Commit | Line | Data |
---|---|---|
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 |
62e6ef33 | 87 | * boundary, and reads the offset directly as a short.] |
a687059c | 88 | */ |
785a26d5 YO |
89 | |
90 | /* This is the stuff that used to live in regexp.h that was truly | |
91 | private to the engine itself. It now lives here. */ | |
92 | ||
28d8d7f4 | 93 | |
785a26d5 YO |
94 | |
95 | typedef struct regexp_internal { | |
1f1031fe | 96 | int name_list_idx; /* Optional data index of an array of paren names */ |
7122b237 YO |
97 | union { |
98 | U32 *offsets; /* offset annotations 20001228 MJD | |
99 | data about mapping the program to the | |
100 | string - | |
101 | offsets[0] is proglen when this is used | |
102 | */ | |
103 | U32 proglen; | |
104 | } u; | |
105 | ||
6d5c990f RGS |
106 | regnode *regstclass; /* Optional startclass as identified or constructed |
107 | by the optimiser */ | |
108 | struct reg_data *data; /* Additional miscellaneous data used by the program. | |
109 | Used to make it easier to clone and free arbitrary | |
110 | data that the regops need. Often the ARG field of | |
111 | a regop is an index into this structure */ | |
3d2bd50a DM |
112 | struct reg_code_block *code_blocks;/* positions of literal (?{}) */ |
113 | int num_code_blocks; /* size of code_blocks[] */ | |
6d5c990f RGS |
114 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
115 | } regexp_internal; | |
116 | ||
117 | #define RXi_SET(x,y) (x)->pprivate = (void*)(y) | |
118 | #define RXi_GET(x) ((regexp_internal *)((x)->pprivate)) | |
119 | #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r) | |
785a26d5 YO |
120 | /* |
121 | * Flags stored in regexp->intflags | |
122 | * These are used only internally to the regexp engine | |
123 | * | |
124 | * See regexp.h for flags used externally to the regexp engine | |
125 | */ | |
e3e400ec YO |
126 | #define RXp_INTFLAGS(rx) ((rx)->intflags) |
127 | #define RX_INTFLAGS(prog) RXp_INTFLAGS(ReANY(prog)) | |
128 | ||
785a26d5 YO |
129 | #define PREGf_SKIP 0x00000001 |
130 | #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */ | |
131 | #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */ | |
132 | #define PREGf_VERBARG_SEEN 0x00000008 | |
133 | #define PREGf_CUTGROUP_SEEN 0x00000010 | |
732caac7 | 134 | #define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */ |
e3e400ec YO |
135 | /* these used to be extflags, but are now intflags */ |
136 | #define PREGf_NOSCAN 0x00000040 | |
33c28ab2 | 137 | /* spare */ |
8e1490ee YO |
138 | #define PREGf_GPOS_SEEN 0x00000100 |
139 | #define PREGf_GPOS_FLOAT 0x00000200 | |
785a26d5 | 140 | |
d3d47aac YO |
141 | #define PREGf_ANCH_MBOL 0x00000400 |
142 | #define PREGf_ANCH_SBOL 0x00000800 | |
143 | #define PREGf_ANCH_GPOS 0x00001000 | |
8e1490ee | 144 | |
d3d47aac YO |
145 | #define PREGf_ANCH \ |
146 | ( PREGf_ANCH_SBOL | PREGf_ANCH_GPOS | PREGf_ANCH_MBOL ) | |
785a26d5 YO |
147 | |
148 | /* this is where the old regcomp.h started */ | |
a687059c | 149 | |
c277df42 | 150 | struct regnode_string { |
cd439c50 | 151 | U8 str_len; |
c277df42 IZ |
152 | U8 type; |
153 | U16 next_off; | |
cd439c50 | 154 | char string[1]; |
c277df42 | 155 | }; |
a687059c | 156 | |
6bda09f9 YO |
157 | /* Argument bearing node - workhorse, |
158 | arg1 is often for the data field */ | |
c277df42 IZ |
159 | struct regnode_1 { |
160 | U8 flags; | |
161 | U8 type; | |
162 | U16 next_off; | |
163 | U32 arg1; | |
164 | }; | |
165 | ||
6bda09f9 YO |
166 | /* Similar to a regnode_1 but with an extra signed argument */ |
167 | struct regnode_2L { | |
168 | U8 flags; | |
169 | U8 type; | |
170 | U16 next_off; | |
171 | U32 arg1; | |
172 | I32 arg2; | |
173 | }; | |
174 | ||
175 | /* 'Two field' -- Two 16 bit unsigned args */ | |
c277df42 IZ |
176 | struct regnode_2 { |
177 | U8 flags; | |
178 | U8 type; | |
179 | U16 next_off; | |
180 | U16 arg1; | |
181 | U16 arg2; | |
182 | }; | |
183 | ||
e0a1ff7a KW |
184 | /* This give the number of code points that can be in the bitmap of an ANYOF |
185 | * node. The shift number must currently be one of: 8..12. It can't be less | |
186 | * than 8 (256) because some code relies on it being at least that. Above 12 | |
187 | * (4096), and you start running into warnings that some data structure widths | |
188 | * have been exceeded, though the test suite as of this writing still passes | |
189 | * for up through 16, which is as high as anyone would ever want to go, | |
190 | * encompassing all of the Unicode BMP, and thus including all the economically | |
191 | * important world scripts. At 12 most of them are: including Arabic, | |
192 | * Cyrillic, Greek, Hebrew, Indian subcontinent, Latin, and Thai; but not Han, | |
193 | * Japanese, nor Korean. (The regarglen structure in regnodes.h is a U8, and | |
194 | * the trie types TRIEC and AHOCORASICKC are larger than U8 for shift values | |
195 | * below above 12.) Be sure to benchmark before changing, as larger sizes do | |
196 | * significantly slow down the test suite */ | |
197 | #define NUM_ANYOF_CODE_POINTS (1 << 8) | |
6bda09f9 | 198 | |
dcb20b36 | 199 | #define ANYOF_BITMAP_SIZE (NUM_ANYOF_CODE_POINTS / 8) /* 8 bits/Byte */ |
936ed897 | 200 | |
1ee208c4 KW |
201 | /* Note that these form structs which are supersets of the next smaller one, by |
202 | * appending fields. Alignment problems can occur if one of those optional | |
203 | * fields requires stricter alignment than the base struct. And formal | |
204 | * parameters that can really be two or more of the structs should be | |
205 | * declared as the smallest one it could be. See commit message for | |
206 | * 7dcac5f6a5195002b55c935ee1d67f67e1df280b. Regnode allocation is done | |
207 | * without regard to alignment, and changing it to would also require changing | |
208 | * the code that inserts and deletes regnodes. The basic single-argument | |
209 | * regnode has a U32, which is what reganode() allocates as a unit. Therefore | |
