This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Re: New file: t/op/regexp_email.t
[perl5.git] / regcomp.h
CommitLineData
a0d0e21e 1/* regcomp.h
d6376244 2 *
4bb101f2 3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
699a97de 4 * 2000, 2001, 2002, 2003, 2005, 2006 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
LW
9 */
10
c277df42
IZ
11typedef OP OP_4tree; /* Will be redefined later. */
12
07be1b83 13
58e23c8d 14/* Convert branch sequences to more efficient trie ops? */
07be1b83 15#define PERL_ENABLE_TRIE_OPTIMISATION 1
58e23c8d
YO
16
17/* Be really agressive about optimising patterns with trie sequences? */
07be1b83 18#define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1
58e23c8d
YO
19
20/* Should the optimiser take positive assertions into account? */
68ba3a3f 21#define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0
58e23c8d
YO
22
23/* Not for production use: */
07be1b83 24#define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0
58e23c8d 25
7122b237
YO
26/* Activate offsets code - set to if 1 to enable */
27#ifdef DEBUGGING
28#define RE_TRACK_PATTERN_OFFSETS
29#endif
30
b9b4dddf
YO
31/* Unless the next line is uncommented it is illegal to combine lazy
32 matching with possessive matching. Frankly it doesn't make much sense
33 to allow it as X*?+ matches nothing, X+?+ matches a single char only,
34 and X{min,max}?+ matches min times only.
35 */
36/* #define REG_ALLOW_MINMOD_SUSPEND */
07be1b83 37
a687059c
LW
38/*
39 * The "internal use only" fields in regexp.h are present to pass info from
40 * compile to execute that permits the execute phase to run lots faster on
41 * simple cases. They are:
42 *
bd61b366 43 * regstart sv that must begin a match; NULL if none obvious
a687059c
LW
44 * reganch is the match anchored (at beginning-of-line only)?
45 * regmust string (pointer into program) that match must include, or NULL
79072805 46 * [regmust changed to SV* for bminstr()--law]
a687059c
LW
47 * regmlen length of regmust string
48 * [regmlen not used currently]
49 *
50 * Regstart and reganch permit very fast decisions on suitable starting points
51 * for a match, cutting down the work a lot. Regmust permits fast rejection
52 * of lines that cannot possibly match. The regmust tests are costly enough
e50aee73 53 * that pregcomp() supplies a regmust only if the r.e. contains something
a687059c
LW
54 * potentially expensive (at present, the only such thing detected is * or +
55 * at the start of the r.e., which can involve a lot of backup). Regmlen is
e50aee73 56 * supplied because the test in pregexec() needs it and pregcomp() is computing
a687059c
LW
57 * it anyway.
58 * [regmust is now supplied always. The tests that use regmust have a
59 * heuristic that disables the test if it usually matches.]
60 *
61 * [In fact, we now use regmust in many cases to locate where the search
62 * starts in the string, so if regback is >= 0, the regmust search is never
63 * wasted effort. The regback variable says how many characters back from
64 * where regmust matched is the earliest possible start of the match.
65 * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.]
66 */
67
68/*
69 * Structure for regexp "program". This is essentially a linear encoding
70 * of a nondeterministic finite-state machine (aka syntax charts or
71 * "railroad normal form" in parsing technology). Each node is an opcode
72 * plus a "next" pointer, possibly plus an operand. "Next" pointers of
73 * all nodes except BRANCH implement concatenation; a "next" pointer with
74 * a BRANCH on both ends of it is connecting two alternatives. (Here we
75 * have one of the subtle syntax dependencies: an individual BRANCH (as
76 * opposed to a collection of them) is never concatenated with anything
77 * because of operator precedence.) The operand of some types of node is
78 * a literal string; for others, it is a node leading into a sub-FSM. In
79 * particular, the operand of a BRANCH node is the first node of the branch.
