This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regex: Add lower bound to ANYOFH nodes UTF-8 byte
[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 */
3dd7db29
JK
10
11#ifndef PERL_REGCOMP_H_
12#define PERL_REGCOMP_H_
13
e1d1eefb 14#include "regcharclass.h"
a687059c 15
58e23c8d 16/* Convert branch sequences to more efficient trie ops? */
07be1b83 17#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d 18
486ec47a 19/* Be really aggressive about optimising patterns with trie sequences? */
07be1b83 20#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d
YO
21
22/* Should the optimiser take positive assertions into account? */
68ba3a3f 23#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
58e23c8d
YO
24
25/* Not for production use: */
07be1b83 26#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 27
7122b237
YO
28/* Activate offsets code - set to if 1 to enable */
29#ifdef DEBUGGING
30#define RE_TRACK_PATTERN_OFFSETS
31#endif
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
62e6ef33 91 * boundary, and reads the offset directly as a short.]
a687059c 92 */
785a26d5
YO
93
94/* This is the stuff that used to live in regexp.h that was truly
95 private to the engine itself. It now lives here. */
96
785a26d5 97 typedef struct regexp_internal {
1f1031fe 98 int name_list_idx; /* Optional data index of an array of paren names */
7122b237
YO
99 union {
100 U32 *offsets; /* offset annotations 20001228 MJD
101 data about mapping the program to the
102 string -
103 offsets[0] is proglen when this is used
104 */
105 U32 proglen;
106 } u;
107
6d5c990f
RGS
108 regnode *regstclass; /* Optional startclass as identified or constructed
109 by the optimiser */
110 struct reg_data *data; /* Additional miscellaneous data used by the program.
111 Used to make it easier to clone and free arbitrary
112 data that the regops need. Often the ARG field of
113 a regop is an index into this structure */
1acab4c5 114 struct reg_code_blocks *code_blocks;/* positions of literal (?{}) */
6d5c990f
RGS
115 regnode program[1]; /* Unwarranted chumminess with compiler. */
116} regexp_internal;
117
118#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
119#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
120#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
121/*
122 * Flags stored in regexp->intflags
123 * These are used only internally to the regexp engine
124 *
125 * See regexp.h for flags used externally to the regexp engine
126 */
e3e400ec
YO
127#define RXp_INTFLAGS(rx) ((rx)->intflags)
128#define RX_INTFLAGS(prog) RXp_INTFLAGS(ReANY(prog))
129
785a26d5
YO
130#define PREGf_SKIP 0x00000001
131#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
132#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
133#define PREGf_VERBARG_SEEN 0x00000008
134#define PREGf_CUTGROUP_SEEN 0x00000010
732caac7 135#define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
e3e400ec
YO
136/* these used to be extflags, but are now intflags */
137#define PREGf_NOSCAN 0x00000040
33c28ab2 138 /* spare */
8e1490ee
YO
139#define PREGf_GPOS_SEEN 0x00000100
140#define PREGf_GPOS_FLOAT 0x00000200
785a26d5 141
d3d47aac
YO
142#define PREGf_ANCH_MBOL 0x00000400
143#define PREGf_ANCH_SBOL 0x00000800
144#define PREGf_ANCH_GPOS 0x00001000
d5a00e4a 145#define PREGf_RECURSE_SEEN 0x00002000
8e1490ee 146
d3d47aac
YO
147#define PREGf_ANCH \
148 ( PREGf_ANCH_SBOL | PREGf_ANCH_GPOS | PREGf_ANCH_MBOL )
785a26d5
YO
149
150/* this is where the old regcomp.h started */
a687059c 151
c277df42 152struct regnode_string {
cd439c50 153 U8 str_len;
c277df42
IZ
154 U8 type;
155 U16 next_off;
cd439c50 156 char string[1];
c277df42 157};
a687059c 158
6bda09f9
YO
159/* Argument bearing node - workhorse,
160 arg1 is often for the data field */
c277df42
IZ
161struct regnode_1 {
162 U8 flags;
163 U8 type;
164 U16 next_off;
165 U32 arg1;
166};
167
6bda09f9
YO
168/* Similar to a regnode_1 but with an extra signed argument */
169struct regnode_2L {
170 U8 flags;
171 U8 type;
172 U16 next_off;
173 U32 arg1;
174 I32 arg2;
175};
176
177/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
178struct regnode_2 {
179 U8 flags;
180 U8 type;
181 U16 next_off;
182 U16 arg1;
183 U16 arg2;
184};
185
e0a1ff7a
KW
186/* This give the number of code points that can be in the bitmap of an ANYOF
187 * node. The shift number must currently be one of: 8..12. It can't be less
188 * than 8 (256) because some code relies on it being at least that. Above 12
189 * (4096), and you start running into warnings that some data structure widths
190 * have been exceeded, though the test suite as of this writing still passes
191 * for up through 16, which is as high as anyone would ever want to go,
192 * encompassing all of the Unicode BMP, and thus including all the economically
193 * important world scripts. At 12 most of them are: including Arabic,
194 * Cyrillic, Greek, Hebrew, Indian subcontinent, Latin, and Thai; but not Han,
195 * Japanese, nor Korean. (The regarglen structure in regnodes.h is a U8, and
196 * the trie types TRIEC and AHOCORASICKC are larger than U8 for shift values
b34f51a7 197 * above 12.) Be sure to benchmark before changing, as larger sizes do
e0a1ff7a
KW
198 * significantly slow down the test suite */
199#define NUM_ANYOF_CODE_POINTS (1 << 8)
6bda09f9 200
dcb20b36 201#define ANYOF_BITMAP_SIZE (NUM_ANYOF_CODE_POINTS / 8) /* 8 bits/Byte */
936ed897 202
1ee208c4
KW
