This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
ANYOF_LARGE is now the same as ANYOF_CLASS
[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
c277df42
IZ
12typedef OP OP_4tree; /* Will be redefined later. */
13
07be1b83 14
58e23c8d 15/* Convert branch sequences to more efficient trie ops? */
07be1b83 16#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d
YO
17
18/* Be really agressive about optimising patterns with trie sequences? */
07be1b83 19#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d 20
da7fcca4
YO
21/* Use old style unicode mappings for perl and posix character classes
22 *
23 * NOTE: Enabling this essentially breaks character class matching against unicode
24 * strings, so that POSIX char classes match when they shouldn't, and \d matches
25 * way more than 10 characters, and sometimes a charclass and its complement either
26 * both match or neither match.
27 * NOTE: Disabling this will cause various backwards compatibility issues to rear
28 * their head, and tests to fail. However it will make the charclass behaviour
29 * consistant regardless of internal string type, and make character class inversions
30 * consistant. The tests that fail in the regex engine are basically broken tests.
31 *
32 * Personally I think 5.12 should disable this for sure. Its a bit more debatable for
33 * 5.10, so for now im leaving it enabled.
ba9ac175 34 * XXX: It is now enabled for 5.11/5.12
da7fcca4
YO
35 *
36 * -demerphq
37 */
2e84be61 38#define PERL_LEGACY_UNICODE_CHARCLASS_MAPPINGS 1
da7fcca4 39
58e23c8d 40/* Should the optimiser take positive assertions into account? */
68ba3a3f 41#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
58e23c8d
YO
42
43/* Not for production use: */
07be1b83 44#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 45
7122b237
YO
46/* Activate offsets code - set to if 1 to enable */
47#ifdef DEBUGGING
48#define RE_TRACK_PATTERN_OFFSETS
49#endif
50
b9b4dddf
YO
51/* Unless the next line is uncommented it is illegal to combine lazy
52 matching with possessive matching. Frankly it doesn't make much sense
53 to allow it as X*?+ matches nothing, X+?+ matches a single char only,
54 and X{min,max}?+ matches min times only.
55 */
56/* #define REG_ALLOW_MINMOD_SUSPEND */
07be1b83 57
a687059c
LW
58/*
59 * The "internal use only" fields in regexp.h are present to pass info from
60 * compile to execute that permits the execute phase to run lots faster on
61 * simple cases. They are:
62 *
bd61b366 63 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
64 * reganch is the match anchored (at beginning-of-line only)?
65 * regmust string (pointer into program) that match must include, or NULL
79072805 66 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
67 * regmlen length of regmust string
68 * [regmlen not used currently]
69 *
70 * Regstart and reganch permit very fast decisions on suitable starting points
71 * for a match, cutting down the work a lot. Regmust permits fast rejection
72 * of lines that cannot possibly match. The regmust tests are costly enough
e50aee73 73 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
74 * potentially expensive (at present, the only such thing detected is * or +
75 * at the start of the r.e., which can involve a lot of backup). Regmlen is
e50aee73 76 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
77 * it anyway.
78 * [regmust is now supplied always. The tests that use regmust have a
79 * heuristic that disables the test if it usually matches.]
80 *
81 * [In fact, we now use regmust in many cases to locate where the search
82 * starts in the string, so if regback is >= 0, the regmust search is never
83 * wasted effort. The regback variable says how many characters back from
84 * where regmust matched is the earliest possible start of the match.
85 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
86 */
87
88/*
89 * Structure for regexp "program". This is essentially a linear encoding
90 * of a nondeterministic finite-state machine (aka syntax charts or
91 * "railroad normal form" in parsing technology). Each node is an opcode
92 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
93 * all nodes except BRANCH implement concatenation; a "next" pointer with
94 * a BRANCH on both ends of it is connecting two alternatives. (Here we
95 * have one of the subtle syntax dependencies: an individual BRANCH (as
96 * opposed to a collection of them) is never concatenated with anything
97 * because of operator precedence.) The operand of some types of node is
98 * a literal string; for others, it is a node leading into a sub-FSM. In
99 * particular, the operand of a BRANCH node is the first node of the branch.