210 | * no field can require stricter alignment than U32. */ | |
211 | ||
786e8c11 | 212 | /* also used by trie */ |
936ed897 IZ |
213 | struct regnode_charclass { |
214 | U8 flags; | |
215 | U8 type; | |
216 | U16 next_off; | |
9da99c5e | 217 | U32 arg1; /* set by set_ANYOF_arg() */ |
19103b91 | 218 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 IZ |
219 | }; |
220 | ||
bf3f2d85 KW |
221 | /* has runtime (locale) \d, \w, ..., [:posix:] classes */ |
222 | struct regnode_charclass_class { | |
93e92956 | 223 | U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */ |
936ed897 IZ |
224 | U8 type; |
225 | U16 next_off; | |
7e817188 | 226 | U32 arg1; |
03fa83ba | 227 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */ |
9cf78fa0 | 228 | U32 classflags; /* and run-time */ |
936ed897 IZ |
229 | }; |
230 | ||
85c8e306 KW |
231 | /* A synthetic start class (SSC); is a regnode_charclass_posixl_fold, plus an |
232 | * extra SV*, used only during its construction and which is not used by | |
233 | * regexec.c. Note that the 'next_off' field is unused, as the SSC stands | |
234 | * alone, so there is never a next node. Also, there is no alignment issue, | |
b8fda935 | 235 | * because these are declared or allocated as a complete unit so the compiler |
85c8e306 KW |
236 | * takes care of alignment. This is unlike the other regnodes which are |
237 | * allocated in terms of multiples of a single-argument regnode. SSC nodes can | |
238 | * have a pointer field because there is no alignment issue, and because it is | |
239 | * set to NULL after construction, before any cloning of the pattern */ | |
b8f7bb16 | 240 | struct regnode_ssc { |
93e92956 | 241 | U8 flags; /* ANYOF_MATCHES_POSIXL bit must go here */ |
b8f7bb16 KW |
242 | U8 type; |
243 | U16 next_off; | |
7e817188 | 244 | U32 arg1; |
03fa83ba | 245 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time ... */ |
631e9161 | 246 | U32 classflags; /* ... and run-time */ |
85c8e306 KW |
247 | |
248 | /* Auxiliary, only used during construction; NULL afterwards: list of code | |
249 | * points matched */ | |
250 | SV* invlist; | |
b8f7bb16 KW |
251 | }; |
252 | ||
71068078 KW |
253 | /* We take advantage of 'next_off' not otherwise being used in the SSC by |
254 | * actually using it: by setting it to 1. This allows us to test and | |
255 | * distinguish between an SSC and other ANYOF node types, as 'next_off' cannot | |
256 | * otherwise be 1, because it is the offset to the next regnode expressed in | |
257 | * units of regnodes. Since an ANYOF node contains extra fields, it adds up | |
258 | * to 12 regnode units on 32-bit systems, (hence the minimum this can be (if | |
259 | * not 0) is 11 there. Even if things get tightly packed on a 64-bit system, | |
260 | * it still would be more than 1. */ | |
261 | #define set_ANYOF_SYNTHETIC(n) STMT_START{ OP(n) = ANYOF; \ | |
262 | NEXT_OFF(n) = 1; \ | |
263 | } STMT_END | |
c1071334 | 264 | #define is_ANYOF_SYNTHETIC(n) (PL_regkind[OP(n)] == ANYOF && NEXT_OFF(n) == 1) |
71068078 | 265 | |
9c7e81e8 AD |
266 | /* XXX fix this description. |
267 | Impose a limit of REG_INFTY on various pattern matching operations | |
268 | to limit stack growth and to avoid "infinite" recursions. | |
269 | */ | |
270 | /* The default size for REG_INFTY is I16_MAX, which is the same as | |
271 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits | |
272 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is | |
273 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is | |
274 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a | |
275 | smaller default. | |
276 | --Andy Dougherty 11 June 1998 | |
277 | */ | |
278 | #if SHORTSIZE > 2 | |
279 | # ifndef REG_INFTY | |
280 | # define REG_INFTY ((1<<15)-1) | |
281 | # endif | |
282 | #endif | |
283 | ||
284 | #ifndef REG_INFTY | |
83cfe48c IZ |
285 | # define REG_INFTY I16_MAX |
286 | #endif | |
a687059c | 287 | |
e91177ed IZ |
288 | #define ARG_VALUE(arg) (arg) |
289 | #define ARG__SET(arg,val) ((arg) = (val)) | |
c277df42 | 290 | |
e8e3b0fb JH |
291 | #undef ARG |
292 | #undef ARG1 | |
293 | #undef ARG2 | |
294 | ||
c277df42 IZ |
295 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
296 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) | |
297 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) | |
6bda09f9 | 298 | #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) |
786e8c11 | 299 | |
c277df42 IZ |
300 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
301 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) | |
302 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) | |
6bda09f9 | 303 | #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) |
c277df42 | 304 | |
e8e3b0fb JH |
305 | #undef NEXT_OFF |
306 | #undef NODE_ALIGN | |
307 | ||
075abff3 AL |
308 | #define NEXT_OFF(p) ((p)->next_off) |
309 | #define NODE_ALIGN(node) | |
aa48e906 YO |
310 | /* the following define was set to 0xde in 075abff3 |
311 | * as part of some linting logic. I have set it to 0 | |
312 | * as otherwise in every place where we /might/ set flags | |
313 | * we have to set it 0 explicitly, which duplicates | |
314 | * assignments and IMO adds an unacceptable level of | |
315 | * surprise to working in the regex engine. If this | |
316 | * is changed from 0 then at the very least make sure | |
317 | * that SBOL for /^/ sets the flags to 0 explicitly. | |
318 | * -- Yves */ | |
319 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0) | |
a687059c | 320 | |
c277df42 IZ |
321 | #define SIZE_ALIGN NODE_ALIGN |
322 | ||
e8e3b0fb JH |
323 | #undef OP |
324 | #undef OPERAND | |
325 | #undef MASK | |
326 | #undef STRING | |
327 | ||
e91177ed | 328 | #define OP(p) ((p)->type) |
b988e673 KW |
329 | #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \ |
330 | regnode types. For some, it's the \ | |
331 | character set of the regnode */ | |
e91177ed | 332 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 IZ |
