Commit | Line | Data |
a0d0e21e |
1 | /* regcomp.h |
d6376244 |
2 | * |
4bb101f2 |
3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
699a97de |
4 | * 2000, 2001, 2002, 2003, 2005, 2006 by Larry Wall and others |
d6376244 |
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? */ |
68ba3a3f |
21 | #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0 |
58e23c8d |
22 | |
23 | /* Not for production use: */ |
07be1b83 |
24 | #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 |
58e23c8d |
25 | |
7122b237 |
26 | /* Activate offsets code - set to if 1 to enable */ |
27 | #ifdef DEBUGGING |
28 | #define RE_TRACK_PATTERN_OFFSETS |
29 | #endif |
30 | |
b9b4dddf |
31 | /* Unless the next line is uncommented it is illegal to combine lazy |
32 | matching with possessive matching. Frankly it doesn't make much sense |
33 | to allow it as X*?+ matches nothing, X+?+ matches a single char only, |
34 | and X{min,max}?+ matches min times only. |
35 | */ |
36 | /* #define REG_ALLOW_MINMOD_SUSPEND */ |
07be1b83 |
37 | |
a687059c |
38 | /* |
39 | * The "internal use only" fields in regexp.h are present to pass info from |
40 | * compile to execute that permits the execute phase to run lots faster on |
41 | * simple cases. They are: |
42 | * |
bd61b366 |
43 | * regstart sv that must begin a match; NULL if none obvious |
a687059c |
44 | * reganch is the match anchored (at beginning-of-line only)? |
45 | * regmust string (pointer into program) that match must include, or NULL |
79072805 |
46 | * [regmust changed to SV* for bminstr()--law] |
a687059c |
47 | * regmlen length of regmust string |
48 | * [regmlen not used currently] |
49 | * |
50 | * Regstart and reganch permit very fast decisions on suitable starting points |
51 | * for a match, cutting down the work a lot. Regmust permits fast rejection |
52 | * of lines that cannot possibly match. The regmust tests are costly enough |
e50aee73 |
53 | * that pregcomp() supplies a regmust only if the r.e. contains something |
a687059c |
54 | * potentially expensive (at present, the only such thing detected is * or + |
55 | * at the start of the r.e., which can involve a lot of backup). Regmlen is |
e50aee73 |
56 | * supplied because the test in pregexec() needs it and pregcomp() is computing |
a687059c |
57 | * it anyway. |
58 | * [regmust is now supplied always. The tests that use regmust have a |
59 | * heuristic that disables the test if it usually matches.] |
60 | * |
61 | * [In fact, we now use regmust in many cases to locate where the search |
62 | * starts in the string, so if regback is >= 0, the regmust search is never |
63 | * wasted effort. The regback variable says how many characters back from |
64 | * where regmust matched is the earliest possible start of the match. |
65 | * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] |
66 | */ |
67 | |
68 | /* |
69 | * Structure for regexp "program". This is essentially a linear encoding |
70 | * of a nondeterministic finite-state machine (aka syntax charts or |
71 | * "railroad normal form" in parsing technology). Each node is an opcode |
72 | * plus a "next" pointer, possibly plus an operand. "Next" pointers of |
73 | * all nodes except BRANCH implement concatenation; a "next" pointer with |
74 | * a BRANCH on both ends of it is connecting two alternatives. (Here we |
75 | * have one of the subtle syntax dependencies: an individual BRANCH (as |
76 | * opposed to a collection of them) is never concatenated with anything |
77 | * because of operator precedence.) The operand of some types of node is |
78 | * a literal string; for others, it is a node leading into a sub-FSM. In |
79 | * particular, the operand of a BRANCH node is the first node of the branch. |
80 | * (NB this is *not* a tree structure: the tail of the branch connects |
a3621e74 |
81 | * to the thing following the set of BRANCHes.) The opcodes are defined |
82 | * in regnodes.h which is generated from regcomp.sym by regcomp.pl. |
a687059c |
83 | */ |
84 | |
a687059c |
85 | /* |
86 | * A node is one char of opcode followed by two chars of "next" pointer. |
87 | * "Next" pointers are stored as two 8-bit pieces, high order first. The |
88 | * value is a positive offset from the opcode of the node containing it. |
