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 | |
58e23c8d |
14 | /* Convert branch sequences to more efficient trie ops? */ |
07be1b83 |
15 | #define PERL_ENABLE_TRIE_OPTIMISATION 1 |
58e23c8d |
16 | |
17 | /* Be really agressive about optimising patterns with trie sequences? */ |
07be1b83 |
18 | #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1 |
58e23c8d |
19 | |
20 | /* Should the optimiser take positive assertions into account? */ |
1de06328 |
21 | #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 1 |
58e23c8d |
22 | |
23 | /* Not for production use: */ |
07be1b83 |
24 | #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 |
58e23c8d |
25 | |
b9b4dddf |
26 | /* Unless the next line is uncommented it is illegal to combine lazy |
27 | matching with possessive matching. Frankly it doesn't make much sense |
28 | to allow it as X*?+ matches nothing, X+?+ matches a single char only, |
29 | and X{min,max}?+ matches min times only. |
30 | */ |
31 | /* #define REG_ALLOW_MINMOD_SUSPEND */ |
07be1b83 |
32 | |
a687059c |
33 | /* |
34 | * The "internal use only" fields in regexp.h are present to pass info from |
35 | * compile to execute that permits the execute phase to run lots faster on |
36 | * simple cases. They are: |
37 | * |
bd61b366 |
38 | * regstart sv that must begin a match; NULL if none obvious |
a687059c |
39 | * reganch is the match anchored (at beginning-of-line only)? |
40 | * regmust string (pointer into program) that match must include, or NULL |
79072805 |
41 | * [regmust changed to SV* for bminstr()--law] |
a687059c |
42 | * regmlen length of regmust string |
43 | * [regmlen not used currently] |
44 | * |
45 | * Regstart and reganch permit very fast decisions on suitable starting points |
46 | * for a match, cutting down the work a lot. Regmust permits fast rejection |
47 | * of lines that cannot possibly match. The regmust tests are costly enough |
e50aee73 |
48 | * that pregcomp() supplies a regmust only if the r.e. contains something |
a687059c |
49 | * potentially expensive (at present, the only such thing detected is * or + |
50 | * at the start of the r.e., which can involve a lot of backup). Regmlen is |
e50aee73 |
51 | * supplied because the test in pregexec() needs it and pregcomp() is computing |
a687059c |
52 | * it anyway. |
53 | * [regmust is now supplied always. The tests that use regmust have a |
54 | * heuristic that disables the test if it usually matches.] |
55 | * |
56 | * [In fact, we now use regmust in many cases to locate where the search |
57 | * starts in the string, so if regback is >= 0, the regmust search is never |
58 | * wasted effort. The regback variable says how many characters back from |
59 | * where regmust matched is the earliest possible start of the match. |
60 | * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] |
61 | */ |
62 | |
63 | /* |
64 | * Structure for regexp "program". This is essentially a linear encoding |
65 | * of a nondeterministic finite-state machine (aka syntax charts or |
66 | * "railroad normal form" in parsing technology). Each node is an opcode |
67 | * plus a "next" pointer, possibly plus an operand. "Next" pointers of |
68 | * all nodes except BRANCH implement concatenation; a "next" pointer with |
69 | * a BRANCH on both ends of it is connecting two alternatives. (Here we |
70 | * have one of the subtle syntax dependencies: an individual BRANCH (as |
71 | * opposed to a collection of them) is never concatenated with anything |
72 | * because of operator precedence.) The operand of some types of node is |
73 | * a literal string; for others, it is a node leading into a sub-FSM. In |
74 | * particular, the operand of a BRANCH node is the first node of the branch. |
75 | * (NB this is *not* a tree structure: the tail of the branch connects |
a3621e74 |
76 | * to the thing following the set of BRANCHes.) The opcodes are defined |
77 | * in regnodes.h which is generated from regcomp.sym by regcomp.pl. |
a687059c |
78 | */ |
79 | |
a687059c |
80 | /* |
81 | * A node is one char of opcode followed by two chars of "next" pointer. |
82 | * "Next" pointers are stored as two 8-bit pieces, high order first. The |
83 | * value is a positive offset from the opcode of the node containing it. |
84 | * An operand, if any, simply follows the node. (Note that much of the |
85 | * code generation knows about this implicit relationship.) |
86 | * |
87 | * Using two bytes for the "next" pointer is vast overkill for most things, |