333 | #define MASK(p) ((char*)OPERAND(p)) |
334 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) | |
335 | #define STRING(p) (((struct regnode_string *)p)->string) | |
336 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) | |
337 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) | |
338 | ||
e8e3b0fb JH |
339 | #undef NODE_ALIGN |
340 | #undef ARG_LOC | |
341 | #undef NEXTOPER | |
342 | #undef PREVOPER | |
343 | ||
e91177ed IZ |
344 | #define NODE_ALIGN(node) |
345 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) | |
346 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) | |
347 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) | |
6bda09f9 | 348 | #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) |
786e8c11 | 349 | |
e91177ed IZ |
350 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
351 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) | |
352 | ||
353 | #define NODE_STEP_B 4 | |
c277df42 IZ |
354 | |
355 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) | |
356 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) | |
357 | ||
e91177ed | 358 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 | 359 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed | 360 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 | 361 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
9990d5ab KW |
362 | #define FILL_ADVANCE_NODE_2L_ARG(ptr, op, arg1, arg2) \ |
363 | STMT_START { \ | |
364 | ARG_SET(ptr, arg1); \ | |
365 | ARG2L_SET(ptr, arg2); \ | |
366 | FILL_ADVANCE_NODE(ptr, op); \ | |
367 | (ptr) += 2; \ | |
368 | } STMT_END | |
a687059c | 369 | |
22c35a8c | 370 | #define REG_MAGIC 0234 |
a687059c | 371 | |
9a81a976 | 372 | #define SIZE_ONLY (RExC_emit == (regnode *) & RExC_emit_dummy) |
6d15ebf3 KW |
373 | #define PASS1 SIZE_ONLY |
374 | #define PASS2 (! SIZE_ONLY) | |
c277df42 | 375 | |
6f16c8da | 376 | /* If the bitmap fully represents what this ANYOF node can match, the |
137165a6 KW |
377 | * ARG is set to this special value (since 0, 1, ... are legal, but will never |
378 | * reach this high). */ | |
93e92956 | 379 | #define ANYOF_ONLY_HAS_BITMAP ((U32) -1) |
137165a6 | 380 | |
d2f465e9 KW |
381 | /* Below are the flags for node->flags of ANYOF. These are in short supply, |
382 | * with none currently available. The ABOVE_BITMAP_ALL bit could be freed up | |
7bc66b18 | 383 | * by resorting to creating a swash containing everything above 255. This |
d2f465e9 KW |
384 | * seems likely to introduce a performance penalty (but actual numbers haven't |
385 | * been done), so its probably better do some of the other possibilities below | |
386 | * in preference to this. | |
387 | * | |
388 | * If just one bit is required, it seems to me (khw) that the best option would | |
389 | * be to turn the ANYOF_LOC_REQ_UTF8 bit into a separate node type: a | |
390 | * specialization of the ANYOFL type, freeing up the currently occupied bit. | |
391 | * When turning a bit into a node type, one has to take into consideration that | |
392 | * a SSC may use that bit -- not just a regular ANYOF[DL]?. In the case of | |
393 | * ANYOF_LOC_REQ_UTF8, the only likely problem is accurately settting the SSC | |
394 | * node-type to the new one, which would likely involve S_ssc_or and S_ssc_and, | |
395 | * and not how the SSC currently gets set to ANYOFL. This bit is a natural | |
396 | * candidate for being a separate node type because it is a specialization of | |
397 | * the current ANYOFL, and because no other ANYOFL-only bits are set when it | |
398 | * is; also most of its uses are actually outside the reginclass() function, so | |
399 | * this could be done with no performance penalty. The other potential bits | |
400 | * seem to me to have a potential issue with a combinatorial explosion of node | |
401 | * types, because of not having that mutual exclusivity, where you may end up | |
402 | * having to have a node type for bitX being set, one for bitY, and one for | |
403 | * both bitXY. | |
404 | * | |
405 | * If you don't want to do this, or two bits are required, one could instead | |
406 | * rename the ANYOF_POSIXL bit to be ANYOFL_LARGE, to mean that the ANYOF node | |
407 | * has an extra 32 bits beyond what a regular one does. That's what it | |
408 | * effectively means now, with the extra space all for the POSIX class bits. | |
409 | * But those classes actually only occupy 30 bits, so the ANYOF_LOC_REQ_BIT (if | |
410 | * an extra node type for it hasn't been created) and/or the ANYOF_LOC_FOLD | |
411 | * bits could be moved there. The downside of this is that ANYOFL nodes with | |
412 | * whichever of the bits get moved would have to have the extra space always | |
413 | * allocated. | |
414 | * | |
415 | * If three bits are required, one could additionally make a node type for | |
416 | * ANYOFL_LARGE, removing that as a bit, and move both the above bits to that | |
417 | * extra word. There isn't an SSC problem as all SSCs are this large anyway, | |
418 | * and the SSC could be set to this node type. REGINCLASS would have to be | |
419 | * modified so that if the node type were this, it would call reginclass(). | |
420 | * as the flag bit that does this now would be gone. | |
a4525e78 | 421 | * |
34fdef84 | 422 | * Several flags are not used in synthetic start class (SSC) nodes, so could be |
93e92956 KW |
423 | * shared should new flags be needed for SSCs, like SSC_MATCHES_EMPTY_STRING |
424 | * now. */ | |
b8c5462f | 425 | |
d2f465e9 KW |
426 | /* If this is set, the result of the match should be complemented. regexec.c |
427 | * is expecting this to be in the low bit. Never in an SSC */ | |
93e92956 | 428 | #define ANYOF_INVERT 0x01 |
1984f6a0 | 429 | |
a0dd4231 | 430 | /* For the SSC node only, which cannot be inverted, so is shared with that bit. |
93e92956 KW |
431 | * This is used only during regex compilation. */ |
432 | #define SSC_MATCHES_EMPTY_STRING ANYOF_INVERT | |