80 * (NB this is *not* a tree structure: the tail of the branch connects
a3621e74
YO
81 * to the thing following the set of BRANCHes.) The opcodes are defined
82 * in regnodes.h which is generated from regcomp.sym by regcomp.pl.
a687059c
LW
83 */
84
a687059c
LW
85/*
86 * A node is one char of opcode followed by two chars of "next" pointer.
87 * "Next" pointers are stored as two 8-bit pieces, high order first. The
88 * value is a positive offset from the opcode of the node containing it.
89 * An operand, if any, simply follows the node. (Note that much of the
90 * code generation knows about this implicit relationship.)
91 *
92 * Using two bytes for the "next" pointer is vast overkill for most things,
93 * but allows patterns to get big without disasters.
94 *
e91177ed 95 * [The "next" pointer is always aligned on an even
a687059c
LW
96 * boundary, and reads the offset directly as a short. Also, there is no
97 * special test to reverse the sign of BACK pointers since the offset is
98 * stored negative.]
99 */
785a26d5
YO
100
101/* This is the stuff that used to live in regexp.h that was truly
102 private to the engine itself. It now lives here. */
103
104/* swap buffer for paren structs */
105typedef struct regexp_paren_ofs {
106 I32 *startp;
107 I32 *endp;
108} regexp_paren_ofs;
109
110 typedef struct regexp_internal {
1f1031fe 111 int name_list_idx; /* Optional data index of an array of paren names */
7122b237
YO
112 union {
113 U32 *offsets; /* offset annotations 20001228 MJD
114 data about mapping the program to the
115 string -
116 offsets[0] is proglen when this is used
117 */
118 U32 proglen;
119 } u;
120
1f1031fe 121 regexp_paren_ofs *swap; /* Swap copy of *startp / *endp */
6d5c990f
RGS
122 regnode *regstclass; /* Optional startclass as identified or constructed
123 by the optimiser */
124 struct reg_data *data; /* Additional miscellaneous data used by the program.
125 Used to make it easier to clone and free arbitrary
126 data that the regops need. Often the ARG field of
127 a regop is an index into this structure */
128 regnode program[1]; /* Unwarranted chumminess with compiler. */
129} regexp_internal;
130
131#define RXi_SET(x,y) (x)->pprivate = (void*)(y)
132#define RXi_GET(x) ((regexp_internal *)((x)->pprivate))
133#define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r)
785a26d5
YO
134/*
135 * Flags stored in regexp->intflags
136 * These are used only internally to the regexp engine
137 *
138 * See regexp.h for flags used externally to the regexp engine
139 */
140#define PREGf_SKIP 0x00000001
141#define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */
142#define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */
143#define PREGf_VERBARG_SEEN 0x00000008
144#define PREGf_CUTGROUP_SEEN 0x00000010
145
146
147/* this is where the old regcomp.h started */
a687059c 148
c277df42 149struct regnode_string {
cd439c50 150 U8 str_len;
c277df42
IZ
151 U8 type;
152 U16 next_off;
cd439c50 153 char string[1];
c277df42 154};
a687059c 155
6bda09f9
YO
156/* Argument bearing node - workhorse,
157 arg1 is often for the data field */
c277df42
IZ
158struct regnode_1 {
159 U8 flags;
160 U8 type;
161 U16 next_off;
162 U32 arg1;
163};
164
6bda09f9
YO
165/* Similar to a regnode_1 but with an extra signed argument */
166struct regnode_2L {
167 U8 flags;
168 U8 type;
169 U16 next_off;
170 U32 arg1;
171 I32 arg2;
172};
173
174/* 'Two field' -- Two 16 bit unsigned args */
c277df42
IZ
175struct regnode_2 {
176 U8 flags;
177 U8 type;
178 U16 next_off;
179 U16 arg1;
180 U16 arg2;
181};
182
6bda09f9 183
936ed897 184#define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */
ffc61ed2 185#define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */
936ed897 186
