This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regexes: Remove uses of ANYOF_LOCALE flag
[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
a687059c
LW
87 * boundary, and reads the offset directly as a short. Also, there is no
88 * special test to reverse the sign of BACK pointers since the offset is
89 * stored negative.]
90 */
785a26d5
YO
91
92/* This is the stuff that used to live in regexp.h that was truly
93 private to the engine itself. It now lives here. */
94
28d8d7f4 95
785a26d5
YO
96
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 */
3d2bd50a
DM
114 struct reg_code_block *code_blocks;/* positions of literal (?{}) */
115 int num_code_blocks; /* size of code_blocks[] */
6d5c990f
RGS
116 regnode program[1]; /* Unwarranted chumminess with compiler. */
117} regexp_internal;
118
119#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
120#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
121#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
122/*
123 * Flags stored in regexp->intflags
124 * These are used only internally to the regexp engine
125 *
126 * See regexp.h for flags used externally to the regexp engine
127 */
e3e400ec
YO
128#define RXp_INTFLAGS(rx) ((rx)->intflags)
129#define RX_INTFLAGS(prog) RXp_INTFLAGS(ReANY(prog))
130
785a26d5
YO
131#define PREGf_SKIP 0x00000001
132#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
133#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
134#define PREGf_VERBARG_SEEN 0x00000008
135#define PREGf_CUTGROUP_SEEN 0x00000010
732caac7 136#define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
e3e400ec
YO
137/* these used to be extflags, but are now intflags */
138#define PREGf_NOSCAN 0x00000040
58430ea8 139#define PREGf_CANY_SEEN 0x00000080
8e1490ee
YO
140#define PREGf_GPOS_SEEN 0x00000100
141#define PREGf_GPOS_FLOAT 0x00000200
785a26d5 142
8e1490ee
YO
143#define PREGf_ANCH_BOL 0x00000400
144#define PREGf_ANCH_MBOL 0x00000800
145#define PREGf_ANCH_SBOL 0x00001000
146#define PREGf_ANCH_GPOS 0x00002000
147
fec6b98a
DM
148#define PREGf_ANCH (PREGf_ANCH_SBOL | PREGf_ANCH_GPOS | \
149 PREGf_ANCH_MBOL | PREGf_ANCH_BOL )
785a26d5
YO
150
151/* this is where the old regcomp.h started */
a687059c 152
c277df42 153struct regnode_string {
cd439c50 154 U8 str_len;
c277df42
IZ
155 U8 type;
156 U16 next_off;
cd439c50 157 char string[1];
c277df42 158};
a687059c 159
6bda09f9
YO
160/* Argument bearing node - workhorse,
161 arg1 is often for the data field */
c277df42
IZ
162struct regnode_1 {
163 U8 flags;
164 U8 type;
165 U16 next_off;
166 U32 arg1;
167};
168
6bda09f9
YO
169/* Similar to a regnode_1 but with an extra signed argument */
170struct regnode_2L {
171 U8 flags;
172 U8 type;
173 U16 next_off;
174 U32 arg1;
175 I32 arg2;
176};
177
178/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
179struct regnode_2 {
180 U8 flags;
181 U8 type;
182 U16 next_off;
183 U16 arg1;
184 U16 arg2;
185};
186
6bda09f9 187
936ed897 188#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
936ed897 189
786e8c11 190/* also used by trie */
936ed897
IZ
191struct regnode_charclass {
192 U8 flags;
193 U8 type;
194 U16 next_off;
7e817188 195 U32 arg1;
19103b91 196 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
197};
198
bf3f2d85
KW
199/* has runtime (locale) \d, \w, ..., [:posix:] classes */
200struct regnode_charclass_class {
8efd3f97 201 U8 flags; /* ANYOF_POSIXL bit must go here */
936ed897
IZ
202 U8 type;
203 U16 next_off;
7e817188 204 U32 arg1;
19103b91 205 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
9cf78fa0 206 U32 classflags; /* and run-time */
936ed897
IZ
207};
208
31f05a37
KW
209/* like above, but also has folds that are used only if the runtime locale is
210 * UTF-8. */
211struct regnode_charclass_posixl_fold {
212 U8 flags; /* ANYOF_POSIXL bit must go here */
213 U8 type;
214 U16 next_off;
7e817188 215 U32 arg1;
31f05a37
KW
216 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
217 U32 classflags; /* and run-time */
218 SV* utf8_locale_list; /* list of code points matched by folds
219 in a UTF-8 locale */
220};
221
222/* A synthetic start class; is a regnode_charclass_posixl_fold, plus an extra
223 * SV*, used only during its construction and which is not used by regexec.c.