a0dd4231 | 433 | |
509d4de0 KW |
434 | /* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This |
435 | * is used for runtime \d, \w, [:posix:], ..., which are used only in locale | |
436 | * and the optimizer's synthetic start class. Non-locale \d, etc are resolved | |
437 | * at compile-time. Only set under /l; can be in SSC */ | |
438 | #define ANYOF_MATCHES_POSIXL 0x02 | |
b63f73da KW |
439 | |
440 | /* The fold is calculated and stored in the bitmap where possible at compile | |
441 | * time. However under locale, the actual folding varies depending on | |
442 | * what the locale is at the time of execution, so it has to be deferred until | |
d2f465e9 | 443 | * then. Only set under /l; never in an SSC */ |
93e92956 | 444 | #define ANYOF_LOC_FOLD 0x04 |
b63f73da | 445 | |
509d4de0 KW |
446 | /* If set, means to warn if runtime locale isn't a UTF-8 one. Only under /l. |
447 | * If set, none of INVERT, LOC_FOLD, POSIXL, HAS_NONBITMAP_NON_UTF8_MATCHES can | |
448 | * be set. Can be in an SSC */ | |
449 | #define ANYOF_LOC_REQ_UTF8 0x08 | |
c8d3cd88 | 450 | |
509d4de0 KW |
451 | /* If set, the node matches every code point NUM_ANYOF_CODE_POINTS and above. |
452 | * Can be in an SSC */ | |
453 | #define ANYOF_MATCHES_ALL_ABOVE_BITMAP 0x10 | |
f4244008 | 454 | |
d2f465e9 KW |
455 | /* If set, the node can match something outside the bitmap that isn't in utf8; |
456 | * never set under /d nor in an SSC */ | |
93e92956 | 457 | #define ANYOF_HAS_NONBITMAP_NON_UTF8_MATCHES 0x20 |
ef87b810 | 458 | |
509d4de0 KW |
459 | /* Are there things outside the bitmap that will match only if the target |
460 | * string is encoded in UTF-8? (This is not set if ANYOF_ABOVE_BITMAP_ALL is | |
461 | * set). Can be in SSC */ | |
462 | #define ANYOF_HAS_UTF8_NONBITMAP_MATCHES 0x40 | |
f240c685 KW |
463 | |
464 | /* Shared bit: | |
d2f465e9 | 465 | * Under /d it means the ANYOFD node matches all non-ASCII Latin1 |
f240c685 KW |
466 | * characters when the target string is not in utf8. |
467 | * When not under /d, it means the ANYOF node should raise a warning if | |
468 | * matching against an above-Unicode code point. | |
469 | * (These uses are mutually exclusive because the warning requires a \p{}, and | |
470 | * \p{} implies /u which deselects /d). An SSC node only has this bit set if | |
471 | * what is meant is the warning. The long macro name is to make sure that you | |
472 | * are cautioned about its shared nature */ | |
473 | #define ANYOF_SHARED_d_MATCHES_ALL_NON_UTF8_NON_ASCII_non_d_WARN_SUPER 0x80 | |
7bc66b18 | 474 | |
3b04b210 | 475 | #define ANYOF_FLAGS_ALL (0xff) |
653099ff | 476 | |
93e92956 | 477 | #define ANYOF_LOCALE_FLAGS (ANYOF_LOC_FOLD | ANYOF_MATCHES_POSIXL) |
ace6b0e4 | 478 | |
eff8b7dc KW |
479 | /* These are the flags that apply to both regular ANYOF nodes and synthetic |
480 | * start class nodes during construction of the SSC. During finalization of | |
d2f465e9 | 481 | * the SSC, other of the flags may get added to it */ |
a0bd1a30 KW |
482 | #define ANYOF_COMMON_FLAGS ( ANYOF_HAS_UTF8_NONBITMAP_MATCHES \ |
483 | |ANYOF_LOC_REQ_UTF8) | |
eff8b7dc | 484 | |
936ed897 | 485 | /* Character classes for node->classflags of ANYOF */ |
653099ff | 486 | /* Should be synchronized with a table in regprop() */ |
575a8a29 | 487 | /* 2n should be the normal one, paired with its complement at 2n+1 */ |
b8c5462f | 488 | |
08ada29e KW |
489 | #define ANYOF_ALPHA ((_CC_ALPHA) * 2) |
490 | #define ANYOF_NALPHA ((ANYOF_ALPHA) + 1) | |
e8d596e0 KW |
491 | #define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */ |
492 | #define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1) | |
08ada29e KW |
493 | #define ANYOF_ASCII ((_CC_ASCII) * 2) |
494 | #define ANYOF_NASCII ((ANYOF_ASCII) + 1) | |
e8d596e0 KW |
495 | #define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */ |
496 | #define ANYOF_NBLANK ((ANYOF_BLANK) + 1) | |
b0d691b2 KW |
497 | #define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or |
498 | [:upper:] under /i */ | |
499 | #define ANYOF_NCASED ((ANYOF_CASED) + 1) | |
08ada29e KW |
500 | #define ANYOF_CNTRL ((_CC_CNTRL) * 2) |
501 | #define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1) | |
e8d596e0 KW |
502 | #define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */ |
503 | #define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1) | |
08ada29e KW |
504 | #define ANYOF_GRAPH ((_CC_GRAPH) * 2) |
505 | #define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1) | |
506 | #define ANYOF_LOWER ((_CC_LOWER) * 2) | |
507 | #define ANYOF_NLOWER ((ANYOF_LOWER) + 1) | |
508 | #define ANYOF_PRINT ((_CC_PRINT) * 2) | |
509 | #define ANYOF_NPRINT ((ANYOF_PRINT) + 1) | |
510 | #define ANYOF_PUNCT ((_CC_PUNCT) * 2) | |
511 | #define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1) | |
e8d596e0 KW |
512 | #define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */ |
513 | #define ANYOF_NSPACE ((ANYOF_SPACE) + 1) | |
08ada29e KW |
514 | #define ANYOF_UPPER ((_CC_UPPER) * 2) |
515 | #define ANYOF_NUPPER ((ANYOF_UPPER) + 1) | |
e8d596e0 KW |
516 | #define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
517 | #define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1) | |
08ada29e KW |
518 | #define ANYOF_XDIGIT ((_CC_XDIGIT) * 2) |
519 | #define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1) | |
b8c5462f | 520 | |
87cad4c0 | 521 | /* pseudo classes below this, not stored in the class bitmap, but used as flags |
e1d1eefb YO |
522 | during compilation of char classes */ |
523 | ||
b2ef2e34 KW |
524 | #define ANYOF_VERTWS ((_CC_VERTSPACE) * 2) |
525 | #define ANYOF_NVERTWS ((ANYOF_VERTWS)+1) | |
e1d1eefb | 526 | |
b2ef2e34 KW |
527 | /* It is best if this is the last one, as all above it are stored as bits in a |
528 | * bitmap, and it isn't part of that bitmap */ | |
529 | #if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC | |
530 | # error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define | |
