This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: Clarify comment
[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 */
128#define PREGf_SKIP 0x00000001
129#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
130#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
131#define PREGf_VERBARG_SEEN 0x00000008
132#define PREGf_CUTGROUP_SEEN 0x00000010
732caac7 133#define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */
785a26d5
YO
134
135
136/* this is where the old regcomp.h started */
a687059c 137
c277df42 138struct regnode_string {
cd439c50 139 U8 str_len;
c277df42
IZ
140 U8 type;
141 U16 next_off;
cd439c50 142 char string[1];
c277df42 143};
a687059c 144
6bda09f9
YO
145/* Argument bearing node - workhorse,
146 arg1 is often for the data field */
c277df42
IZ
147struct regnode_1 {
148 U8 flags;
149 U8 type;
150 U16 next_off;
151 U32 arg1;
152};
153
6bda09f9
YO
154/* Similar to a regnode_1 but with an extra signed argument */
155struct regnode_2L {
156 U8 flags;
157 U8 type;
158 U16 next_off;
159 U32 arg1;
160 I32 arg2;
161};
162
163/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
164struct regnode_2 {
165 U8 flags;
166 U8 type;
167 U16 next_off;
168 U16 arg1;
169 U16 arg2;
170};
171
6bda09f9 172
936ed897 173#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
936ed897 174
786e8c11 175/* also used by trie */
936ed897
IZ
176struct regnode_charclass {
177 U8 flags;
178 U8 type;
179 U16 next_off;
bf3f2d85 180 U32 arg1; /* used as ptr in S_regclass */
19103b91 181 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
182};
183
bf3f2d85
KW
184/* has runtime (locale) \d, \w, ..., [:posix:] classes */
185struct regnode_charclass_class {
186 U8 flags; /* ANYOF_CLASS bit must go here */
936ed897
IZ
187 U8 type;
188 U16 next_off;
bf3f2d85 189 U32 arg1; /* used as ptr in S_regclass */
19103b91 190 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
9cf78fa0 191 U32 classflags; /* and run-time */
936ed897
IZ
192};
193
b8f7bb16
KW
194/* Synthetic start class, is a regnode_charclass_class plus an SV* */
195struct regnode_ssc {
196 U8 flags; /* ANYOF_CLASS bit must go here */
197 U8 type;
198 U16 next_off;
199 U32 arg1; /* used as ptr in S_regclass */
200 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
201 U32 classflags; /* and run-time */
202 SV* invlist; /* list of code points matched */
203};
204
9c7e81e8
AD
205/* XXX fix this description.
206 Impose a limit of REG_INFTY on various pattern matching operations
207 to limit stack growth and to avoid "infinite" recursions.
208*/
209/* The default size for REG_INFTY is I16_MAX, which is the same as
210 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
211 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
212 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
213 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
214 smaller default.
215 --Andy Dougherty 11 June 1998
216*/
217#if SHORTSIZE > 2
218# ifndef REG_INFTY
219# define REG_INFTY ((1<<15)-1)
220# endif
221#endif
222
223#ifndef REG_INFTY
83cfe48c
IZ
224# define REG_INFTY I16_MAX
225#endif
a687059c 226
e91177ed
IZ
227#define ARG_VALUE(arg) (arg)
228#define ARG__SET(arg,val) ((arg) = (val))
c277df42 229
e8e3b0fb
JH
230#undef ARG
231#undef ARG1
232#undef ARG2
233
c277df42
IZ
234#define ARG(p) ARG_VALUE(ARG_LOC(p))
235#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
236#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 237#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 238
c277df42
IZ
239#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
240#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
241#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 242#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 243
e8e3b0fb
JH
244#undef NEXT_OFF
245#undef NODE_ALIGN
246
075abff3
AL