100 * (NB this is *not* a tree structure: the tail of the branch connects
a3621e74
YO
101 * to the thing following the set of BRANCHes.) The opcodes are defined
102 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
a687059c
LW
103 */
104
a687059c
LW
105/*
106 * A node is one char of opcode followed by two chars of "next" pointer.
107 * "Next" pointers are stored as two 8-bit pieces, high order first. The
108 * value is a positive offset from the opcode of the node containing it.
109 * An operand, if any, simply follows the node. (Note that much of the
110 * code generation knows about this implicit relationship.)
111 *
112 * Using two bytes for the "next" pointer is vast overkill for most things,
113 * but allows patterns to get big without disasters.
114 *
e91177ed 115 * [The "next" pointer is always aligned on an even
a687059c
LW
116 * boundary, and reads the offset directly as a short. Also, there is no
117 * special test to reverse the sign of BACK pointers since the offset is
118 * stored negative.]
119 */
785a26d5
YO
120
121/* This is the stuff that used to live in regexp.h that was truly
122 private to the engine itself. It now lives here. */
123
28d8d7f4 124
785a26d5
YO
125
126 typedef struct regexp_internal {
1f1031fe 127 int name_list_idx; /* Optional data index of an array of paren names */
7122b237
YO
128 union {
129 U32 *offsets; /* offset annotations 20001228 MJD
130 data about mapping the program to the
131 string -
132 offsets[0] is proglen when this is used
133 */
134 U32 proglen;
135 } u;
136
6d5c990f
RGS
137 regnode *regstclass; /* Optional startclass as identified or constructed
138 by the optimiser */
139 struct reg_data *data; /* Additional miscellaneous data used by the program.
140 Used to make it easier to clone and free arbitrary
141 data that the regops need. Often the ARG field of
142 a regop is an index into this structure */
143 regnode program[1]; /* Unwarranted chumminess with compiler. */
144} regexp_internal;
145
146#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
147#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
148#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
149/*
150 * Flags stored in regexp->intflags
151 * These are used only internally to the regexp engine
152 *
153 * See regexp.h for flags used externally to the regexp engine
154 */
155#define PREGf_SKIP 0x00000001
156#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
157#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
158#define PREGf_VERBARG_SEEN 0x00000008
159#define PREGf_CUTGROUP_SEEN 0x00000010
160
161
162/* this is where the old regcomp.h started */
a687059c 163
c277df42 164struct regnode_string {
cd439c50 165 U8 str_len;
c277df42
IZ
166 U8 type;
167 U16 next_off;
cd439c50 168 char string[1];
c277df42 169};
a687059c 170
6bda09f9
YO
171/* Argument bearing node - workhorse,
172 arg1 is often for the data field */
c277df42
IZ
173struct regnode_1 {
174 U8 flags;
175 U8 type;
176 U16 next_off;
177 U32 arg1;
178};
179
6bda09f9
YO
180/* Similar to a regnode_1 but with an extra signed argument */
181struct regnode_2L {
182 U8 flags;
183 U8 type;
184 U16 next_off;
185 U32 arg1;
186 I32 arg2;
187};
188
189/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
190struct regnode_2 {
191 U8 flags;
192 U8 type;
193 U16 next_off;
194 U16 arg1;
195 U16 arg2;
196};
197
6bda09f9 198
936ed897 199#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
d1eb3177 200#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 201
786e8c11 202/* also used by trie */
936ed897
IZ
203struct regnode_charclass {
204 U8 flags;
205 U8 type;
206 U16 next_off;
ffc61ed2 207 U32 arg1;
19103b91 208 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
209};
210
19103b91
JH
211struct regnode_charclass_class { /* has [[:blah:]] classes */
212 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
213 U8 type;
214 U16 next_off;
ffc61ed2 215 U32 arg1;
19103b91
JH
216 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
217 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
218};
219
9c7e81e8
AD
220/* XXX fix this description.
221 Impose a limit of REG_INFTY on various pattern matching operations
222 to limit stack growth and to avoid "infinite" recursions.
223*/
224/* The default size for REG_INFTY is I16_MAX, which is the same as
225 SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits
226 (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is
227 ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is
228 ((1<<63)-1). To limit stack growth to reasonable sizes, supply a
229 smaller default.
