Commit | Line | Data |
a0d0e21e |
1 | /* regcomp.h |
d6376244 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
1d325971 |
4 | * 2000, 2001, 2002, 2003, 2005 by Larry Wall and others |
d6376244 |
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 | */ |
10 | |
c277df42 |
11 | typedef OP OP_4tree; /* Will be redefined later. */ |
12 | |
07be1b83 |
13 | |
14 | #define PERL_ENABLE_TRIE_OPTIMISATION 1 |
15 | #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1 |
16 | #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 |
17 | |
a687059c |
18 | /* |
19 | * The "internal use only" fields in regexp.h are present to pass info from |
20 | * compile to execute that permits the execute phase to run lots faster on |
21 | * simple cases. They are: |
22 | * |
bd61b366 |
23 | * regstart sv that must begin a match; NULL if none obvious |
a687059c |
24 | * reganch is the match anchored (at beginning-of-line only)? |
25 | * regmust string (pointer into program) that match must include, or NULL |
79072805 |
26 | * [regmust changed to SV* for bminstr()--law] |
a687059c |
27 | * regmlen length of regmust string |
28 | * [regmlen not used currently] |
29 | * |
30 | * Regstart and reganch permit very fast decisions on suitable starting points |
31 | * for a match, cutting down the work a lot. Regmust permits fast rejection |
32 | * of lines that cannot possibly match. The regmust tests are costly enough |
e50aee73 |
33 | * that pregcomp() supplies a regmust only if the r.e. contains something |
a687059c |
34 | * potentially expensive (at present, the only such thing detected is * or + |
35 | * at the start of the r.e., which can involve a lot of backup). Regmlen is |
e50aee73 |
36 | * supplied because the test in pregexec() needs it and pregcomp() is computing |
a687059c |
37 | * it anyway. |
38 | * [regmust is now supplied always. The tests that use regmust have a |
39 | * heuristic that disables the test if it usually matches.] |
40 | * |
41 | * [In fact, we now use regmust in many cases to locate where the search |
42 | * starts in the string, so if regback is >= 0, the regmust search is never |
43 | * wasted effort. The regback variable says how many characters back from |
44 | * where regmust matched is the earliest possible start of the match. |
45 | * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] |
46 | */ |
47 | |
48 | /* |
49 | * Structure for regexp "program". This is essentially a linear encoding |
50 | * of a nondeterministic finite-state machine (aka syntax charts or |
51 | * "railroad normal form" in parsing technology). Each node is an opcode |
52 | * plus a "next" pointer, possibly plus an operand. "Next" pointers of |
53 | * all nodes except BRANCH implement concatenation; a "next" pointer with |
54 | * a BRANCH on both ends of it is connecting two alternatives. (Here we |
55 | * have one of the subtle syntax dependencies: an individual BRANCH (as |
56 | * opposed to a collection of them) is never concatenated with anything |
57 | * because of operator precedence.) The operand of some types of node is |
58 | * a literal string; for others, it is a node leading into a sub-FSM. In |
59 | * particular, the operand of a BRANCH node is the first node of the branch. |
60 | * (NB this is *not* a tree structure: the tail of the branch connects |
a3621e74 |
61 | * to the thing following the set of BRANCHes.) The opcodes are defined |
62 | * in regnodes.h which is generated from regcomp.sym by regcomp.pl. |
a687059c |
63 | */ |
64 | |
a687059c |
65 | /* |
66 | * A node is one char of opcode followed by two chars of "next" pointer. |
67 | * "Next" pointers are stored as two 8-bit pieces, high order first. The |
68 | * value is a positive offset from the opcode of the node containing it. |
69 | * An operand, if any, simply follows the node. (Note that much of the |
70 | * code generation knows about this implicit relationship.) |
71 | * |
72 | * Using two bytes for the "next" pointer is vast overkill for most things, |
73 | * but allows patterns to get big without disasters. |
74 | * |
e91177ed |
75 | * [The "next" pointer is always aligned on an even |
a687059c |
76 | * boundary, and reads the offset directly as a short. Also, there is no |
77 | * special test to reverse the sign of BACK pointers since the offset is |
78 | * stored negative.] |
79 | */ |
80 | |