88 | * but allows patterns to get big without disasters. |
89 | * |
e91177ed |
90 | * [The "next" pointer is always aligned on an even |
a687059c |
91 | * boundary, and reads the offset directly as a short. Also, there is no |
92 | * special test to reverse the sign of BACK pointers since the offset is |
93 | * stored negative.] |
94 | */ |
95 | |
c277df42 |
96 | struct regnode_string { |
cd439c50 |
97 | U8 str_len; |
c277df42 |
98 | U8 type; |
99 | U16 next_off; |
cd439c50 |
100 | char string[1]; |
c277df42 |
101 | }; |
a687059c |
102 | |
6bda09f9 |
103 | /* Argument bearing node - workhorse, |
104 | arg1 is often for the data field */ |
c277df42 |
105 | struct regnode_1 { |
106 | U8 flags; |
107 | U8 type; |
108 | U16 next_off; |
109 | U32 arg1; |
110 | }; |
111 | |
6bda09f9 |
112 | /* Similar to a regnode_1 but with an extra signed argument */ |
113 | struct regnode_2L { |
114 | U8 flags; |
115 | U8 type; |
116 | U16 next_off; |
117 | U32 arg1; |
118 | I32 arg2; |
119 | }; |
120 | |
121 | /* 'Two field' -- Two 16 bit unsigned args */ |
c277df42 |
122 | struct regnode_2 { |
123 | U8 flags; |
124 | U8 type; |
125 | U16 next_off; |
126 | U16 arg1; |
127 | U16 arg2; |
128 | }; |
129 | |
6bda09f9 |
130 | |
936ed897 |
131 | #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ |
ffc61ed2 |
132 | #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ |
936ed897 |
133 | |
786e8c11 |
134 | /* also used by trie */ |
936ed897 |
135 | struct regnode_charclass { |
136 | U8 flags; |
137 | U8 type; |
138 | U16 next_off; |
ffc61ed2 |
139 | U32 arg1; |
19103b91 |
140 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 |
141 | }; |
142 | |
19103b91 |
143 | struct regnode_charclass_class { /* has [[:blah:]] classes */ |
144 | U8 flags; /* should have ANYOF_CLASS here */ |
936ed897 |
145 | U8 type; |
146 | U16 next_off; |
ffc61ed2 |
147 | U32 arg1; |
19103b91 |
148 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
149 | char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ |
936ed897 |
150 | }; |
151 | |
9c7e81e8 |
152 | /* XXX fix this description. |
153 | Impose a limit of REG_INFTY on various pattern matching operations |
154 | to limit stack growth and to avoid "infinite" recursions. |
155 | */ |
156 | /* The default size for REG_INFTY is I16_MAX, which is the same as |
157 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits |
158 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is |
159 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is |
160 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a |
161 | smaller default. |
162 | --Andy Dougherty 11 June 1998 |
163 | */ |
164 | #if SHORTSIZE > 2 |
165 | # ifndef REG_INFTY |
166 | # define REG_INFTY ((1<<15)-1) |
167 | # endif |
168 | #endif |
169 | |
170 | #ifndef REG_INFTY |
83cfe48c |
171 | # define REG_INFTY I16_MAX |
172 | #endif |
a687059c |
173 | |
e91177ed |
174 | #define ARG_VALUE(arg) (arg) |
175 | #define ARG__SET(arg,val) ((arg) = (val)) |
c277df42 |
176 | |
e8e3b0fb |
177 | #undef ARG |
178 | #undef ARG1 |
179 | #undef ARG2 |
180 | |
c277df42 |
181 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
182 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) |
183 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) |
6bda09f9 |
184 | #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) |
786e8c11 |
185 | |
c277df42 |
186 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
187 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) |
188 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) |
6bda09f9 |
189 | #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) |
c277df42 |
190 | |
e8e3b0fb |
191 | #undef NEXT_OFF |
192 | #undef NODE_ALIGN |
193 | |
075abff3 |
194 | #define NEXT_OFF(p) ((p)->next_off) |
195 | #define NODE_ALIGN(node) |
196 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ |
a687059c |
197 | |
c277df42 |
198 | #define SIZE_ALIGN NODE_ALIGN |
199 | |
e8e3b0fb |
200 | #undef OP |
201 | #undef OPERAND |
202 | #undef MASK |
203 | #undef STRING |
204 | |
e91177ed |
205 | #define OP(p) ((p)->type) |
206 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 |
207 | #define MASK(p) ((char*)OPERAND(p)) |
208 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) |
209 | #define STRING(p) (((struct regnode_string *)p)->string) |
210 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) |
211 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) |
212 | |
e8e3b0fb |
213 | #undef NODE_ALIGN |
214 | #undef ARG_LOC |
215 | #undef NEXTOPER |
216 | #undef PREVOPER |