247#define NEXT_OFF(p) ((p)->next_off)
248#define NODE_ALIGN(node)
249#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 250
c277df42
IZ
251#define SIZE_ALIGN NODE_ALIGN
252
e8e3b0fb
JH
253#undef OP
254#undef OPERAND
255#undef MASK
256#undef STRING
257
e91177ed 258#define OP(p) ((p)->type)
b988e673
KW
259#define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \
260 regnode types. For some, it's the \
261 character set of the regnode */
e91177ed 262#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
263#define MASK(p) ((char*)OPERAND(p))
264#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
265#define STRING(p) (((struct regnode_string *)p)->string)
266#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
267#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
268
e8e3b0fb
JH
269#undef NODE_ALIGN
270#undef ARG_LOC
271#undef NEXTOPER
272#undef PREVOPER
273
e91177ed
IZ
274#define NODE_ALIGN(node)
275#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
276#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
277#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 278#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 279
e91177ed
IZ
280#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
281#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
282
283#define NODE_STEP_B 4
c277df42
IZ
284
285#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
286#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
287
e91177ed 288#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 289 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 290#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 291 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 292
22c35a8c 293#define REG_MAGIC 0234
a687059c 294
f7c7e32a 295#define SIZE_ONLY (RExC_emit == &RExC_emit_dummy)
6d15ebf3
KW
296#define PASS1 SIZE_ONLY
297#define PASS2 (! SIZE_ONLY)
c277df42 298
137165a6
KW
299/* If the bitmap doesn't fully represent what this ANYOF node can match, the
300 * ARG is set to this special value (since 0, 1, ... are legal, but will never
301 * reach this high). */
302#define ANYOF_NONBITMAP_EMPTY ((U32) -1)
303
304/* The information used to be stored as as combination of the ANYOF_UTF8 and
305 * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there
306 * to free up a bit for other uses. This tries to hide the change from
307 * existing code as much as possible. Now, the data structure that goes in ARG
308 * is not allocated unless it is needed, and that is what is used to determine
309 * if there is something outside the bitmap. The code now assumes that if
310 * that structure exists, that any UTF-8 encoded string should be tried against
311 * it, but a non-UTF8-encoded string will be tried only if the
312 * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */
313#define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY)
314
899d20b9
KW
315/* Flags for node->flags of ANYOF. These are in short supply, but there is one
316 * currently available. If more than this are needed, the ANYOF_LOCALE and
9ffd0d41 317 * ANYOF_CLASS bits could be shared, making a space penalty for all locale nodes.
899d20b9
KW
318 * Also, the UNICODE_ALL bit could be freed up by resorting to creating a swash
319 * containing everything above 255. This introduces a performance penalty.
320 * Better would be to split it off into a separate node, which actually would
321 * improve performance a bit by allowing regexec.c to test for a UTF-8
322 * character being above 255 without having to call a function nor calculate
323 * its code point value. However, this solution might need to have a second
324 * node type, ANYOF_SYNTHETIC_ABOVE_LATIN1_ALL */
b8c5462f 325
a4e790c1 326#define ANYOF_LOCALE 0x01 /* /l modifier */
39065660
KW
327