c277df42 |
81 | struct regnode_string { |
cd439c50 |
82 | U8 str_len; |
c277df42 |
83 | U8 type; |
84 | U16 next_off; |
cd439c50 |
85 | char string[1]; |
c277df42 |
86 | }; |
a687059c |
87 | |
c277df42 |
88 | struct regnode_1 { |
89 | U8 flags; |
90 | U8 type; |
91 | U16 next_off; |
92 | U32 arg1; |
93 | }; |
94 | |
95 | struct regnode_2 { |
96 | U8 flags; |
97 | U8 type; |
98 | U16 next_off; |
99 | U16 arg1; |
100 | U16 arg2; |
101 | }; |
102 | |
936ed897 |
103 | #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ |
ffc61ed2 |
104 | #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ |
936ed897 |
105 | |
106 | struct regnode_charclass { |
107 | U8 flags; |
108 | U8 type; |
109 | U16 next_off; |
ffc61ed2 |
110 | U32 arg1; |
19103b91 |
111 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 |
112 | }; |
113 | |
19103b91 |
114 | struct regnode_charclass_class { /* has [[:blah:]] classes */ |
115 | U8 flags; /* should have ANYOF_CLASS here */ |
936ed897 |
116 | U8 type; |
117 | U16 next_off; |
ffc61ed2 |
118 | U32 arg1; |
19103b91 |
119 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
120 | char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ |
936ed897 |
121 | }; |
122 | |
9c7e81e8 |
123 | /* XXX fix this description. |
124 | Impose a limit of REG_INFTY on various pattern matching operations |
125 | to limit stack growth and to avoid "infinite" recursions. |
126 | */ |
127 | /* The default size for REG_INFTY is I16_MAX, which is the same as |
128 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits |
129 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is |
130 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is |
131 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a |
132 | smaller default. |
133 | --Andy Dougherty 11 June 1998 |
134 | */ |
135 | #if SHORTSIZE > 2 |
136 | # ifndef REG_INFTY |
137 | # define REG_INFTY ((1<<15)-1) |
138 | # endif |
139 | #endif |
140 | |
141 | #ifndef REG_INFTY |
83cfe48c |
142 | # define REG_INFTY I16_MAX |
143 | #endif |
a687059c |
144 | |
e91177ed |
145 | #define ARG_VALUE(arg) (arg) |
146 | #define ARG__SET(arg,val) ((arg) = (val)) |
c277df42 |
147 | |
e8e3b0fb |
148 | #undef ARG |
149 | #undef ARG1 |
150 | #undef ARG2 |
151 | |
c277df42 |
152 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
153 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) |
154 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) |
155 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
156 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) |
157 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) |
158 | |
e8e3b0fb |
159 | #undef NEXT_OFF |
160 | #undef NODE_ALIGN |
161 | |
075abff3 |
162 | #define NEXT_OFF(p) ((p)->next_off) |
163 | #define NODE_ALIGN(node) |
164 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ |
a687059c |
165 | |
c277df42 |
166 | #define SIZE_ALIGN NODE_ALIGN |
167 | |
e8e3b0fb |
168 | #undef OP |
169 | #undef OPERAND |
170 | #undef MASK |
171 | #undef STRING |
172 | |
e91177ed |
173 | #define OP(p) ((p)->type) |
174 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 |
175 | #define MASK(p) ((char*)OPERAND(p)) |
176 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) |
177 | #define STRING(p) (((struct regnode_string *)p)->string) |
178 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) |
179 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) |
180 | |
e8e3b0fb |
181 | #undef NODE_ALIGN |
182 | #undef ARG_LOC |
183 | #undef NEXTOPER |
184 | #undef PREVOPER |
185 | |
e91177ed |
186 | #define NODE_ALIGN(node) |
187 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) |
188 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) |
189 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) |
190 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
191 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) |
192 | |
193 | #define NODE_STEP_B 4 |
c277df42 |
194 | |
195 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) |
196 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) |
197 | |
e91177ed |
198 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 |
199 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed |
200 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 |
201 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c |
202 | |
22c35a8c |
203 | #define REG_MAGIC 0234 |