230 --Andy Dougherty 11 June 1998
231*/
232#if SHORTSIZE > 2
233# ifndef REG_INFTY
234# define REG_INFTY ((1<<15)-1)
235# endif
236#endif
237
238#ifndef REG_INFTY
83cfe48c
IZ
239# define REG_INFTY I16_MAX
240#endif
a687059c 241
e91177ed
IZ
242#define ARG_VALUE(arg) (arg)
243#define ARG__SET(arg,val) ((arg) = (val))
c277df42 244
e8e3b0fb
JH
245#undef ARG
246#undef ARG1
247#undef ARG2
248
c277df42
IZ
249#define ARG(p) ARG_VALUE(ARG_LOC(p))
250#define ARG1(p) ARG_VALUE(ARG1_LOC(p))
251#define ARG2(p) ARG_VALUE(ARG2_LOC(p))
6bda09f9 252#define ARG2L(p) ARG_VALUE(ARG2L_LOC(p))
786e8c11 253
c277df42
IZ
254#define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val))
255#define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val))
256#define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val))
6bda09f9 257#define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val))
c277df42 258
e8e3b0fb
JH
259#undef NEXT_OFF
260#undef NODE_ALIGN
261
075abff3
AL
262#define NEXT_OFF(p) ((p)->next_off)
263#define NODE_ALIGN(node)
264#define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */
a687059c 265
c277df42
IZ
266#define SIZE_ALIGN NODE_ALIGN
267
e8e3b0fb
JH
268#undef OP
269#undef OPERAND
270#undef MASK
271#undef STRING
272
e91177ed 273#define OP(p) ((p)->type)
b4eac3bb
KW
274#define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all
275 regnode types */
e91177ed 276#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
277#define MASK(p) ((char*)OPERAND(p))
278#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
279#define STRING(p) (((struct regnode_string *)p)->string)
280#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
281#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
282
e8e3b0fb
JH
283#undef NODE_ALIGN
284#undef ARG_LOC
285#undef NEXTOPER
286#undef PREVOPER
287
e91177ed
IZ
288#define NODE_ALIGN(node)
289#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
290#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
291#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 292#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 293
e91177ed
IZ
294#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
295#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
296
297#define NODE_STEP_B 4
c277df42
IZ
298
299#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
300#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
301
e91177ed 302#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 303 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 304#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 305 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 306
22c35a8c 307#define REG_MAGIC 0234
a687059c 308
830247a4 309#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 310
a12cf05f
KW
311/* Flags for node->flags of several of the node types */
312#define USE_UNI 0x01
313
936ed897 314/* Flags for node->flags of ANYOF */
b8c5462f 315
eb91e300
KW
316#define ANYOF_CLASS 0x08 /* has runtime \d, \w, [:posix:], ... */
317#define ANYOF_LARGE ANYOF_CLASS /* Same; name retained for back compat */
ffc61ed2
JH
318#define ANYOF_INVERT 0x04
319#define ANYOF_FOLD 0x02
320#define ANYOF_LOCALE 0x01
b8c5462f 321
653099ff 322/* Used for regstclass only */
ffc61ed2
JH
323#define ANYOF_EOS 0x10 /* Can match an empty string too */
324
325/* There is a character or a range past 0xff */
326#define ANYOF_UNICODE 0x20
1aa99e6b 327#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2
JH
328
329/* Are there any runtime flags on in this node? */
330#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
331
332#define ANYOF_FLAGS_ALL 0xff
653099ff 333
936ed897 334/* Character classes for node->classflags of ANYOF */
653099ff
GS
335/* Should be synchronized with a table in regprop() */
336/* 2n should pair with 2n+1 */
b8c5462f 337
289d4f09 338#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 339#define ANYOF_NALNUM 1
aaa51d5e 340#define ANYOF_SPACE 2 /* \s */
b8c5462f 341#define ANYOF_NSPACE 3
24d786f4 342#define ANYOF_DIGIT 4 /* \d */
b8c5462f 343#define ANYOF_NDIGIT 5
24d786f4 344#define ANYOF_ALNUMC 6 /* [[:alnum:]] isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
345#define ANYOF_NALNUMC 7
346#define ANYOF_ALPHA 8
347#define ANYOF_NALPHA 9
348#define ANYOF_ASCII 10