203/* Note that these form structs which are supersets of the next smaller one, by
204 * appending fields. Alignment problems can occur if one of those optional
205 * fields requires stricter alignment than the base struct. And formal
206 * parameters that can really be two or more of the structs should be
207 * declared as the smallest one it could be. See commit message for
208 * 7dcac5f6a5195002b55c935ee1d67f67e1df280b. Regnode allocation is done
209 * without regard to alignment, and changing it to would also require changing
210 * the code that inserts and deletes regnodes. The basic single-argument
211 * regnode has a U32, which is what reganode() allocates as a unit. Therefore
212 * no field can require stricter alignment than U32. */
213
786e8c11 214/* also used by trie */
936ed897
IZ
215struct regnode_charclass {
216 U8 flags;
217 U8 type;
218 U16 next_off;
9da99c5e 219 U32 arg1; /* set by set_ANYOF_arg() */
19103b91 220 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
221};
222
bf3f2d85 223/* has runtime (locale) \d, \w, ..., [:posix:] classes */
ea461dd6 224struct regnode_charclass_posixl {
93e92956 225 U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */
936ed897
IZ
226 U8 type;
227 U16 next_off;
7e817188 228 U32 arg1;
03fa83ba 229 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */
9cf78fa0 230 U32 classflags; /* and run-time */
936ed897
IZ
231};
232
85c8e306
KW
233/* A synthetic start class (SSC); is a regnode_charclass_posixl_fold, plus an
234 * extra SV*, used only during its construction and which is not used by
235 * regexec.c. Note that the 'next_off' field is unused, as the SSC stands
236 * alone, so there is never a next node. Also, there is no alignment issue,
b8fda935 237 * because these are declared or allocated as a complete unit so the compiler
85c8e306
KW
238 * takes care of alignment. This is unlike the other regnodes which are
239 * allocated in terms of multiples of a single-argument regnode. SSC nodes can
240 * have a pointer field because there is no alignment issue, and because it is
241 * set to NULL after construction, before any cloning of the pattern */
b8f7bb16 242struct regnode_ssc {
93e92956 243 U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */
b8f7bb16
KW
244 U8 type;
245 U16 next_off;
7e817188 246 U32 arg1;
03fa83ba 247 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */
631e9161 248 U32 classflags; /* ... and run-time */
85c8e306
KW
249
250 /* Auxiliary, only used during construction; NULL afterwards: list of code
251 * points matched */
252 SV* invlist;
b8f7bb16
KW
253};
254
71068078
KW
255/* We take advantage of 'next_off' not otherwise being used in the SSC by
256 * actually using it: by setting it to 1. This allows us to test and
257 * distinguish between an SSC and other ANYOF node types, as 'next_off' cannot
258 * otherwise be 1, because it is the offset to the next regnode expressed in
259 * units of regnodes. Since an ANYOF node contains extra fields, it adds up
260 * to 12 regnode units on 32-bit systems, (hence the minimum this can be (if
261 * not 0) is 11 there. Even if things get tightly packed on a 64-bit system,
262 * it still would be more than 1. */
263#define set_ANYOF_SYNTHETIC(n) STMT_START{ OP(n) = ANYOF; \
264 NEXT_OFF(n) = 1; \
265 } STMT_END
c1071334 266#define is_ANYOF_SYNTHETIC(n) (PL_regkind[OP(n)] == ANYOF && NEXT_OFF(n) == 1)
71068078 267
9c7e81e8
AD
268/* XXX fix this description.
269 Impose a limit of REG_INFTY on various pattern matching operations
270 to limit stack growth and to avoid "infinite" recursions.
271*/
4288c5b9
KW
272/* The default size for REG_INFTY is U16_MAX, which is the same as
273 USHORT_MAX (see perl.h). Unfortunately U16 isn't necessarily 16 bits
274 (see handy.h). On the Cray C90, sizeof(short)==4 and hence U16_MAX is
275 ((1<<32)-1), while on the Cray T90, sizeof(short)==8 and U16_MAX is
276 ((1<<64)-1). To limit stack growth to reasonable sizes, supply a
9c7e81e8
AD
277 smaller default.
278 --Andy Dougherty 11 June 1998
279*/
280#if SHORTSIZE > 2
281# ifndef REG_INFTY
4288c5b9 282# define REG_INFTY ((1<<16)-1)
9c7e81e8
AD
283# endif
284#endif
285
286#ifndef REG_INFTY
4288c5b9 287# define REG_INFTY U16_MAX
83cfe48c 288#endif
a687059c 289
e91177ed
IZ
290#define ARG_VALUE(arg) (arg)
291#define ARG__SET(arg,val) ((arg) = (val))
c277df42 292
e8e3b0fb
JH
293#undef ARG
294#undef ARG1
295#undef ARG2
296
c277df42
IZ
297#define ARG(p) ARG_VALUE(ARG_LOC(p))
298#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
299#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 300#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 301
c277df42
IZ
302#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
303#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
304#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 305#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 306
e8e3b0fb
JH
307#undef NEXT_OFF
308#undef NODE_ALIGN
309
075abff3
AL
310#define NEXT_OFF(p) ((p)->next_off)
311#define NODE_ALIGN(node)
aa48e906
YO
312/* the following define was set to 0xde in 075abff3
313 * as part of some linting logic. I have set it to 0
314 * as otherwise in every place where we /might/ set flags
315 * we have to set it 0 explicitly, which duplicates
316 * assignments and IMO adds an unacceptable level of
317 * surprise to working in the regex engine. If this
318 * is changed from 0 then at the very least make sure
319 * that SBOL for /^/ sets the flags to 0 explicitly.
