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 | |
786e8c11 |
106 | /* also used by trie */ |
936ed897 |
107 | struct regnode_charclass { |
108 | U8 flags; |
109 | U8 type; |
110 | U16 next_off; |
ffc61ed2 |
111 | U32 arg1; |
19103b91 |
112 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 |
113 | }; |
114 | |
19103b91 |
115 | struct regnode_charclass_class { /* has [[:blah:]] classes */ |
116 | U8 flags; /* should have ANYOF_CLASS here */ |
936ed897 |
117 | U8 type; |
118 | U16 next_off; |
ffc61ed2 |
119 | U32 arg1; |
19103b91 |
120 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
121 | char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ |
936ed897 |
122 | }; |
123 | |
9c7e81e8 |
124 | /* XXX fix this description. |
125 | Impose a limit of REG_INFTY on various pattern matching operations |
126 | to limit stack growth and to avoid "infinite" recursions. |
127 | */ |
128 | /* The default size for REG_INFTY is I16_MAX, which is the same as |
129 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits |
130 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is |
131 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is |
132 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a |
133 | smaller default. |
134 | --Andy Dougherty 11 June 1998 |
135 | */ |
136 | #if SHORTSIZE > 2 |
137 | # ifndef REG_INFTY |
138 | # define REG_INFTY ((1<<15)-1) |
139 | # endif |
140 | #endif |
141 | |
142 | #ifndef REG_INFTY |
83cfe48c |
143 | # define REG_INFTY I16_MAX |
144 | #endif |
a687059c |
145 | |
e91177ed |
146 | #define ARG_VALUE(arg) (arg) |
147 | #define ARG__SET(arg,val) ((arg) = (val)) |
c277df42 |
148 | |
e8e3b0fb |
149 | #undef ARG |
150 | #undef ARG1 |
151 | #undef ARG2 |
152 | |
c277df42 |
153 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
154 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) |
155 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) |
786e8c11 |
156 | |
c277df42 |
157 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
158 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) |
159 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) |
160 | |
e8e3b0fb |
161 | #undef NEXT_OFF |
162 | #undef NODE_ALIGN |
163 | |
075abff3 |
164 | #define NEXT_OFF(p) ((p)->next_off) |
165 | #define NODE_ALIGN(node) |
166 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ |
a687059c |
167 | |
c277df42 |
168 | #define SIZE_ALIGN NODE_ALIGN |
169 | |
e8e3b0fb |
170 | #undef OP |
171 | #undef OPERAND |
172 | #undef MASK |
173 | #undef STRING |
174 | |
e91177ed |
175 | #define OP(p) ((p)->type) |
176 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 |
177 | #define MASK(p) ((char*)OPERAND(p)) |
178 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) |
179 | #define STRING(p) (((struct regnode_string *)p)->string) |
180 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) |
181 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) |
182 | |
e8e3b0fb |
183 | #undef NODE_ALIGN |
184 | #undef ARG_LOC |
185 | #undef NEXTOPER |
186 | #undef PREVOPER |
187 | |
e91177ed |
188 | #define NODE_ALIGN(node) |
189 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) |
190 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) |
191 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) |
786e8c11 |
192 | |
193 | |
e91177ed |
194 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
195 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) |
196 | |
197 | #define NODE_STEP_B 4 |
c277df42 |
198 | |
199 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) |
200 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) |
201 | |
e91177ed |
202 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 |
203 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed |
204 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 |
205 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c |
206 | |
22c35a8c |
207 | #define REG_MAGIC 0234 |
a687059c |
208 | |
830247a4 |
209 | #define SIZE_ONLY (RExC_emit == &PL_regdummy) |
c277df42 |
210 | |
936ed897 |
211 | /* Flags for node->flags of ANYOF */ |
b8c5462f |
212 | |
19103b91 |
213 | #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ |
ffc61ed2 |
214 | #define ANYOF_INVERT 0x04 |
215 | #define ANYOF_FOLD 0x02 |
216 | #define ANYOF_LOCALE 0x01 |
b8c5462f |
217 | |
653099ff |
218 | /* Used for regstclass only */ |
ffc61ed2 |
219 | #define ANYOF_EOS 0x10 /* Can match an empty string too */ |
220 | |
221 | /* There is a character or a range past 0xff */ |
222 | #define ANYOF_UNICODE 0x20 |
1aa99e6b |
223 | #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ |
ffc61ed2 |
224 | |
4f66b38d |