786e8c11 187/* also used by trie */
936ed897
IZ
188struct regnode_charclass {
189 U8 flags;
190 U8 type;
191 U16 next_off;
ffc61ed2 192 U32 arg1;
19103b91 193 char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */
936ed897
IZ
194};
195
19103b91
JH
196struct regnode_charclass_class { /* has [[:blah:]] classes */
197 U8 flags; /* should have ANYOF_CLASS here */
936ed897
IZ
198 U8 type;
199 U16 next_off;
ffc61ed2 200 U32 arg1;
19103b91
JH
201 char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */
202 char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */
936ed897
IZ
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
IZ
258#define OP(p) ((p)->type)
259#define OPERAND(p) (((struct regnode_string *)p)->string)
cd439c50
IZ
260#define MASK(p) ((char*)OPERAND(p))
261#define STR_LEN(p) (((struct regnode_string *)p)->str_len)
262#define STRING(p) (((struct regnode_string *)p)->string)
263#define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode))
264#define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1)
265
e8e3b0fb
JH
266#undef NODE_ALIGN
267#undef ARG_LOC
268#undef NEXTOPER
269#undef PREVOPER
270
e91177ed
IZ
271#define NODE_ALIGN(node)
272#define ARG_LOC(p) (((struct regnode_1 *)p)->arg1)
273#define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1)
274#define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2)
6bda09f9 275#define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2)
786e8c11 276
e91177ed
IZ
277#define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */
278#define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2)
279
280#define NODE_STEP_B 4
c277df42
IZ
281
282#define NEXTOPER(p) ((p) + NODE_STEP_REGNODE)
283#define PREVOPER(p) ((p) - NODE_STEP_REGNODE)
284
e91177ed 285#define FILL_ADVANCE_NODE(ptr, op) STMT_START { \
c277df42 286 (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END
e91177ed 287#define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \
c277df42 288 ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END
a687059c 289
22c35a8c 290#define REG_MAGIC 0234
a687059c 291
830247a4 292#define SIZE_ONLY (RExC_emit == &PL_regdummy)
c277df42 293
936ed897 294/* Flags for node->flags of ANYOF */
b8c5462f 295
19103b91 296#define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */
ffc61ed2
JH
297#define ANYOF_INVERT 0x04
298#define ANYOF_FOLD 0x02
299#define ANYOF_LOCALE 0x01
b8c5462f 300
653099ff 301/* Used for regstclass only */
ffc61ed2
JH
302#define ANYOF_EOS 0x10 /* Can match an empty string too */
303
304/* There is a character or a range past 0xff */
305#define ANYOF_UNICODE 0x20
1aa99e6b 306#define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */
ffc61ed2 307
4f66b38d
HS
308/* size of node is large (includes class pointer) */
309#define ANYOF_LARGE 0x80
310
ffc61ed2
JH
311/* Are there any runtime flags on in this node? */
312#define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f)
313
314#define ANYOF_FLAGS_ALL 0xff
653099ff 315
936ed897 316/* Character classes for node->classflags of ANYOF */
653099ff
GS
317/* Should be synchronized with a table in regprop() */
318/* 2n should pair with 2n+1 */
b8c5462f 319
289d4f09 320#define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */
b8c5462f 321#define ANYOF_NALNUM 1
aaa51d5e 322#define ANYOF_SPACE 2 /* \s */
b8c5462f
JH
323#define ANYOF_NSPACE 3
324#define ANYOF_DIGIT 4
325#define ANYOF_NDIGIT 5
289d4f09 326#define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */
b8c5462f
JH
327#define ANYOF_NALNUMC 7
328#define ANYOF_ALPHA 8
329#define ANYOF_NALPHA 9
330#define ANYOF_ASCII 10
331#define ANYOF_NASCII 11
332#define ANYOF_CNTRL 12
333#define ANYOF_NCNTRL 13
334#define ANYOF_GRAPH 14
335#define ANYOF_NGRAPH 15
336#define ANYOF_LOWER 16