a687059c |
204 | |
830247a4 |
205 | #define SIZE_ONLY (RExC_emit == &PL_regdummy) |
c277df42 |
206 | |
936ed897 |
207 | /* Flags for node->flags of ANYOF */ |
b8c5462f |
208 | |
19103b91 |
209 | #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ |
ffc61ed2 |
210 | #define ANYOF_INVERT 0x04 |
211 | #define ANYOF_FOLD 0x02 |
212 | #define ANYOF_LOCALE 0x01 |
b8c5462f |
213 | |
653099ff |
214 | /* Used for regstclass only */ |
ffc61ed2 |
215 | #define ANYOF_EOS 0x10 /* Can match an empty string too */ |
216 | |
217 | /* There is a character or a range past 0xff */ |
218 | #define ANYOF_UNICODE 0x20 |
1aa99e6b |
219 | #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ |
ffc61ed2 |
220 | |
4f66b38d |
221 | /* size of node is large (includes class pointer) */ |
222 | #define ANYOF_LARGE 0x80 |
223 | |
ffc61ed2 |
224 | /* Are there any runtime flags on in this node? */ |
225 | #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) |
226 | |
227 | #define ANYOF_FLAGS_ALL 0xff |
653099ff |
228 | |
936ed897 |
229 | /* Character classes for node->classflags of ANYOF */ |
653099ff |
230 | /* Should be synchronized with a table in regprop() */ |
231 | /* 2n should pair with 2n+1 */ |
b8c5462f |
232 | |
289d4f09 |
233 | #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
b8c5462f |
234 | #define ANYOF_NALNUM 1 |
aaa51d5e |
235 | #define ANYOF_SPACE 2 /* \s */ |
b8c5462f |
236 | #define ANYOF_NSPACE 3 |
237 | #define ANYOF_DIGIT 4 |
238 | #define ANYOF_NDIGIT 5 |
289d4f09 |
239 | #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ |
b8c5462f |
240 | #define ANYOF_NALNUMC 7 |
241 | #define ANYOF_ALPHA 8 |
242 | #define ANYOF_NALPHA 9 |
243 | #define ANYOF_ASCII 10 |
244 | #define ANYOF_NASCII 11 |
245 | #define ANYOF_CNTRL 12 |
246 | #define ANYOF_NCNTRL 13 |
247 | #define ANYOF_GRAPH 14 |
248 | #define ANYOF_NGRAPH 15 |
249 | #define ANYOF_LOWER 16 |
250 | #define ANYOF_NLOWER 17 |
251 | #define ANYOF_PRINT 18 |
252 | #define ANYOF_NPRINT 19 |
253 | #define ANYOF_PUNCT 20 |
254 | #define ANYOF_NPUNCT 21 |
255 | #define ANYOF_UPPER 22 |
256 | #define ANYOF_NUPPER 23 |
257 | #define ANYOF_XDIGIT 24 |
258 | #define ANYOF_NXDIGIT 25 |
aaa51d5e |
259 | #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ |
260 | #define ANYOF_NPSXSPC 27 |
ffc61ed2 |
261 | #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ |
289d4f09 |
262 | #define ANYOF_NBLANK 29 |
b8c5462f |
263 | |
aaa51d5e |
264 | #define ANYOF_MAX 32 |
b8c5462f |
265 | |
266 | /* Backward source code compatibility. */ |
267 | |
268 | #define ANYOF_ALNUML ANYOF_ALNUM |
269 | #define ANYOF_NALNUML ANYOF_NALNUM |
270 | #define ANYOF_SPACEL ANYOF_SPACE |
271 | #define ANYOF_NSPACEL ANYOF_NSPACE |
272 | |
273 | /* Utility macros for the bitmap and classes of ANYOF */ |
274 | |
936ed897 |
275 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
276 | #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) |
b8c5462f |
277 | |
936ed897 |
278 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f |
279 | |
280 | #define ANYOF_BIT(c) (1 << ((c) & 7)) |
281 | |
936ed897 |
282 | #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) |
b8c5462f |
283 | #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) |
284 | #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) |
285 | #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) |
286 | |
936ed897 |
287 | #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) |
288 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
289 | |
290 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
291 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[((c) >> 3) & 31]) |
b8c5462f |
292 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
293 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) |
294 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) |
295 | |
f8bef550 |
296 | #define ANYOF_BITMAP_SETALL(p) \ |
297 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) |
298 | #define ANYOF_BITMAP_CLEARALL(p) \ |
299 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) |
300 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ |
301 | #define ANYOF_BITMAP_TESTALLSET(p) \ |
302 | 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) |
303 | |
936ed897 |
304 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
305 | #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) |