225 | /* size of node is large (includes class pointer) */ |
226 | #define ANYOF_LARGE 0x80 |
227 | |
ffc61ed2 |
228 | /* Are there any runtime flags on in this node? */ |
229 | #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) |
230 | |
231 | #define ANYOF_FLAGS_ALL 0xff |
653099ff |
232 | |
936ed897 |
233 | /* Character classes for node->classflags of ANYOF */ |
653099ff |
234 | /* Should be synchronized with a table in regprop() */ |
235 | /* 2n should pair with 2n+1 */ |
b8c5462f |
236 | |
289d4f09 |
237 | #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
b8c5462f |
238 | #define ANYOF_NALNUM 1 |
aaa51d5e |
239 | #define ANYOF_SPACE 2 /* \s */ |
b8c5462f |
240 | #define ANYOF_NSPACE 3 |
241 | #define ANYOF_DIGIT 4 |
242 | #define ANYOF_NDIGIT 5 |
289d4f09 |
243 | #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ |
b8c5462f |
244 | #define ANYOF_NALNUMC 7 |
245 | #define ANYOF_ALPHA 8 |
246 | #define ANYOF_NALPHA 9 |
247 | #define ANYOF_ASCII 10 |
248 | #define ANYOF_NASCII 11 |
249 | #define ANYOF_CNTRL 12 |
250 | #define ANYOF_NCNTRL 13 |
251 | #define ANYOF_GRAPH 14 |
252 | #define ANYOF_NGRAPH 15 |
253 | #define ANYOF_LOWER 16 |
254 | #define ANYOF_NLOWER 17 |
255 | #define ANYOF_PRINT 18 |
256 | #define ANYOF_NPRINT 19 |
257 | #define ANYOF_PUNCT 20 |
258 | #define ANYOF_NPUNCT 21 |
259 | #define ANYOF_UPPER 22 |
260 | #define ANYOF_NUPPER 23 |
261 | #define ANYOF_XDIGIT 24 |
262 | #define ANYOF_NXDIGIT 25 |
aaa51d5e |
263 | #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ |
264 | #define ANYOF_NPSXSPC 27 |
ffc61ed2 |
265 | #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ |
289d4f09 |
266 | #define ANYOF_NBLANK 29 |
b8c5462f |
267 | |
aaa51d5e |
268 | #define ANYOF_MAX 32 |
b8c5462f |
269 | |
270 | /* Backward source code compatibility. */ |
271 | |
272 | #define ANYOF_ALNUML ANYOF_ALNUM |
273 | #define ANYOF_NALNUML ANYOF_NALNUM |
274 | #define ANYOF_SPACEL ANYOF_SPACE |
275 | #define ANYOF_NSPACEL ANYOF_NSPACE |
276 | |
277 | /* Utility macros for the bitmap and classes of ANYOF */ |
278 | |
936ed897 |
279 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
280 | #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) |
b8c5462f |
281 | |
936ed897 |
282 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f |
283 | |
284 | #define ANYOF_BIT(c) (1 << ((c) & 7)) |
285 | |
936ed897 |
286 | #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) |
b8c5462f |
287 | #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) |
288 | #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) |
289 | #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) |
290 | |
936ed897 |
291 | #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) |
292 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
293 | |
294 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
786e8c11 |
295 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f |
296 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
297 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) |
298 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) |
299 | |
f8bef550 |
300 | #define ANYOF_BITMAP_SETALL(p) \ |
301 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) |
302 | #define ANYOF_BITMAP_CLEARALL(p) \ |
303 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) |
304 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ |
305 | #define ANYOF_BITMAP_TESTALLSET(p) \ |
306 | 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) |
307 | |
936ed897 |
308 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
309 | #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) |
310 | #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) |
c277df42 |
311 | |
786e8c11 |
312 | |
a687059c |
313 | /* |
314 | * Utility definitions. |
315 | */ |
2304df62 |
316 | #ifndef CHARMASK |
075abff3 |
317 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c |
318 | #else |
075abff3 |
319 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c |
320 | #endif |
a687059c |
321 | |
c277df42 |
322 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
323 | |
3dab1dad |
324 | #define REG_SEEN_ZERO_LEN 0x00000001 |
325 | #define REG_SEEN_LOOKBEHIND 0x00000002 |
326 | #define REG_SEEN_GPOS 0x00000004 |
327 | #define REG_SEEN_EVAL 0x00000008 |
328 | #define REG_SEEN_CANY 0x00000010 |
cce850e4 |
329 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
d09b2d29 |
330 | |
73c4f7a1 |
331 | START_EXTERN_C |
332 | |
be8e71aa |
333 | #ifdef PLUGGABLE_RE_EXTENSION |
334 | #include "re_nodes.h" |
335 | #else |
d09b2d29 |
336 | #include "regnodes.h" |
be8e71aa |
337 | #endif |
d09b2d29 |
338 | |
4d61ec05 |