337#define ANYOF_NLOWER 17
338#define ANYOF_PRINT 18
339#define ANYOF_NPRINT 19
340#define ANYOF_PUNCT 20
341#define ANYOF_NPUNCT 21
342#define ANYOF_UPPER 22
343#define ANYOF_NUPPER 23
344#define ANYOF_XDIGIT 24
345#define ANYOF_NXDIGIT 25
aaa51d5e
JF
346#define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */
347#define ANYOF_NPSXSPC 27
ffc61ed2 348#define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */
289d4f09 349#define ANYOF_NBLANK 29
b8c5462f 350
aaa51d5e 351#define ANYOF_MAX 32
b8c5462f
JH
352
353/* Backward source code compatibility. */
354
355#define ANYOF_ALNUML ANYOF_ALNUM
356#define ANYOF_NALNUML ANYOF_NALNUM
357#define ANYOF_SPACEL ANYOF_SPACE
358#define ANYOF_NSPACEL ANYOF_NSPACE
359
360/* Utility macros for the bitmap and classes of ANYOF */
361
936ed897
IZ
362#define ANYOF_SIZE (sizeof(struct regnode_charclass))
363#define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class))
b8c5462f 364
936ed897 365#define ANYOF_FLAGS(p) ((p)->flags)
b8c5462f
JH
366
367#define ANYOF_BIT(c) (1 << ((c) & 7))
368
936ed897 369#define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3])
b8c5462f
JH
370#define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c))
371#define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c))
372#define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c))
373
936ed897
IZ
374#define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char)
375#define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char)
376
377#define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap)
786e8c11 378#define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31])
b8c5462f
JH
379#define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c))
380#define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c))
381#define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c))
382
f8bef550
NC
383#define ANYOF_BITMAP_SETALL(p) \
384 memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE)
385#define ANYOF_BITMAP_CLEARALL(p) \
386 Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE)
387/* Check that all 256 bits are all set. Used in S_cl_is_anything() */
388#define ANYOF_BITMAP_TESTALLSET(p) \
389 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)
390
936ed897
IZ
391#define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode))
392#define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode))
393#define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP)
c277df42 394
786e8c11 395
a687059c
LW
396/*
397 * Utility definitions.
398 */
2304df62 399#ifndef CHARMASK
075abff3 400# define UCHARAT(p) ((int)*(const U8*)(p))
a687059c 401#else
075abff3 402# define UCHARAT(p) ((int)*(p)&CHARMASK)
a687059c 403#endif
a687059c 404
c277df42
IZ
405#define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode))
406
3dab1dad
YO
407#define REG_SEEN_ZERO_LEN 0x00000001
408#define REG_SEEN_LOOKBEHIND 0x00000002
409#define REG_SEEN_GPOS 0x00000004
410#define REG_SEEN_EVAL 0x00000008
411#define REG_SEEN_CANY 0x00000010
cce850e4 412#define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */
6bda09f9 413#define REG_SEEN_RECURSE 0x00000020
eaf3ca90 414#define REG_TOP_LEVEL_BRANCHES 0x00000040
e2e6a0f1 415#define REG_SEEN_VERBARG 0x00000080
5d458dd8 416#define REG_SEEN_CUTGROUP 0x00000100
bcdf7404 417#define REG_SEEN_RUN_ON_COMMENT 0x00000200
d09b2d29 418
73c4f7a1
GS
419START_EXTERN_C
420
be8e71aa
YO
421#ifdef PLUGGABLE_RE_EXTENSION
422#include "re_nodes.h"
423#else
d09b2d29 424#include "regnodes.h"
be8e71aa 425#endif
d09b2d29 426
4d61ec05 427/* The following have no fixed length. U8 so we can do strchr() on it. */
d09b2d29 428#ifndef DOINIT
4d61ec05 429EXTCONST U8 PL_varies[];
d09b2d29 430#else
4d61ec05 431EXTCONST U8 PL_varies[] = {