89 | * An operand, if any, simply follows the node. (Note that much of the |
90 | * code generation knows about this implicit relationship.) |
91 | * |
92 | * Using two bytes for the "next" pointer is vast overkill for most things, |
93 | * but allows patterns to get big without disasters. |
94 | * |
e91177ed |
95 | * [The "next" pointer is always aligned on an even |
a687059c |
96 | * boundary, and reads the offset directly as a short. Also, there is no |
97 | * special test to reverse the sign of BACK pointers since the offset is |
98 | * stored negative.] |
99 | */ |
785a26d5 |
100 | |
101 | /* This is the stuff that used to live in regexp.h that was truly |
102 | private to the engine itself. It now lives here. */ |
103 | |
28d8d7f4 |
104 | |
785a26d5 |
105 | |
106 | typedef struct regexp_internal { |
1f1031fe |
107 | int name_list_idx; /* Optional data index of an array of paren names */ |
7122b237 |
108 | union { |
109 | U32 *offsets; /* offset annotations 20001228 MJD |
110 | data about mapping the program to the |
111 | string - |
112 | offsets[0] is proglen when this is used |
113 | */ |
114 | U32 proglen; |
115 | } u; |
116 | |
6d5c990f |
117 | regnode *regstclass; /* Optional startclass as identified or constructed |
118 | by the optimiser */ |
119 | struct reg_data *data; /* Additional miscellaneous data used by the program. |
120 | Used to make it easier to clone and free arbitrary |
121 | data that the regops need. Often the ARG field of |
122 | a regop is an index into this structure */ |
123 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
124 | } regexp_internal; |
125 | |
126 | #define RXi_SET(x,y) (x)->pprivate = (void*)(y) |
127 | #define RXi_GET(x) ((regexp_internal *)((x)->pprivate)) |
128 | #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r) |
785a26d5 |
129 | /* |
130 | * Flags stored in regexp->intflags |
131 | * These are used only internally to the regexp engine |
132 | * |
133 | * See regexp.h for flags used externally to the regexp engine |
134 | */ |
135 | #define PREGf_SKIP 0x00000001 |
136 | #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */ |
137 | #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */ |
138 | #define PREGf_VERBARG_SEEN 0x00000008 |
139 | #define PREGf_CUTGROUP_SEEN 0x00000010 |
140 | |
141 | |
142 | /* this is where the old regcomp.h started */ |
a687059c |
143 | |
c277df42 |
144 | struct regnode_string { |
cd439c50 |
145 | U8 str_len; |
c277df42 |
146 | U8 type; |
147 | U16 next_off; |
cd439c50 |
148 | char string[1]; |
c277df42 |
149 | }; |
a687059c |
150 | |
6bda09f9 |
151 | /* Argument bearing node - workhorse, |
152 | arg1 is often for the data field */ |
c277df42 |
153 | struct regnode_1 { |
154 | U8 flags; |
155 | U8 type; |
156 | U16 next_off; |
157 | U32 arg1; |
158 | }; |
159 | |
6bda09f9 |
160 | /* Similar to a regnode_1 but with an extra signed argument */ |
161 | struct regnode_2L { |
162 | U8 flags; |
163 | U8 type; |
164 | U16 next_off; |
165 | U32 arg1; |
166 | I32 arg2; |
167 | }; |
168 | |
169 | /* 'Two field' -- Two 16 bit unsigned args */ |
c277df42 |
170 | struct regnode_2 { |
171 | U8 flags; |
172 | U8 type; |
173 | U16 next_off; |
174 | U16 arg1; |
175 | U16 arg2; |
176 | }; |
177 | |
6bda09f9 |
178 | |
936ed897 |
179 | #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ |
ffc61ed2 |
180 | #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ |
936ed897 |
181 | |
786e8c11 |
182 | /* also used by trie */ |
936ed897 |
183 | struct regnode_charclass { |
184 | U8 flags; |
185 | U8 type; |
186 | U16 next_off; |
ffc61ed2 |
187 | U32 arg1; |
19103b91 |
188 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 |
189 | }; |
190 | |
19103b91 |
191 | struct regnode_charclass_class { /* has [[:blah:]] classes */ |
192 | U8 flags; /* should have ANYOF_CLASS here */ |
936ed897 |
193 | U8 type; |
194 | U16 next_off; |
ffc61ed2 |
195 | U32 arg1; |
19103b91 |
196 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
197 | char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ |
936ed897 |
198 | }; |
199 | |
9c7e81e8 |
200 | /* XXX fix this description. |
201 | Impose a limit of REG_INFTY on various pattern matching operations |