320 * -- Yves */
321#define NODE_ALIGN_FILL(node) ((node)->flags = 0)
a687059c 322
c277df42
IZ
323#define SIZE_ALIGN NODE_ALIGN
324
e8e3b0fb
JH
325#undef OP
326#undef OPERAND
327#undef MASK
328#undef STRING
329
e91177ed 330#define OP(p) ((p)->type)
b988e673
KW
331#define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
332 regnode types. For some, it's the \
333 character set of the regnode */
e91177ed 334#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
335#define MASK(p) ((char*)OPERAND(p))
336#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
337#define STRING(p) (((struct regnode_string *)p)->string)
338#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
339#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
340
e8e3b0fb
JH
341#undef NODE_ALIGN
342#undef ARG_LOC
343#undef NEXTOPER
344#undef PREVOPER
345
e91177ed
IZ
346#define NODE_ALIGN(node)
347#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
348#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
349#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 350#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 351
e91177ed
IZ
352#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
353#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
354
355#define NODE_STEP_B 4
c277df42
IZ
356
357#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
358#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
359
f55b7b26
KW
360#define FILL_NODE(offset, op) \
361 STMT_START { \
362 OP(REGNODE_p(offset)) = op; \
363 NEXT_OFF(REGNODE_p(offset)) = 0; \
d2a88850 364 } STMT_END
f55b7b26
KW
365#define FILL_ADVANCE_NODE(offset, op) \
366 STMT_START { \
367 FILL_NODE(offset, op); \
368 (offset)++; \
d2a88850 369 } STMT_END
f55b7b26
KW
370#define FILL_ADVANCE_NODE_ARG(offset, op, arg) \
371 STMT_START { \
372 ARG_SET(REGNODE_p(offset), arg); \
373 FILL_ADVANCE_NODE(offset, op); \
374 /* This is used generically for other operations \
375 * that have a longer argument */ \
376 (offset) += regarglen[op]; \
93539530 377 } STMT_END
f55b7b26
KW
378#define FILL_ADVANCE_NODE_2L_ARG(offset, op, arg1, arg2) \
379 STMT_START { \
380 ARG_SET(REGNODE_p(offset), arg1); \
381 ARG2L_SET(REGNODE_p(offset), arg2); \
382 FILL_ADVANCE_NODE(offset, op); \
383 (offset) += 2; \
93539530 384 } STMT_END
a687059c 385
22c35a8c 386#define REG_MAGIC 0234
a687059c 387
fe8d1b7c
KW
388/* An ANYOF node is basically a bitmap with the index being a code point. If
389 * the bit for that code point is 1, the code point matches; if 0, it doesn't
390 * match (complemented if inverted). There is an additional mechanism to deal
391 * with cases where the bitmap is insufficient in and of itself. This #define
392 * indicates if the bitmap does fully represent what this ANYOF node can match.
393 * The ARG is set to this special value (since 0, 1, ... are legal, but will
394 * never reach this high). */
93e92956 395#define ANYOF_ONLY_HAS_BITMAP ((U32) -1)
137165a6 396
93539530 397/* When the bitmap isn't completely sufficient for handling the ANYOF node,
fe8d1b7c
KW
398 * flags (in node->flags of the ANYOF node) get set to indicate this. These
399 * are perennially in short supply. Beyond several cases where warnings need
400 * to be raised under certain circumstances, currently, there are six cases
401 * where the bitmap alone isn't sufficient. We could use six flags to
402 * represent the 6 cases, but to save flags bits, we play some games. The
403 * cases are:
d2f465e9 404 *
fe8d1b7c
KW
405 * 1) The bitmap has a compiled-in very finite size. So something else needs
406 * to be used to specify if a code point that is too large for the bitmap
7a94a5fe 407 * actually matches. The mechanism currently is an inversion
fe8d1b7c
KW
408 * list. ANYOF_ONLY_HAS_BITMAP, described above, being TRUE indicates
409 * there are no matches of too-large code points. But if it is FALSE,
410 * then almost certainly there are matches too large for the bitmap. (The
411 * other cases, described below, either imply this one or are extremely
412 * rare in practice.) So we can just assume that a too-large code point
413 * will need something beyond the bitmap if ANYOF_ONLY_HAS_BITMAP is
414 * FALSE, instead of having a separate flag for this.
415 * 2) A subset of item 1) is if all possible code points outside the bitmap
416 * match. This is a common occurrence when the class is complemented,
417 * like /[^ij]/. Therefore a bit is reserved to indicate this,
7a94a5fe 418 * rather than having a more expensive inversion list created,
d1c40ef5 419 * ANYOF_MATCHES_ALL_ABOVE_BITMAP.
fe8d1b7c
KW
420 * 3) Under /d rules, it can happen that code points that are in the upper
421 * latin1 range (\x80-\xFF or their equivalents on EBCDIC platforms) match
422 * only if the runtime target string being matched against is UTF-8. For
423 * example /[\w[:punct:]]/d. This happens only for posix classes (with a
d1c40ef5
KW
424 * couple of exceptions, like \d where it doesn't happen), and all such
425 * ones also have above-bitmap matches. Thus, 3) implies 1) as well.
426 * Note that /d rules are no longer encouraged; 'use 5.14' or higher
427 * deselects them. But a flag is required so that they can be properly
428 * handled. But it can be a shared flag: see 5) below.
429 * 4) Also under /d rules, something like /[\Wfoo]/ will match everything in
fe8d1b7c 430 * the \x80-\xFF range, unless the string being matched against is UTF-8.
7a94a5fe
KW
431 * An inversion list could be created for this case, but this is
432 * relatively common, and it turns out that it's all or nothing: if any
433 * one of these code points matches, they all do. Hence a single bit
434 * suffices. We use a shared flag that doesn't take up space by itself:
435 * ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER. This
436 * also implies 1), with one exception: [:^cntrl:].
fe8d1b7c
KW
437 * 5) A user-defined \p{} property may not have been defined by the time the
438 * regex is compiled. In this case, we don't know until runtime what it
439 * will match, so we have to assume it could match anything, including
440 * code points that ordinarily would be in the bitmap. A flag bit is
441 * necessary to indicate this , though it can be shared with the item 3)
442 * flag, as that only occurs under /d, and this only occurs under non-d.
443 * This case is quite uncommon in the field, and the /(?[ ...])/ construct
444 * is a better way to accomplish what this feature does. This case also
445 * implies 1).