0111c4fd 432 BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL,
99ed40ff
YO
433 WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP,
434 NREF, NREFF, NREFFL,
435 0
c277df42 436};
d09b2d29 437#endif
c277df42 438
4d61ec05
GS
439/* The following always have a length of 1. U8 we can do strchr() on it. */
440/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
d09b2d29 441#ifndef DOINIT
4d61ec05 442EXTCONST U8 PL_simple[];
d09b2d29 443#else
4d61ec05 444EXTCONST U8 PL_simple[] = {
f33976b4 445 REG_ANY, SANY, CANY,
ffc61ed2
JH
446 ANYOF,
447 ALNUM, ALNUML,
448 NALNUM, NALNUML,
449 SPACE, SPACEL,
450 NSPACE, NSPACEL,
451 DIGIT, NDIGIT,
452 0
c277df42
IZ
453};
454#endif
455
f9f4320a
YO
456#ifndef PLUGGABLE_RE_EXTENSION
457#ifndef DOINIT
458EXTCONST regexp_engine PL_core_reg_engine;
459#else /* DOINIT */
460EXTCONST regexp_engine PL_core_reg_engine = {
3c6f95d1 461 Perl_re_compile,
f9f4320a
YO
462 Perl_regexec_flags,
463 Perl_re_intuit_start,
464 Perl_re_intuit_string,
f8149455 465 Perl_regfree_internal,
93b32b6d
YO
466 Perl_reg_numbered_buff_get,
467 Perl_reg_named_buff_get,
f9f4320a 468#if defined(USE_ITHREADS)
f8149455 469 Perl_regdupe_internal
f9f4320a
YO
470#endif
471};
472#endif /* DOINIT */
473#endif /* PLUGGABLE_RE_EXTENSION */
474
475
73c4f7a1 476END_EXTERN_C
cad2e5aa 477
cad2e5aa 478
261faec3
MJD
479/* .what is a character array with one character for each member of .data
480 * The character describes the function of the corresponding .data item:
0111c4fd 481 * f - start-class data for regstclass optimization
261faec3
MJD
482 * n - Root of op tree for (?{EVAL}) item
483 * o - Start op for (?{EVAL}) item
3dab1dad 484 * p - Pad for (?{EVAL}) item
9e55ce06
JH
485 * s - swash for unicode-style character class, and the multicharacter
486 * strings resulting from casefolding the single-character entries
487 * in the character class
a3621e74 488 * t - trie struct
55eed653 489 * u - trie struct's widecharmap (a HV, so can't share, must dup)
2b8b4781 490 * also used for revcharmap and words under DEBUGGING
07be1b83 491 * T - aho-trie struct
81714fb9 492 * S - sv for named capture lookup
261faec3 493 * 20010712 mjd@plover.com
c3613ed7 494 * (Remember to update re_dup() and pregfree() if you add any items.)
261faec3 495 */
cad2e5aa
JH
496struct reg_data {
497 U32 count;
498 U8 *what;
499 void* data[1];
500};
501
a1cac82e
NC
502/* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses
503 anchored* and float* via array indexes 0 and 1. */
cad2e5aa 504#define anchored_substr substrs->data[0].substr
33b8afdf 505#define anchored_utf8 substrs->data[0].utf8_substr
cad2e5aa 506#define anchored_offset substrs->data[0].min_offset
1de06328
YO
507#define anchored_end_shift substrs->data[0].end_shift
508
cad2e5aa 509#define float_substr substrs->data[1].substr
33b8afdf 510#define float_utf8 substrs->data[1].utf8_substr
cad2e5aa
JH
511#define float_min_offset substrs->data[1].min_offset
512#define float_max_offset substrs->data[1].max_offset
1de06328
YO
513#define float_end_shift substrs->data[1].end_shift
514
cad2e5aa 515#define check_substr substrs->data[2].substr
33b8afdf 516#define check_utf8 substrs->data[2].utf8_substr
cad2e5aa
JH
517#define check_offset_min substrs->data[2].min_offset
518#define check_offset_max substrs->data[2].max_offset
1de06328 519#define check_end_shift substrs->data[2].end_shift
a3621e74
YO
520
521
522
523/* trie related stuff */
5d9a96ca 524
a3621e74
YO
525/* a transition record for the state machine. the
526 check field determines which state "owns" the
527 transition. the char the transition is for is
528 determined by offset from the owning states base
529 field. the next field determines which state
530 is to be transitioned to if any.