202 | to limit stack growth and to avoid "infinite" recursions. |
203 | */ |
204 | /* The default size for REG_INFTY is I16_MAX, which is the same as |
205 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits |
206 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is |
207 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is |
208 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a |
209 | smaller default. |
210 | --Andy Dougherty 11 June 1998 |
211 | */ |
212 | #if SHORTSIZE > 2 |
213 | # ifndef REG_INFTY |
214 | # define REG_INFTY ((1<<15)-1) |
215 | # endif |
216 | #endif |
217 | |
218 | #ifndef REG_INFTY |
83cfe48c |
219 | # define REG_INFTY I16_MAX |
220 | #endif |
a687059c |
221 | |
e91177ed |
222 | #define ARG_VALUE(arg) (arg) |
223 | #define ARG__SET(arg,val) ((arg) = (val)) |
c277df42 |
224 | |
e8e3b0fb |
225 | #undef ARG |
226 | #undef ARG1 |
227 | #undef ARG2 |
228 | |
c277df42 |
229 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
230 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) |
231 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) |
6bda09f9 |
232 | #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) |
786e8c11 |
233 | |
c277df42 |
234 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
235 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) |
236 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) |
6bda09f9 |
237 | #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) |
c277df42 |
238 | |
e8e3b0fb |
239 | #undef NEXT_OFF |
240 | #undef NODE_ALIGN |
241 | |
075abff3 |
242 | #define NEXT_OFF(p) ((p)->next_off) |
243 | #define NODE_ALIGN(node) |
244 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ |
a687059c |
245 | |
c277df42 |
246 | #define SIZE_ALIGN NODE_ALIGN |
247 | |
e8e3b0fb |
248 | #undef OP |
249 | #undef OPERAND |
250 | #undef MASK |
251 | #undef STRING |
252 | |
e91177ed |
253 | #define OP(p) ((p)->type) |
254 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 |
255 | #define MASK(p) ((char*)OPERAND(p)) |
256 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) |
257 | #define STRING(p) (((struct regnode_string *)p)->string) |
258 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) |
259 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) |
260 | |
e8e3b0fb |
261 | #undef NODE_ALIGN |
262 | #undef ARG_LOC |
263 | #undef NEXTOPER |
264 | #undef PREVOPER |
265 | |
e91177ed |
266 | #define NODE_ALIGN(node) |
267 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) |
268 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) |
269 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) |
6bda09f9 |
270 | #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) |
786e8c11 |
271 | |
e91177ed |
272 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
273 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) |
274 | |
275 | #define NODE_STEP_B 4 |
c277df42 |
276 | |
277 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) |
278 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) |
279 | |
e91177ed |
280 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 |
281 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed |
282 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 |
283 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c |
284 | |
22c35a8c |
285 | #define REG_MAGIC 0234 |
a687059c |
286 | |
830247a4 |
287 | #define SIZE_ONLY (RExC_emit == &PL_regdummy) |
c277df42 |
288 | |
936ed897 |
289 | /* Flags for node->flags of ANYOF */ |
b8c5462f |
290 | |
19103b91 |
291 | #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ |
ffc61ed2 |
292 | #define ANYOF_INVERT 0x04 |
293 | #define ANYOF_FOLD 0x02 |
294 | #define ANYOF_LOCALE 0x01 |
b8c5462f |
295 | |
653099ff |
296 | /* Used for regstclass only */ |
ffc61ed2 |
297 | #define ANYOF_EOS 0x10 /* Can match an empty string too */ |
298 | |
299 | /* There is a character or a range past 0xff */ |
300 | #define ANYOF_UNICODE 0x20 |
1aa99e6b |
301 | #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ |
ffc61ed2 |
302 | |
4f66b38d |
303 | /* size of node is large (includes class pointer) */ |
304 | #define ANYOF_LARGE 0x80 |
305 | |
ffc61ed2 |
306 | /* Are there any runtime flags on in this node? */ |
