dealt with in pregfree.
745 refcount is first in both this and _reg_ac_data to allow a space
746 optimisation in Perl_regdupe. */
a3621e74 747struct _reg_trie_data {
23eab42c 748 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
749 U32 lasttrans; /* last valid transition element */
750 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
751 reg_trie_state *states; /* state data */
752 reg_trie_trans *trans; /* array of transition elements */
753 char *bitmap; /* stclass bitmap */
786e8c11
YO
754 U16 *jump; /* optional 1 indexed array of offsets before tail
755 for the node following a given word. */
2e64971a 756 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
757 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
758 U32 startstate; /* initial state - used for common prefix optimisation */
759 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
760 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 761 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
762 U32 statecount; /* Build only - number of states in the states array
763 (including the unused zero state) */
786e8c11 764 U32 wordcount; /* Build only */
a3621e74 765#ifdef DEBUGGING
786e8c11 766 STRLEN charcount; /* Build only */
a3621e74
YO
767#endif
768};
2b8b4781
NC
769/* There is one (3 under DEBUGGING) pointers that logically belong in this
770 structure, but are held outside as they need duplication on thread cloning,
771 whereas the rest of the structure can be read only:
772 HV *widecharmap; code points > 255 to charid
773#ifdef DEBUGGING
774 AV *words; Array of words contained in trie, for dumping
775 AV *revcharmap; Map of each charid back to its character representation
776#endif
777*/
778
779#define TRIE_WORDS_OFFSET 2
780
a3621e74
YO
781typedef struct _reg_trie_data reg_trie_data;
782
23eab42c
NC
783/* refcount is first in both this and _reg_trie_data to allow a space
784 optimisation in Perl_regdupe. */
07be1b83 785struct _reg_ac_data {
23eab42c 786 U32 refcount;
1b5c067c 787 U32 trie;
07be1b83
YO
788 U32 *fail;
789 reg_trie_state *states;
07be1b83
YO
790};
791typedef struct _reg_ac_data reg_ac_data;
792
486ec47a 793/* ANY_BIT doesn't use the structure, so we can borrow it here.
3dab1dad
YO
794 This is simpler than refactoring all of it as wed end up with
795 three different sets... */
796
797#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 798#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
799#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
800#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
801#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 802
1de06328
YO
803#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
804#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
805
806
786e8c11
YO
807#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
808#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 809
a3621e74
YO
810/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
811#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
812#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
813#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
814
3dab1dad 815#ifdef DEBUGGING
3dab1dad 816#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 817#else
3dab1dad 818#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 819#endif
a3621e74 820
0111c4fd
RGS
821#define RE_TRIE_MAXBUF_INIT 65536
822#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
823#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
824
be8e71aa
YO
825/*
826
827RE_DEBUG_FLAGS is used to control what debug output is emitted
828its divided into three groups of options, some of which interact.
829The three groups are: Compile, Execute, Extra. There is room for a
830further group, as currently only the low three bytes are used.
831
832 Compile Options:
07be1b83 833
be8e71aa
YO
834 PARSE
835 PEEP
836 TRIE
837 PROGRAM
838 OFFSETS
a3621e74 839
be8e71aa 840 Execute Options:
a3621e74 841
be8e71aa
YO
842 INTUIT
843 MATCH
844 TRIE
a3621e74 845
be8e71aa 846 Extra Options
a3621e74 847
be8e71aa
YO
848 TRIE
849 OFFSETS
850
851If you modify any of these make sure you make corresponding changes to
852re.pm, especially to the documentation.
a3621e74 853
be8e71aa
YO
854*/
855
856
857/* Compile */
858#define RE_DEBUG_COMPILE_MASK 0x0000FF
859#define RE_DEBUG_COMPILE_PARSE 0x000001
860#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
861#define RE_DEBUG_COMPILE_TRIE 0x000004
862#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 863#define RE_DEBUG_COMPILE_FLAGS 0x000010
d9a72fcc 864#define RE_DEBUG_COMPILE_TEST 0x000020
be8e71aa
YO
865
866/* Execute */
867#define RE_DEBUG_EXECUTE_MASK 0x00FF00
868#define RE_DEBUG_EXECUTE_INTUIT 0x000100
869#define RE_DEBUG_EXECUTE_MATCH 0x000200
870#define RE_DEBUG_EXECUTE_TRIE 0x000400
871
872/* Extra */
873#define RE_DEBUG_EXTRA_MASK 0xFF0000
874#define RE_DEBUG_EXTRA_TRIE 0x010000
875#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
876#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
877#define RE_DEBUG_EXTRA_STATE 0x080000
878#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 879#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 880#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
881/* combined */
882#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 883
e68ec53f 884#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
885/* Compile */
886#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 887 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 888#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 889 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 890#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 891 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 892#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 893 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 894#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 895 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 896#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 897 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
f7819f85
A
898#define DEBUG_FLAGS_r(x) DEBUG_r( \
899 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
d9a72fcc
YO
900#define DEBUG_TEST_r(x) DEBUG_r( \
901 if (re_debug_flags & RE_DEBUG_COMPILE_TEST) x )
be8e71aa
YO
902/* Execute */
903#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 904 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 905#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 906 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 907#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 908 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 909#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 910 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
911
912/* Extra */
913#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 914 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
915#define DEBUG_OFFSETS_r(x) DEBUG_r( \
916 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
917#define DEBUG_STATE_r(x) DEBUG_r( \
918 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
919#define DEBUG_STACK_r(x) DEBUG_r( \
920 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
921#define DEBUG_BUFFERS_r(x) DEBUG_r( \
922 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
923
a5ca303d
YO
924#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
925 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
926 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 927#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 928 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
929 Perl_warn_nocontext x )
930#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 931 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 932#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 933 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
934
935#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 936 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa 937 | RE_DEBUG_EXECUTE_TRIE )) x )
2c296965
YO
938#define DEBUG_GPOS_r(x) DEBUG_r( \
939 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
be8e71aa
YO
940
941/* initialization */
d2d8c711 942/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
943#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
944 SV * re_debug_flags_sv = NULL; \
945 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
946 if (re_debug_flags_sv) { \
8e9a8a48 947 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 948 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 949 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
950 }\
951})
a3621e74
YO
952
953#ifdef DEBUGGING
ab3bbdeb 954
cb65b687
DM
955#define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
956 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
957
958#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
959 const char * const rpv = \
ddc5bc0f 960 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 961 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 962 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 963 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
964
965#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
966 const char * const rpv = \
ddc5bc0f 967 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 968 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 969 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
970
971#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
972 const char * const rpv = \
ddc5bc0f 973 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 974 PL_colors[0], PL_colors[1], \
c89df6cf 975 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 976 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 977 )
ab3bbdeb
YO
978
979#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
980#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
981
982#else /* if not DEBUGGING */
983
a3621e74 984#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
985#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
986#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
987#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
988#define RE_SV_DUMPLEN(ItEm)
989#define RE_SV_TAIL(ItEm)
990
991#endif /* DEBUG RELATED DEFINES */
a3621e74 992
64935bc6
KW
993typedef enum {
994 TRADITIONAL_BOUND = _CC_WORDCHAR,
ae3bb8ea 995 GCB_BOUND,
06ae2722 996 SB_BOUND,
ae3bb8ea 997 WB_BOUND
64935bc6
KW
998} bound_type;
999
e9a8c099 1000/*
14d04a33 1001 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 1002 */