224 * Note that the 'next_off' field is unused, as the SSC stands alone, so there
225 * is never a next node. */
b8f7bb16 226struct regnode_ssc {
8efd3f97 227 U8 flags; /* ANYOF_POSIXL bit must go here */
b8f7bb16
KW
228 U8 type;
229 U16 next_off;
7e817188 230 U32 arg1;
b8f7bb16
KW
231 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
232 U32 classflags; /* and run-time */
31f05a37
KW
233 SV* utf8_locale_list; /* list of code points matched by folds
234 in a UTF-8 locale */
b8f7bb16
KW
235 SV* invlist; /* list of code points matched */
236};
237
71068078
KW
238/* We take advantage of 'next_off' not otherwise being used in the SSC by
239 * actually using it: by setting it to 1. This allows us to test and
240 * distinguish between an SSC and other ANYOF node types, as 'next_off' cannot
241 * otherwise be 1, because it is the offset to the next regnode expressed in
242 * units of regnodes. Since an ANYOF node contains extra fields, it adds up
243 * to 12 regnode units on 32-bit systems, (hence the minimum this can be (if
244 * not 0) is 11 there. Even if things get tightly packed on a 64-bit system,
245 * it still would be more than 1. */
246#define set_ANYOF_SYNTHETIC(n) STMT_START{ OP(n) = ANYOF; \
247 NEXT_OFF(n) = 1; \
248 } STMT_END
249#define is_ANYOF_SYNTHETIC(n) (OP(n) == ANYOF && NEXT_OFF(n) == 1)
250
9c7e81e8
AD
251/* XXX fix this description.
252 Impose a limit of REG_INFTY on various pattern matching operations
253 to limit stack growth and to avoid "infinite" recursions.
254*/
255/* The default size for REG_INFTY is I16_MAX, which is the same as
256 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
257 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
258 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
259 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
260 smaller default.
261 --Andy Dougherty 11 June 1998
262*/
263#if SHORTSIZE > 2
264# ifndef REG_INFTY
265# define REG_INFTY ((1<<15)-1)
266# endif
267#endif
268
269#ifndef REG_INFTY
83cfe48c
IZ
270# define REG_INFTY I16_MAX
271#endif
a687059c 272
e91177ed
IZ
273#define ARG_VALUE(arg) (arg)
274#define ARG__SET(arg,val) ((arg) = (val))
c277df42 275
e8e3b0fb
JH
276#undef ARG
277#undef ARG1
278#undef ARG2
279
c277df42
IZ
280#define ARG(p) ARG_VALUE(ARG_LOC(p))
281#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
282#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 283#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 284
c277df42
IZ
285#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
286#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
287#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 288#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 289
e8e3b0fb
JH
290#undef NEXT_OFF
291#undef NODE_ALIGN
292
075abff3
AL
293#define NEXT_OFF(p) ((p)->next_off)
294#define NODE_ALIGN(node)
295#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 296
c277df42
IZ
297#define SIZE_ALIGN NODE_ALIGN
298
e8e3b0fb
JH
299#undef OP
300#undef OPERAND
301#undef MASK
302#undef STRING
303
e91177ed 304#define OP(p) ((p)->type)
b988e673
KW
305#define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
306 regnode types. For some, it's the \
307 character set of the regnode */
e91177ed 308#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
309#define MASK(p) ((char*)OPERAND(p))
310#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
311#define STRING(p) (((struct regnode_string *)p)->string)
312#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
313#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
314
e8e3b0fb
JH
315#undef NODE_ALIGN
316#undef ARG_LOC
317#undef NEXTOPER
318#undef PREVOPER
319
e91177ed
IZ
320#define NODE_ALIGN(node)
321#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
322#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
323#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 324#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 325
e91177ed
IZ
326#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
327#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
328
329#define NODE_STEP_B 4
c277df42
IZ
330
331#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
332#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
333