339 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
d09b2d29 |
340 | #ifndef DOINIT |
4d61ec05 |
341 | EXTCONST U8 PL_varies[]; |
d09b2d29 |
342 | #else |
4d61ec05 |
343 | EXTCONST U8 PL_varies[] = { |
0111c4fd |
344 | BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, |
a0ed51b3 |
345 | WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0 |
c277df42 |
346 | }; |
d09b2d29 |
347 | #endif |
c277df42 |
348 | |
4d61ec05 |
349 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
350 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ |
d09b2d29 |
351 | #ifndef DOINIT |
4d61ec05 |
352 | EXTCONST U8 PL_simple[]; |
d09b2d29 |
353 | #else |
4d61ec05 |
354 | EXTCONST U8 PL_simple[] = { |
f33976b4 |
355 | REG_ANY, SANY, CANY, |
ffc61ed2 |
356 | ANYOF, |
357 | ALNUM, ALNUML, |
358 | NALNUM, NALNUML, |
359 | SPACE, SPACEL, |
360 | NSPACE, NSPACEL, |
361 | DIGIT, NDIGIT, |
362 | 0 |
c277df42 |
363 | }; |
364 | #endif |
365 | |
73c4f7a1 |
366 | END_EXTERN_C |
cad2e5aa |
367 | |
368 | typedef struct re_scream_pos_data_s |
369 | { |
370 | char **scream_olds; /* match pos */ |
371 | I32 *scream_pos; /* Internal iterator of scream. */ |
372 | } re_scream_pos_data; |
373 | |
261faec3 |
374 | /* .what is a character array with one character for each member of .data |
375 | * The character describes the function of the corresponding .data item: |
0111c4fd |
376 | * f - start-class data for regstclass optimization |
261faec3 |
377 | * n - Root of op tree for (?{EVAL}) item |
378 | * o - Start op for (?{EVAL}) item |
3dab1dad |
379 | * p - Pad for (?{EVAL}) item |
9e55ce06 |
380 | * s - swash for unicode-style character class, and the multicharacter |
381 | * strings resulting from casefolding the single-character entries |
382 | * in the character class |
a3621e74 |
383 | * t - trie struct |
07be1b83 |
384 | * T - aho-trie struct |
261faec3 |
385 | * 20010712 mjd@plover.com |
c3613ed7 |
386 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 |
387 | */ |
cad2e5aa |
388 | struct reg_data { |
389 | U32 count; |
390 | U8 *what; |
391 | void* data[1]; |
392 | }; |
393 | |
394 | struct reg_substr_datum { |
395 | I32 min_offset; |
396 | I32 max_offset; |
33b8afdf |
397 | SV *substr; /* non-utf8 variant */ |
398 | SV *utf8_substr; /* utf8 variant */ |
cad2e5aa |
399 | }; |
400 | |
401 | struct reg_substr_data { |
402 | struct reg_substr_datum data[3]; /* Actual array */ |
403 | }; |
404 | |
405 | #define anchored_substr substrs->data[0].substr |
33b8afdf |
406 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa |
407 | #define anchored_offset substrs->data[0].min_offset |
408 | #define float_substr substrs->data[1].substr |
33b8afdf |
409 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa |
410 | #define float_min_offset substrs->data[1].min_offset |
411 | #define float_max_offset substrs->data[1].max_offset |
412 | #define check_substr substrs->data[2].substr |
33b8afdf |
413 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa |
414 | #define check_offset_min substrs->data[2].min_offset |
415 | #define check_offset_max substrs->data[2].max_offset |
a3621e74 |
416 | |
417 | |
418 | |
419 | /* trie related stuff */ |
5d9a96ca |
420 | |
a3621e74 |
421 | /* a transition record for the state machine. the |
422 | check field determines which state "owns" the |
423 | transition. the char the transition is for is |
424 | determined by offset from the owning states base |
425 | field. the next field determines which state |
426 | is to be transitioned to if any. |
427 | */ |
428 | struct _reg_trie_trans { |
429 | U32 next; |
430 | U32 check; |
431 | }; |
432 | |
433 | /* a transition list element for the list based representation */ |
434 | struct _reg_trie_trans_list_elem { |
435 | U16 forid; |
436 | U32 newstate; |
437 | }; |
438 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; |
439 | |
440 | /* a state for compressed nodes. base is an offset |
441 | into an array of reg_trie_trans array. If wordnum is |
442 | nonzero the state is accepting. if base is zero then |
443 | the state has no children (and will be accepting) |
444 | */ |
445 | struct _reg_trie_state { |
446 | U16 wordnum; |
447 | union { |
448 | U32 base; |
449 | reg_trie_trans_le* list; |
450 | } trans; |
451 | }; |
452 | |
453 | |
454 | |
a3621e74 |
455 | typedef struct _reg_trie_state reg_trie_state; |
456 | typedef struct _reg_trie_trans reg_trie_trans; |
457 | |
458 | |
459 | /* anything in here that needs to be freed later |
8e11feef |
460 | should be dealt with in pregfree */ |
a3621e74 |
461 | struct _reg_trie_data { |
786e8c11 |
462 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