531 | #endif | |
532 | ||
8efd3f97 KW |
533 | #define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written: |
534 | * '< ANYOF_MAX' | |
535 | * Hence doesn't include VERTWS, as that | |
536 | * is a pseudo class */ | |
537 | #define ANYOF_MAX ANYOF_POSIXL_MAX | |
538 | ||
539 | #if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */ | |
b2ef2e34 | 540 | # error Problem with handy.h _CC_foo #defines |
a0947d7b KW |
541 | #endif |
542 | ||
8efd3f97 KW |
543 | #define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */ |
544 | #define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3) | |
87cad4c0 | 545 | |
8efd3f97 KW |
546 | #define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode |
547 | property: \p{} or \P{} */ | |
f39a4dc2 | 548 | |
b8c5462f JH |
549 | /* Backward source code compatibility. */ |
550 | ||
551 | #define ANYOF_ALNUML ANYOF_ALNUM | |
552 | #define ANYOF_NALNUML ANYOF_NALNUM | |
553 | #define ANYOF_SPACEL ANYOF_SPACE | |
554 | #define ANYOF_NSPACEL ANYOF_NSPACE | |
8ebfc29a KW |
555 | #define ANYOF_ALNUM ANYOF_WORDCHAR |
556 | #define ANYOF_NALNUM ANYOF_NWORDCHAR | |
b8c5462f JH |
557 | |
558 | /* Utility macros for the bitmap and classes of ANYOF */ | |
559 | ||
936ed897 | 560 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f | 561 | |
bc51fd78 | 562 | #define ANYOF_BIT(c) (1U << ((c) & 7)) |
b8c5462f | 563 | |
d1d040e5 | 564 | #define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c))) |
8efd3f97 KW |
565 | #define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c)) |
566 | ||
d1d040e5 | 567 | #define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c))) |
8efd3f97 KW |
568 | #define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c)) |
569 | ||
d1d040e5 | 570 | #define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c))) |
8efd3f97 | 571 | #define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c)) |
b8c5462f | 572 | |
d1d040e5 | 573 | #define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END |
8efd3f97 | 574 | #define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret) |
9cf78fa0 KW |
575 | |
576 | /* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */ | |
d1d040e5 | 577 | #define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END |
8efd3f97 | 578 | #define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret) |
9cf78fa0 | 579 | |
8efd3f97 | 580 | #define ANYOF_POSIXL_TEST_ANY_SET(p) \ |
93e92956 | 581 | ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \ |
d1d040e5 | 582 | && (((regnode_charclass_posixl*)(p))->classflags)) |
8efd3f97 | 583 | #define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p) |
e867ce45 | 584 | |
e0e1be5f KW |
585 | /* Since an SSC always has this field, we don't have to test for that; nor do |
586 | * we want to because the bit isn't set for SSC during its construction */ | |
587 | #define ANYOF_POSIXL_SSC_TEST_ANY_SET(p) \ | |
588 | cBOOL(((regnode_ssc*)(p))->classflags) | |
589 | #define ANYOF_POSIXL_SSC_TEST_ALL_SET(p) /* Are all bits set? */ \ | |
590 | (((regnode_ssc*) (p))->classflags \ | |
591 | == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1) | |
592 | ||
593 | #define ANYOF_POSIXL_TEST_ALL_SET(p) \ | |
93e92956 | 594 | ((ANYOF_FLAGS(p) & ANYOF_MATCHES_POSIXL) \ |
e0e1be5f KW |
595 | && ((regnode_charclass_posixl*) (p))->classflags \ |
596 | == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1) | |
fb38762f | 597 | |
b27fd97a | 598 | #define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END |
8efd3f97 | 599 | #define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest)) |
936ed897 | 600 | |
cdd87c1d KW |
601 | #define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END |
602 | ||
9cf78fa0 | 603 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
936ed897 | 604 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
786e8c11 | 605 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f JH |
606 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
607 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) | |
bc51fd78 | 608 | #define ANYOF_BITMAP_TEST(p, c) cBOOL(ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) |
b8c5462f | 609 | |
f8bef550 NC |
610 | #define ANYOF_BITMAP_SETALL(p) \ |
611 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) | |
612 | #define ANYOF_BITMAP_CLEARALL(p) \ | |
613 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) | |
f8bef550 | 614 | |
888ee784 KW |
615 | #define ANYOF_SKIP EXTRA_SIZE(struct regnode_charclass) |
616 | #define ANYOF_POSIXL_SKIP EXTRA_SIZE(regnode_charclass_posixl) | |
3a15e693 | 617 | |
a687059c LW |
618 | /* |
619 | * Utility definitions. | |
620 | */ | |
2304df62 | 621 | #ifndef CHARMASK |
075abff3 | 622 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c | 623 | #else |
075abff3 | 624 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c | 625 | #endif |
a687059c | 626 | |
c277df42 IZ |
627 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
628 | ||
e384d5c1 YO |
629 | #define REG_ZERO_LEN_SEEN 0x00000001 |
630 | #define REG_LOOKBEHIND_SEEN 0x00000002 | |
631 | #define REG_GPOS_SEEN 0x00000004 | |
471f5387 | 632 | /* spare */ |
e384d5c1 YO |
633 | #define REG_RECURSE_SEEN 0x00000020 |
634 | #define REG_TOP_LEVEL_BRANCHES_SEEN 0x00000040 | |
635 | #define REG_VERBARG_SEEN 0x00000080 | |
636 | #define REG_CUTGROUP_SEEN 0x00000100 | |
637 | #define REG_RUN_ON_COMMENT_SEEN 0x00000200 | |
638 | #define REG_UNFOLDED_MULTI_SEEN 0x00000400 | |
639 | #define REG_GOSTART_SEEN 0x00000800 | |
ee273784 YO |
640 | #define REG_UNBOUNDED_QUANTIFIER_SEEN 0x00001000 |
641 | ||
d09b2d29 | 642 | |
73c4f7a1 GS |
643 | START_EXTERN_C |
644 | ||
be8e71aa YO |
645 | #ifdef PLUGGABLE_RE_EXTENSION |
646 | #include "re_nodes.h" | |
647 | #else | |
d09b2d29 | 648 | #include "regnodes.h" |
be8e71aa | 649 | #endif |
d09b2d29 | 650 | |
f9f4320a YO |
651 | #ifndef PLUGGABLE_RE_EXTENSION |
652 | #ifndef DOINIT | |
