This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regen/regcharclass.pl: Don't generate unused macros
[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
4c1877fc 188#define ANYOF_BITMAP_SIZE (256 / 8) /* 8 bits/Byte */
936ed897 189
1ee208c4
KW
190/* Note that these form structs which are supersets of the next smaller one, by
191 * appending fields. Alignment problems can occur if one of those optional
192 * fields requires stricter alignment than the base struct. And formal
193 * parameters that can really be two or more of the structs should be
194 * declared as the smallest one it could be. See commit message for
195 * 7dcac5f6a5195002b55c935ee1d67f67e1df280b. Regnode allocation is done
196 * without regard to alignment, and changing it to would also require changing
197 * the code that inserts and deletes regnodes. The basic single-argument
198 * regnode has a U32, which is what reganode() allocates as a unit. Therefore
199 * no field can require stricter alignment than U32. */
200
786e8c11 201/* also used by trie */
936ed897
IZ
202struct regnode_charclass {
203 U8 flags;
204 U8 type;
205 U16 next_off;
7e817188 206 U32 arg1;
19103b91 207 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
208};
209
bf3f2d85
KW
210/* has runtime (locale) \d, \w, ..., [:posix:] classes */
211struct regnode_charclass_class {
8efd3f97 212 U8 flags; /* ANYOF_POSIXL bit must go here */
936ed897
IZ
213 U8 type;
214 U16 next_off;
7e817188 215 U32 arg1;
19103b91 216 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
9cf78fa0 217 U32 classflags; /* and run-time */
936ed897
IZ
218};
219
31f05a37
KW
220/* A synthetic start class; is a regnode_charclass_posixl_fold, plus an extra
221 * SV*, used only during its construction and which is not used by regexec.c.
222 * Note that the 'next_off' field is unused, as the SSC stands alone, so there
1ee208c4
KW
223 * is never a next node. Also, there is no alignment issue, becase these are
224 * declared or allocated as a complete unit so the compiler takes care of
225 * alignment. This is unlike ithe other regnodes which are allocated in terms
226 * of multiples of a single-argument regnode. Because there is no alignment
227 * issue, these can have a pointer field */
b8f7bb16 228struct regnode_ssc {
8efd3f97 229 U8 flags; /* ANYOF_POSIXL bit must go here */
b8f7bb16
KW
230 U8 type;
231 U16 next_off;
7e817188 232 U32 arg1;
b8f7bb16
KW
233 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
234 U32 classflags; /* and run-time */
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
3b04b210 350/* Flags for node->flags of ANYOF. These are in short supply, with none
8f536e26 351 * currently available. The ABOVE_LATIN1_ALL bit could be freed up
7bc66b18
KW
352 * by resorting to creating a swash containing everything above 255. This
353 * introduces a performance penalty. An option that wouldn't slow things down
8f536e26 354 * would be to split one of the two LOC flags out into a separate
7bc66b18
KW
355 * node, like what was done with ANYOF_NON_UTF8_NON_ASCII_ALL in commit
356 * 34fdef848b1687b91892ba55e9e0c3430e0770f6 (but which was reverted because it
357 * wasn't the best option available at the time), and using a LOC flag is
358 * probably better than that commit anyway. But it could be reinstated if we
359 * need a bit. The LOC flags are only for /l nodes; the reverted commit was
8f536e26
KW
360 * only for /d, so there are no combinatorial issues. The LOC flag to use is
361 * probably the POSIXL one.
34fdef84
KW
362 * Several flags are not used in synthetic start class (SSC) nodes, so could be
363 * shared should new flags be needed for SSCs, like ANYOF_EMPTY_STRING now. */
b8c5462f 364
5dbb0c08
KW
365/* regexec.c is expecting this to be in the low bit */
366#define ANYOF_INVERT 0x01
1984f6a0 367
a0dd4231
KW
368/* For the SSC node only, which cannot be inverted, so is shared with that bit.