446 * ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP
e2f5e63d 447 * is the shared flag.
fe8d1b7c
KW
448 * 6) /[foo]/il may have folds that are only valid if the runtime locale is a
449 * UTF-8 one. These are quite rare, so it would be good to avoid the
450 * expense of looking for them. But /l matching is slow anyway, and we've
e2f5e63d 451 * traditionally not worried too much about its performance. And this
037715a6 452 * condition requires the ANYOFL_FOLD flag to be set, so testing for
fe8d1b7c 453 * that flag would be sufficient to rule out most cases of this. So it is
d1c40ef5
KW
454 * unclear if this should have a flag or not. But, this flag can be
455 * shared with another, so it doesn't occupy extra space.
d2f465e9 456 *
d1c40ef5 457 * At the moment, there is one spare bit, but this could be increased by
93539530 458 * various tricks:
e2f5e63d 459 *
93539530 460 * If just one more bit is needed, as of this writing it seems to khw that the
d1c40ef5
KW
461 * best choice would be to make ANYOF_MATCHES_ALL_ABOVE_BITMAP not a flag, but
462 * something like
d2f465e9 463 *
d1c40ef5 464 * #define ANYOF_MATCHES_ALL_ABOVE_BITMAP ((U32) -2)
fe8d1b7c 465 *
d1c40ef5
KW
466 * and access it through the ARG like ANYOF_ONLY_HAS_BITMAP is. This flag is
467 * used by all ANYOF node types, and it could be used to avoid calling the
468 * handler function, as the macro REGINCLASS in regexec.c does now for other
469 * cases.
470 *
46fc0c43
KW
471 * Another possibility is based on the fact that ANYOF_MATCHES_POSIXL is
472 * redundant with the node type ANYOFPOSIXL. That flag could be removed, but
473 * at the expense of extra code in regexec.c. The flag has been retained
474 * because it allows us to see if we need to call reginsert, or just use the
475 * bitmap in one test.
fe8d1b7c 476 *
46fc0c43
KW
477 * If this is done, an extension would be to make all ANYOFL nodes contain the
478 * extra 32 bits that ANYOFPOSIXL ones do. The posix flags only occupy 30
479 * bits, so the ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD flags
480 * and ANYOFL_FOLD could be moved to that extra space, but it would mean extra
481 * instructions, as there are currently places in the code that assume those
482 * two bits are zero.
d1c40ef5
KW
483 *
484 * All told, 5 bits could be available for other uses if all of the above were
485 * done.
a4525e78 486 *
d1c40ef5 487 * Some flags are not used in synthetic start class (SSC) nodes, so could be
93e92956
KW
488 * shared should new flags be needed for SSCs, like SSC_MATCHES_EMPTY_STRING
489 * now. */
b8c5462f 490
d2f465e9
KW
491/* If this is set, the result of the match should be complemented. regexec.c
492 * is expecting this to be in the low bit. Never in an SSC */
93e92956 493#define ANYOF_INVERT 0x01
1984f6a0 494
a0dd4231 495/* For the SSC node only, which cannot be inverted, so is shared with that bit.
93e92956
KW
496 * This is used only during regex compilation. */
497#define SSC_MATCHES_EMPTY_STRING ANYOF_INVERT
a0dd4231 498
509d4de0
KW
499/* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This
500 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
501 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
502 * at compile-time. Only set under /l; can be in SSC */
503#define ANYOF_MATCHES_POSIXL 0x02
b63f73da
KW
504
505/* The fold is calculated and stored in the bitmap where possible at compile
506 * time. However under locale, the actual folding varies depending on
507 * what the locale is at the time of execution, so it has to be deferred until
d2f465e9 508 * then. Only set under /l; never in an SSC */
037715a6 509#define ANYOFL_FOLD 0x04
b63f73da 510
d1c40ef5 511/* Shared bit set only with ANYOFL and SSC nodes:
269bb65b
KW
512 * If ANYOFL_FOLD is set, this flag indicates there are potential matches
513 * valid only if the locale is a UTF-8 one.
514 * If ANYOFL_FOLD is NOT set, this flag means to warn if the runtime locale
d1c40ef5
KW
515 * isn't a UTF-8 one (and the generated node assumes a UTF-8 locale).
516 * None of INVERT, POSIXL,
517 * ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP
518 * can be set. */
519#define ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD 0x08
520
521/* Convenience macros for teasing apart the meanings when reading the above bit
522 * */
523#define ANYOFL_SOME_FOLDS_ONLY_IN_UTF8_LOCALE(flags) \
524 ((flags & ( ANYOFL_FOLD /* Both bits are set */ \
525 |ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD)) \
526 == ( ANYOFL_FOLD \
527 |ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD))
528
529#define ANYOFL_UTF8_LOCALE_REQD(flags) \
530 ((flags & ( ANYOFL_FOLD /* Only REQD bit is set */ \
531 |ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD)) \
532 == ANYOFL_SHARED_UTF8_LOCALE_fold_HAS_MATCHES_nonfold_REQD)
533
1744d18d 534/* Spare: Be sure to change ANYOF_FLAGS_ALL if this gets used 0x10 */
c8d3cd88 535
509d4de0
KW
536/* If set, the node matches every code point NUM_ANYOF_CODE_POINTS and above.
537 * Can be in an SSC */
b3b1cf17 538#define ANYOF_MATCHES_ALL_ABOVE_BITMAP 0x20
ef87b810 539
108316fb
KW
540/* Shared bit:
541 * Under /d it means the ANYOFD node matches more things if the target
542 * string is encoded in UTF-8; any such things will be non-ASCII,
7a94a5fe
KW
543 * characters that are < 256, and can be accessed via the inversion
544 * list.
108316fb
KW
545 * When not under /d, it means the ANYOF node contains a user-defined
546 * property that wasn't yet defined at the time the regex was compiled,
7a94a5fe 547 * and so must be looked up at runtime, by creating an inversion list.
108316fb
KW
548 * (These uses are mutually exclusive because a user-defined property is
549 * specified by \p{}, and \p{} implies /u which deselects /d). The long macro
550 * name is to make sure that you are cautioned about its shared nature. Only
551 * the non-/d meaning can be in an SSC */
552#define ANYOF_SHARED_d_UPPER_LATIN1_UTF8_STRING_MATCHES_non_d_RUNTIME_USER_PROP 0x40
f240c685
KW
553
554/* Shared bit:
d2f465e9 555 * Under /d it means the ANYOFD node matches all non-ASCII Latin1
f240c685
KW
556 * characters when the target string is not in utf8.