653 | EXTCONST regexp_engine PL_core_reg_engine; | |
654 | #else /* DOINIT */ | |
655 | EXTCONST regexp_engine PL_core_reg_engine = { | |
fe578d7f | 656 | Perl_re_compile, |
2fdbfb4d | 657 | Perl_regexec_flags, |
f9f4320a YO |
658 | Perl_re_intuit_start, |
659 | Perl_re_intuit_string, | |
2fdbfb4d AB |
660 | Perl_regfree_internal, |
661 | Perl_reg_numbered_buff_fetch, | |
662 | Perl_reg_numbered_buff_store, | |
663 | Perl_reg_numbered_buff_length, | |
192b9cd1 AB |
664 | Perl_reg_named_buff, |
665 | Perl_reg_named_buff_iter, | |
49d7dfbc | 666 | Perl_reg_qr_package, |
f9f4320a | 667 | #if defined(USE_ITHREADS) |
3c13cae6 | 668 | Perl_regdupe_internal, |
f9f4320a | 669 | #endif |
3c13cae6 | 670 | Perl_re_op_compile |
f9f4320a YO |
671 | }; |
672 | #endif /* DOINIT */ | |
673 | #endif /* PLUGGABLE_RE_EXTENSION */ | |
674 | ||
675 | ||
73c4f7a1 | 676 | END_EXTERN_C |
cad2e5aa | 677 | |
cad2e5aa | 678 | |
261faec3 MJD |
679 | /* .what is a character array with one character for each member of .data |
680 | * The character describes the function of the corresponding .data item: | |
af534a04 | 681 | * a - AV for paren_name_list under DEBUGGING |
0111c4fd | 682 | * f - start-class data for regstclass optimization |
68e2671b | 683 | * l - start op for literal (?{EVAL}) item |
d63c20f2 | 684 | * L - start op for literal (?{EVAL}) item, with separate CV (qr//) |
b30fcab9 | 685 | * r - pointer to an embedded code-containing qr, e.g. /ab$qr/ |
38a44b82 | 686 | * s - swash for Unicode-style character class, and the multicharacter |
9e55ce06 JH |
687 | * strings resulting from casefolding the single-character entries |
688 | * in the character class | |
a3621e74 | 689 | * t - trie struct |
55eed653 | 690 | * u - trie struct's widecharmap (a HV, so can't share, must dup) |
2b8b4781 | 691 | * also used for revcharmap and words under DEBUGGING |
07be1b83 | 692 | * T - aho-trie struct |
81714fb9 | 693 | * S - sv for named capture lookup |
261faec3 | 694 | * 20010712 mjd@plover.com |
c3613ed7 | 695 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 | 696 | */ |
cad2e5aa JH |
697 | struct reg_data { |
698 | U32 count; | |
699 | U8 *what; | |
700 | void* data[1]; | |
701 | }; | |
702 | ||
a1cac82e NC |
703 | /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses |
704 | anchored* and float* via array indexes 0 and 1. */ | |
cad2e5aa | 705 | #define anchored_substr substrs->data[0].substr |
33b8afdf | 706 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa | 707 | #define anchored_offset substrs->data[0].min_offset |
1de06328 YO |
708 | #define anchored_end_shift substrs->data[0].end_shift |
709 | ||
cad2e5aa | 710 | #define float_substr substrs->data[1].substr |
33b8afdf | 711 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa JH |
712 | #define float_min_offset substrs->data[1].min_offset |
713 | #define float_max_offset substrs->data[1].max_offset | |
1de06328 YO |
714 | #define float_end_shift substrs->data[1].end_shift |
715 | ||
cad2e5aa | 716 | #define check_substr substrs->data[2].substr |
33b8afdf | 717 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa JH |
718 | #define check_offset_min substrs->data[2].min_offset |
719 | #define check_offset_max substrs->data[2].max_offset | |
1de06328 | 720 | #define check_end_shift substrs->data[2].end_shift |
a3621e74 | 721 | |
8d919b0a FC |
722 | #define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr) |
723 | #define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8) | |
724 | #define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr) | |
725 | #define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8) | |
a3621e74 YO |
726 | |
727 | /* trie related stuff */ | |
5d9a96ca | 728 | |
a3621e74 YO |
729 | /* a transition record for the state machine. the |
730 | check field determines which state "owns" the | |
731 | transition. the char the transition is for is | |
732 | determined by offset from the owning states base | |
733 | field. the next field determines which state | |
734 | is to be transitioned to if any. | |
735 | */ | |
736 | struct _reg_trie_trans { | |
737 | U32 next; | |
738 | U32 check; | |
739 | }; | |
740 | ||
741 | /* a transition list element for the list based representation */ | |
742 | struct _reg_trie_trans_list_elem { | |
743 | U16 forid; | |
744 | U32 newstate; | |
745 | }; | |
746 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; | |
747 | ||
748 | /* a state for compressed nodes. base is an offset | |
749 | into an array of reg_trie_trans array. If wordnum is | |
750 | nonzero the state is accepting. if base is zero then | |
751 | the state has no children (and will be accepting) | |
752 | */ | |
753 | struct _reg_trie_state { | |
754 | U16 wordnum; | |
755 | union { | |
756 | U32 base; | |
757 | reg_trie_trans_le* list; | |
758 | } trans; | |
759 | }; | |
760 | ||
2e64971a DM |
761 | /* info per word; indexed by wordnum */ |
762 | typedef struct { | |
763 | U16 prev; /* previous word in acceptance chain; eg in | |
764 | * zzz|abc|ab/ after matching the chars abc, the | |
765 | * accepted word is #2, and the previous accepted | |
766 | * word is #3 */ | |
767 | U32 len; /* how many chars long is this word? */ | |
768 | U32 accept; /* accept state for this word */ | |
769 | } reg_trie_wordinfo; | |
a3621e74 YO |
770 | |
771 | ||
a3621e74 YO |
772 | typedef struct _reg_trie_state reg_trie_state; |
773 | typedef struct _reg_trie_trans reg_trie_trans; | |
774 | ||
775 | ||
776 | /* anything in here that needs to be freed later | |
23eab42c NC |
777 | should be dealt with in pregfree. |
778 | refcount is first in both this and _reg_ac_data to allow a space | |
779 | optimisation in Perl_regdupe. */ | |
a3621e74 | 780 | struct _reg_trie_data { |
23eab42c | 781 | U32 refcount; /* number of times this trie is referenced */ |
786e8c11 YO |
782 | U32 lasttrans; /* last valid transition element */ |
783 | U16 *charmap; /* byte to charid lookup array */ | |
786e8c11 YO |