307 | #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) |
308 | |
309 | #define ANYOF_FLAGS_ALL 0xff |
653099ff |
310 | |
936ed897 |
311 | /* Character classes for node->classflags of ANYOF */ |
653099ff |
312 | /* Should be synchronized with a table in regprop() */ |
313 | /* 2n should pair with 2n+1 */ |
b8c5462f |
314 | |
289d4f09 |
315 | #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
b8c5462f |
316 | #define ANYOF_NALNUM 1 |
aaa51d5e |
317 | #define ANYOF_SPACE 2 /* \s */ |
b8c5462f |
318 | #define ANYOF_NSPACE 3 |
319 | #define ANYOF_DIGIT 4 |
320 | #define ANYOF_NDIGIT 5 |
289d4f09 |
321 | #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ |
b8c5462f |
322 | #define ANYOF_NALNUMC 7 |
323 | #define ANYOF_ALPHA 8 |
324 | #define ANYOF_NALPHA 9 |
325 | #define ANYOF_ASCII 10 |
326 | #define ANYOF_NASCII 11 |
327 | #define ANYOF_CNTRL 12 |
328 | #define ANYOF_NCNTRL 13 |
329 | #define ANYOF_GRAPH 14 |
330 | #define ANYOF_NGRAPH 15 |
331 | #define ANYOF_LOWER 16 |
332 | #define ANYOF_NLOWER 17 |
333 | #define ANYOF_PRINT 18 |
334 | #define ANYOF_NPRINT 19 |
335 | #define ANYOF_PUNCT 20 |
336 | #define ANYOF_NPUNCT 21 |
337 | #define ANYOF_UPPER 22 |
338 | #define ANYOF_NUPPER 23 |
339 | #define ANYOF_XDIGIT 24 |
340 | #define ANYOF_NXDIGIT 25 |
aaa51d5e |
341 | #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ |
342 | #define ANYOF_NPSXSPC 27 |
ffc61ed2 |
343 | #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ |
289d4f09 |
344 | #define ANYOF_NBLANK 29 |
b8c5462f |
345 | |
aaa51d5e |
346 | #define ANYOF_MAX 32 |
b8c5462f |
347 | |
348 | /* Backward source code compatibility. */ |
349 | |
350 | #define ANYOF_ALNUML ANYOF_ALNUM |
351 | #define ANYOF_NALNUML ANYOF_NALNUM |
352 | #define ANYOF_SPACEL ANYOF_SPACE |
353 | #define ANYOF_NSPACEL ANYOF_NSPACE |
354 | |
355 | /* Utility macros for the bitmap and classes of ANYOF */ |
356 | |
936ed897 |
357 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
358 | #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) |
b8c5462f |
359 | |
936ed897 |
360 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f |
361 | |
362 | #define ANYOF_BIT(c) (1 << ((c) & 7)) |
363 | |
936ed897 |
364 | #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) |
b8c5462f |
365 | #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) |
366 | #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) |
367 | #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) |
368 | |
936ed897 |
369 | #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) |
370 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
371 | |
372 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
786e8c11 |
373 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f |
374 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
375 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) |
376 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) |
377 | |
f8bef550 |
378 | #define ANYOF_BITMAP_SETALL(p) \ |
379 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) |
380 | #define ANYOF_BITMAP_CLEARALL(p) \ |
381 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) |
382 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ |
383 | #define ANYOF_BITMAP_TESTALLSET(p) \ |
384 | 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) |
385 | |
936ed897 |
386 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
387 | #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) |
388 | #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) |
c277df42 |
389 | |
786e8c11 |
390 | |
a687059c |
391 | /* |
392 | * Utility definitions. |
393 | */ |
2304df62 |
394 | #ifndef CHARMASK |
075abff3 |
395 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c |
396 | #else |
075abff3 |
397 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c |
398 | #endif |
a687059c |
399 | |
c277df42 |
400 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
401 | |
3dab1dad |
402 | #define REG_SEEN_ZERO_LEN 0x00000001 |
403 | #define REG_SEEN_LOOKBEHIND 0x00000002 |
404 | #define REG_SEEN_GPOS 0x00000004 |
405 | #define REG_SEEN_EVAL 0x00000008 |
406 | #define REG_SEEN_CANY 0x00000010 |
cce850e4 |
407 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
6bda09f9 |