217 | |
e91177ed |
218 | #define NODE_ALIGN(node) |
219 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) |
220 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) |
221 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) |
6bda09f9 |
222 | #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) |
786e8c11 |
223 | |
e91177ed |
224 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
225 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) |
226 | |
227 | #define NODE_STEP_B 4 |
c277df42 |
228 | |
229 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) |
230 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) |
231 | |
e91177ed |
232 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 |
233 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed |
234 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 |
235 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c |
236 | |
22c35a8c |
237 | #define REG_MAGIC 0234 |
a687059c |
238 | |
830247a4 |
239 | #define SIZE_ONLY (RExC_emit == &PL_regdummy) |
c277df42 |
240 | |
936ed897 |
241 | /* Flags for node->flags of ANYOF */ |
b8c5462f |
242 | |
19103b91 |
243 | #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ |
ffc61ed2 |
244 | #define ANYOF_INVERT 0x04 |
245 | #define ANYOF_FOLD 0x02 |
246 | #define ANYOF_LOCALE 0x01 |
b8c5462f |
247 | |
653099ff |
248 | /* Used for regstclass only */ |
ffc61ed2 |
249 | #define ANYOF_EOS 0x10 /* Can match an empty string too */ |
250 | |
251 | /* There is a character or a range past 0xff */ |
252 | #define ANYOF_UNICODE 0x20 |
1aa99e6b |
253 | #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ |
ffc61ed2 |
254 | |
4f66b38d |
255 | /* size of node is large (includes class pointer) */ |
256 | #define ANYOF_LARGE 0x80 |
257 | |
ffc61ed2 |
258 | /* Are there any runtime flags on in this node? */ |
259 | #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) |
260 | |
261 | #define ANYOF_FLAGS_ALL 0xff |
653099ff |
262 | |
936ed897 |
263 | /* Character classes for node->classflags of ANYOF */ |
653099ff |
264 | /* Should be synchronized with a table in regprop() */ |
265 | /* 2n should pair with 2n+1 */ |
b8c5462f |
266 | |
289d4f09 |
267 | #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
b8c5462f |
268 | #define ANYOF_NALNUM 1 |
aaa51d5e |
269 | #define ANYOF_SPACE 2 /* \s */ |
b8c5462f |
270 | #define ANYOF_NSPACE 3 |
271 | #define ANYOF_DIGIT 4 |
272 | #define ANYOF_NDIGIT 5 |
289d4f09 |
273 | #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ |
b8c5462f |
274 | #define ANYOF_NALNUMC 7 |
275 | #define ANYOF_ALPHA 8 |
276 | #define ANYOF_NALPHA 9 |
277 | #define ANYOF_ASCII 10 |
278 | #define ANYOF_NASCII 11 |
279 | #define ANYOF_CNTRL 12 |
280 | #define ANYOF_NCNTRL 13 |
281 | #define ANYOF_GRAPH 14 |
282 | #define ANYOF_NGRAPH 15 |
283 | #define ANYOF_LOWER 16 |
284 | #define ANYOF_NLOWER 17 |
285 | #define ANYOF_PRINT 18 |
286 | #define ANYOF_NPRINT 19 |
287 | #define ANYOF_PUNCT 20 |
288 | #define ANYOF_NPUNCT 21 |
289 | #define ANYOF_UPPER 22 |
290 | #define ANYOF_NUPPER 23 |
291 | #define ANYOF_XDIGIT 24 |
292 | #define ANYOF_NXDIGIT 25 |
aaa51d5e |
293 | #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ |
294 | #define ANYOF_NPSXSPC 27 |
ffc61ed2 |
295 | #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ |
289d4f09 |
296 | #define ANYOF_NBLANK 29 |
b8c5462f |
297 | |
aaa51d5e |
298 | #define ANYOF_MAX 32 |
b8c5462f |
299 | |
300 | /* Backward source code compatibility. */ |
301 | |
302 | #define ANYOF_ALNUML ANYOF_ALNUM |
303 | #define ANYOF_NALNUML ANYOF_NALNUM |
304 | #define ANYOF_SPACEL ANYOF_SPACE |
305 | #define ANYOF_NSPACEL ANYOF_NSPACE |
306 | |
307 | /* Utility macros for the bitmap and classes of ANYOF */ |
308 | |
936ed897 |
309 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
310 | #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) |
b8c5462f |
311 | |
936ed897 |
312 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f |
313 | |
314 | #define ANYOF_BIT(c) (1 << ((c) & 7)) |
315 | |
936ed897 |
316 | #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) |
b8c5462f |
317 | #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) |
318 | #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) |
319 | #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) |
320 | |
936ed897 |
321 | #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) |
322 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
323 | |
324 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
786e8c11 |