e91177ed 334#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 335 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 336#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 337 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 338
22c35a8c 339#define REG_MAGIC 0234
a687059c 340
9a81a976 341#define SIZE_ONLY (RExC_emit == (regnode *) & RExC_emit_dummy)
6d15ebf3
KW
342#define PASS1 SIZE_ONLY
343#define PASS2 (! SIZE_ONLY)
c277df42 344
137165a6
KW
345/* If the bitmap doesn't fully represent what this ANYOF node can match, the
346 * ARG is set to this special value (since 0, 1, ... are legal, but will never
347 * reach this high). */
348#define ANYOF_NONBITMAP_EMPTY ((U32) -1)
349
350/* The information used to be stored as as combination of the ANYOF_UTF8 and
351 * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there
352 * to free up a bit for other uses. This tries to hide the change from
353 * existing code as much as possible. Now, the data structure that goes in ARG
354 * is not allocated unless it is needed, and that is what is used to determine
355 * if there is something outside the bitmap. The code now assumes that if
356 * that structure exists, that any UTF-8 encoded string should be tried against
357 * it, but a non-UTF8-encoded string will be tried only if the
358 * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */
359#define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY)
360
34fdef84
KW
361/* Flags for node->flags of ANYOF. These are in short supply, with one
362 * currently available. If more are needed, the ABOVE_LATIN1_ALL bit could be
363 * freed up by resorting to creating a swash containing everything above 255.
7bc66b18
KW
364 * This introduces a performance penalty.
365 * shared should new flags be needed for SSCs, like ANYOF_EMPTY_STRING now. */
366
367/* Flags for node->flags of ANYOF. These are in short supply, with none
368 * currently available. The easiest solution for one more flag is to eliminate
369 * the ANYOF_LOCALE flag, because it doesn't add any extra information beyond
370 * the other two LOC flags. Also, the ABOVE_LATIN1_ALL bit could be freed up
371 * by resorting to creating a swash containing everything above 255. This
372 * introduces a performance penalty. An option that wouldn't slow things down
373 * would be to split one of the two remaining LOC flags out into a separate
374 * node, like what was done with ANYOF_NON_UTF8_NON_ASCII_ALL in commit
375 * 34fdef848b1687b91892ba55e9e0c3430e0770f6 (but which was reverted because it
376 * wasn't the best option available at the time), and using a LOC flag is
377 * probably better than that commit anyway. But it could be reinstated if we
378 * need a bit. The LOC flags are only for /l nodes; the reverted commit was
379 * only for /d, so there are no combinatorial issues.
34fdef84
KW
380 * Several flags are not used in synthetic start class (SSC) nodes, so could be
381 * shared should new flags be needed for SSCs, like ANYOF_EMPTY_STRING now. */
b8c5462f 382
5dbb0c08
KW
383/* regexec.c is expecting this to be in the low bit */
384#define ANYOF_INVERT 0x01
1984f6a0 385
a0dd4231
KW
386/* For the SSC node only, which cannot be inverted, so is shared with that bit.
387 * This means "Does this SSC match an empty string?" This is used only during
388 * regex compilation. */
389#define ANYOF_EMPTY_STRING ANYOF_INVERT
390
b63f73da
KW
391#define ANYOF_LOCALE 0x02 /* /l modifier */
392
393/* The fold is calculated and stored in the bitmap where possible at compile
394 * time. However under locale, the actual folding varies depending on
395 * what the locale is at the time of execution, so it has to be deferred until
396 * then */
397#define ANYOF_LOC_FOLD 0x04
398
d1d040e5 399/* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This
dd58aee1
KW
400 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
401 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
9ffd0d41
KW
402 * at compile-time. Could be shared with ANYOF_LOCALE, forcing all locale
403 * nodes to be large */
8efd3f97
KW
404#define ANYOF_POSIXL 0x08
405#define ANYOF_CLASS ANYOF_POSIXL
406#define ANYOF_LARGE ANYOF_POSIXL
c8d3cd88 407
ae986089
KW
408/* Should we raise a warning if matching against an above-Unicode code point?