784 | reg_trie_state *states; /* state data */ |
785 | reg_trie_trans *trans; /* array of transition elements */ | |
786 | char *bitmap; /* stclass bitmap */ | |
786e8c11 YO |
787 | U16 *jump; /* optional 1 indexed array of offsets before tail |
788 | for the node following a given word. */ | |
2e64971a | 789 | reg_trie_wordinfo *wordinfo; /* array of info per word */ |
1b5c067c NC |
790 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
791 | U32 startstate; /* initial state - used for common prefix optimisation */ | |
792 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ | |
793 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ | |
2e64971a | 794 | U32 prefixlen; /* #chars in common prefix */ |
1e2e3d02 YO |
795 | U32 statecount; /* Build only - number of states in the states array |
796 | (including the unused zero state) */ | |
786e8c11 | 797 | U32 wordcount; /* Build only */ |
a3621e74 | 798 | #ifdef DEBUGGING |
786e8c11 | 799 | STRLEN charcount; /* Build only */ |
a3621e74 YO |
800 | #endif |
801 | }; | |
2b8b4781 NC |
802 | /* There is one (3 under DEBUGGING) pointers that logically belong in this |
803 | structure, but are held outside as they need duplication on thread cloning, | |
804 | whereas the rest of the structure can be read only: | |
805 | HV *widecharmap; code points > 255 to charid | |
806 | #ifdef DEBUGGING | |
807 | AV *words; Array of words contained in trie, for dumping | |
808 | AV *revcharmap; Map of each charid back to its character representation | |
809 | #endif | |
810 | */ | |
811 | ||
812 | #define TRIE_WORDS_OFFSET 2 | |
813 | ||
a3621e74 YO |
814 | typedef struct _reg_trie_data reg_trie_data; |
815 | ||
23eab42c NC |
816 | /* refcount is first in both this and _reg_trie_data to allow a space |
817 | optimisation in Perl_regdupe. */ | |
07be1b83 | 818 | struct _reg_ac_data { |
23eab42c | 819 | U32 refcount; |
1b5c067c | 820 | U32 trie; |
07be1b83 YO |
821 | U32 *fail; |
822 | reg_trie_state *states; | |
07be1b83 YO |
823 | }; |
824 | typedef struct _reg_ac_data reg_ac_data; | |
825 | ||
486ec47a | 826 | /* ANY_BIT doesn't use the structure, so we can borrow it here. |
3dab1dad YO |
827 | This is simpler than refactoring all of it as wed end up with |
828 | three different sets... */ | |
829 | ||
830 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) | |
786e8c11 | 831 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f YO |
832 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
833 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) | |
834 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 835 | |
1de06328 YO |
836 | #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) |
837 | #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) | |
838 | ||
839 | ||
786e8c11 YO |
840 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
841 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 842 | |
a3621e74 YO |
843 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
844 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) | |
845 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) | |
846 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) | |
847 | ||
3dab1dad | 848 | #ifdef DEBUGGING |
3dab1dad | 849 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad | 850 | #else |
3dab1dad | 851 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad | 852 | #endif |
a3621e74 | 853 | |
0111c4fd RGS |
854 | #define RE_TRIE_MAXBUF_INIT 65536 |
855 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" | |
a3621e74 YO |
856 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
857 | ||
be8e71aa YO |
858 | /* |
859 | ||
860 | RE_DEBUG_FLAGS is used to control what debug output is emitted | |
861 | its divided into three groups of options, some of which interact. | |
862 | The three groups are: Compile, Execute, Extra. There is room for a | |
863 | further group, as currently only the low three bytes are used. | |
864 | ||
865 | Compile Options: | |
07be1b83 | 866 | |
be8e71aa YO |
867 | PARSE |
868 | PEEP | |
869 | TRIE | |
870 | PROGRAM | |
871 | OFFSETS | |
a3621e74 | 872 | |
be8e71aa | 873 | Execute Options: |
a3621e74 | 874 | |
be8e71aa YO |
875 | INTUIT |
876 | MATCH | |
877 | TRIE | |
a3621e74 | 878 | |
be8e71aa | 879 | Extra Options |
a3621e74 | 880 | |
be8e71aa YO |
881 | TRIE |
882 | OFFSETS | |
883 | ||
884 | If you modify any of these make sure you make corresponding changes to | |
885 | re.pm, especially to the documentation. | |
a3621e74 | 886 | |
be8e71aa YO |
887 | */ |
888 | ||
889 | ||
890 | /* Compile */ | |
891 | #define RE_DEBUG_COMPILE_MASK 0x0000FF | |
892 | #define RE_DEBUG_COMPILE_PARSE 0x000001 | |
893 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 | |
894 | #define RE_DEBUG_COMPILE_TRIE 0x000004 | |
895 | #define RE_DEBUG_COMPILE_DUMP 0x000008 | |
f7819f85 | 896 | #define RE_DEBUG_COMPILE_FLAGS 0x000010 |
d9a72fcc | 897 | #define RE_DEBUG_COMPILE_TEST 0x000020 |
be8e71aa YO |
898 | |
899 | /* Execute */ | |
900 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 | |
901 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 | |
902 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 | |
903 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 | |
904 | ||
905 | /* Extra */ | |
906 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 | |
907 | #define RE_DEBUG_EXTRA_TRIE 0x010000 | |
908 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 | |
a5ca303d YO |
909 | #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 |
910 | #define RE_DEBUG_EXTRA_STATE 0x080000 | |
911 | #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 | |
e7707071 | 912 | #define RE_DEBUG_EXTRA_BUFFERS 0x400000 |
2c296965 | 913 | #define RE_DEBUG_EXTRA_GPOS 0x800000 |
24b23f37 YO |
914 | /* combined */ |
915 | #define RE_DEBUG_EXTRA_STACK 0x280000 | |
be8e71aa | 916 | |
e68ec53f | 917 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa YO |
918 | /* Compile */ |