325 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f |
326 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
327 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) |
328 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) |
329 | |
f8bef550 |
330 | #define ANYOF_BITMAP_SETALL(p) \ |
331 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) |
332 | #define ANYOF_BITMAP_CLEARALL(p) \ |
333 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) |
334 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ |
335 | #define ANYOF_BITMAP_TESTALLSET(p) \ |
336 | 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) |
337 | |
936ed897 |
338 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
339 | #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) |
340 | #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) |
c277df42 |
341 | |
786e8c11 |
342 | |
a687059c |
343 | /* |
344 | * Utility definitions. |
345 | */ |
2304df62 |
346 | #ifndef CHARMASK |
075abff3 |
347 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c |
348 | #else |
075abff3 |
349 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c |
350 | #endif |
a687059c |
351 | |
c277df42 |
352 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
353 | |
3dab1dad |
354 | #define REG_SEEN_ZERO_LEN 0x00000001 |
355 | #define REG_SEEN_LOOKBEHIND 0x00000002 |
356 | #define REG_SEEN_GPOS 0x00000004 |
357 | #define REG_SEEN_EVAL 0x00000008 |
358 | #define REG_SEEN_CANY 0x00000010 |
cce850e4 |
359 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
6bda09f9 |
360 | #define REG_SEEN_RECURSE 0x00000020 |
eaf3ca90 |
361 | #define REG_TOP_LEVEL_BRANCHES 0x00000040 |
e2e6a0f1 |
362 | #define REG_SEEN_VERBARG 0x00000080 |
5d458dd8 |
363 | #define REG_SEEN_CUTGROUP 0x00000100 |
d09b2d29 |
364 | |
73c4f7a1 |
365 | START_EXTERN_C |
366 | |
be8e71aa |
367 | #ifdef PLUGGABLE_RE_EXTENSION |
368 | #include "re_nodes.h" |
369 | #else |
d09b2d29 |
370 | #include "regnodes.h" |
be8e71aa |
371 | #endif |
d09b2d29 |
372 | |
4d61ec05 |
373 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
d09b2d29 |
374 | #ifndef DOINIT |
4d61ec05 |
375 | EXTCONST U8 PL_varies[]; |
d09b2d29 |
376 | #else |
4d61ec05 |
377 | EXTCONST U8 PL_varies[] = { |
0111c4fd |
378 | BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, |
a0ed51b3 |
379 | WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0 |
c277df42 |
380 | }; |
d09b2d29 |
381 | #endif |
c277df42 |
382 | |
4d61ec05 |
383 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
384 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ |
d09b2d29 |
385 | #ifndef DOINIT |
4d61ec05 |
386 | EXTCONST U8 PL_simple[]; |
d09b2d29 |
387 | #else |
4d61ec05 |
388 | EXTCONST U8 PL_simple[] = { |
f33976b4 |
389 | REG_ANY, SANY, CANY, |
ffc61ed2 |
390 | ANYOF, |
391 | ALNUM, ALNUML, |
392 | NALNUM, NALNUML, |
393 | SPACE, SPACEL, |
394 | NSPACE, NSPACEL, |
395 | DIGIT, NDIGIT, |
396 | 0 |
c277df42 |
397 | }; |
398 | #endif |
399 | |
f9f4320a |
400 | #ifndef PLUGGABLE_RE_EXTENSION |
401 | #ifndef DOINIT |
402 | EXTCONST regexp_engine PL_core_reg_engine; |
403 | #else /* DOINIT */ |
404 | EXTCONST regexp_engine PL_core_reg_engine = { |
405 | Perl_pregcomp, |
406 | Perl_regexec_flags, |
407 | Perl_re_intuit_start, |
408 | Perl_re_intuit_string, |
409 | Perl_pregfree, |
de8c5301 |
410 | Perl_reg_stringify, |
f9f4320a |
411 | #if defined(USE_ITHREADS) |
412 | Perl_regdupe |
413 | #endif |
414 | }; |
415 | #endif /* DOINIT */ |
416 | #endif /* PLUGGABLE_RE_EXTENSION */ |
417 | |
418 | |
73c4f7a1 |
419 | END_EXTERN_C |
cad2e5aa |
420 | |
cad2e5aa |
421 | |
261faec3 |
422 | /* .what is a character array with one character for each member of .data |
423 | * The character describes the function of the corresponding .data item: |
0111c4fd |
424 | * f - start-class data for regstclass optimization |
261faec3 |
425 | * n - Root of op tree for (?{EVAL}) item |
426 | * o - Start op for (?{EVAL}) item |
3dab1dad |
427 | * p - Pad for (?{EVAL}) item |
9e55ce06 |
428 | * s - swash for unicode-style character class, and the multicharacter |
429 | * strings resulting from casefolding the single-character entries |
430 | * in the character class |
a3621e74 |
431 | * t - trie struct |
55eed653 |
432 | * u - trie struct's widecharmap (a HV, so can't share, must dup) |
2b8b4781 |
433 | * also used for revcharmap and words under DEBUGGING |
07be1b83 |
434 | * T - aho-trie struct |
81714fb9 |
435 | * S - sv for named capture lookup |
261faec3 |
436 | * 20010712 mjd@plover.com |