349#define ANYOF_NASCII 11
350#define ANYOF_CNTRL 12
351#define ANYOF_NCNTRL 13
352#define ANYOF_GRAPH 14
353#define ANYOF_NGRAPH 15
354#define ANYOF_LOWER 16
355#define ANYOF_NLOWER 17
356#define ANYOF_PRINT 18
357#define ANYOF_NPRINT 19
358#define ANYOF_PUNCT 20
359#define ANYOF_NPUNCT 21
360#define ANYOF_UPPER 22
361#define ANYOF_NUPPER 23
362#define ANYOF_XDIGIT 24
363#define ANYOF_NXDIGIT 25
aaa51d5e
JF
364#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
365#define ANYOF_NPSXSPC 27
ffc61ed2 366#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 367#define ANYOF_NBLANK 29
b8c5462f 368
aaa51d5e 369#define ANYOF_MAX 32
b8c5462f 370
e1d1eefb
YO
371/* pseudo classes, not stored in the class bitmap, but used as flags
372 during compilation of char classes */
373
374#define ANYOF_VERTWS (ANYOF_MAX+1)
375#define ANYOF_NVERTWS (ANYOF_MAX+2)
376#define ANYOF_HORIZWS (ANYOF_MAX+3)
377#define ANYOF_NHORIZWS (ANYOF_MAX+4)
378
b8c5462f
JH
379/* Backward source code compatibility. */
380
381#define ANYOF_ALNUML ANYOF_ALNUM
382#define ANYOF_NALNUML ANYOF_NALNUM
383#define ANYOF_SPACEL ANYOF_SPACE
384#define ANYOF_NSPACEL ANYOF_NSPACE
385
386/* Utility macros for the bitmap and classes of ANYOF */
387
936ed897
IZ
388#define ANYOF_SIZE (sizeof(struct regnode_charclass))
389#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 390
936ed897 391#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
392
393#define ANYOF_BIT(c) (1 << ((c) & 7))
394
936ed897 395#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
396#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
397#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
398#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
399
936ed897
IZ
400#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
401#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
402
403#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 404#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
405#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
406#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
407#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
408
f8bef550
NC
409#define ANYOF_BITMAP_SETALL(p) \
410 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
411#define ANYOF_BITMAP_CLEARALL(p) \
412 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
413/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
414#define ANYOF_BITMAP_TESTALLSET(p) \
415 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)
416
936ed897
IZ
417#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
418#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
419#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 420
786e8c11 421
a687059c
LW
422/*
423 * Utility definitions.
424 */
2304df62 425#ifndef CHARMASK
075abff3 426# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 427#else
075abff3 428# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 429#endif
a687059c 430
c277df42
IZ
431#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
432
3dab1dad
YO
433#define REG_SEEN_ZERO_LEN 0x00000001
434#define REG_SEEN_LOOKBEHIND 0x00000002
435#define REG_SEEN_GPOS 0x00000004
436#define REG_SEEN_EVAL 0x00000008
437#define REG_SEEN_CANY 0x00000010
cce850e4 438#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 439#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 440#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 441#define REG_SEEN_VERBARG 0x00000080
5d458dd8 442#define REG_SEEN_CUTGROUP 0x00000100
bcdf7404 443#define REG_SEEN_RUN_ON_COMMENT 0x00000200
d09b2d29 444
73c4f7a1
GS
445START_EXTERN_C
446
be8e71aa
YO
447#ifdef PLUGGABLE_RE_EXTENSION
448#include "re_nodes.h"
449#else
d09b2d29 450#include "regnodes.h"
be8e71aa 451#endif
d09b2d29 452
f9f4320a
YO
453#ifndef PLUGGABLE_RE_EXTENSION
454#ifndef DOINIT
455EXTCONST regexp_engine PL_core_reg_engine;
456#else /* DOINIT */
457EXTCONST regexp_engine PL_core_reg_engine = {
fe578d7f 458 Perl_re_compile,
2fdbfb4d 459 Perl_regexec_flags,
f9f4320a
YO
460 Perl_re_intuit_start,
461 Perl_re_intuit_string,
2fdbfb4d
AB
462 Perl_regfree_internal,
463 Perl_reg_numbered_buff_fetch,
464 Perl_reg_numbered_buff_store,
465 Perl_reg_numbered_buff_length,
192b9cd1
AB
466 Perl_reg_named_buff,
467 Perl_reg_named_buff_iter,