409 * */
410#define ANYOF_WARN_SUPER 0x10
f4244008 411
ef87b810
KW
412/* Can match something outside the bitmap that isn't in utf8 */
413#define ANYOF_NONBITMAP_NON_UTF8 0x20
414
61ef9872 415/* Matches every code point 0x100 and above*/
9b1eb067
KW
416#define ANYOF_ABOVE_LATIN1_ALL 0x40
417#define ANYOF_UNICODE_ALL ANYOF_ABOVE_LATIN1_ALL
ffc61ed2 418
7bc66b18
KW
419/* Match all Latin1 characters that aren't ASCII when the target string is not
420 * in utf8. */
421#define ANYOF_NON_UTF8_NON_ASCII_ALL 0x80
422
423#define ANYOF_FLAGS_ALL (0xff)
653099ff 424
1462525b 425#define ANYOF_LOCALE_FLAGS (ANYOF_LOC_FOLD | ANYOF_POSIXL)
ace6b0e4 426
eff8b7dc
KW
427/* These are the flags that apply to both regular ANYOF nodes and synthetic
428 * start class nodes during construction of the SSC. During finalization of
429 * the SSC, other of the flags could be added to it */
1462525b 430#define ANYOF_COMMON_FLAGS (ANYOF_WARN_SUPER)
eff8b7dc 431
936ed897 432/* Character classes for node->classflags of ANYOF */
653099ff 433/* Should be synchronized with a table in regprop() */
575a8a29 434/* 2n should be the normal one, paired with its complement at 2n+1 */
b8c5462f 435
08ada29e
KW
436#define ANYOF_ALPHA ((_CC_ALPHA) * 2)
437#define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
e8d596e0
KW
438#define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
439#define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
08ada29e
KW
440#define ANYOF_ASCII ((_CC_ASCII) * 2)
441#define ANYOF_NASCII ((ANYOF_ASCII) + 1)
e8d596e0
KW
442#define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
443#define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
b0d691b2
KW
444#define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
445 [:upper:] under /i */
446#define ANYOF_NCASED ((ANYOF_CASED) + 1)
08ada29e
KW
447#define ANYOF_CNTRL ((_CC_CNTRL) * 2)
448#define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
e8d596e0
KW
449#define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
450#define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
08ada29e
KW
451#define ANYOF_GRAPH ((_CC_GRAPH) * 2)
452#define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
453#define ANYOF_LOWER ((_CC_LOWER) * 2)
454#define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
455#define ANYOF_PRINT ((_CC_PRINT) * 2)
456#define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
e8d596e0
KW
457#define ANYOF_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */
458#define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1)
08ada29e
KW
459#define ANYOF_PUNCT ((_CC_PUNCT) * 2)
460#define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
e8d596e0
KW
461#define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
462#define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
08ada29e
KW
463#define ANYOF_UPPER ((_CC_UPPER) * 2)
464#define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
e8d596e0
KW
465#define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
466#define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
08ada29e
KW
467#define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
468#define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
b8c5462f 469
87cad4c0 470/* pseudo classes below this, not stored in the class bitmap, but used as flags
e1d1eefb
YO
471 during compilation of char classes */
472
b2ef2e34
KW
473#define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
474#define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
e1d1eefb 475
b2ef2e34
KW
476/* It is best if this is the last one, as all above it are stored as bits in a
477 * bitmap, and it isn't part of that bitmap */
478#if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
479# error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
480#endif
481
8efd3f97
KW
482#define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
483 * '< ANYOF_MAX'
484 * Hence doesn't include VERTWS, as that
485 * is a pseudo class */
486#define ANYOF_MAX ANYOF_POSIXL_MAX
487
488#if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */
b2ef2e34 489# error Problem with handy.h _CC_foo #defines
a0947d7b
KW
490#endif
491
8efd3f97
KW
492#define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
493#define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3)