328/* The fold is calculated and stored in the bitmap where possible at compile
538b546e
KW
329 * time. However under locale, the actual folding varies depending on
330 * what the locale is at the time of execution, so it has to be deferred until
331 * then */
acf67047 332#define ANYOF_LOC_FOLD 0x02
39065660
KW
333
334#define ANYOF_INVERT 0x04
1984f6a0 335
dd58aee1
KW
336/* Set if this is a struct regnode_charclass_class vs a regnode_charclass. This
337 * is used for runtime \d, \w, [:posix:], ..., which are used only in locale
338 * and the optimizer's synthetic start class. Non-locale \d, etc are resolved
9ffd0d41
KW
339 * at compile-time. Could be shared with ANYOF_LOCALE, forcing all locale
340 * nodes to be large */
341#define ANYOF_CLASS 0x08
acf67047 342#define ANYOF_LARGE ANYOF_CLASS /* Same; name retained for back compat */
c8d3cd88 343
899d20b9 344/* Unused: 0x10. When using, be sure to change ANYOF_FLAGS_ALL below */
f4244008 345
ef87b810
KW
346/* Can match something outside the bitmap that isn't in utf8 */
347#define ANYOF_NONBITMAP_NON_UTF8 0x20
348
61ef9872
KW
349/* Matches every code point 0x100 and above*/
350#define ANYOF_UNICODE_ALL 0x40
ffc61ed2 351
11454c59
KW
352/* Match all Latin1 characters that aren't ASCII when the target string is not
353 * in utf8. */
354#define ANYOF_NON_UTF8_LATIN1_ALL 0x80
355
899d20b9 356#define ANYOF_FLAGS_ALL (0xff & ~0x10)
653099ff 357
ace6b0e4
KW
358/* These are the flags that ANYOF_INVERT being set or not doesn't affect
359 * whether they are operative or not. e.g., the node still has LOCALE
360 * regardless of being inverted; whereas ANYOF_UNICODE_ALL means something
361 * different if inverted */
362#define INVERSION_UNAFFECTED_FLAGS (ANYOF_LOCALE \
538b546e 363 |ANYOF_LOC_FOLD \
ace6b0e4 364 |ANYOF_CLASS \
ace6b0e4
KW
365 |ANYOF_NONBITMAP_NON_UTF8)
366
936ed897 367/* Character classes for node->classflags of ANYOF */
653099ff 368/* Should be synchronized with a table in regprop() */
575a8a29 369/* 2n should be the normal one, paired with its complement at 2n+1 */
b8c5462f 370
08ada29e
KW
371#define ANYOF_ALPHA ((_CC_ALPHA) * 2)
372#define ANYOF_NALPHA ((ANYOF_ALPHA) + 1)
e8d596e0
KW
373#define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */
374#define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1)
08ada29e
KW
375#define ANYOF_ASCII ((_CC_ASCII) * 2)
376#define ANYOF_NASCII ((ANYOF_ASCII) + 1)
e8d596e0
KW
377#define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */
378#define ANYOF_NBLANK ((ANYOF_BLANK) + 1)
b0d691b2
KW
379#define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or
380 [:upper:] under /i */
381#define ANYOF_NCASED ((ANYOF_CASED) + 1)
08ada29e
KW
382#define ANYOF_CNTRL ((_CC_CNTRL) * 2)
383#define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1)
e8d596e0
KW
384#define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */
385#define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1)
08ada29e
KW
386#define ANYOF_GRAPH ((_CC_GRAPH) * 2)
387#define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1)
388#define ANYOF_LOWER ((_CC_LOWER) * 2)
389#define ANYOF_NLOWER ((ANYOF_LOWER) + 1)
390#define ANYOF_PRINT ((_CC_PRINT) * 2)
391#define ANYOF_NPRINT ((ANYOF_PRINT) + 1)
e8d596e0
KW
392#define ANYOF_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */
393#define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1)
08ada29e
KW
394#define ANYOF_PUNCT ((_CC_PUNCT) * 2)
395#define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1)
e8d596e0
KW
396#define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */
397#define ANYOF_NSPACE ((ANYOF_SPACE) + 1)
08ada29e
KW
398#define ANYOF_UPPER ((_CC_UPPER) * 2)
399#define ANYOF_NUPPER ((ANYOF_UPPER) + 1)
e8d596e0
KW
400#define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
401#define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1)