919 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ | |
e68ec53f | 920 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa | 921 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 922 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 923 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f | 924 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa | 925 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 926 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 927 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f | 928 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa | 929 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f | 930 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
f7819f85 A |
931 | #define DEBUG_FLAGS_r(x) DEBUG_r( \ |
932 | if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x ) | |
d9a72fcc YO |
933 | #define DEBUG_TEST_r(x) DEBUG_r( \ |
934 | if (re_debug_flags & RE_DEBUG_COMPILE_TEST) x ) | |
be8e71aa YO |
935 | /* Execute */ |
936 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ | |
e68ec53f | 937 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa | 938 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f | 939 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa | 940 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f | 941 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa | 942 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f | 943 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa YO |
944 | |
945 | /* Extra */ | |
946 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ | |
e68ec53f | 947 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
a5ca303d YO |
948 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
949 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) | |
ab3bbdeb YO |
950 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
951 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) | |
24b23f37 YO |
952 | #define DEBUG_STACK_r(x) DEBUG_r( \ |
953 | if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x ) | |
e7707071 YO |
954 | #define DEBUG_BUFFERS_r(x) DEBUG_r( \ |
955 | if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x ) | |
956 | ||
a5ca303d YO |
957 | #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ |
958 | if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ | |
959 | (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) | |
07be1b83 | 960 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
a5ca303d | 961 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ |
be8e71aa YO |
962 | Perl_warn_nocontext x ) |
963 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ | |
e68ec53f | 964 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa | 965 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f | 966 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa YO |
967 | |
968 | #define DEBUG_TRIE_r(x) DEBUG_r( \ | |
e68ec53f | 969 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa | 970 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
2c296965 YO |
971 | #define DEBUG_GPOS_r(x) DEBUG_r( \ |
972 | if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x ) | |
be8e71aa YO |
973 | |
974 | /* initialization */ | |
d2d8c711 | 975 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f YO |
976 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
977 | SV * re_debug_flags_sv = NULL; \ | |
978 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ | |
979 | if (re_debug_flags_sv) { \ | |
8e9a8a48 | 980 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f | 981 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 | 982 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f YO |
983 | }\ |
984 | }) | |
a3621e74 YO |
985 | |
986 | #ifdef DEBUGGING | |
ab3bbdeb | 987 | |
cb65b687 DM |
988 | #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \ |
989 | PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS; | |
ab3bbdeb YO |
990 | |
991 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ | |
992 | const char * const rpv = \ | |
ddc5bc0f | 993 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb | 994 | PL_colors[(c1)],PL_colors[(c2)], \ |
c89df6cf | 995 | PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ |
0df25f3d | 996 | const int rlen = SvCUR(dsv) |
ab3bbdeb YO |
997 | |
998 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ | |
999 | const char * const rpv = \ | |
ddc5bc0f | 1000 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb | 1001 | PL_colors[(c1)],PL_colors[(c2)], \ |
c89df6cf | 1002 | PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) |
ab3bbdeb YO |
1003 | |
1004 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ | |
1005 | const char * const rpv = \ | |
ddc5bc0f | 1006 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb | 1007 | PL_colors[0], PL_colors[1], \ |
c89df6cf | 1008 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \ |
ab3bbdeb | 1009 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ |
0a0e9afc | 1010 | ) |
ab3bbdeb YO |
1011 | |
1012 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) | |
1013 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") | |
1014 | ||
1015 | #else /* if not DEBUGGING */ | |
1016 | ||
a3621e74 | 1017 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb YO |
1018 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
1019 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) | |
1020 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) | |
1021 | #define RE_SV_DUMPLEN(ItEm) | |
1022 | #define RE_SV_TAIL(ItEm) | |
1023 | ||
1024 | #endif /* DEBUG RELATED DEFINES */ | |
a3621e74 | 1025 | |
64935bc6 KW |
1026 | typedef enum { |
1027 | TRADITIONAL_BOUND = _CC_WORDCHAR, | |
ae3bb8ea | 1028 | GCB_BOUND, |
06ae2722 | 1029 | SB_BOUND, |
ae3bb8ea | 1030 | WB_BOUND |
64935bc6 KW |
1031 | } bound_type; |
1032 | ||
e9a8c099 | 1033 | /* |
14d04a33 | 1034 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 1035 | */ |