87cad4c0 494
8efd3f97
KW
495#define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode
496 property: \p{} or \P{} */
f39a4dc2 497
b8c5462f
JH
498/* Backward source code compatibility. */
499
500#define ANYOF_ALNUML ANYOF_ALNUM
501#define ANYOF_NALNUML ANYOF_NALNUM
502#define ANYOF_SPACEL ANYOF_SPACE
503#define ANYOF_NSPACEL ANYOF_NSPACE
8ebfc29a
KW
504#define ANYOF_ALNUM ANYOF_WORDCHAR
505#define ANYOF_NALNUM ANYOF_NWORDCHAR
b8c5462f
JH
506
507/* Utility macros for the bitmap and classes of ANYOF */
508
936ed897 509#define ANYOF_SIZE (sizeof(struct regnode_charclass))
d1d040e5 510#define ANYOF_POSIXL_SIZE (sizeof(regnode_charclass_posixl))
8efd3f97 511#define ANYOF_CLASS_SIZE ANYOF_POSIXL_SIZE
31f05a37 512#define ANYOF_POSIXL_FOLD_SIZE (sizeof(regnode_charclass_posixl_fold))
b8c5462f 513
936ed897 514#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
515
516#define ANYOF_BIT(c) (1 << ((c) & 7))
517
d1d040e5 518#define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c)))
8efd3f97
KW
519#define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c))
520
d1d040e5 521#define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c)))
8efd3f97
KW
522#define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c))
523
d1d040e5 524#define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c)))
8efd3f97 525#define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c))
b8c5462f 526
d1d040e5 527#define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END
8efd3f97 528#define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret)
9cf78fa0
KW
529
530/* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
d1d040e5 531#define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END
8efd3f97 532#define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret)
9cf78fa0 533
8efd3f97
KW
534#define ANYOF_POSIXL_TEST_ANY_SET(p) \
535 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
d1d040e5 536 && (((regnode_charclass_posixl*)(p))->classflags))
8efd3f97 537#define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p)
e867ce45 538
e0e1be5f
KW
539/* Since an SSC always has this field, we don't have to test for that; nor do
540 * we want to because the bit isn't set for SSC during its construction */
541#define ANYOF_POSIXL_SSC_TEST_ANY_SET(p) \
542 cBOOL(((regnode_ssc*)(p))->classflags)
543#define ANYOF_POSIXL_SSC_TEST_ALL_SET(p) /* Are all bits set? */ \
544 (((regnode_ssc*) (p))->classflags \
545 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
546
547#define ANYOF_POSIXL_TEST_ALL_SET(p) \
548 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
549 && ((regnode_charclass_posixl*) (p))->classflags \
550 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
fb38762f 551
b27fd97a 552#define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END
8efd3f97 553#define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest))
936ed897 554
cdd87c1d
KW
555#define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END
556
9cf78fa0 557#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
936ed897 558#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 559#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
560#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
561#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
562#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
563
f8bef550
NC
564#define ANYOF_BITMAP_SETALL(p) \
565 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
566#define ANYOF_BITMAP_CLEARALL(p) \
567 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
568/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
c355c09a 569#define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
f8bef550
NC
570 memEQ (ANYOF_BITMAP(p), "\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377\377", ANYOF_BITMAP_SIZE)
571
936ed897 572#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
8efd3f97 573#define ANYOF_POSIXL_SKIP ((ANYOF_POSIXL_SIZE - 1)/sizeof(regnode))
31f05a37 574#define ANYOF_POSIXL_FOLD_SKIP ((ANYOF_POSIXL_FOLD_SIZE - 1)/sizeof(regnode))
8efd3f97 575#define ANYOF_CLASS_SKIP ANYOF_POSIXL_SKIP
3a15e693 576
31f05a37
KW
577#define ANYOF_UTF8_LOCALE_INVLIST(node) (((regnode_charclass_posixl_fold*) (node))->utf8_locale_list)
578
a687059c
LW
579/*
580 * Utility definitions.