08ada29e
KW
402#define ANYOF_XDIGIT ((_CC_XDIGIT) * 2)
403#define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1)
b8c5462f 404
87cad4c0 405/* pseudo classes below this, not stored in the class bitmap, but used as flags
e1d1eefb
YO
406 during compilation of char classes */
407
b2ef2e34
KW
408#define ANYOF_VERTWS ((_CC_VERTSPACE) * 2)
409#define ANYOF_NVERTWS ((ANYOF_VERTWS)+1)
e1d1eefb 410
b2ef2e34
KW
411/* It is best if this is the last one, as all above it are stored as bits in a
412 * bitmap, and it isn't part of that bitmap */
413#if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC
414# error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define
415#endif
416
417#define ANYOF_MAX (ANYOF_VERTWS) /* So upper loop limit is written:
418 * '< ANYOF_MAX'
419 * Hence doesn't include VERTWS, as that
420 * is a pseudo class */
421#if (ANYOF_MAX > 32) /* Must fit in 32-bit word */
422# error Problem with handy.h _CC_foo #defines
a0947d7b
KW
423#endif
424
b2ef2e34 425#define ANYOF_HORIZWS ((ANYOF_MAX)+2) /* = (ANYOF_NVERTWS + 1) */
87cad4c0
KW
426#define ANYOF_NHORIZWS ((ANYOF_MAX)+3)
427
428#define ANYOF_UNIPROP ((ANYOF_MAX)+4) /* Used to indicate a Unicode
429 property: \p{} or \P{} */
f39a4dc2 430
b8c5462f
JH
431/* Backward source code compatibility. */
432
433#define ANYOF_ALNUML ANYOF_ALNUM
434#define ANYOF_NALNUML ANYOF_NALNUM
435#define ANYOF_SPACEL ANYOF_SPACE
436#define ANYOF_NSPACEL ANYOF_NSPACE
8ebfc29a
KW
437#define ANYOF_ALNUM ANYOF_WORDCHAR
438#define ANYOF_NALNUM ANYOF_NWORDCHAR
b8c5462f
JH
439
440/* Utility macros for the bitmap and classes of ANYOF */
441
936ed897
IZ
442#define ANYOF_SIZE (sizeof(struct regnode_charclass))
443#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 444
936ed897 445#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
446
447#define ANYOF_BIT(c) (1 << ((c) & 7))
448
9cf78fa0
KW
449#define ANYOF_CLASS_SET(p, c) (((struct regnode_charclass_class*) (p))->classflags |= (1U << (c)))
450#define ANYOF_CLASS_CLEAR(p, c) (((struct regnode_charclass_class*) (p))->classflags &= ~ (1U <<(c)))
451#define ANYOF_CLASS_TEST(p, c) (((struct regnode_charclass_class*) (p))->classflags & (1U << (c)))
b8c5462f 452
9cf78fa0
KW
453#define ANYOF_CLASS_ZERO(ret) STMT_START { ((struct regnode_charclass_class*) (ret))->classflags = 0; } STMT_END
454
455/* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */
456#define ANYOF_CLASS_SETALL(ret) STMT_START { ((struct regnode_charclass_class*) (ret))->classflags = ((1U << ((ANYOF_MAX) - 1))) - 1; } STMT_END
457
e867ce45
KW
458#define ANYOF_CLASS_TEST_ANY_SET(p) \
459 ((ANYOF_FLAGS(p) & ANYOF_CLASS) \
460 && (((struct regnode_charclass_class*)(p))->classflags))
461
9cf78fa0 462#define ANYOF_CLASS_OR(source, dest) STMT_START { (dest)->classflags |= source->classflags ; } STMT_END
936ed897 463
9cf78fa0 464#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
936ed897 465#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 466#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
467#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
468#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
469#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
470
f8bef550
NC
471#define ANYOF_BITMAP_SETALL(p) \
472 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
473#define ANYOF_BITMAP_CLEARALL(p) \
474 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
475/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
c355c09a 476#define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \
f8bef550
NC
477 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)
478
936ed897
IZ
479#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
480#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
3a15e693 481
a687059c
LW
482/*
483 * Utility definitions.