49d7dfbc 468 Perl_reg_qr_package,
f9f4320a 469#if defined(USE_ITHREADS)
f8149455 470 Perl_regdupe_internal
f9f4320a
YO
471#endif
472};
473#endif /* DOINIT */
474#endif /* PLUGGABLE_RE_EXTENSION */
475
476
73c4f7a1 477END_EXTERN_C
cad2e5aa 478
cad2e5aa 479
261faec3
MJD
480/* .what is a character array with one character for each member of .data
481 * The character describes the function of the corresponding .data item:
af534a04 482 * a - AV for paren_name_list under DEBUGGING
0111c4fd 483 * f - start-class data for regstclass optimization
261faec3
MJD
484 * n - Root of op tree for (?{EVAL}) item
485 * o - Start op for (?{EVAL}) item
3dab1dad 486 * p - Pad for (?{EVAL}) item
38a44b82 487 * s - swash for Unicode-style character class, and the multicharacter
9e55ce06
JH
488 * strings resulting from casefolding the single-character entries
489 * in the character class
a3621e74 490 * t - trie struct
55eed653 491 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 492 * also used for revcharmap and words under DEBUGGING
07be1b83 493 * T - aho-trie struct
81714fb9 494 * S - sv for named capture lookup
261faec3 495 * 20010712 mjd@plover.com
c3613ed7 496 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 497 */
cad2e5aa
JH
498struct reg_data {
499 U32 count;
500 U8 *what;
501 void* data[1];
502};
503
a1cac82e
NC
504/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
505 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 506#define anchored_substr substrs->data[0].substr
33b8afdf 507#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 508#define anchored_offset substrs->data[0].min_offset
1de06328
YO
509#define anchored_end_shift substrs->data[0].end_shift
510
cad2e5aa 511#define float_substr substrs->data[1].substr
33b8afdf 512#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
513#define float_min_offset substrs->data[1].min_offset
514#define float_max_offset substrs->data[1].max_offset
1de06328
YO
515#define float_end_shift substrs->data[1].end_shift
516
cad2e5aa 517#define check_substr substrs->data[2].substr
33b8afdf 518#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
519#define check_offset_min substrs->data[2].min_offset
520#define check_offset_max substrs->data[2].max_offset
1de06328 521#define check_end_shift substrs->data[2].end_shift
a3621e74 522
288b8c02
NC
523#define RX_ANCHORED_SUBSTR(rx) (((struct regexp *)SvANY(rx))->anchored_substr)
524#define RX_ANCHORED_UTF8(rx) (((struct regexp *)SvANY(rx))->anchored_utf8)
525#define RX_FLOAT_SUBSTR(rx) (((struct regexp *)SvANY(rx))->float_substr)
526#define RX_FLOAT_UTF8(rx) (((struct regexp *)SvANY(rx))->float_utf8)
a3621e74
YO
527
528/* trie related stuff */
5d9a96ca 529
a3621e74
YO
530/* a transition record for the state machine. the
531 check field determines which state "owns" the
532 transition. the char the transition is for is
533 determined by offset from the owning states base
534 field. the next field determines which state
535 is to be transitioned to if any.
536*/
537struct _reg_trie_trans {
538 U32 next;
539 U32 check;
540};
541
542/* a transition list element for the list based representation */
543struct _reg_trie_trans_list_elem {
544 U16 forid;
545 U32 newstate;
546};
547typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
548
549/* a state for compressed nodes. base is an offset
550 into an array of reg_trie_trans array. If wordnum is
551 nonzero the state is accepting. if base is zero then
552 the state has no children (and will be accepting)
553*/
554struct _reg_trie_state {
555 U16 wordnum;
556 union {
557 U32 base;
558 reg_trie_trans_le* list;
559 } trans;
560};
561
2e64971a
DM
562/* info per word; indexed by wordnum */
563typedef struct {
564 U16 prev; /* previous word in acceptance chain; eg in
565 * zzz|abc|ab/ after matching the chars abc, the
566 * accepted word is #2, and the previous accepted
567 * word is #3 */
568 U32 len; /* how many chars long is this word? */
569 U32 accept; /* accept state for this word */
570} reg_trie_wordinfo;
a3621e74
YO
571
572
a3621e74
YO
573typedef struct _reg_trie_state reg_trie_state;
574typedef struct _reg_trie_trans reg_trie_trans;
575
576
577/* anything in here that needs to be freed later
23eab42c
NC
578 should be dealt with in pregfree.