408 | #define REG_SEEN_RECURSE 0x00000020 |
eaf3ca90 |
409 | #define REG_TOP_LEVEL_BRANCHES 0x00000040 |
e2e6a0f1 |
410 | #define REG_SEEN_VERBARG 0x00000080 |
5d458dd8 |
411 | #define REG_SEEN_CUTGROUP 0x00000100 |
bcdf7404 |
412 | #define REG_SEEN_RUN_ON_COMMENT 0x00000200 |
d09b2d29 |
413 | |
73c4f7a1 |
414 | START_EXTERN_C |
415 | |
be8e71aa |
416 | #ifdef PLUGGABLE_RE_EXTENSION |
417 | #include "re_nodes.h" |
418 | #else |
d09b2d29 |
419 | #include "regnodes.h" |
be8e71aa |
420 | #endif |
d09b2d29 |
421 | |
4d61ec05 |
422 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
d09b2d29 |
423 | #ifndef DOINIT |
4d61ec05 |
424 | EXTCONST U8 PL_varies[]; |
d09b2d29 |
425 | #else |
4d61ec05 |
426 | EXTCONST U8 PL_varies[] = { |
0111c4fd |
427 | BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, |
99ed40ff |
428 | WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, |
429 | NREF, NREFF, NREFFL, |
430 | 0 |
c277df42 |
431 | }; |
d09b2d29 |
432 | #endif |
c277df42 |
433 | |
4d61ec05 |
434 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
435 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ |
d09b2d29 |
436 | #ifndef DOINIT |
4d61ec05 |
437 | EXTCONST U8 PL_simple[]; |
d09b2d29 |
438 | #else |
4d61ec05 |
439 | EXTCONST U8 PL_simple[] = { |
f33976b4 |
440 | REG_ANY, SANY, CANY, |
ffc61ed2 |
441 | ANYOF, |
442 | ALNUM, ALNUML, |
443 | NALNUM, NALNUML, |
444 | SPACE, SPACEL, |
445 | NSPACE, NSPACEL, |
446 | DIGIT, NDIGIT, |
447 | 0 |
c277df42 |
448 | }; |
449 | #endif |
450 | |
f9f4320a |
451 | #ifndef PLUGGABLE_RE_EXTENSION |
452 | #ifndef DOINIT |
453 | EXTCONST regexp_engine PL_core_reg_engine; |
454 | #else /* DOINIT */ |
455 | EXTCONST regexp_engine PL_core_reg_engine = { |
3c6f95d1 |
456 | Perl_re_compile, |
f9f4320a |
457 | Perl_regexec_flags, |
458 | Perl_re_intuit_start, |
459 | Perl_re_intuit_string, |
f8149455 |
460 | Perl_regfree_internal, |
93b32b6d |
461 | Perl_reg_numbered_buff_get, |
462 | Perl_reg_named_buff_get, |
f9f4320a |
463 | #if defined(USE_ITHREADS) |
f8149455 |
464 | Perl_regdupe_internal |
f9f4320a |
465 | #endif |
466 | }; |
467 | #endif /* DOINIT */ |
468 | #endif /* PLUGGABLE_RE_EXTENSION */ |
469 | |
470 | |
73c4f7a1 |
471 | END_EXTERN_C |
cad2e5aa |
472 | |
cad2e5aa |
473 | |
261faec3 |
474 | /* .what is a character array with one character for each member of .data |
475 | * The character describes the function of the corresponding .data item: |
0111c4fd |
476 | * f - start-class data for regstclass optimization |
261faec3 |
477 | * n - Root of op tree for (?{EVAL}) item |
478 | * o - Start op for (?{EVAL}) item |
3dab1dad |
479 | * p - Pad for (?{EVAL}) item |
9e55ce06 |
480 | * s - swash for unicode-style character class, and the multicharacter |
481 | * strings resulting from casefolding the single-character entries |
482 | * in the character class |
a3621e74 |
483 | * t - trie struct |
55eed653 |
484 | * u - trie struct's widecharmap (a HV, so can't share, must dup) |
2b8b4781 |
485 | * also used for revcharmap and words under DEBUGGING |
07be1b83 |
486 | * T - aho-trie struct |
81714fb9 |
487 | * S - sv for named capture lookup |
261faec3 |
488 | * 20010712 mjd@plover.com |
c3613ed7 |
489 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 |
490 | */ |
cad2e5aa |
491 | struct reg_data { |
492 | U32 count; |
493 | U8 *what; |
494 | void* data[1]; |
495 | }; |
496 | |
a1cac82e |
497 | /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses |
498 | anchored* and float* via array indexes 0 and 1. */ |
cad2e5aa |
499 | #define anchored_substr substrs->data[0].substr |
33b8afdf |
500 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa |
501 | #define anchored_offset substrs->data[0].min_offset |
1de06328 |
502 | #define anchored_end_shift substrs->data[0].end_shift |
503 | |
cad2e5aa |
504 | #define float_substr substrs->data[1].substr |
33b8afdf |
505 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa |
506 | #define float_min_offset substrs->data[1].min_offset |
507 | #define float_max_offset substrs->data[1].max_offset |
1de06328 |
508 | #define float_end_shift substrs->data[1].end_shift |
509 | |
cad2e5aa |
510 | #define check_substr substrs->data[2].substr |
33b8afdf |