306 | #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) |
c277df42 |
307 | |
a687059c |
308 | /* |
309 | * Utility definitions. |
310 | */ |
2304df62 |
311 | #ifndef CHARMASK |
075abff3 |
312 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c |
313 | #else |
075abff3 |
314 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c |
315 | #endif |
a687059c |
316 | |
c277df42 |
317 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
318 | |
3dab1dad |
319 | #define REG_SEEN_ZERO_LEN 0x00000001 |
320 | #define REG_SEEN_LOOKBEHIND 0x00000002 |
321 | #define REG_SEEN_GPOS 0x00000004 |
322 | #define REG_SEEN_EVAL 0x00000008 |
323 | #define REG_SEEN_CANY 0x00000010 |
cce850e4 |
324 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
d09b2d29 |
325 | |
73c4f7a1 |
326 | START_EXTERN_C |
327 | |
be8e71aa |
328 | #ifdef PLUGGABLE_RE_EXTENSION |
329 | #include "re_nodes.h" |
330 | #else |
d09b2d29 |
331 | #include "regnodes.h" |
be8e71aa |
332 | #endif |
d09b2d29 |
333 | |
4d61ec05 |
334 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
d09b2d29 |
335 | #ifndef DOINIT |
4d61ec05 |
336 | EXTCONST U8 PL_varies[]; |
d09b2d29 |
337 | #else |
4d61ec05 |
338 | EXTCONST U8 PL_varies[] = { |
0111c4fd |
339 | BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, |
a0ed51b3 |
340 | WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0 |
c277df42 |
341 | }; |
d09b2d29 |
342 | #endif |
c277df42 |
343 | |
4d61ec05 |
344 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
345 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ |
d09b2d29 |
346 | #ifndef DOINIT |
4d61ec05 |
347 | EXTCONST U8 PL_simple[]; |
d09b2d29 |
348 | #else |
4d61ec05 |
349 | EXTCONST U8 PL_simple[] = { |
f33976b4 |
350 | REG_ANY, SANY, CANY, |
ffc61ed2 |
351 | ANYOF, |
352 | ALNUM, ALNUML, |
353 | NALNUM, NALNUML, |
354 | SPACE, SPACEL, |
355 | NSPACE, NSPACEL, |
356 | DIGIT, NDIGIT, |
357 | 0 |
c277df42 |
358 | }; |
359 | #endif |
360 | |
73c4f7a1 |
361 | END_EXTERN_C |
cad2e5aa |
362 | |
363 | typedef struct re_scream_pos_data_s |
364 | { |
365 | char **scream_olds; /* match pos */ |
366 | I32 *scream_pos; /* Internal iterator of scream. */ |
367 | } re_scream_pos_data; |
368 | |
261faec3 |
369 | /* .what is a character array with one character for each member of .data |
370 | * The character describes the function of the corresponding .data item: |
0111c4fd |
371 | * f - start-class data for regstclass optimization |
261faec3 |
372 | * n - Root of op tree for (?{EVAL}) item |
373 | * o - Start op for (?{EVAL}) item |
3dab1dad |
374 | * p - Pad for (?{EVAL}) item |
9e55ce06 |
375 | * s - swash for unicode-style character class, and the multicharacter |
376 | * strings resulting from casefolding the single-character entries |
377 | * in the character class |
a3621e74 |
378 | * t - trie struct |
07be1b83 |
379 | * T - aho-trie struct |
261faec3 |
380 | * 20010712 mjd@plover.com |
c3613ed7 |
381 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 |
382 | */ |
cad2e5aa |
383 | struct reg_data { |
384 | U32 count; |
385 | U8 *what; |
386 | void* data[1]; |
387 | }; |
388 | |
389 | struct reg_substr_datum { |
390 | I32 min_offset; |
391 | I32 max_offset; |
33b8afdf |
392 | SV *substr; /* non-utf8 variant */ |
393 | SV *utf8_substr; /* utf8 variant */ |
cad2e5aa |
394 | }; |
395 | |
396 | struct reg_substr_data { |
397 | struct reg_substr_datum data[3]; /* Actual array */ |
398 | }; |
399 | |
400 | #define anchored_substr substrs->data[0].substr |
33b8afdf |
401 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa |
402 | #define anchored_offset substrs->data[0].min_offset |
403 | #define float_substr substrs->data[1].substr |
33b8afdf |
404 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa |
405 | #define float_min_offset substrs->data[1].min_offset |
406 | #define float_max_offset substrs->data[1].max_offset |
407 | #define check_substr substrs->data[2].substr |
33b8afdf |
408 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa |
409 | #define check_offset_min substrs->data[2].min_offset |
410 | #define check_offset_max substrs->data[2].max_offset |
a3621e74 |
411 | |
412 | |
413 | |
414 | /* trie related stuff */ |
5d9a96ca |
415 | |
a3621e74 |
416 | /* a transition record for the state machine. the |
417 | check field determines which state "owns" the |