369 * This means "Does this SSC match an empty string?" This is used only during
370 * regex compilation. */
371#define ANYOF_EMPTY_STRING ANYOF_INVERT
372
3b04b210
KW
373/* Are there things that will match only if the target string is encoded in
374 * UTF-8? (This is not set if ANYOF_AOVE_LATIN1_ALL is set) */
375#define ANYOF_UTF8 0x02
b63f73da
KW
376
377/* The fold is calculated and stored in the bitmap where possible at compile
378 * time. However under locale, the actual folding varies depending on
379 * what the locale is at the time of execution, so it has to be deferred until
380 * then */
381#define ANYOF_LOC_FOLD 0x04
382
d1d040e5 383/* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This
dd58aee1
KW
384 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
385 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
8f536e26 386 * at compile-time */
8efd3f97
KW
387#define ANYOF_POSIXL 0x08
388#define ANYOF_CLASS ANYOF_POSIXL
389#define ANYOF_LARGE ANYOF_POSIXL
c8d3cd88 390
ae986089
KW
391/* Should we raise a warning if matching against an above-Unicode code point?
392 * */
393#define ANYOF_WARN_SUPER 0x10
f4244008 394
ef87b810
KW
395/* Can match something outside the bitmap that isn't in utf8 */
396#define ANYOF_NONBITMAP_NON_UTF8 0x20
397
61ef9872 398/* Matches every code point 0x100 and above*/
9b1eb067
KW
399#define ANYOF_ABOVE_LATIN1_ALL 0x40
400#define ANYOF_UNICODE_ALL ANYOF_ABOVE_LATIN1_ALL
ffc61ed2 401
7bc66b18
KW
402/* Match all Latin1 characters that aren't ASCII when the target string is not
403 * in utf8. */
404#define ANYOF_NON_UTF8_NON_ASCII_ALL 0x80
405
3b04b210 406#define ANYOF_FLAGS_ALL (0xff)
653099ff 407
1462525b 408#define ANYOF_LOCALE_FLAGS (ANYOF_LOC_FOLD | ANYOF_POSIXL)
ace6b0e4 409
eff8b7dc
KW
410/* These are the flags that apply to both regular ANYOF nodes and synthetic
411 * start class nodes during construction of the SSC. During finalization of
412 * the SSC, other of the flags could be added to it */
3b04b210 413#define ANYOF_COMMON_FLAGS (ANYOF_WARN_SUPER|ANYOF_UTF8)
eff8b7dc 414
936ed897 415/* Character classes for node->classflags of ANYOF */
653099ff 416/* Should be synchronized with a table in regprop() */
575a8a29 417/* 2n should be the normal one, paired with its complement at 2n+1 */
b8c5462f 418
08ada29e
KW
419#define ANYOF_ALPHA ((_CC_ALPHA) * 2)
420#define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
e8d596e0
KW
421#define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
422#define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
08ada29e
KW
423#define ANYOF_ASCII ((_CC_ASCII) * 2)
424#define ANYOF_NASCII ((ANYOF_ASCII) + 1)
e8d596e0
KW
425#define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
426#define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
b0d691b2
KW
427#define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
428 [:upper:] under /i */
429#define ANYOF_NCASED ((ANYOF_CASED) + 1)
08ada29e
KW
430#define ANYOF_CNTRL ((_CC_CNTRL) * 2)
431#define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
e8d596e0
KW
432#define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
433#define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
08ada29e
KW
434#define ANYOF_GRAPH ((_CC_GRAPH) * 2)
435#define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
436#define ANYOF_LOWER ((_CC_LOWER) * 2)
437#define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
438#define ANYOF_PRINT ((_CC_PRINT) * 2)
439#define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
e8d596e0
KW
440#define ANYOF_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */
441#define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1)
08ada29e
KW
442#define ANYOF_PUNCT ((_CC_PUNCT) * 2)
443#define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
e8d596e0
KW
444#define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
445#define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
08ada29e
KW
446#define ANYOF_UPPER ((_CC_UPPER) * 2)
447#define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
e8d596e0
KW
448#define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
449#define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
08ada29e
KW
450#define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
451#define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
b8c5462f 452
87cad4c0 453/* pseudo classes below this, not stored in the class bitmap, but used as flags
e1d1eefb
YO
454 during compilation of char classes */
455
b2ef2e34
KW
456#define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
457#define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
e1d1eefb 458
b2ef2e34
KW