531*/
532struct _reg_trie_trans {
533 U32 next;
534 U32 check;
535};
536
537/* a transition list element for the list based representation */
538struct _reg_trie_trans_list_elem {
539 U16 forid;
540 U32 newstate;
541};
542typedef struct _reg_trie_trans_list_elem reg_trie_trans_le;
543
544/* a state for compressed nodes. base is an offset
545 into an array of reg_trie_trans array. If wordnum is
546 nonzero the state is accepting. if base is zero then
547 the state has no children (and will be accepting)
548*/
549struct _reg_trie_state {
550 U16 wordnum;
551 union {
552 U32 base;
553 reg_trie_trans_le* list;
554 } trans;
555};
556
557
558
a3621e74
YO
559typedef struct _reg_trie_state reg_trie_state;
560typedef struct _reg_trie_trans reg_trie_trans;
561
562
563/* anything in here that needs to be freed later
23eab42c
NC
564 should be dealt with in pregfree.
565 refcount is first in both this and _reg_ac_data to allow a space
566 optimisation in Perl_regdupe. */
a3621e74 567struct _reg_trie_data {
23eab42c 568 U32 refcount; /* number of times this trie is referenced */
786e8c11
YO
569 U16 uniquecharcount; /* unique chars in trie (width of trans table) */
570 U32 lasttrans; /* last valid transition element */
571 U16 *charmap; /* byte to charid lookup array */
786e8c11
YO
572 reg_trie_state *states; /* state data */
573 reg_trie_trans *trans; /* array of transition elements */
574 char *bitmap; /* stclass bitmap */
786e8c11
YO
575 U32 startstate; /* initial state - used for common prefix optimisation */
576 STRLEN minlen; /* minimum length of words in trie - build/opt only? */
577 STRLEN maxlen; /* maximum length of words in trie - build/opt only? */
578 U32 *wordlen; /* array of lengths of words */
579 U16 *jump; /* optional 1 indexed array of offsets before tail
580 for the node following a given word. */
581 U16 *nextword; /* optional 1 indexed array to support linked list
582 of duplicate wordnums */
1e2e3d02
YO
583 U32 statecount; /* Build only - number of states in the states array
584 (including the unused zero state) */
786e8c11 585 U32 wordcount; /* Build only */
a3621e74 586#ifdef DEBUGGING
786e8c11 587 STRLEN charcount; /* Build only */
a3621e74
YO
588#endif
589};
2b8b4781
NC
590/* There is one (3 under DEBUGGING) pointers that logically belong in this
591 structure, but are held outside as they need duplication on thread cloning,
592 whereas the rest of the structure can be read only:
593 HV *widecharmap; code points > 255 to charid
594#ifdef DEBUGGING
595 AV *words; Array of words contained in trie, for dumping
596 AV *revcharmap; Map of each charid back to its character representation
597#endif
598*/
599
600#define TRIE_WORDS_OFFSET 2
601
a3621e74
YO
602typedef struct _reg_trie_data reg_trie_data;
603
23eab42c
NC
604/* refcount is first in both this and _reg_trie_data to allow a space
605 optimisation in Perl_regdupe. */
07be1b83 606struct _reg_ac_data {
23eab42c 607 U32 refcount;
07be1b83
YO
608 U32 *fail;
609 reg_trie_state *states;
3251b653 610 U32 trie;
07be1b83
YO
611};
612typedef struct _reg_ac_data reg_ac_data;
613
3dab1dad
YO
614/* ANY_BIT doesnt use the structure, so we can borrow it here.