557 * When not under /d, it means the ANYOF node should raise a warning if
558 * matching against an above-Unicode code point.
559 * (These uses are mutually exclusive because the warning requires a \p{}, and
560 * \p{} implies /u which deselects /d). An SSC node only has this bit set if
561 * what is meant is the warning. The long macro name is to make sure that you
562 * are cautioned about its shared nature */
563#define ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER 0x80
7bc66b18 564
b271ede1 565#define ANYOF_FLAGS_ALL (0xff & ~0x10)
653099ff 566
037715a6 567#define ANYOF_LOCALE_FLAGS (ANYOFL_FOLD | ANYOF_MATCHES_POSIXL)
ace6b0e4 568
eff8b7dc
KW
569/* These are the flags that apply to both regular ANYOF nodes and synthetic
570 * start class nodes during construction of the SSC. During finalization of
d2f465e9 571 * the SSC, other of the flags may get added to it */
d1c40ef5 572#define ANYOF_COMMON_FLAGS 0
eff8b7dc 573
936ed897 574/* Character classes for node->classflags of ANYOF */
653099ff 575/* Should be synchronized with a table in regprop() */
575a8a29 576/* 2n should be the normal one, paired with its complement at 2n+1 */
b8c5462f 577
08ada29e
KW
578#define ANYOF_ALPHA ((_CC_ALPHA) * 2)
579#define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
e8d596e0
KW
580#define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
581#define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
08ada29e
KW
582#define ANYOF_ASCII ((_CC_ASCII) * 2)
583#define ANYOF_NASCII ((ANYOF_ASCII) + 1)
e8d596e0
KW
584#define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
585#define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
b0d691b2
KW
586#define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
587 [:upper:] under /i */
588#define ANYOF_NCASED ((ANYOF_CASED) + 1)
08ada29e
KW
589#define ANYOF_CNTRL ((_CC_CNTRL) * 2)
590#define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
e8d596e0
KW
591#define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
592#define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
08ada29e
KW
593#define ANYOF_GRAPH ((_CC_GRAPH) * 2)
594#define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
595#define ANYOF_LOWER ((_CC_LOWER) * 2)
596#define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
597#define ANYOF_PRINT ((_CC_PRINT) * 2)
598#define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
599#define ANYOF_PUNCT ((_CC_PUNCT) * 2)
600#define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
e8d596e0
KW
601#define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
602#define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
08ada29e
KW
603#define ANYOF_UPPER ((_CC_UPPER) * 2)
604#define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
e8d596e0
KW
605#define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
606#define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
08ada29e
KW
607#define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
608#define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
b8c5462f 609
87cad4c0 610/* pseudo classes below this, not stored in the class bitmap, but used as flags
e1d1eefb
YO
611 during compilation of char classes */
612
b2ef2e34
KW
613#define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
614#define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
e1d1eefb 615
b2ef2e34
KW
616/* It is best if this is the last one, as all above it are stored as bits in a
617 * bitmap, and it isn't part of that bitmap */
618#if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
619# error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
620#endif
621
8efd3f97
KW
622#define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
623 * '< ANYOF_MAX'
624 * Hence doesn't include VERTWS, as that
625 * is a pseudo class */
626#define ANYOF_MAX ANYOF_POSIXL_MAX
627
628#if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */
b2ef2e34 629# error Problem with handy.h _CC_foo #defines
a0947d7b
KW
630#endif
631
8efd3f97
KW
632#define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
633#define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3)
87cad4c0 634
8efd3f97
KW
635#define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode
636 property: \p{} or \P{} */
f39a4dc2 637
b8c5462f
JH
638/* Backward source code compatibility. */
639
640#define ANYOF_ALNUML ANYOF_ALNUM
641#define ANYOF_NALNUML ANYOF_NALNUM
642#define ANYOF_SPACEL ANYOF_SPACE
643#define ANYOF_NSPACEL ANYOF_NSPACE
8ebfc29a
KW
644#define ANYOF_ALNUM ANYOF_WORDCHAR
645#define ANYOF_NALNUM ANYOF_NWORDCHAR
b8c5462f
JH
646
647/* Utility macros for the bitmap and classes of ANYOF */
648
936ed897 649#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f 650
bc51fd78 651#define ANYOF_BIT(c) (1U << ((c) & 7))
b8c5462f 652
14ed74f6
KW
653#define POSIXL_SET(field, c) ((field) |= (1U << (c)))
654#define ANYOF_POSIXL_SET(p, c) POSIXL_SET(((regnode_charclass_posixl*) (p))->classflags, (c))
8efd3f97 655
14ed74f6
KW
656#define POSIXL_CLEAR(field, c) ((field) &= ~ (1U <<(c)))
657#define ANYOF_POSIXL_CLEAR(p, c) POSIXL_CLEAR(((regnode_charclass_posixl*) (p))->classflags, (c))
8efd3f97 658
14ed74f6
KW
659#define POSIXL_TEST(field, c) ((field) & (1U << (c)))
660#define ANYOF_POSIXL_TEST(p, c) POSIXL_TEST(((regnode_charclass_posixl*) (p))->classflags, (c))
b8c5462f 661
14ed74f6
KW
662#define POSIXL_ZERO(field) STMT_START { (field) = 0; } STMT_END
663#define ANYOF_POSIXL_ZERO(ret) POSIXL_ZERO(((regnode_charclass_posixl*) (ret))->classflags)
664
665#define ANYOF_POSIXL_SET_TO_BITMAP(p, bits) \
666 STMT_START { \
667 ((regnode_charclass_posixl*) (p))->classflags = (bits); \
668 } STMT_END
9cf78fa0
KW
669
670/* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
d1d040e5 671#define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END
8efd3f97 672#define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret)
9cf78fa0 673
8efd3f97 674#define ANYOF_POSIXL_TEST_ANY_SET(p) \
93e92956 675 ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \
d1d040e5 676 && (((regnode_charclass_posixl*)(p))->classflags))
8efd3f97 677#define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p)
e867ce45 678
e0e1be5f
KW
679/* Since an SSC always has this field, we don't have to test for that; nor do
680 * we want to because the bit isn't set for SSC during its construction */
681#define ANYOF_POSIXL_SSC_TEST_ANY_SET(p) \
682 cBOOL(((regnode_ssc*)(p))->classflags)
683#define ANYOF_POSIXL_SSC_TEST_ALL_SET(p) /* Are all bits set? */ \
684 (((regnode_ssc*) (p))->classflags \
685 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
686
687#define ANYOF_POSIXL_TEST_ALL_SET(p) \
93e92956 688 ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \
e0e1be5f
KW
689 && ((regnode_charclass_posixl*) (p))->classflags \
690 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
fb38762f 691
b27fd97a 692#define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END
8efd3f97 693#define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest))
936ed897 694
cdd87c1d
KW
695#define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END
696
428c1aff
KW
697#define ANYOF_BITMAP_ZERO(ret) Zero(((regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
698#define ANYOF_BITMAP(p) ((regnode_charclass*)(p))->bitmap
4cbce0a6 699#define ANYOF_BITMAP_BYTE(p, c) BITMAP_BYTE(ANYOF_BITMAP(p), c)
b8c5462f
JH
700#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
701#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
bc51fd78 702#define ANYOF_BITMAP_TEST(p, c) cBOOL(ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
b8c5462f 703
f8bef550
NC
704#define ANYOF_BITMAP_SETALL(p) \
705 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
706#define ANYOF_BITMAP_CLEARALL(p) \
707 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
f8bef550 708
a687059c
LW
709/*
710 * Utility definitions.