459/* It is best if this is the last one, as all above it are stored as bits in a
460 * bitmap, and it isn't part of that bitmap */
461#if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
462# error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
463#endif
464
8efd3f97
KW
465#define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
466 * '< ANYOF_MAX'
467 * Hence doesn't include VERTWS, as that
468 * is a pseudo class */
469#define ANYOF_MAX ANYOF_POSIXL_MAX
470
471#if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */
b2ef2e34 472# error Problem with handy.h _CC_foo #defines
a0947d7b
KW
473#endif
474
8efd3f97
KW
475#define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
476#define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3)
87cad4c0 477
8efd3f97
KW
478#define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode
479 property: \p{} or \P{} */
f39a4dc2 480
b8c5462f
JH
481/* Backward source code compatibility. */
482
483#define ANYOF_ALNUML ANYOF_ALNUM
484#define ANYOF_NALNUML ANYOF_NALNUM
485#define ANYOF_SPACEL ANYOF_SPACE
486#define ANYOF_NSPACEL ANYOF_NSPACE
8ebfc29a
KW
487#define ANYOF_ALNUM ANYOF_WORDCHAR
488#define ANYOF_NALNUM ANYOF_NWORDCHAR
b8c5462f
JH
489
490/* Utility macros for the bitmap and classes of ANYOF */
491
936ed897 492#define ANYOF_SIZE (sizeof(struct regnode_charclass))
d1d040e5 493#define ANYOF_POSIXL_SIZE (sizeof(regnode_charclass_posixl))
8efd3f97 494#define ANYOF_CLASS_SIZE ANYOF_POSIXL_SIZE
b8c5462f 495
936ed897 496#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
497
498#define ANYOF_BIT(c) (1 << ((c) & 7))
499
d1d040e5 500#define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c)))
8efd3f97
KW
501#define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c))
502
d1d040e5 503#define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c)))
8efd3f97
KW
504#define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c))
505
d1d040e5 506#define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c)))
8efd3f97 507#define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c))
b8c5462f 508
d1d040e5 509#define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END
8efd3f97 510#define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret)
9cf78fa0
KW
511
512/* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
d1d040e5 513#define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END
8efd3f97 514#define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret)
9cf78fa0 515
8efd3f97
KW
516#define ANYOF_POSIXL_TEST_ANY_SET(p) \
517 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
d1d040e5 518 && (((regnode_charclass_posixl*)(p))->classflags))
8efd3f97 519#define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p)
e867ce45 520
e0e1be5f
KW
521/* Since an SSC always has this field, we don't have to test for that; nor do
522 * we want to because the bit isn't set for SSC during its construction */
523#define ANYOF_POSIXL_SSC_TEST_ANY_SET(p) \
524 cBOOL(((regnode_ssc*)(p))->classflags)
525#define ANYOF_POSIXL_SSC_TEST_ALL_SET(p) /* Are all bits set? */ \
526 (((regnode_ssc*) (p))->classflags \
527 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
528
529#define ANYOF_POSIXL_TEST_ALL_SET(p) \
530 ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \
531 && ((regnode_charclass_posixl*) (p))->classflags \
532 == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1)
fb38762f 533
b27fd97a 534#define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END
8efd3f97 535#define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest))
936ed897 536
cdd87c1d
KW
537#define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END
538
9cf78fa0 539#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
936ed897 540#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 541#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
542#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
543#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
544#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
545
f8bef550
NC
546#define ANYOF_BITMAP_SETALL(p) \
547 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
548#define ANYOF_BITMAP_CLEARALL(p) \
549 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
550/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
c355c09a 551#define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
f8bef550
NC
552 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)
553
936ed897 554#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
8efd3f97
KW
555#define ANYOF_POSIXL_SKIP ((ANYOF_POSIXL_SIZE - 1)/sizeof(regnode))
556#define ANYOF_CLASS_SKIP ANYOF_POSIXL_SKIP
3a15e693 557
a687059c
LW
558/*
559 * Utility definitions.