484 */
2304df62 485#ifndef CHARMASK
075abff3 486# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 487#else
075abff3 488# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 489#endif
a687059c 490
c277df42
IZ
491#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
492
3dab1dad
YO
493#define REG_SEEN_ZERO_LEN 0x00000001
494#define REG_SEEN_LOOKBEHIND 0x00000002
495#define REG_SEEN_GPOS 0x00000004
471f5387 496/* spare */
3dab1dad 497#define REG_SEEN_CANY 0x00000010
cce850e4 498#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 499#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 500#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 501#define REG_SEEN_VERBARG 0x00000080
5d458dd8 502#define REG_SEEN_CUTGROUP 0x00000100
bcdf7404 503#define REG_SEEN_RUN_ON_COMMENT 0x00000200
bb914485 504#define REG_SEEN_EXACTF_SHARP_S 0x00000400
d09b2d29 505
73c4f7a1
GS
506START_EXTERN_C
507
be8e71aa
YO
508#ifdef PLUGGABLE_RE_EXTENSION
509#include "re_nodes.h"
510#else
d09b2d29 511#include "regnodes.h"
be8e71aa 512#endif
d09b2d29 513
f9f4320a
YO
514#ifndef PLUGGABLE_RE_EXTENSION
515#ifndef DOINIT
516EXTCONST regexp_engine PL_core_reg_engine;
517#else /* DOINIT */
518EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 519 Perl_re_compile,
2fdbfb4d 520 Perl_regexec_flags,
f9f4320a
YO
521 Perl_re_intuit_start,
522 Perl_re_intuit_string,
2fdbfb4d
AB
523 Perl_regfree_internal,
524 Perl_reg_numbered_buff_fetch,
525 Perl_reg_numbered_buff_store,
526 Perl_reg_numbered_buff_length,
192b9cd1
AB
527 Perl_reg_named_buff,
528 Perl_reg_named_buff_iter,
49d7dfbc 529 Perl_reg_qr_package,
f9f4320a 530#if defined(USE_ITHREADS)
3c13cae6 531 Perl_regdupe_internal,
f9f4320a 532#endif
3c13cae6 533 Perl_re_op_compile
f9f4320a
YO
534};
535#endif /* DOINIT */
536#endif /* PLUGGABLE_RE_EXTENSION */
537
538
73c4f7a1 539END_EXTERN_C
cad2e5aa 540
cad2e5aa 541
261faec3
MJD
542/* .what is a character array with one character for each member of .data
543 * The character describes the function of the corresponding .data item:
af534a04 544 * a - AV for paren_name_list under DEBUGGING
0111c4fd 545 * f - start-class data for regstclass optimization
68e2671b 546 * l - start op for literal (?{EVAL}) item
d63c20f2 547 * L - start op for literal (?{EVAL}) item, with separate CV (qr//)
b30fcab9 548 * r - pointer to an embedded code-containing qr, e.g. /ab$qr/
38a44b82 549 * s - swash for Unicode-style character class, and the multicharacter
9e55ce06
JH
550 * strings resulting from casefolding the single-character entries
551 * in the character class
a3621e74 552 * t - trie struct
55eed653 553 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 554 * also used for revcharmap and words under DEBUGGING
07be1b83 555 * T - aho-trie struct
81714fb9 556 * S - sv for named capture lookup
261faec3 557 * 20010712 mjd@plover.com
c3613ed7 558 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 559 */
cad2e5aa
JH
560struct reg_data {
561 U32 count;
562 U8 *what;
563 void* data[1];
564};
565
a1cac82e
NC
566/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
567 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 568#define anchored_substr substrs->data[0].substr
33b8afdf 569#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 570#define anchored_offset substrs->data[0].min_offset
1de06328
YO
571#define anchored_end_shift substrs->data[0].end_shift
572
cad2e5aa 573#define float_substr substrs->data[1].substr
33b8afdf 574#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
575#define float_min_offset substrs->data[1].min_offset
576#define float_max_offset substrs->data[1].max_offset
1de06328
YO
577#define float_end_shift substrs->data[1].end_shift
578
cad2e5aa 579#define check_substr substrs->data[2].substr
33b8afdf 580#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
581#define check_offset_min substrs->data[2].min_offset
582#define check_offset_max substrs->data[2].max_offset
1de06328 583#define check_end_shift substrs->data[2].end_shift
a3621e74 584
8d919b0a
FC
585#define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr)
586#define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8)
587#define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr)
588#define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8)
a3621e74
YO
589
590/* trie related stuff */
5d9a96ca 591
a3621e74
YO
592/* a transition record for the state machine. the
593 check field determines which state "owns" the
594 transition. the char the transition is for is
595 determined by offset from the owning states base
596 field. the next field determines which state
597 is to be transitioned to if any.