711 */
2304df62 712#ifndef CHARMASK
075abff3 713# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 714#else
075abff3 715# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 716#endif
a687059c 717
c277df42
IZ
718#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
719
e384d5c1
YO
720#define REG_ZERO_LEN_SEEN 0x00000001
721#define REG_LOOKBEHIND_SEEN 0x00000002
722#define REG_GPOS_SEEN 0x00000004
471f5387 723/* spare */
e384d5c1
YO
724#define REG_RECURSE_SEEN 0x00000020
725#define REG_TOP_LEVEL_BRANCHES_SEEN 0x00000040
726#define REG_VERBARG_SEEN 0x00000080
727#define REG_CUTGROUP_SEEN 0x00000100
728#define REG_RUN_ON_COMMENT_SEEN 0x00000200
729#define REG_UNFOLDED_MULTI_SEEN 0x00000400
d5a00e4a 730/* spare */
ee273784
YO
731#define REG_UNBOUNDED_QUANTIFIER_SEEN 0x00001000
732
d09b2d29 733
73c4f7a1
GS
734START_EXTERN_C
735
be8e71aa
YO
736#ifdef PLUGGABLE_RE_EXTENSION
737#include "re_nodes.h"
738#else
d09b2d29 739#include "regnodes.h"
be8e71aa 740#endif
d09b2d29 741
f9f4320a
YO
742#ifndef PLUGGABLE_RE_EXTENSION
743#ifndef DOINIT
744EXTCONST regexp_engine PL_core_reg_engine;
745#else /* DOINIT */
746EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 747 Perl_re_compile,
2fdbfb4d 748 Perl_regexec_flags,
f9f4320a
YO
749 Perl_re_intuit_start,
750 Perl_re_intuit_string,
2fdbfb4d
AB
751 Perl_regfree_internal,
752 Perl_reg_numbered_buff_fetch,
753 Perl_reg_numbered_buff_store,
754 Perl_reg_numbered_buff_length,
192b9cd1
AB
755 Perl_reg_named_buff,
756 Perl_reg_named_buff_iter,
49d7dfbc 757 Perl_reg_qr_package,
f9f4320a 758#if defined(USE_ITHREADS)
3c13cae6 759 Perl_regdupe_internal,
f9f4320a 760#endif
3c13cae6 761 Perl_re_op_compile
f9f4320a
YO
762};
763#endif /* DOINIT */
764#endif /* PLUGGABLE_RE_EXTENSION */
765
766
73c4f7a1 767END_EXTERN_C
cad2e5aa 768
cad2e5aa 769
261faec3
MJD
770/* .what is a character array with one character for each member of .data
771 * The character describes the function of the corresponding .data item:
af534a04 772 * a - AV for paren_name_list under DEBUGGING
0111c4fd 773 * f - start-class data for regstclass optimization
68e2671b 774 * l - start op for literal (?{EVAL}) item
d63c20f2 775 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
b30fcab9 776 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
7a94a5fe
KW
777 * s - inversion list for Unicode-style character class, and the
778 * multicharacter strings resulting from casefolding the single-character
779 * entries in the character class
a3621e74 780 * t - trie struct
55eed653 781 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 782 * also used for revcharmap and words under DEBUGGING
07be1b83 783 * T - aho-trie struct
81714fb9 784 * S - sv for named capture lookup
261faec3 785 * 20010712 mjd@plover.com
c3613ed7 786 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 787 */
cad2e5aa
JH
788struct reg_data {
789 U32 count;
790 U8 *what;
791 void* data[1];
792};
793
a1cac82e
NC
794/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
795 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 796#define anchored_substr substrs->data[0].substr
33b8afdf 797#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 798#define anchored_offset substrs->data[0].min_offset
1de06328
YO
799#define anchored_end_shift substrs->data[0].end_shift
800
cad2e5aa 801#define float_substr substrs->data[1].substr
33b8afdf 802#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
803#define float_min_offset substrs->data[1].min_offset
804#define float_max_offset substrs->data[1].max_offset
1de06328
YO
805#define float_end_shift substrs->data[1].end_shift
806
cad2e5aa 807#define check_substr substrs->data[2].substr
33b8afdf 808#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
809#define check_offset_min substrs->data[2].min_offset
810#define check_offset_max substrs->data[2].max_offset
1de06328 811#define check_end_shift substrs->data[2].end_shift
a3621e74 812
8d919b0a
FC
813#define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
814#define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
815#define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
816#define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
a3621e74
YO
817
818/* trie related stuff */
5d9a96ca 819
a3621e74
YO
820/* a transition record for the state machine. the
821 check field determines which state "owns" the
822 transition. the char the transition is for is
823 determined by offset from the owning states base
824 field. the next field determines which state
825 is to be transitioned to if any.