463 | U32 lasttrans; /* last valid transition element */ |
464 | U16 *charmap; /* byte to charid lookup array */ |
465 | HV *widecharmap; /* code points > 255 to charid */ |
466 | reg_trie_state *states; /* state data */ |
467 | reg_trie_trans *trans; /* array of transition elements */ |
468 | char *bitmap; /* stclass bitmap */ |
469 | U32 refcount; /* number of times this trie is referenced */ |
470 | U32 startstate; /* initial state - used for common prefix optimisation */ |
471 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ |
472 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ |
473 | U32 *wordlen; /* array of lengths of words */ |
474 | U16 *jump; /* optional 1 indexed array of offsets before tail |
475 | for the node following a given word. */ |
476 | U16 *nextword; /* optional 1 indexed array to support linked list |
477 | of duplicate wordnums */ |
478 | U32 laststate; /* Build only */ |
479 | U32 wordcount; /* Build only */ |
a3621e74 |
480 | #ifdef DEBUGGING |
786e8c11 |
481 | STRLEN charcount; /* Build only */ |
482 | AV *words; /* Array of words contained in trie, for dumping */ |
483 | AV *revcharmap; /* Map of each charid back to its character representation */ |
a3621e74 |
484 | #endif |
485 | }; |
a3621e74 |
486 | typedef struct _reg_trie_data reg_trie_data; |
487 | |
07be1b83 |
488 | struct _reg_ac_data { |
489 | U32 *fail; |
490 | reg_trie_state *states; |
491 | reg_trie_data *trie; |
492 | U32 refcount; |
493 | }; |
494 | typedef struct _reg_ac_data reg_ac_data; |
495 | |
3dab1dad |
496 | /* ANY_BIT doesnt use the structure, so we can borrow it here. |
497 | This is simpler than refactoring all of it as wed end up with |
498 | three different sets... */ |
499 | |
500 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) |
786e8c11 |
501 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f |
502 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
503 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) |
504 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
505 | |
786e8c11 |
506 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
507 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
508 | |
a3621e74 |
509 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
510 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) |
511 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) |
512 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) |
513 | |
3dab1dad |
514 | #ifdef DEBUGGING |
3dab1dad |
515 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad |
516 | #define TRIE_REVCHARMAP(trie) ((trie)->revcharmap) |
517 | #else |
3dab1dad |
518 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad |
519 | #define TRIE_REVCHARMAP(trie) (trie_revcharmap) |
520 | #endif |
a3621e74 |
521 | |
0111c4fd |
522 | #define RE_TRIE_MAXBUF_INIT 65536 |
523 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" |
a3621e74 |
524 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
525 | |
be8e71aa |
526 | /* |
527 | |
528 | RE_DEBUG_FLAGS is used to control what debug output is emitted |
529 | its divided into three groups of options, some of which interact. |
530 | The three groups are: Compile, Execute, Extra. There is room for a |
531 | further group, as currently only the low three bytes are used. |
532 | |
533 | Compile Options: |
07be1b83 |
534 | |
be8e71aa |
535 | PARSE |
536 | PEEP |
537 | TRIE |
538 | PROGRAM |
539 | OFFSETS |
a3621e74 |
540 | |
be8e71aa |
541 | Execute Options: |
a3621e74 |
542 | |
be8e71aa |
543 | INTUIT |
544 | MATCH |
545 | TRIE |
a3621e74 |
546 | |
be8e71aa |
547 | Extra Options |
a3621e74 |
548 | |
be8e71aa |
549 | TRIE |
550 | OFFSETS |
551 | |
552 | If you modify any of these make sure you make corresponding changes to |
553 | re.pm, especially to the documentation. |
a3621e74 |
554 | |
be8e71aa |
555 | */ |
556 | |
557 | |
558 | /* Compile */ |
559 | #define RE_DEBUG_COMPILE_MASK 0x0000FF |
560 | #define RE_DEBUG_COMPILE_PARSE 0x000001 |
561 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 |
562 | #define RE_DEBUG_COMPILE_TRIE 0x000004 |
563 | #define RE_DEBUG_COMPILE_DUMP 0x000008 |
564 | #define RE_DEBUG_COMPILE_OFFSETS 0x000010 |
565 | |
566 | /* Execute */ |
567 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 |
568 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 |
569 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 |
570 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 |
571 | |
572 | /* Extra */ |
573 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 |
574 | #define RE_DEBUG_EXTRA_TRIE 0x010000 |
575 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 |