581 */
2304df62 582#ifndef CHARMASK
075abff3 583# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 584#else
075abff3 585# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 586#endif
a687059c 587
c277df42
IZ
588#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
589
e384d5c1
YO
590#define REG_ZERO_LEN_SEEN 0x00000001
591#define REG_LOOKBEHIND_SEEN 0x00000002
592#define REG_GPOS_SEEN 0x00000004
471f5387 593/* spare */
e384d5c1
YO
594#define REG_CANY_SEEN 0x00000010
595#define REG_RECURSE_SEEN 0x00000020
596#define REG_TOP_LEVEL_BRANCHES_SEEN 0x00000040
597#define REG_VERBARG_SEEN 0x00000080
598#define REG_CUTGROUP_SEEN 0x00000100
599#define REG_RUN_ON_COMMENT_SEEN 0x00000200
600#define REG_UNFOLDED_MULTI_SEEN 0x00000400
601#define REG_GOSTART_SEEN 0x00000800
ee273784
YO
602#define REG_UNBOUNDED_QUANTIFIER_SEEN 0x00001000
603
d09b2d29 604
73c4f7a1
GS
605START_EXTERN_C
606
be8e71aa
YO
607#ifdef PLUGGABLE_RE_EXTENSION
608#include "re_nodes.h"
609#else
d09b2d29 610#include "regnodes.h"
be8e71aa 611#endif
d09b2d29 612
f9f4320a
YO
613#ifndef PLUGGABLE_RE_EXTENSION
614#ifndef DOINIT
615EXTCONST regexp_engine PL_core_reg_engine;
616#else /* DOINIT */
617EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 618 Perl_re_compile,
2fdbfb4d 619 Perl_regexec_flags,
f9f4320a
YO
620 Perl_re_intuit_start,
621 Perl_re_intuit_string,
2fdbfb4d
AB
622 Perl_regfree_internal,
623 Perl_reg_numbered_buff_fetch,
624 Perl_reg_numbered_buff_store,
625 Perl_reg_numbered_buff_length,
192b9cd1
AB
626 Perl_reg_named_buff,
627 Perl_reg_named_buff_iter,
49d7dfbc 628 Perl_reg_qr_package,
f9f4320a 629#if defined(USE_ITHREADS)
3c13cae6 630 Perl_regdupe_internal,
f9f4320a 631#endif
3c13cae6 632 Perl_re_op_compile
f9f4320a
YO
633};
634#endif /* DOINIT */
635#endif /* PLUGGABLE_RE_EXTENSION */
636
637
73c4f7a1 638END_EXTERN_C
cad2e5aa 639
cad2e5aa 640
261faec3
MJD
641/* .what is a character array with one character for each member of .data
642 * The character describes the function of the corresponding .data item:
af534a04 643 * a - AV for paren_name_list under DEBUGGING
0111c4fd 644 * f - start-class data for regstclass optimization
68e2671b 645 * l - start op for literal (?{EVAL}) item
d63c20f2 646 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
b30fcab9 647 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
38a44b82 648 * s - swash for Unicode-style character class, and the multicharacter
9e55ce06
JH
649 * strings resulting from casefolding the single-character entries
650 * in the character class
a3621e74 651 * t - trie struct
55eed653 652 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 653 * also used for revcharmap and words under DEBUGGING
07be1b83 654 * T - aho-trie struct
81714fb9 655 * S - sv for named capture lookup
261faec3 656 * 20010712 mjd@plover.com
c3613ed7 657 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 658 */
cad2e5aa
JH
659struct reg_data {
660 U32 count;
661 U8 *what;
662 void* data[1];
663};
664
a1cac82e
NC
665/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
666 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 667#define anchored_substr substrs->data[0].substr
33b8afdf 668#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 669#define anchored_offset substrs->data[0].min_offset
1de06328
YO
670#define anchored_end_shift substrs->data[0].end_shift
671
cad2e5aa 672#define float_substr substrs->data[1].substr
33b8afdf 673#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
674#define float_min_offset substrs->data[1].min_offset
675#define float_max_offset substrs->data[1].max_offset
1de06328
YO
676#define float_end_shift substrs->data[1].end_shift
677
cad2e5aa 678#define check_substr substrs->data[2].substr
33b8afdf 679#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
680#define check_offset_min substrs->data[2].min_offset
681#define check_offset_max substrs->data[2].max_offset
1de06328 682#define check_end_shift substrs->data[2].end_shift
a3621e74 683
8d919b0a
FC
684#define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
685#define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
686#define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
687#define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
a3621e74
YO
688
689/* trie related stuff */
5d9a96ca 690
a3621e74
YO
691/* a transition record for the state machine. the
692 check field determines which state "owns" the
693 transition. the char the transition is for is
694 determined by offset from the owning states base
695 field. the next field determines which state
696 is to be transitioned to if any.