418 | transition. the char the transition is for is |
419 | determined by offset from the owning states base |
420 | field. the next field determines which state |
421 | is to be transitioned to if any. |
422 | */ |
423 | struct _reg_trie_trans { |
424 | U32 next; |
425 | U32 check; |
426 | }; |
427 | |
428 | /* a transition list element for the list based representation */ |
429 | struct _reg_trie_trans_list_elem { |
430 | U16 forid; |
431 | U32 newstate; |
432 | }; |
433 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; |
434 | |
435 | /* a state for compressed nodes. base is an offset |
436 | into an array of reg_trie_trans array. If wordnum is |
437 | nonzero the state is accepting. if base is zero then |
438 | the state has no children (and will be accepting) |
439 | */ |
440 | struct _reg_trie_state { |
441 | U16 wordnum; |
442 | union { |
443 | U32 base; |
444 | reg_trie_trans_le* list; |
445 | } trans; |
446 | }; |
447 | |
448 | |
449 | |
a3621e74 |
450 | typedef struct _reg_trie_state reg_trie_state; |
451 | typedef struct _reg_trie_trans reg_trie_trans; |
452 | |
453 | |
454 | /* anything in here that needs to be freed later |
8e11feef |
455 | should be dealt with in pregfree */ |
a3621e74 |
456 | struct _reg_trie_data { |
457 | U16 uniquecharcount; |
cc601c31 |
458 | U32 lasttrans; |
a3621e74 |
459 | U16 *charmap; |
460 | HV *widecharmap; |
461 | reg_trie_state *states; |
462 | reg_trie_trans *trans; |
3dab1dad |
463 | char *bitmap; |
a3621e74 |
464 | U32 refcount; |
3dab1dad |
465 | U32 startstate; |
8e11feef |
466 | STRLEN minlen; |
467 | STRLEN maxlen; |
07be1b83 |
468 | U32 *wordlen; |
469 | U32 laststate; /* Build only */ |
a3621e74 |
470 | #ifdef DEBUGGING |
3dab1dad |
471 | U16 wordcount; /* Build only */ |
472 | STRLEN charcount; /* Build only */ |
a3621e74 |
473 | AV *words; |
474 | AV *revcharmap; |
475 | #endif |
476 | }; |
a3621e74 |
477 | typedef struct _reg_trie_data reg_trie_data; |
478 | |
07be1b83 |
479 | struct _reg_ac_data { |
480 | U32 *fail; |
481 | reg_trie_state *states; |
482 | reg_trie_data *trie; |
483 | U32 refcount; |
484 | }; |
485 | typedef struct _reg_ac_data reg_ac_data; |
486 | |
3dab1dad |
487 | /* ANY_BIT doesnt use the structure, so we can borrow it here. |
488 | This is simpler than refactoring all of it as wed end up with |
489 | three different sets... */ |
490 | |
491 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) |
5dfe062f |
492 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)c) >> 3) & 31]) |
493 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
494 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) |
495 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
496 | |
497 | |
a3621e74 |
498 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
499 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) |
500 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) |
501 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) |
502 | |
3dab1dad |
503 | #ifdef DEBUGGING |
504 | #define TRIE_WORDCOUNT(trie) ((trie)->wordcount) |
505 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
506 | #define TRIE_LASTSTATE(trie) ((trie)->laststate) |
507 | #define TRIE_REVCHARMAP(trie) ((trie)->revcharmap) |
508 | #else |
509 | #define TRIE_WORDCOUNT(trie) (trie_wordcount) |
510 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
07be1b83 |
511 | /*#define TRIE_LASTSTATE(trie) (trie_laststate)*/ |
512 | #define TRIE_LASTSTATE(trie) ((trie)->laststate) |
3dab1dad |
513 | #define TRIE_REVCHARMAP(trie) (trie_revcharmap) |
514 | #endif |
a3621e74 |
515 | |
0111c4fd |
516 | #define RE_TRIE_MAXBUF_INIT 65536 |
517 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" |
a3621e74 |
518 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
519 | |
be8e71aa |
520 | /* |
521 | |
522 | RE_DEBUG_FLAGS is used to control what debug output is emitted |
523 | its divided into three groups of options, some of which interact. |
524 | The three groups are: Compile, Execute, Extra. There is room for a |
525 | further group, as currently only the low three bytes are used. |
526 | |
527 | Compile Options: |
07be1b83 |
528 | |
be8e71aa |
529 | PARSE |
530 | PEEP |
531 | TRIE |
532 | PROGRAM |
533 | OFFSETS |
a3621e74 |
534 | |
be8e71aa |
535 | Execute Options: |
a3621e74 |
536 | |
be8e71aa |
537 | INTUIT |
538 | MATCH |
539 | TRIE |
a3621e74 |
540 | |