826*/
827struct _reg_trie_trans {
828 U32 next;
829 U32 check;
830};
831
832/* a transition list element for the list based representation */
833struct _reg_trie_trans_list_elem {
834 U16 forid;
835 U32 newstate;
836};
837typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
838
839/* a state for compressed nodes. base is an offset
840 into an array of reg_trie_trans array. If wordnum is
841 nonzero the state is accepting. if base is zero then
842 the state has no children (and will be accepting)
843*/
844struct _reg_trie_state {
845 U16 wordnum;
846 union {
847 U32 base;
848 reg_trie_trans_le* list;
849 } trans;
850};
851
2e64971a
DM
852/* info per word; indexed by wordnum */
853typedef struct {
854 U16 prev; /* previous word in acceptance chain; eg in
855 * zzz|abc|ab/ after matching the chars abc, the
856 * accepted word is #2, and the previous accepted
857 * word is #3 */
858 U32 len; /* how many chars long is this word? */
859 U32 accept; /* accept state for this word */
860} reg_trie_wordinfo;
a3621e74
YO
861
862
a3621e74
YO
863typedef struct _reg_trie_state reg_trie_state;
864typedef struct _reg_trie_trans reg_trie_trans;
865
866
867/* anything in here that needs to be freed later
23eab42c
NC
868 should be dealt with in pregfree.
869 refcount is first in both this and _reg_ac_data to allow a space
870 optimisation in Perl_regdupe. */
a3621e74 871struct _reg_trie_data {
23eab42c 872 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
873 U32 lasttrans; /* last valid transition element */
874 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
875 reg_trie_state *states; /* state data */
876 reg_trie_trans *trans; /* array of transition elements */
877 char *bitmap; /* stclass bitmap */
786e8c11
YO
878 U16 *jump; /* optional 1 indexed array of offsets before tail
879 for the node following a given word. */
2e64971a 880 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
881 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
882 U32 startstate; /* initial state - used for common prefix optimisation */
883 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
884 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 885 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
886 U32 statecount; /* Build only - number of states in the states array
887 (including the unused zero state) */
786e8c11 888 U32 wordcount; /* Build only */
a3621e74 889#ifdef DEBUGGING
786e8c11 890 STRLEN charcount; /* Build only */
a3621e74
YO
891#endif
892};
2b8b4781
NC
893/* There is one (3 under DEBUGGING) pointers that logically belong in this
894 structure, but are held outside as they need duplication on thread cloning,
895 whereas the rest of the structure can be read only:
896 HV *widecharmap; code points > 255 to charid
897#ifdef DEBUGGING
898 AV *words; Array of words contained in trie, for dumping
899 AV *revcharmap; Map of each charid back to its character representation
900#endif
901*/
902
903#define TRIE_WORDS_OFFSET 2
904
a3621e74
YO
905typedef struct _reg_trie_data reg_trie_data;
906
23eab42c
NC
907/* refcount is first in both this and _reg_trie_data to allow a space
908 optimisation in Perl_regdupe. */
07be1b83 909struct _reg_ac_data {
23eab42c 910 U32 refcount;
1b5c067c 911 U32 trie;
07be1b83
YO
912 U32 *fail;
913 reg_trie_state *states;
07be1b83
YO
914};
915typedef struct _reg_ac_data reg_ac_data;
916
486ec47a 917/* ANY_BIT doesn't use the structure, so we can borrow it here.
3dab1dad
YO
918 This is simpler than refactoring all of it as wed end up with
919 three different sets... */
920
921#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
4cbce0a6 922#define TRIE_BITMAP_BYTE(p, c) BITMAP_BYTE(TRIE_BITMAP(p), c)
5dfe062f
YO
923#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
924#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
925#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 926
1de06328
YO
927#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
928#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
929
930
0effac60 931#define BITMAP_BYTE(p, c) (( (U8*) p) [ ( ( (UV) (c)) >> 3) ] )
786e8c11 932#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 933
a3621e74
YO
934/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
935#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
936#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
937#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
938
3dab1dad 939#ifdef DEBUGGING
3dab1dad 940#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 941#else
3dab1dad 942#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 943#endif
a3621e74 944
0111c4fd
RGS
945#define RE_TRIE_MAXBUF_INIT 65536
946#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
947#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
948
6ef7fe53
KW
949#define RE_COMPILE_RECURSION_INIT 1000
950#define RE_COMPILE_RECURSION_LIMIT "\022E_COMPILE_RECURSION_LIMIT"
951
be8e71aa
YO
952/*
953
954RE_DEBUG_FLAGS is used to control what debug output is emitted
955its divided into three groups of options, some of which interact.
956The three groups are: Compile, Execute, Extra. There is room for a
957further group, as currently only the low three bytes are used.
958
959 Compile Options:
07be1b83 960
be8e71aa
YO
961 PARSE
962 PEEP
963 TRIE
964 PROGRAM
965 OFFSETS
a3621e74 966
be8e71aa 967 Execute Options:
a3621e74 968
be8e71aa
YO
969 INTUIT
970 MATCH
971 TRIE
a3621e74 972
be8e71aa 973 Extra Options
a3621e74 974
be8e71aa
YO
975 TRIE
976 OFFSETS
977
978If you modify any of these make sure you make corresponding changes to
979re.pm, especially to the documentation.