615 This is simpler than refactoring all of it as wed end up with
616 three different sets... */
617
618#define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap)
786e8c11 619#define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31])
5dfe062f
YO
620#define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c))
621#define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c))
622#define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 623
1de06328
YO
624#define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC)
625#define IS_TRIE_AC(op) ((op)>=AHOCORASICK)
626
627
786e8c11
YO
628#define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31])
629#define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c))
3dab1dad 630
a3621e74
YO
631/* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */
632#define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1)
633#define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state))
634#define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state))
635
3dab1dad 636#ifdef DEBUGGING
3dab1dad 637#define TRIE_CHARCOUNT(trie) ((trie)->charcount)
3dab1dad 638#else
3dab1dad 639#define TRIE_CHARCOUNT(trie) (trie_charcount)
3dab1dad 640#endif
a3621e74 641
0111c4fd
RGS
642#define RE_TRIE_MAXBUF_INIT 65536
643#define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF"
a3621e74
YO
644#define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS"
645
be8e71aa
YO
646/*
647
648RE_DEBUG_FLAGS is used to control what debug output is emitted
649its divided into three groups of options, some of which interact.
650The three groups are: Compile, Execute, Extra. There is room for a
651further group, as currently only the low three bytes are used.
652
653 Compile Options:
07be1b83 654
be8e71aa
YO
655 PARSE
656 PEEP
657 TRIE
658 PROGRAM
659 OFFSETS
a3621e74 660
be8e71aa 661 Execute Options:
a3621e74 662
be8e71aa
YO
663 INTUIT
664 MATCH
665 TRIE
a3621e74 666
be8e71aa 667 Extra Options
a3621e74 668
be8e71aa
YO
669 TRIE
670 OFFSETS
671
672If you modify any of these make sure you make corresponding changes to
673re.pm, especially to the documentation.
a3621e74 674
be8e71aa
YO
675*/
676
677
678/* Compile */
679#define RE_DEBUG_COMPILE_MASK 0x0000FF
680#define RE_DEBUG_COMPILE_PARSE 0x000001
681#define RE_DEBUG_COMPILE_OPTIMISE 0x000002
682#define RE_DEBUG_COMPILE_TRIE 0x000004
683#define RE_DEBUG_COMPILE_DUMP 0x000008
be8e71aa
YO
684
685/* Execute */
686#define RE_DEBUG_EXECUTE_MASK 0x00FF00
687#define RE_DEBUG_EXECUTE_INTUIT 0x000100
688#define RE_DEBUG_EXECUTE_MATCH 0x000200
689#define RE_DEBUG_EXECUTE_TRIE 0x000400
690
691/* Extra */
692#define RE_DEBUG_EXTRA_MASK 0xFF0000
693#define RE_DEBUG_EXTRA_TRIE 0x010000
694#define RE_DEBUG_EXTRA_OFFSETS 0x020000
a5ca303d
YO
695#define RE_DEBUG_EXTRA_OFFDEBUG 0x040000
696#define RE_DEBUG_EXTRA_STATE 0x080000
697#define RE_DEBUG_EXTRA_OPTIMISE 0x100000
e7707071 698#define RE_DEBUG_EXTRA_BUFFERS 0x400000
24b23f37
YO
699/* combined */
700#define RE_DEBUG_EXTRA_STACK 0x280000
be8e71aa 701
e68ec53f 702#define RE_DEBUG_FLAG(x) (re_debug_flags & x)
be8e71aa
YO
703/* Compile */
704#define DEBUG_COMPILE_r(x) DEBUG_r( \
e68ec53f 705 if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x )
be8e71aa 706#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 707 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 708#define DEBUG_OPTIMISE_r(x) DEBUG_r( \
e68ec53f 709 if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x )
be8e71aa 710#define DEBUG_PARSE_r(x) DEBUG_r( \
e68ec53f 711 if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x )
be8e71aa 712#define DEBUG_DUMP_r(x) DEBUG_r( \
e68ec53f 713 if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x )
be8e71aa 714#define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \
e68ec53f 715 if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x )
be8e71aa
YO
716
717/* Execute */
718#define DEBUG_EXECUTE_r(x) DEBUG_r( \
e68ec53f 719 if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x )