579 refcount is first in both this and _reg_ac_data to allow a space
580 optimisation in Perl_regdupe. */
a3621e74 581struct _reg_trie_data {
23eab42c 582 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
583 U32 lasttrans; /* last valid transition element */
584 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
585 reg_trie_state *states; /* state data */
586 reg_trie_trans *trans; /* array of transition elements */
587 char *bitmap; /* stclass bitmap */
786e8c11
YO
588 U16 *jump; /* optional 1 indexed array of offsets before tail
589 for the node following a given word. */
2e64971a 590 reg_trie_wordinfo *wordinfo; /* array of info per word */
1b5c067c
NC
591 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
592 U32 startstate; /* initial state - used for common prefix optimisation */
593 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
594 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
2e64971a 595 U32 prefixlen; /* #chars in common prefix */
1e2e3d02
YO
596 U32 statecount; /* Build only - number of states in the states array
597 (including the unused zero state) */
786e8c11 598 U32 wordcount; /* Build only */
a3621e74 599#ifdef DEBUGGING
786e8c11 600 STRLEN charcount; /* Build only */
a3621e74
YO
601#endif
602};
2b8b4781
NC
603/* There is one (3 under DEBUGGING) pointers that logically belong in this
604 structure, but are held outside as they need duplication on thread cloning,
605 whereas the rest of the structure can be read only:
606 HV *widecharmap; code points > 255 to charid
607#ifdef DEBUGGING
608 AV *words; Array of words contained in trie, for dumping
609 AV *revcharmap; Map of each charid back to its character representation
610#endif
611*/
612
613#define TRIE_WORDS_OFFSET 2
614
a3621e74
YO
615typedef struct _reg_trie_data reg_trie_data;
616
23eab42c
NC
617/* refcount is first in both this and _reg_trie_data to allow a space
618 optimisation in Perl_regdupe. */
07be1b83 619struct _reg_ac_data {
23eab42c 620 U32 refcount;
1b5c067c 621 U32 trie;
07be1b83
YO
622 U32 *fail;
623 reg_trie_state *states;
07be1b83
YO
624};
625typedef struct _reg_ac_data reg_ac_data;
626
3dab1dad
YO
627/* ANY_BIT doesnt use the structure, so we can borrow it here.
628 This is simpler than refactoring all of it as wed end up with
629 three different sets... */
630
631#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 632#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
633#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
634#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
635#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 636
1de06328
YO
637#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
638#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
639
640
786e8c11
YO
641#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
642#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 643
a3621e74
YO
644/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
645#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
646#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
647#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
648
3dab1dad 649#ifdef DEBUGGING
3dab1dad 650#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 651#else
3dab1dad 652#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 653#endif
a3621e74 654
0111c4fd
RGS
655#define RE_TRIE_MAXBUF_INIT 65536
656#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
657#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
658
be8e71aa
YO
659/*
660
661RE_DEBUG_FLAGS is used to control what debug output is emitted
662its divided into three groups of options, some of which interact.
663The three groups are: Compile, Execute, Extra. There is room for a
664further group, as currently only the low three bytes are used.