be8e71aa |
541 | Extra Options |
a3621e74 |
542 | |
be8e71aa |
543 | TRIE |
544 | OFFSETS |
545 | |
546 | If you modify any of these make sure you make corresponding changes to |
547 | re.pm, especially to the documentation. |
a3621e74 |
548 | |
be8e71aa |
549 | */ |
550 | |
551 | |
552 | /* Compile */ |
553 | #define RE_DEBUG_COMPILE_MASK 0x0000FF |
554 | #define RE_DEBUG_COMPILE_PARSE 0x000001 |
555 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 |
556 | #define RE_DEBUG_COMPILE_TRIE 0x000004 |
557 | #define RE_DEBUG_COMPILE_DUMP 0x000008 |
558 | #define RE_DEBUG_COMPILE_OFFSETS 0x000010 |
559 | |
560 | /* Execute */ |
561 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 |
562 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 |
563 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 |
564 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 |
565 | |
566 | /* Extra */ |
567 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 |
568 | #define RE_DEBUG_EXTRA_TRIE 0x010000 |
569 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 |
570 | |
e68ec53f |
571 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa |
572 | /* Compile */ |
573 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
574 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa |
575 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
576 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
577 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f |
578 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa |
579 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
580 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
581 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f |
582 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa |
583 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
e68ec53f |
584 | if (re_debug_flags & RE_DEBUG_COMPILE_OFFSETS) x ) |
be8e71aa |
585 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
586 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
be8e71aa |
587 | |
588 | /* Execute */ |
589 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
590 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa |
591 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f |
592 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa |
593 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f |
594 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa |
595 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
596 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa |
597 | |
598 | /* Extra */ |
599 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ |
e68ec53f |
600 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
07be1b83 |
601 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
e68ec53f |
602 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) \ |
be8e71aa |
603 | Perl_warn_nocontext x ) |
604 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ |
e68ec53f |
605 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
606 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f |
607 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
608 | |
609 | #define DEBUG_TRIE_r(x) DEBUG_r( \ |
e68ec53f |
610 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa |
611 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
612 | |
613 | /* initialization */ |
614 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f |
615 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
616 | SV * re_debug_flags_sv = NULL; \ |
617 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ |
618 | if (re_debug_flags_sv) { \ |
8e9a8a48 |
619 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f |
620 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 |
621 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f |
622 | }\ |
623 | }) |
a3621e74 |
624 | |
625 | #ifdef DEBUGGING |
8e9a8a48 |
626 | #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; |
0df25f3d |
627 | #define RE_PV_DISPLAY_DECL(rpv,rlen,isuni,dsv,pv,l,m) \ |
628 | const char * const rpv = (isuni) ? \ |
629 | pv_uni_display(dsv, (U8*)(pv), l, m, UNI_DISPLAY_REGEX) : \ |
630 | pv_escape(dsv, pv, l, m, 0); \ |
631 | const int rlen = SvCUR(dsv) |
a3621e74 |
632 | #else |
633 | #define GET_RE_DEBUG_FLAGS_DECL |
0df25f3d |
634 | #define RE_PV_DISPLAY_DECL |
a3621e74 |
635 | #endif |
636 | |