560 */
2304df62 561#ifndef CHARMASK
075abff3 562# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 563#else
075abff3 564# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 565#endif
a687059c 566
c277df42
IZ
567#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
568
e384d5c1
YO
569#define REG_ZERO_LEN_SEEN 0x00000001
570#define REG_LOOKBEHIND_SEEN 0x00000002
571#define REG_GPOS_SEEN 0x00000004
471f5387 572/* spare */
e384d5c1
YO
573#define REG_CANY_SEEN 0x00000010
574#define REG_RECURSE_SEEN 0x00000020
575#define REG_TOP_LEVEL_BRANCHES_SEEN 0x00000040
576#define REG_VERBARG_SEEN 0x00000080
577#define REG_CUTGROUP_SEEN 0x00000100
578#define REG_RUN_ON_COMMENT_SEEN 0x00000200
579#define REG_UNFOLDED_MULTI_SEEN 0x00000400
580#define REG_GOSTART_SEEN 0x00000800
ee273784
YO
581#define REG_UNBOUNDED_QUANTIFIER_SEEN 0x00001000
582
d09b2d29 583
73c4f7a1
GS
584START_EXTERN_C
585
be8e71aa
YO
586#ifdef PLUGGABLE_RE_EXTENSION
587#include "re_nodes.h"
588#else
d09b2d29 589#include "regnodes.h"
be8e71aa 590#endif
d09b2d29 591
f9f4320a
YO
592#ifndef PLUGGABLE_RE_EXTENSION
593#ifndef DOINIT
594EXTCONST regexp_engine PL_core_reg_engine;
595#else /* DOINIT */
596EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 597 Perl_re_compile,
2fdbfb4d 598 Perl_regexec_flags,
f9f4320a
YO
599 Perl_re_intuit_start,
600 Perl_re_intuit_string,
2fdbfb4d
AB
601 Perl_regfree_internal,
602 Perl_reg_numbered_buff_fetch,
603 Perl_reg_numbered_buff_store,
604 Perl_reg_numbered_buff_length,
192b9cd1
AB
605 Perl_reg_named_buff,
606 Perl_reg_named_buff_iter,
49d7dfbc 607 Perl_reg_qr_package,
f9f4320a 608#if defined(USE_ITHREADS)
3c13cae6 609 Perl_regdupe_internal,
f9f4320a 610#endif
3c13cae6 611 Perl_re_op_compile
f9f4320a
YO
612};
613#endif /* DOINIT */
614#endif /* PLUGGABLE_RE_EXTENSION */
615
616
73c4f7a1 617END_EXTERN_C
cad2e5aa 618
cad2e5aa 619
261faec3
MJD
620/* .what is a character array with one character for each member of .data
621 * The character describes the function of the corresponding .data item:
af534a04 622 * a - AV for paren_name_list under DEBUGGING
0111c4fd 623 * f - start-class data for regstclass optimization
68e2671b 624 * l - start op for literal (?{EVAL}) item
d63c20f2 625 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
b30fcab9 626 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
38a44b82 627 * s - swash for Unicode-style character class, and the multicharacter
9e55ce06
JH
628 * strings resulting from casefolding the single-character entries
629 * in the character class
a3621e74 630 * t - trie struct
55eed653 631 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 632 * also used for revcharmap and words under DEBUGGING
07be1b83 633 * T - aho-trie struct
81714fb9 634 * S - sv for named capture lookup
261faec3 635 * 20010712 mjd@plover.com
c3613ed7 636 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 637 */
cad2e5aa
JH
638struct reg_data {
639 U32 count;
640 U8 *what;
641 void* data[1];
642};
643
a1cac82e
NC
644/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
645 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 646#define anchored_substr substrs->data[0].substr
33b8afdf 647#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 648#define anchored_offset substrs->data[0].min_offset
1de06328
YO
649#define anchored_end_shift substrs->data[0].end_shift
650
cad2e5aa 651#define float_substr substrs->data[1].substr
33b8afdf 652#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
653#define float_min_offset substrs->data[1].min_offset
654#define float_max_offset substrs->data[1].max_offset
1de06328
YO
655#define float_end_shift substrs->data[1].end_shift
656
cad2e5aa 657#define check_substr substrs->data[2].substr
33b8afdf 658#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
659#define check_offset_min substrs->data[2].min_offset
660#define check_offset_max substrs->data[2].max_offset
1de06328 661#define check_end_shift substrs->data[2].end_shift
a3621e74 662
8d919b0a
FC
663#define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
664#define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
665#define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
666#define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
a3621e74
YO
667
668/* trie related stuff */
5d9a96ca 669
a3621e74
YO
670/* a transition record for the state machine. the
671 check field determines which state "owns" the
672 transition. the char the transition is for is
673 determined by offset from the owning states base
674 field. the next field determines which state
675 is to be transitioned to if any.