598*/
599struct _reg_trie_trans {
600 U32 next;
601 U32 check;
602};
603
604/* a transition list element for the list based representation */
605struct _reg_trie_trans_list_elem {
606 U16 forid;
607 U32 newstate;
608};
609typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
610
611/* a state for compressed nodes. base is an offset
612 into an array of reg_trie_trans array. If wordnum is
613 nonzero the state is accepting. if base is zero then
614 the state has no children (and will be accepting)
615*/
616struct _reg_trie_state {
617 U16 wordnum;
618 union {
619 U32 base;
620 reg_trie_trans_le* list;
621 } trans;
622};
623
2e64971a
DM
624/* info per word; indexed by wordnum */
625typedef struct {
626 U16 prev; /* previous word in acceptance chain; eg in
627 * zzz|abc|ab/ after matching the chars abc, the
628 * accepted word is #2, and the previous accepted
629 * word is #3 */
630 U32 len; /* how many chars long is this word? */
631 U32 accept; /* accept state for this word */
632} reg_trie_wordinfo;
a3621e74
YO
633
634
a3621e74
YO
635typedef struct _reg_trie_state reg_trie_state;
636typedef struct _reg_trie_trans reg_trie_trans;
637
638
639/* anything in here that needs to be freed later
23eab42c
NC
640 should be dealt with in pregfree.
641 refcount is first in both this and _reg_ac_data to allow a space
642 optimisation in Perl_regdupe. */
a3621e74 643struct _reg_trie_data {
23eab42c 644 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
645 U32 lasttrans; /* last valid transition element */
646 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
647 reg_trie_state *states; /* state data */
648 reg_trie_trans *trans; /* array of transition elements */
649 char *bitmap; /* stclass bitmap */
786e8c11
YO
650 U16 *jump; /* optional 1 indexed array of offsets before tail
651 for the node following a given word. */
2e64971a 652 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
653 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
654 U32 startstate; /* initial state - used for common prefix optimisation */
655 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
656 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 657 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
658 U32 statecount; /* Build only - number of states in the states array
659 (including the unused zero state) */
786e8c11 660 U32 wordcount; /* Build only */
a3621e74 661#ifdef DEBUGGING
786e8c11 662 STRLEN charcount; /* Build only */
a3621e74
YO
663#endif
664};
2b8b4781
NC
665/* There is one (3 under DEBUGGING) pointers that logically belong in this
666 structure, but are held outside as they need duplication on thread cloning,
667 whereas the rest of the structure can be read only:
668 HV *widecharmap; code points > 255 to charid
669#ifdef DEBUGGING
670 AV *words; Array of words contained in trie, for dumping
671 AV *revcharmap; Map of each charid back to its character representation
672#endif
673*/
674
675#define TRIE_WORDS_OFFSET 2
676
a3621e74
YO
677typedef struct _reg_trie_data reg_trie_data;
678
23eab42c
NC
679/* refcount is first in both this and _reg_trie_data to allow a space
680 optimisation in Perl_regdupe. */
07be1b83 681struct _reg_ac_data {
23eab42c 682 U32 refcount;
1b5c067c 683 U32 trie;
07be1b83
YO
684 U32 *fail;
685 reg_trie_state *states;
07be1b83
YO
686};
687typedef struct _reg_ac_data reg_ac_data;
688
486ec47a 689/* ANY_BIT doesn't use the structure, so we can borrow it here.
3dab1dad
YO
690 This is simpler than refactoring all of it as wed end up with
691 three different sets... */
692
693#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 694#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
695#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
696#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
697#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 698
1de06328
YO
699#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
700#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
701
702
786e8c11
YO
703#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
704#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 705
a3621e74
YO
706/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
707#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
708#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
709#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
710
3dab1dad 711#ifdef DEBUGGING
3dab1dad 712#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 713#else
3dab1dad 714#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 715#endif
a3621e74 716
0111c4fd
RGS
717#define RE_TRIE_MAXBUF_INIT 65536
718#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
719#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
720
be8e71aa
YO
721/*
722
723RE_DEBUG_FLAGS is used to control what debug output is emitted
724its divided into three groups of options, some of which interact.
725The three groups are: Compile, Execute, Extra. There is room for a
726further group, as currently only the low three bytes are used.