be8e71aa 720#define DEBUG_INTUIT_r(x) DEBUG_r( \
e68ec53f 721 if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x )
be8e71aa 722#define DEBUG_MATCH_r(x) DEBUG_r( \
e68ec53f 723 if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x )
be8e71aa 724#define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \
e68ec53f 725 if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x )
be8e71aa
YO
726
727/* Extra */
728#define DEBUG_EXTRA_r(x) DEBUG_r( \
e68ec53f 729 if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x )
a5ca303d
YO
730#define DEBUG_OFFSETS_r(x) DEBUG_r( \
731 if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x )
ab3bbdeb
YO
732#define DEBUG_STATE_r(x) DEBUG_r( \
733 if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x )
24b23f37
YO
734#define DEBUG_STACK_r(x) DEBUG_r( \
735 if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x )
e7707071
YO
736#define DEBUG_BUFFERS_r(x) DEBUG_r( \
737 if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x )
738
a5ca303d
YO
739#define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \
740 if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \
741 (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x )
07be1b83 742#define MJD_OFFSET_DEBUG(x) DEBUG_r( \
a5ca303d 743 if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \
be8e71aa
YO
744 Perl_warn_nocontext x )
745#define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \
e68ec53f 746 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa 747#define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \
e68ec53f 748 if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x )
be8e71aa
YO
749
750#define DEBUG_TRIE_r(x) DEBUG_r( \
e68ec53f 751 if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \
be8e71aa
YO
752 | RE_DEBUG_EXECUTE_TRIE )) x )
753
754/* initialization */
755/* get_sv() can return NULL during global destruction. */
e68ec53f
YO
756#define GET_RE_DEBUG_FLAGS DEBUG_r({ \
757 SV * re_debug_flags_sv = NULL; \
758 re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \
759 if (re_debug_flags_sv) { \
8e9a8a48 760 if (!SvIOK(re_debug_flags_sv)) \
e68ec53f 761 sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \
8e9a8a48 762 re_debug_flags=SvIV(re_debug_flags_sv); \
e68ec53f
YO
763 }\
764})
a3621e74
YO
765
766#ifdef DEBUGGING
ab3bbdeb 767
8e9a8a48 768#define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS;
ab3bbdeb
YO
769
770#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \
771 const char * const rpv = \
ddc5bc0f 772 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 773 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 774 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \
0df25f3d 775 const int rlen = SvCUR(dsv)
ab3bbdeb
YO
776
777#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \
778 const char * const rpv = \
ddc5bc0f 779 pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \
ab3bbdeb 780 PL_colors[(c1)],PL_colors[(c2)], \
0a0e9afc 781 PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) )
ab3bbdeb
YO
782
783#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \
784 const char * const rpv = \
ddc5bc0f 785 pv_pretty((dsv), (pv), (l), (m), \
ab3bbdeb 786 PL_colors[0], PL_colors[1], \
0a0e9afc 787 ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELIPSES | \
ab3bbdeb 788 ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \
0a0e9afc 789 )
ab3bbdeb
YO
790
791#define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0))
792#define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "")
793
794#else /* if not DEBUGGING */
795
a3621e74 796#define GET_RE_DEBUG_FLAGS_DECL
ab3bbdeb
YO
797#define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2)
798#define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m)
799#define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m)
800#define RE_SV_DUMPLEN(ItEm)
801#define RE_SV_TAIL(ItEm)
802
803#endif /* DEBUG RELATED DEFINES */
a3621e74 804
99ed40ff 805