c3613ed7 |
437 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 |
438 | */ |
cad2e5aa |
439 | struct reg_data { |
440 | U32 count; |
441 | U8 *what; |
442 | void* data[1]; |
443 | }; |
444 | |
445 | struct reg_substr_datum { |
446 | I32 min_offset; |
447 | I32 max_offset; |
33b8afdf |
448 | SV *substr; /* non-utf8 variant */ |
449 | SV *utf8_substr; /* utf8 variant */ |
1de06328 |
450 | I32 end_shift; |
cad2e5aa |
451 | }; |
452 | |
453 | struct reg_substr_data { |
454 | struct reg_substr_datum data[3]; /* Actual array */ |
455 | }; |
456 | |
457 | #define anchored_substr substrs->data[0].substr |
33b8afdf |
458 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa |
459 | #define anchored_offset substrs->data[0].min_offset |
1de06328 |
460 | #define anchored_end_shift substrs->data[0].end_shift |
461 | |
cad2e5aa |
462 | #define float_substr substrs->data[1].substr |
33b8afdf |
463 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa |
464 | #define float_min_offset substrs->data[1].min_offset |
465 | #define float_max_offset substrs->data[1].max_offset |
1de06328 |
466 | #define float_end_shift substrs->data[1].end_shift |
467 | |
cad2e5aa |
468 | #define check_substr substrs->data[2].substr |
33b8afdf |
469 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa |
470 | #define check_offset_min substrs->data[2].min_offset |
471 | #define check_offset_max substrs->data[2].max_offset |
1de06328 |
472 | #define check_end_shift substrs->data[2].end_shift |
a3621e74 |
473 | |
474 | |
475 | |
476 | /* trie related stuff */ |
5d9a96ca |
477 | |
a3621e74 |
478 | /* a transition record for the state machine. the |
479 | check field determines which state "owns" the |
480 | transition. the char the transition is for is |
481 | determined by offset from the owning states base |
482 | field. the next field determines which state |
483 | is to be transitioned to if any. |
484 | */ |
485 | struct _reg_trie_trans { |
486 | U32 next; |
487 | U32 check; |
488 | }; |
489 | |
490 | /* a transition list element for the list based representation */ |
491 | struct _reg_trie_trans_list_elem { |
492 | U16 forid; |
493 | U32 newstate; |
494 | }; |
495 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; |
496 | |
497 | /* a state for compressed nodes. base is an offset |
498 | into an array of reg_trie_trans array. If wordnum is |
499 | nonzero the state is accepting. if base is zero then |
500 | the state has no children (and will be accepting) |
501 | */ |
502 | struct _reg_trie_state { |
503 | U16 wordnum; |
504 | union { |
505 | U32 base; |
506 | reg_trie_trans_le* list; |
507 | } trans; |
508 | }; |
509 | |
510 | |
511 | |
a3621e74 |
512 | typedef struct _reg_trie_state reg_trie_state; |
513 | typedef struct _reg_trie_trans reg_trie_trans; |
514 | |
515 | |
516 | /* anything in here that needs to be freed later |
23eab42c |
517 | should be dealt with in pregfree. |
518 | refcount is first in both this and _reg_ac_data to allow a space |
519 | optimisation in Perl_regdupe. */ |
a3621e74 |
520 | struct _reg_trie_data { |
23eab42c |
521 | U32 refcount; /* number of times this trie is referenced */ |
786e8c11 |
522 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
523 | U32 lasttrans; /* last valid transition element */ |
524 | U16 *charmap; /* byte to charid lookup array */ |
786e8c11 |
525 | reg_trie_state *states; /* state data */ |
526 | reg_trie_trans *trans; /* array of transition elements */ |
527 | char *bitmap; /* stclass bitmap */ |
786e8c11 |
528 | U32 startstate; /* initial state - used for common prefix optimisation */ |
529 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ |
530 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ |
531 | U32 *wordlen; /* array of lengths of words */ |
532 | U16 *jump; /* optional 1 indexed array of offsets before tail |
533 | for the node following a given word. */ |
534 | U16 *nextword; /* optional 1 indexed array to support linked list |
535 | of duplicate wordnums */ |
1e2e3d02 |
536 | U32 statecount; /* Build only - number of states in the states array |
537 | (including the unused zero state) */ |
786e8c11 |
538 | U32 wordcount; /* Build only */ |
a3621e74 |
539 | #ifdef DEBUGGING |
786e8c11 |
540 | STRLEN charcount; /* Build only */ |
a3621e74 |
541 | #endif |
542 | }; |
2b8b4781 |
543 | /* There is one (3 under DEBUGGING) pointers that logically belong in this |
544 | structure, but are held outside as they need duplication on thread cloning, |
545 | whereas the rest of the structure can be read only: |