676*/
677struct _reg_trie_trans {
678 U32 next;
679 U32 check;
680};
681
682/* a transition list element for the list based representation */
683struct _reg_trie_trans_list_elem {
684 U16 forid;
685 U32 newstate;
686};
687typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
688
689/* a state for compressed nodes. base is an offset
690 into an array of reg_trie_trans array. If wordnum is
691 nonzero the state is accepting. if base is zero then
692 the state has no children (and will be accepting)
693*/
694struct _reg_trie_state {
695 U16 wordnum;
696 union {
697 U32 base;
698 reg_trie_trans_le* list;
699 } trans;
700};
701
2e64971a
DM
702/* info per word; indexed by wordnum */
703typedef struct {
704 U16 prev; /* previous word in acceptance chain; eg in
705 * zzz|abc|ab/ after matching the chars abc, the
706 * accepted word is #2, and the previous accepted
707 * word is #3 */
708 U32 len; /* how many chars long is this word? */
709 U32 accept; /* accept state for this word */
710} reg_trie_wordinfo;
a3621e74
YO
711
712
a3621e74
YO
713typedef struct _reg_trie_state reg_trie_state;
714typedef struct _reg_trie_trans reg_trie_trans;
715
716
717/* anything in here that needs to be freed later
23eab42c
NC
718 should be dealt with in pregfree.
719 refcount is first in both this and _reg_ac_data to allow a space
720 optimisation in Perl_regdupe. */
a3621e74 721struct _reg_trie_data {
23eab42c 722 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
723 U32 lasttrans; /* last valid transition element */
724 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
725 reg_trie_state *states; /* state data */
726 reg_trie_trans *trans; /* array of transition elements */
727 char *bitmap; /* stclass bitmap */
786e8c11
YO
728 U16 *jump; /* optional 1 indexed array of offsets before tail
729 for the node following a given word. */
2e64971a 730 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
731 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
732 U32 startstate; /* initial state - used for common prefix optimisation */
733 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
734 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 735 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
736 U32 statecount; /* Build only - number of states in the states array
737 (including the unused zero state) */
786e8c11 738 U32 wordcount; /* Build only */
a3621e74 739#ifdef DEBUGGING
786e8c11 740 STRLEN charcount; /* Build only */
a3621e74
YO
741#endif
742};
2b8b4781
NC
743/* There is one (3 under DEBUGGING) pointers that logically belong in this
744 structure, but are held outside as they need duplication on thread cloning,
745 whereas the rest of the structure can be read only:
746 HV *widecharmap; code points > 255 to charid
747#ifdef DEBUGGING
748 AV *words; Array of words contained in trie, for dumping
749 AV *revcharmap; Map of each charid back to its character representation
750#endif
751*/
752
753#define TRIE_WORDS_OFFSET 2
754
a3621e74
YO
755typedef struct _reg_trie_data reg_trie_data;
756
23eab42c
NC
757/* refcount is first in both this and _reg_trie_data to allow a space
758 optimisation in Perl_regdupe. */
07be1b83 759struct _reg_ac_data {
23eab42c 760 U32 refcount;
1b5c067c 761 U32 trie;
07be1b83
YO
762 U32 *fail;
763 reg_trie_state *states;
07be1b83
YO
764};
765typedef struct _reg_ac_data reg_ac_data;
766
486ec47a 767/* ANY_BIT doesn't use the structure, so we can borrow it here.
3dab1dad
YO
768 This is simpler than refactoring all of it as wed end up with
769 three different sets... */
770
771#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 772#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
773#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
774#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
775#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 776
1de06328
YO
777#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
778#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
779
780
786e8c11
YO
781#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
782#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 783
a3621e74
YO
784/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
785#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
786#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
787#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
788
3dab1dad 789#ifdef DEBUGGING
3dab1dad 790#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 791#else
3dab1dad 792#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 793#endif
a3621e74 794
0111c4fd
RGS
795#define RE_TRIE_MAXBUF_INIT 65536
796#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
797#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
798
be8e71aa
YO
799/*
800
801RE_DEBUG_FLAGS is used to control what debug output is emitted
802its divided into three groups of options, some of which interact.