511 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa |
512 | #define check_offset_min substrs->data[2].min_offset |
513 | #define check_offset_max substrs->data[2].max_offset |
1de06328 |
514 | #define check_end_shift substrs->data[2].end_shift |
a3621e74 |
515 | |
516 | |
517 | |
518 | /* trie related stuff */ |
5d9a96ca |
519 | |
a3621e74 |
520 | /* a transition record for the state machine. the |
521 | check field determines which state "owns" the |
522 | transition. the char the transition is for is |
523 | determined by offset from the owning states base |
524 | field. the next field determines which state |
525 | is to be transitioned to if any. |
526 | */ |
527 | struct _reg_trie_trans { |
528 | U32 next; |
529 | U32 check; |
530 | }; |
531 | |
532 | /* a transition list element for the list based representation */ |
533 | struct _reg_trie_trans_list_elem { |
534 | U16 forid; |
535 | U32 newstate; |
536 | }; |
537 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; |
538 | |
539 | /* a state for compressed nodes. base is an offset |
540 | into an array of reg_trie_trans array. If wordnum is |
541 | nonzero the state is accepting. if base is zero then |
542 | the state has no children (and will be accepting) |
543 | */ |
544 | struct _reg_trie_state { |
545 | U16 wordnum; |
546 | union { |
547 | U32 base; |
548 | reg_trie_trans_le* list; |
549 | } trans; |
550 | }; |
551 | |
552 | |
553 | |
a3621e74 |
554 | typedef struct _reg_trie_state reg_trie_state; |
555 | typedef struct _reg_trie_trans reg_trie_trans; |
556 | |
557 | |
558 | /* anything in here that needs to be freed later |
23eab42c |
559 | should be dealt with in pregfree. |
560 | refcount is first in both this and _reg_ac_data to allow a space |
561 | optimisation in Perl_regdupe. */ |
a3621e74 |
562 | struct _reg_trie_data { |
23eab42c |
563 | U32 refcount; /* number of times this trie is referenced */ |
786e8c11 |
564 | U32 lasttrans; /* last valid transition element */ |
565 | U16 *charmap; /* byte to charid lookup array */ |
786e8c11 |
566 | reg_trie_state *states; /* state data */ |
567 | reg_trie_trans *trans; /* array of transition elements */ |
568 | char *bitmap; /* stclass bitmap */ |
786e8c11 |
569 | U32 *wordlen; /* array of lengths of words */ |
570 | U16 *jump; /* optional 1 indexed array of offsets before tail |
571 | for the node following a given word. */ |
572 | U16 *nextword; /* optional 1 indexed array to support linked list |
573 | of duplicate wordnums */ |
1b5c067c |
574 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
575 | U32 startstate; /* initial state - used for common prefix optimisation */ |
576 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ |
577 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ |
1e2e3d02 |
578 | U32 statecount; /* Build only - number of states in the states array |
579 | (including the unused zero state) */ |
786e8c11 |
580 | U32 wordcount; /* Build only */ |
a3621e74 |
581 | #ifdef DEBUGGING |
786e8c11 |
582 | STRLEN charcount; /* Build only */ |
a3621e74 |
583 | #endif |
584 | }; |
2b8b4781 |
585 | /* There is one (3 under DEBUGGING) pointers that logically belong in this |
586 | structure, but are held outside as they need duplication on thread cloning, |
587 | whereas the rest of the structure can be read only: |
588 | HV *widecharmap; code points > 255 to charid |
589 | #ifdef DEBUGGING |
590 | AV *words; Array of words contained in trie, for dumping |
591 | AV *revcharmap; Map of each charid back to its character representation |
592 | #endif |
593 | */ |
594 | |
595 | #define TRIE_WORDS_OFFSET 2 |
596 | |
a3621e74 |
597 | typedef struct _reg_trie_data reg_trie_data; |
598 | |
23eab42c |
599 | /* refcount is first in both this and _reg_trie_data to allow a space |
600 | optimisation in Perl_regdupe. */ |
07be1b83 |
601 | struct _reg_ac_data { |
23eab42c |
602 | U32 refcount; |
1b5c067c |
603 | U32 trie; |
07be1b83 |
604 | U32 *fail; |
605 | reg_trie_state *states; |
07be1b83 |
606 | }; |
607 | typedef struct _reg_ac_data reg_ac_data; |
608 | |
3dab1dad |
609 | /* ANY_BIT doesnt use the structure, so we can borrow it here. |
610 | This is simpler than refactoring all of it as wed end up with |
611 | three different sets... */ |