727
728 Compile Options:
07be1b83 729
be8e71aa
YO
730 PARSE
731 PEEP
732 TRIE
733 PROGRAM
734 OFFSETS
a3621e74 735
be8e71aa 736 Execute Options:
a3621e74 737
be8e71aa
YO
738 INTUIT
739 MATCH
740 TRIE
a3621e74 741
be8e71aa 742 Extra Options
a3621e74 743
be8e71aa
YO
744 TRIE
745 OFFSETS
746
747If you modify any of these make sure you make corresponding changes to
748re.pm, especially to the documentation.
a3621e74 749
be8e71aa
YO
750*/
751
752
753/* Compile */
754#define RE_DEBUG_COMPILE_MASK 0x0000FF
755#define RE_DEBUG_COMPILE_PARSE 0x000001
756#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
757#define RE_DEBUG_COMPILE_TRIE 0x000004
758#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 759#define RE_DEBUG_COMPILE_FLAGS 0x000010
be8e71aa
YO
760
761/* Execute */
762#define RE_DEBUG_EXECUTE_MASK 0x00FF00
763#define RE_DEBUG_EXECUTE_INTUIT 0x000100
764#define RE_DEBUG_EXECUTE_MATCH 0x000200
765#define RE_DEBUG_EXECUTE_TRIE 0x000400
766
767/* Extra */
768#define RE_DEBUG_EXTRA_MASK 0xFF0000
769#define RE_DEBUG_EXTRA_TRIE 0x010000
770#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
771#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
772#define RE_DEBUG_EXTRA_STATE 0x080000
773#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 774#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 775#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
776/* combined */
777#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 778
e68ec53f 779#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
780/* Compile */
781#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 782 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 783#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 784 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 785#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 786 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 787#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 788 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 789#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 790 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 791#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 792 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
f7819f85
A
793#define DEBUG_FLAGS_r(x) DEBUG_r( \
794 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
be8e71aa
YO
795/* Execute */
796#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 797 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 798#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 799 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 800#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 801 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 802#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 803 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
804
805/* Extra */
806#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 807 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
808#define DEBUG_OFFSETS_r(x) DEBUG_r( \
809 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
810#define DEBUG_STATE_r(x) DEBUG_r( \
811 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
812#define DEBUG_STACK_r(x) DEBUG_r( \
813 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
814#define DEBUG_BUFFERS_r(x) DEBUG_r( \
815 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
816
a5ca303d
YO
817#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
818 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
819 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 820#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 821 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
822 Perl_warn_nocontext x )
823#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 824 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 825#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 826 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
827
828#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 829 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa 830 | RE_DEBUG_EXECUTE_TRIE )) x )
2c296965
YO
831#define DEBUG_GPOS_r(x) DEBUG_r( \
832 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
be8e71aa
YO
833
834/* initialization */
d2d8c711 835/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
836#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
837 SV * re_debug_flags_sv = NULL; \
838 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
839 if (re_debug_flags_sv) { \
8e9a8a48 840 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 841 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 842 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
843 }\
844})
a3621e74
YO
845
846#ifdef DEBUGGING
ab3bbdeb 847
cb65b687
DM
848#define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \
849 PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
850
851#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
852 const char * const rpv = \
ddc5bc0f 853 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 854 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 855 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 856 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
857
858#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
859 const char * const rpv = \
ddc5bc0f 860 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 861 PL_colors[(c1)],PL_colors[(c2)], \
c89df6cf 862 PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
863
864#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
865 const char * const rpv = \
ddc5bc0f 866 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 867 PL_colors[0], PL_colors[1], \
c89df6cf 868 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 869 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 870 )
ab3bbdeb
YO
871
872#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
873#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
874
875#else /* if not DEBUGGING */
876
a3621e74 877#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
878#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
879#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
880#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
881#define RE_SV_DUMPLEN(ItEm)
882#define RE_SV_TAIL(ItEm)
883
884#endif /* DEBUG RELATED DEFINES */
a3621e74 885
e9a8c099
MHM
886/*
887 * Local variables:
888 * c-indentation-style: bsd
889 * c-basic-offset: 4
14d04a33 890 * indent-tabs-mode: nil
e9a8c099
MHM
891 * End:
892 *
14d04a33 893 * ex: set ts=8 sts=4 sw=4 et:
e9a8c099 894 */