803The three groups are: Compile, Execute, Extra. There is room for a
804further group, as currently only the low three bytes are used.
805
806 Compile Options:
07be1b83 807
be8e71aa
YO
808 PARSE
809 PEEP
810 TRIE
811 PROGRAM
812 OFFSETS
a3621e74 813
be8e71aa 814 Execute Options:
a3621e74 815
be8e71aa
YO
816 INTUIT
817 MATCH
818 TRIE
a3621e74 819
be8e71aa 820 Extra Options
a3621e74 821
be8e71aa
YO
822 TRIE
823 OFFSETS
824
825If you modify any of these make sure you make corresponding changes to
826re.pm, especially to the documentation.
a3621e74 827
be8e71aa
YO
828*/
829
830
831/* Compile */
832#define RE_DEBUG_COMPILE_MASK 0x0000FF
833#define RE_DEBUG_COMPILE_PARSE 0x000001
834#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
835#define RE_DEBUG_COMPILE_TRIE 0x000004
836#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 837#define RE_DEBUG_COMPILE_FLAGS 0x000010
be8e71aa
YO
838
839/* Execute */
840#define RE_DEBUG_EXECUTE_MASK 0x00FF00
841#define RE_DEBUG_EXECUTE_INTUIT 0x000100
842#define RE_DEBUG_EXECUTE_MATCH 0x000200
843#define RE_DEBUG_EXECUTE_TRIE 0x000400
844
845/* Extra */
846#define RE_DEBUG_EXTRA_MASK 0xFF0000
847#define RE_DEBUG_EXTRA_TRIE 0x010000
848#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
849#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
850#define RE_DEBUG_EXTRA_STATE 0x080000
851#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 852#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 853#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
854/* combined */
855#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 856
e68ec53f 857#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
858/* Compile */
859#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 860 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 861#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 862 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 863#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 864 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 865#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 866 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 867#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 868 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 869#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 870 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
f7819f85
A
871#define DEBUG_FLAGS_r(x) DEBUG_r( \
872 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
be8e71aa
YO
873/* Execute */
874#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 875 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 876#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 877 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 878#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 879 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 880#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 881 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
882
883/* Extra */
884#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 885 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
886#define DEBUG_OFFSETS_r(x) DEBUG_r( \
887 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
888#define DEBUG_STATE_r(x) DEBUG_r( \
889 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
890#define DEBUG_STACK_r(x) DEBUG_r( \
891 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
892#define DEBUG_BUFFERS_r(x) DEBUG_r( \
893 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
894
a5ca303d
YO
895#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
896 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
897 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 898#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 899 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
900 Perl_warn_nocontext x )
901#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 902 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 903#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 904 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
905
906#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 907 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa 908 | RE_DEBUG_EXECUTE_TRIE )) x )
2c296965
YO
909#define DEBUG_GPOS_r(x) DEBUG_r( \
910 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
be8e71aa
YO
911
912/* initialization */
d2d8c711 913/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
914#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
915 SV * re_debug_flags_sv = NULL; \
916 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
917 if (re_debug_flags_sv) { \
8e9a8a48 918 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 919 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 920 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
921 }\
922})
a3621e74
YO
923
924#ifdef DEBUGGING
ab3bbdeb 925
cb65b687
DM
926#define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
927 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
928
929#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
930 const char * const rpv = \
ddc5bc0f 931 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 932 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 933 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 934 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
935
936#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
937 const char * const rpv = \
ddc5bc0f 938 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 939 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 940 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
941
942#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
943 const char * const rpv = \
ddc5bc0f 944 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 945 PL_colors[0], PL_colors[1], \
c89df6cf 946 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 947 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 948 )
ab3bbdeb
YO
949
950#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
951#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
952
953#else /* if not DEBUGGING */
954
a3621e74 955#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
956#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
957#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
958#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
959#define RE_SV_DUMPLEN(ItEm)
960#define RE_SV_TAIL(ItEm)
961
962#endif /* DEBUG RELATED DEFINES */
a3621e74 963
e9a8c099
MHM
964/*
965 * Local variables:
966 * c-indentation-style: bsd
967 * c-basic-offset: 4
14d04a33 968 * indent-tabs-mode: nil
e9a8c099
MHM
969 * End:
970 *
14d04a33 971 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 972 */