612 | |
613 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) |
786e8c11 |
614 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f |
615 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
616 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) |
617 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
618 | |
1de06328 |
619 | #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) |
620 | #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) |
621 | |
622 | |
786e8c11 |
623 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
624 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) |
3dab1dad |
625 | |
a3621e74 |
626 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
627 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) |
628 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) |
629 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) |
630 | |
3dab1dad |
631 | #ifdef DEBUGGING |
3dab1dad |
632 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad |
633 | #else |
3dab1dad |
634 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad |
635 | #endif |
a3621e74 |
636 | |
0111c4fd |
637 | #define RE_TRIE_MAXBUF_INIT 65536 |
638 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" |
a3621e74 |
639 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
640 | |
be8e71aa |
641 | /* |
642 | |
643 | RE_DEBUG_FLAGS is used to control what debug output is emitted |
644 | its divided into three groups of options, some of which interact. |
645 | The three groups are: Compile, Execute, Extra. There is room for a |
646 | further group, as currently only the low three bytes are used. |
647 | |
648 | Compile Options: |
07be1b83 |
649 | |
be8e71aa |
650 | PARSE |
651 | PEEP |
652 | TRIE |
653 | PROGRAM |
654 | OFFSETS |
a3621e74 |
655 | |
be8e71aa |
656 | Execute Options: |
a3621e74 |
657 | |
be8e71aa |
658 | INTUIT |
659 | MATCH |
660 | TRIE |
a3621e74 |
661 | |
be8e71aa |
662 | Extra Options |
a3621e74 |
663 | |
be8e71aa |
664 | TRIE |
665 | OFFSETS |
666 | |
667 | If you modify any of these make sure you make corresponding changes to |
668 | re.pm, especially to the documentation. |
a3621e74 |
669 | |
be8e71aa |
670 | */ |
671 | |
672 | |
673 | /* Compile */ |
674 | #define RE_DEBUG_COMPILE_MASK 0x0000FF |
675 | #define RE_DEBUG_COMPILE_PARSE 0x000001 |
676 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 |
677 | #define RE_DEBUG_COMPILE_TRIE 0x000004 |
678 | #define RE_DEBUG_COMPILE_DUMP 0x000008 |
be8e71aa |
679 | |
680 | /* Execute */ |
681 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 |
682 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 |
683 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 |
684 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 |
685 | |
686 | /* Extra */ |
687 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 |
688 | #define RE_DEBUG_EXTRA_TRIE 0x010000 |
689 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 |
a5ca303d |
690 | #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 |
691 | #define RE_DEBUG_EXTRA_STATE 0x080000 |
692 | #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 |
e7707071 |
693 | #define RE_DEBUG_EXTRA_BUFFERS 0x400000 |
24b23f37 |
694 | /* combined */ |
695 | #define RE_DEBUG_EXTRA_STACK 0x280000 |
be8e71aa |
696 | |
e68ec53f |
697 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa |
698 | /* Compile */ |
699 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
700 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa |
701 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
702 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
703 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f |
704 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa |
705 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f |
706 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa |
707 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f |
708 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa |
709 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f |
710 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
be8e71aa |
711 | |
712 | /* Execute */ |
713 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