697*/
698struct _reg_trie_trans {
699 U32 next;
700 U32 check;
701};
702
703/* a transition list element for the list based representation */
704struct _reg_trie_trans_list_elem {
705 U16 forid;
706 U32 newstate;
707};
708typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
709
710/* a state for compressed nodes. base is an offset
711 into an array of reg_trie_trans array. If wordnum is
712 nonzero the state is accepting. if base is zero then
713 the state has no children (and will be accepting)
714*/
715struct _reg_trie_state {
716 U16 wordnum;
717 union {
718 U32 base;
719 reg_trie_trans_le* list;
720 } trans;
721};
722
2e64971a
DM
723/* info per word; indexed by wordnum */
724typedef struct {
725 U16 prev; /* previous word in acceptance chain; eg in
726 * zzz|abc|ab/ after matching the chars abc, the
727 * accepted word is #2, and the previous accepted
728 * word is #3 */
729 U32 len; /* how many chars long is this word? */
730 U32 accept; /* accept state for this word */
731} reg_trie_wordinfo;
a3621e74
YO
732
733
a3621e74
YO
734typedef struct _reg_trie_state reg_trie_state;
735typedef struct _reg_trie_trans reg_trie_trans;
736
737
738/* anything in here that needs to be freed later
23eab42c
NC
739 should be dealt with in pregfree.
740 refcount is first in both this and _reg_ac_data to allow a space
741 optimisation in Perl_regdupe. */
a3621e74 742struct _reg_trie_data {
23eab42c 743 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
744 U32 lasttrans; /* last valid transition element */
745 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
746 reg_trie_state *states; /* state data */
747 reg_trie_trans *trans; /* array of transition elements */
748 char *bitmap; /* stclass bitmap */
786e8c11
YO
749 U16 *jump; /* optional 1 indexed array of offsets before tail
750 for the node following a given word. */
2e64971a 751 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
752 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
753 U32 startstate; /* initial state - used for common prefix optimisation */
754 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
755 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 756 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
757 U32 statecount; /* Build only - number of states in the states array
758 (including the unused zero state) */
786e8c11 759 U32 wordcount; /* Build only */
a3621e74 760#ifdef DEBUGGING
786e8c11 761 STRLEN charcount; /* Build only */
a3621e74
YO
762#endif
763};
2b8b4781
NC
764/* There is one (3 under DEBUGGING) pointers that logically belong in this
765 structure, but are held outside as they need duplication on thread cloning,
766 whereas the rest of the structure can be read only:
767 HV *widecharmap; code points > 255 to charid
768#ifdef DEBUGGING
769 AV *words; Array of words contained in trie, for dumping
770 AV *revcharmap; Map of each charid back to its character representation
771#endif
772*/
773
774#define TRIE_WORDS_OFFSET 2
775
a3621e74
YO
776typedef struct _reg_trie_data reg_trie_data;
777
23eab42c
NC
778/* refcount is first in both this and _reg_trie_data to allow a space
779 optimisation in Perl_regdupe. */
07be1b83 780struct _reg_ac_data {
23eab42c 781 U32 refcount;
1b5c067c 782 U32 trie;
07be1b83
YO
783 U32 *fail;
784 reg_trie_state *states;
07be1b83
YO
785};
786typedef struct _reg_ac_data reg_ac_data;
787
486ec47a 788/* ANY_BIT doesn't use the structure, so we can borrow it here.
3dab1dad
YO
789 This is simpler than refactoring all of it as wed end up with
790 three different sets... */
791
792#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 793#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
794#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
795#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
796#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 797
1de06328
YO
798#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
799#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
800
801
786e8c11
YO
802#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
803#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 804
a3621e74
YO
805/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
806#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
807#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
808#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
809
3dab1dad 810#ifdef DEBUGGING
3dab1dad 811#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 812#else
3dab1dad 813#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 814#endif
a3621e74 815
0111c4fd
RGS
816#define RE_TRIE_MAXBUF_INIT 65536
817#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
818#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
819
be8e71aa
YO
820/*
821
822RE_DEBUG_FLAGS is used to control what debug output is emitted
823its divided into three groups of options, some of which interact.
824The three groups are: Compile, Execute, Extra. There is room for a
825further group, as currently only the low three bytes are used.