a3621e74 980
be8e71aa
YO
981*/
982
983
984/* Compile */
985#define RE_DEBUG_COMPILE_MASK 0x0000FF
986#define RE_DEBUG_COMPILE_PARSE 0x000001
987#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
988#define RE_DEBUG_COMPILE_TRIE 0x000004
989#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 990#define RE_DEBUG_COMPILE_FLAGS 0x000010
d9a72fcc 991#define RE_DEBUG_COMPILE_TEST 0x000020
be8e71aa
YO
992
993/* Execute */
994#define RE_DEBUG_EXECUTE_MASK 0x00FF00
995#define RE_DEBUG_EXECUTE_INTUIT 0x000100
996#define RE_DEBUG_EXECUTE_MATCH 0x000200
997#define RE_DEBUG_EXECUTE_TRIE 0x000400
998
999/* Extra */
1000#define RE_DEBUG_EXTRA_MASK 0xFF0000
1001#define RE_DEBUG_EXTRA_TRIE 0x010000
1002#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
1003#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
1004#define RE_DEBUG_EXTRA_STATE 0x080000
1005#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 1006#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 1007#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
1008/* combined */
1009#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 1010
e68ec53f 1011#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
1012/* Compile */
1013#define DEBUG_COMPILE_r(x) DEBUG_r( \
dafb2544 1014 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_MASK)) x )
be8e71aa 1015#define DEBUG_PARSE_r(x) DEBUG_r( \
dafb2544 1016 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_PARSE)) x )
be8e71aa 1017#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
dafb2544 1018 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE)) x )
be8e71aa 1019#define DEBUG_DUMP_r(x) DEBUG_r( \
dafb2544 1020 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_DUMP)) x )
be8e71aa 1021#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
dafb2544 1022 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_TRIE)) x )
f7819f85 1023#define DEBUG_FLAGS_r(x) DEBUG_r( \
dafb2544 1024 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_FLAGS)) x )
d9a72fcc 1025#define DEBUG_TEST_r(x) DEBUG_r( \
dafb2544 1026 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_COMPILE_TEST)) x )
be8e71aa
YO
1027/* Execute */
1028#define DEBUG_EXECUTE_r(x) DEBUG_r( \
dafb2544 1029 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXECUTE_MASK)) x )
be8e71aa 1030#define DEBUG_INTUIT_r(x) DEBUG_r( \
dafb2544 1031 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT)) x )
be8e71aa 1032#define DEBUG_MATCH_r(x) DEBUG_r( \
dafb2544 1033 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXECUTE_MATCH)) x )
be8e71aa 1034#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
dafb2544 1035 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXECUTE_TRIE)) x )
be8e71aa
YO
1036
1037/* Extra */
1038#define DEBUG_EXTRA_r(x) DEBUG_r( \
dafb2544 1039 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_MASK)) x )
a5ca303d 1040#define DEBUG_OFFSETS_r(x) DEBUG_r( \
dafb2544 1041 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS)) x )
ab3bbdeb 1042#define DEBUG_STATE_r(x) DEBUG_r( \
dafb2544 1043 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_STATE)) x )
24b23f37 1044#define DEBUG_STACK_r(x) DEBUG_r( \
dafb2544 1045 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_STACK)) x )
e7707071 1046#define DEBUG_BUFFERS_r(x) DEBUG_r( \
dafb2544 1047 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS)) x )
e7707071 1048
a5ca303d 1049#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
dafb2544
KW
1050 if (DEBUG_v_TEST || ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
1051 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)))) x )
07be1b83 1052#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
dafb2544 1053 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG)) \
be8e71aa
YO
1054 Perl_warn_nocontext x )
1055#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
dafb2544 1056 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_TRIE)) x )
be8e71aa 1057#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
dafb2544 1058 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_TRIE)) x )
be8e71aa
YO
1059
1060#define DEBUG_TRIE_r(x) DEBUG_r( \
dafb2544
KW
1061 if (DEBUG_v_TEST || (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
1062 | RE_DEBUG_EXECUTE_TRIE ))) x )
2c296965 1063#define DEBUG_GPOS_r(x) DEBUG_r( \
dafb2544 1064 if (DEBUG_v_TEST || (re_debug_flags & RE_DEBUG_EXTRA_GPOS)) x )
be8e71aa
YO
1065
1066/* initialization */
d2d8c711 1067/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
1068#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
1069 SV * re_debug_flags_sv = NULL; \
243a3b9c 1070 re_debug_flags_sv = PL_curcop ? get_sv(RE_DEBUG_FLAGS, GV_ADD) : NULL; \
e68ec53f 1071 if (re_debug_flags_sv) { \
8e9a8a48 1072 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 1073 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 1074 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
1075 }\
1076})
a3621e74
YO
1077
1078#ifdef DEBUGGING
ab3bbdeb 1079
8162b70e 1080#define GET_RE_DEBUG_FLAGS_DECL volatile IV re_debug_flags = 0; \
cb65b687 1081 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
ab3bbdeb 1082
eecd4d11
KW
1083#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
1084 const char * const rpv = \
1085 pv_pretty((dsv), (pv), (l), (m), \
1086 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 1087 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 1088 const int rlen = SvCUR(dsv)
ab3bbdeb 1089
8ec77c3c 1090/* This is currently unsed in the core */
eecd4d11
KW
1091#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
1092 const char * const rpv = \
1093 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
1094 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 1095 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb 1096
eecd4d11
KW
1097#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
1098 const char * const rpv = \
1099 pv_pretty((dsv), (pv), (l), (m), \
1100 PL_colors[0], PL_colors[1], \
c89df6cf 1101 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 1102 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 1103 )
ab3bbdeb
YO
1104
1105#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
1106#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
1107
1108#else /* if not DEBUGGING */
1109
a3621e74 1110#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
1111#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
1112#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
1113#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
1114#define RE_SV_DUMPLEN(ItEm)
1115#define RE_SV_TAIL(ItEm)
1116
1117#endif /* DEBUG RELATED DEFINES */
a3621e74 1118
0bc0e44f 1119#define FIRST_NON_ASCII_DECIMAL_DIGIT 0x660 /* ARABIC_INDIC_DIGIT_ZERO */
2ca74f27 1120
64935bc6
KW
1121typedef enum {
1122 TRADITIONAL_BOUND = _CC_WORDCHAR,
ae3bb8ea 1123 GCB_BOUND,
6b659339 1124 LB_BOUND,
06ae2722 1125 SB_BOUND,
ae3bb8ea 1126 WB_BOUND
64935bc6
KW
1127} bound_type;
1128
3dd7db29
JK
1129#endif /* PERL_REGCOMP_H_ */
1130
e9a8c099 1131/*
14d04a33 1132 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 1133 */