546 | HV *widecharmap; code points > 255 to charid |
547 | #ifdef DEBUGGING |
548 | AV *words; Array of words contained in trie, for dumping |
549 | AV *revcharmap; Map of each charid back to its character representation |
550 | #endif |
551 | */ |
552 | |
553 | #define TRIE_WORDS_OFFSET 2 |
554 | |
a3621e74 |
555 | typedef struct _reg_trie_data reg_trie_data; |
556 | |
23eab42c |
557 | /* refcount is first in both this and _reg_trie_data to allow a space |
558 | optimisation in Perl_regdupe. */ |
07be1b83 |
559 | struct _reg_ac_data { |
23eab42c |
560 | U32 refcount; |
07be1b83 |
561 | U32 *fail; |
562 | reg_trie_state *states; |
3251b653 |
563 | U32 trie; |
07be1b83 |
564 | }; |
565 | typedef struct _reg_ac_data reg_ac_data; |
566 | |
3dab1dad |
567 | /* ANY_BIT doesnt use the structure, so we can borrow it here. |
568 | This is simpler than refactoring all of it as wed end up with |
569 | three different sets... */ |
570 | |
571 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) |
786e8c11 |
572 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f |
573 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
574 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) |
575 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
576 | |
1de06328 |
577 | #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) |
578 | #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) |
579 | |
580 | |
786e8c11 |
581 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
582 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
583 | |
a3621e74 |
584 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
585 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) |
586 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) |
587 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) |
588 | |
3dab1dad |
589 | #ifdef DEBUGGING |
3dab1dad |
590 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad |
591 | #else |
3dab1dad |
592 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad |
593 | #endif |
a3621e74 |
594 | |
0111c4fd |
595 | #define RE_TRIE_MAXBUF_INIT 65536 |
596 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" |
a3621e74 |
597 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
598 | |
be8e71aa |
599 | /* |
600 | |
601 | RE_DEBUG_FLAGS is used to control what debug output is emitted |
602 | its divided into three groups of options, some of which interact. |
603 | The three groups are: Compile, Execute, Extra. There is room for a |
604 | further group, as currently only the low three bytes are used. |
605 | |
606 | Compile Options: |
07be1b83 |
607 | |
be8e71aa |
608 | PARSE |
609 | PEEP |
610 | TRIE |
611 | PROGRAM |
612 | OFFSETS |
a3621e74 |
613 | |
be8e71aa |
614 | Execute Options: |
a3621e74 |
615 | |
be8e71aa |
616 | INTUIT |
617 | MATCH |
618 | TRIE |
a3621e74 |
619 | |
be8e71aa |
620 | Extra Options |
a3621e74 |
621 | |
be8e71aa |
622 | TRIE |
623 | OFFSETS |
624 | |
625 | If you modify any of these make sure you make corresponding changes to |
626 | re.pm, especially to the documentation. |
a3621e74 |
627 | |
be8e71aa |
628 | */ |
629 | |
630 | |
631 | /* Compile */ |
632 | #define RE_DEBUG_COMPILE_MASK 0x0000FF |
633 | #define RE_DEBUG_COMPILE_PARSE 0x000001 |
634 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 |
635 | #define RE_DEBUG_COMPILE_TRIE 0x000004 |
636 | #define RE_DEBUG_COMPILE_DUMP 0x000008 |
be8e71aa |
637 | |
638 | /* Execute */ |
639 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 |
640 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 |
641 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 |
642 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 |
643 | |
644 | /* Extra */ |
645 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 |
646 | #define RE_DEBUG_EXTRA_TRIE 0x010000 |
647 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 |
a5ca303d |
648 | #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 |
649 | #define RE_DEBUG_EXTRA_STATE 0x080000 |
650 | #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 |
24b23f37 |
651 | /* combined */ |
652 | #define RE_DEBUG_EXTRA_STACK 0x280000 |
be8e71aa |
653 | |
e68ec53f |
654 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa |
655 | /* Compile */ |
656 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
657 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa |
658 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
659 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