826
827 Compile Options:
07be1b83 828
be8e71aa
YO
829 PARSE
830 PEEP
831 TRIE
832 PROGRAM
833 OFFSETS
a3621e74 834
be8e71aa 835 Execute Options:
a3621e74 836
be8e71aa
YO
837 INTUIT
838 MATCH
839 TRIE
a3621e74 840
be8e71aa 841 Extra Options
a3621e74 842
be8e71aa
YO
843 TRIE
844 OFFSETS
845
846If you modify any of these make sure you make corresponding changes to
847re.pm, especially to the documentation.
a3621e74 848
be8e71aa
YO
849*/
850
851
852/* Compile */
853#define RE_DEBUG_COMPILE_MASK 0x0000FF
854#define RE_DEBUG_COMPILE_PARSE 0x000001
855#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
856#define RE_DEBUG_COMPILE_TRIE 0x000004
857#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 858#define RE_DEBUG_COMPILE_FLAGS 0x000010
be8e71aa
YO
859
860/* Execute */
861#define RE_DEBUG_EXECUTE_MASK 0x00FF00
862#define RE_DEBUG_EXECUTE_INTUIT 0x000100
863#define RE_DEBUG_EXECUTE_MATCH 0x000200
864#define RE_DEBUG_EXECUTE_TRIE 0x000400
865
866/* Extra */
867#define RE_DEBUG_EXTRA_MASK 0xFF0000
868#define RE_DEBUG_EXTRA_TRIE 0x010000
869#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
870#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
871#define RE_DEBUG_EXTRA_STATE 0x080000
872#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 873#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 874#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
875/* combined */
876#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 877
e68ec53f 878#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
879/* Compile */
880#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 881 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 882#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 883 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 884#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 885 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 886#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 887 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 888#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 889 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 890#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 891 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
f7819f85
A
892#define DEBUG_FLAGS_r(x) DEBUG_r( \
893 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
be8e71aa
YO
894/* Execute */
895#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 896 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 897#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 898 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 899#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 900 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 901#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 902 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
903
904/* Extra */
905#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 906 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
907#define DEBUG_OFFSETS_r(x) DEBUG_r( \
908 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
909#define DEBUG_STATE_r(x) DEBUG_r( \
910 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
911#define DEBUG_STACK_r(x) DEBUG_r( \
912 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
913#define DEBUG_BUFFERS_r(x) DEBUG_r( \
914 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
915
a5ca303d
YO
916#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
917 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
918 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 919#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 920 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
921 Perl_warn_nocontext x )
922#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 923 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 924#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 925 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
926
927#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 928 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa 929 | RE_DEBUG_EXECUTE_TRIE )) x )
2c296965
YO
930#define DEBUG_GPOS_r(x) DEBUG_r( \
931 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
be8e71aa
YO
932
933/* initialization */
d2d8c711 934/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
935#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
936 SV * re_debug_flags_sv = NULL; \
937 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
938 if (re_debug_flags_sv) { \
8e9a8a48 939 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 940 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 941 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
942 }\
943})
a3621e74
YO
944
945#ifdef DEBUGGING
ab3bbdeb 946
cb65b687
DM
947#define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
948 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
949
950#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
951 const char * const rpv = \
ddc5bc0f 952 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 953 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 954 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 955 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
956
957#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
958 const char * const rpv = \
ddc5bc0f 959 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 960 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 961 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
962
963#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
964 const char * const rpv = \
ddc5bc0f 965 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 966 PL_colors[0], PL_colors[1], \
c89df6cf 967 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 968 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 969 )
ab3bbdeb
YO
970
971#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
972#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
973
974#else /* if not DEBUGGING */
975
a3621e74 976#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
977#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
978#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
979#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
980#define RE_SV_DUMPLEN(ItEm)
981#define RE_SV_TAIL(ItEm)
982
983#endif /* DEBUG RELATED DEFINES */
a3621e74 984
e9a8c099
MHM
985/*
986 * Local variables:
987 * c-indentation-style: bsd
988 * c-basic-offset: 4
14d04a33 989 * indent-tabs-mode: nil
e9a8c099
MHM
990 * End:
991 *
14d04a33 992 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 993 */