665
666 Compile Options:
07be1b83 667
be8e71aa
YO
668 PARSE
669 PEEP
670 TRIE
671 PROGRAM
672 OFFSETS
a3621e74 673
be8e71aa 674 Execute Options:
a3621e74 675
be8e71aa
YO
676 INTUIT
677 MATCH
678 TRIE
a3621e74 679
be8e71aa 680 Extra Options
a3621e74 681
be8e71aa
YO
682 TRIE
683 OFFSETS
684
685If you modify any of these make sure you make corresponding changes to
686re.pm, especially to the documentation.
a3621e74 687
be8e71aa
YO
688*/
689
690
691/* Compile */
692#define RE_DEBUG_COMPILE_MASK 0x0000FF
693#define RE_DEBUG_COMPILE_PARSE 0x000001
694#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
695#define RE_DEBUG_COMPILE_TRIE 0x000004
696#define RE_DEBUG_COMPILE_DUMP 0x000008
f7819f85 697#define RE_DEBUG_COMPILE_FLAGS 0x000010
be8e71aa
YO
698
699/* Execute */
700#define RE_DEBUG_EXECUTE_MASK 0x00FF00
701#define RE_DEBUG_EXECUTE_INTUIT 0x000100
702#define RE_DEBUG_EXECUTE_MATCH 0x000200
703#define RE_DEBUG_EXECUTE_TRIE 0x000400
704
705/* Extra */
706#define RE_DEBUG_EXTRA_MASK 0xFF0000
707#define RE_DEBUG_EXTRA_TRIE 0x010000
708#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
709#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
710#define RE_DEBUG_EXTRA_STATE 0x080000
711#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 712#define RE_DEBUG_EXTRA_BUFFERS 0x400000
2c296965 713#define RE_DEBUG_EXTRA_GPOS 0x800000
24b23f37
YO
714/* combined */
715#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 716
e68ec53f 717#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
718/* Compile */
719#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 720 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 721#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 722 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 723#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 724 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 725#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 726 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 727#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 728 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 729#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 730 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
f7819f85
A
731#define DEBUG_FLAGS_r(x) DEBUG_r( \
732 if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x )
be8e71aa
YO
733/* Execute */
734#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 735 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 736#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 737 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 738#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 739 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 740#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 741 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
742
743/* Extra */
744#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 745 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
746#define DEBUG_OFFSETS_r(x) DEBUG_r( \
747 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
748#define DEBUG_STATE_r(x) DEBUG_r( \
749 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
750#define DEBUG_STACK_r(x) DEBUG_r( \
751 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
752#define DEBUG_BUFFERS_r(x) DEBUG_r( \
753 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
754
a5ca303d
YO
755#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
756 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
757 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 758#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 759 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
760 Perl_warn_nocontext x )
761#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 762 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 763#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 764 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
765
766#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 767 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa 768 | RE_DEBUG_EXECUTE_TRIE )) x )
2c296965
YO
769#define DEBUG_GPOS_r(x) DEBUG_r( \
770 if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x )
be8e71aa
YO
771
772/* initialization */
595598ee
KW
773/* get_sv() can return NULL during global destruction. re_debug_flags can get
774 * clobbered by a longjmp, so must be initialized */
e68ec53f
YO
775#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
776 SV * re_debug_flags_sv = NULL; \
595598ee 777 re_debug_flags = 0; \
e68ec53f
YO
778 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
779 if (re_debug_flags_sv) { \
8e9a8a48 780 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 781 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 782 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
783 }\
784})
a3621e74
YO
785
786#ifdef DEBUGGING
ab3bbdeb 787
d53d27f9 788#define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
789
790#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
791 const char * const rpv = \
ddc5bc0f 792 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 793 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 794 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 795 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
796
797#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
798 const char * const rpv = \
ddc5bc0f 799 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 800 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 801 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
802
803#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
804 const char * const rpv = \
ddc5bc0f 805 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 806 PL_colors[0], PL_colors[1], \
95b611b0 807 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELLIPSES | \
ab3bbdeb 808 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 809 )
ab3bbdeb
YO
810
811#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
812#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
813
814#else /* if not DEBUGGING */
815
a3621e74 816#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
817#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
818#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
819#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
820#define RE_SV_DUMPLEN(ItEm)
821#define RE_SV_TAIL(ItEm)
822
823#endif /* DEBUG RELATED DEFINES */
a3621e74 824
e9a8c099
MHM
825/*
826 * Local variables:
827 * c-indentation-style: bsd
828 * c-basic-offset: 4
829 * indent-tabs-mode: t
830 * End:
831 *
832 * ex: set ts=8 sts=4 sw=4 noet:
833 */