ab3bbdeb |
576 | #define RE_DEBUG_EXTRA_STATE 0x040000 |
be8e71aa |
577 | |
e68ec53f |
578 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa |
579 | /* Compile */ |
580 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
581 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa |
582 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
583 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
584 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f |
585 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa |
586 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
587 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
588 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f |
589 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa |
590 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
e68ec53f |
591 | if (re_debug_flags & RE_DEBUG_COMPILE_OFFSETS) x ) |
be8e71aa |
592 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
593 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
be8e71aa |
594 | |
595 | /* Execute */ |
596 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
597 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa |
598 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f |
599 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa |
600 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f |
601 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa |
602 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
603 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa |
604 | |
605 | /* Extra */ |
606 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ |
e68ec53f |
607 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
ab3bbdeb |
608 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
609 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) |
07be1b83 |
610 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
e68ec53f |
611 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) \ |
be8e71aa |
612 | Perl_warn_nocontext x ) |
613 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ |
e68ec53f |
614 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
615 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f |
616 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
617 | |
618 | #define DEBUG_TRIE_r(x) DEBUG_r( \ |
e68ec53f |
619 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa |
620 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
621 | |
622 | /* initialization */ |
623 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f |
624 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
625 | SV * re_debug_flags_sv = NULL; \ |
626 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ |
627 | if (re_debug_flags_sv) { \ |
8e9a8a48 |
628 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f |
629 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 |
630 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f |
631 | }\ |
632 | }) |
a3621e74 |
633 | |
634 | #ifdef DEBUGGING |
ab3bbdeb |
635 | |
8e9a8a48 |
636 | #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; |
ab3bbdeb |
637 | |
638 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ |
639 | const char * const rpv = \ |
ddc5bc0f |
640 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
641 | PL_colors[(c1)],PL_colors[(c2)], \ |
642 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ |
0df25f3d |
643 | const int rlen = SvCUR(dsv) |
ab3bbdeb |
644 | |
645 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ |
646 | const char * const rpv = \ |
ddc5bc0f |
647 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb |
648 | PL_colors[(c1)],PL_colors[(c2)], \ |
649 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) |
650 | |
651 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ |
652 | const char * const rpv = \ |
ddc5bc0f |
653 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
654 | PL_colors[0], PL_colors[1], \ |
655 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \ |
656 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ |
657 | ) |
658 | |
659 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) |
660 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") |
661 | |
662 | #else /* if not DEBUGGING */ |
663 | |
a3621e74 |
664 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb |
665 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
666 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) |
667 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) |
668 | #define RE_SV_DUMPLEN(ItEm) |
669 | #define RE_SV_TAIL(ItEm) |
670 | |
671 | #endif /* DEBUG RELATED DEFINES */ |
a3621e74 |
672 | |