714 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa |
715 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f |
716 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa |
717 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f |
718 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa |
719 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f |
720 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa |
721 | |
722 | /* Extra */ |
723 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ |
e68ec53f |
724 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
a5ca303d |
725 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
726 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) |
ab3bbdeb |
727 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
728 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) |
24b23f37 |
729 | #define DEBUG_STACK_r(x) DEBUG_r( \ |
730 | if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x ) |
e7707071 |
731 | #define DEBUG_BUFFERS_r(x) DEBUG_r( \ |
732 | if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x ) |
733 | |
a5ca303d |
734 | #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ |
735 | if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ |
736 | (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) |
07be1b83 |
737 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
a5ca303d |
738 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ |
be8e71aa |
739 | Perl_warn_nocontext x ) |
740 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ |
e68ec53f |
741 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
742 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f |
743 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa |
744 | |
745 | #define DEBUG_TRIE_r(x) DEBUG_r( \ |
e68ec53f |
746 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa |
747 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
748 | |
749 | /* initialization */ |
750 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f |
751 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
752 | SV * re_debug_flags_sv = NULL; \ |
753 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ |
754 | if (re_debug_flags_sv) { \ |
8e9a8a48 |
755 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f |
756 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 |
757 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f |
758 | }\ |
759 | }) |
a3621e74 |
760 | |
761 | #ifdef DEBUGGING |
ab3bbdeb |
762 | |
8e9a8a48 |
763 | #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; |
ab3bbdeb |
764 | |
765 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ |
766 | const char * const rpv = \ |
ddc5bc0f |
767 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
768 | PL_colors[(c1)],PL_colors[(c2)], \ |
0a0e9afc |
769 | PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ |
0df25f3d |
770 | const int rlen = SvCUR(dsv) |
ab3bbdeb |
771 | |
772 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ |
773 | const char * const rpv = \ |
ddc5bc0f |
774 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb |
775 | PL_colors[(c1)],PL_colors[(c2)], \ |
0a0e9afc |
776 | PERL_PV_ESCAPE_RE |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) |
ab3bbdeb |
777 | |
778 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ |
779 | const char * const rpv = \ |
ddc5bc0f |
780 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb |
781 | PL_colors[0], PL_colors[1], \ |
0a0e9afc |
782 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_PRETTY_ELIPSES | \ |
ab3bbdeb |
783 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ |
0a0e9afc |
784 | ) |
ab3bbdeb |
785 | |
786 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) |
787 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") |
788 | |
789 | #else /* if not DEBUGGING */ |
790 | |
a3621e74 |
791 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb |
792 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
793 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) |
794 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) |
795 | #define RE_SV_DUMPLEN(ItEm) |
796 | #define RE_SV_TAIL(ItEm) |
797 | |
798 | #endif /* DEBUG RELATED DEFINES */ |
a3621e74 |
799 | |
99ed40ff |
800 | |