660 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f |
661 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa |
662 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
663 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
664 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f |
665 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa |
666 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
667 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
be8e71aa |
668 | |
669 | /* Execute */ |
670 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
671 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa |
672 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f |
673 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa |
674 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f |
675 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa |
676 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
677 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa |
678 | |
679 | /* Extra */ |
680 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ |
e68ec53f |
681 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
a5ca303d |
682 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
683 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) |
ab3bbdeb |
684 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
685 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) |
24b23f37 |
686 | #define DEBUG_STACK_r(x) DEBUG_r( \ |
687 | if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x ) |
a5ca303d |
688 | #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ |
689 | if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ |
690 | (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) |
07be1b83 |
691 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
a5ca303d |
692 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ |
be8e71aa |
693 | Perl_warn_nocontext x ) |
694 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ |
e68ec53f |
695 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
696 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f |
697 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
698 | |
699 | #define DEBUG_TRIE_r(x) DEBUG_r( \ |
e68ec53f |
700 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa |
701 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
702 | |
703 | /* initialization */ |
704 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f |
705 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
706 | SV * re_debug_flags_sv = NULL; \ |
707 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ |
708 | if (re_debug_flags_sv) { \ |
8e9a8a48 |
709 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f |
710 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 |
711 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f |
712 | }\ |
713 | }) |
a3621e74 |
714 | |
715 | #ifdef DEBUGGING |
ab3bbdeb |
716 | |
8e9a8a48 |
717 | #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; |
ab3bbdeb |
718 | |
719 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ |
720 | const char * const rpv = \ |
ddc5bc0f |
721 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
722 | PL_colors[(c1)],PL_colors[(c2)], \ |
723 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ |
0df25f3d |
724 | const int rlen = SvCUR(dsv) |
ab3bbdeb |
725 | |
726 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ |
727 | const char * const rpv = \ |
ddc5bc0f |
728 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb |
729 | PL_colors[(c1)],PL_colors[(c2)], \ |
730 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) |
731 | |
732 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ |
733 | const char * const rpv = \ |
ddc5bc0f |
734 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
735 | PL_colors[0], PL_colors[1], \ |
736 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \ |
737 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ |
738 | ) |
739 | |
740 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) |
741 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") |
742 | |
743 | #else /* if not DEBUGGING */ |
744 | |
a3621e74 |
745 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb |
746 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
747 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) |
748 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) |
749 | #define RE_SV_DUMPLEN(ItEm) |
750 | #define RE_SV_TAIL(ItEm) |
751 | |
752 | #endif /* DEBUG RELATED DEFINES */ |
a3621e74 |
753 | |