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 JH |
5 | * |
6 | * You may distribute under the terms of either the GNU General Public | |
7 | * License or the Artistic License, as specified in the README file. | |
8 | * | |
a687059c LW |
9 | */ |
10 | ||
c277df42 IZ |
11 | typedef OP OP_4tree; /* Will be redefined later. */ |
12 | ||
07be1b83 YO |
13 | |
14 | #define PERL_ENABLE_TRIE_OPTIMISATION 1 | |
15 | #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1 | |
1de06328 | 16 | #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 1 |
07be1b83 YO |
17 | #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 |
18 | ||
a687059c LW |
19 | /* |
20 | * The "internal use only" fields in regexp.h are present to pass info from | |
21 | * compile to execute that permits the execute phase to run lots faster on | |
22 | * simple cases. They are: | |
23 | * | |
bd61b366 | 24 | * regstart sv that must begin a match; NULL if none obvious |
a687059c LW |
25 | * reganch is the match anchored (at beginning-of-line only)? |
26 | * regmust string (pointer into program) that match must include, or NULL | |
79072805 | 27 | * [regmust changed to SV* for bminstr()--law] |
a687059c LW |
28 | * regmlen length of regmust string |
29 | * [regmlen not used currently] | |
30 | * | |
31 | * Regstart and reganch permit very fast decisions on suitable starting points | |
32 | * for a match, cutting down the work a lot. Regmust permits fast rejection | |
33 | * of lines that cannot possibly match. The regmust tests are costly enough | |
e50aee73 | 34 | * that pregcomp() supplies a regmust only if the r.e. contains something |
a687059c LW |
35 | * potentially expensive (at present, the only such thing detected is * or + |
36 | * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
e50aee73 | 37 | * supplied because the test in pregexec() needs it and pregcomp() is computing |
a687059c LW |
38 | * it anyway. |
39 | * [regmust is now supplied always. The tests that use regmust have a | |
40 | * heuristic that disables the test if it usually matches.] | |
41 | * | |
42 | * [In fact, we now use regmust in many cases to locate where the search | |
43 | * starts in the string, so if regback is >= 0, the regmust search is never | |
44 | * wasted effort. The regback variable says how many characters back from | |
45 | * where regmust matched is the earliest possible start of the match. | |
46 | * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] | |
47 | */ | |
48 | ||
49 | /* | |
50 | * Structure for regexp "program". This is essentially a linear encoding | |
51 | * of a nondeterministic finite-state machine (aka syntax charts or | |
52 | * "railroad normal form" in parsing technology). Each node is an opcode | |
53 | * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
54 | * all nodes except BRANCH implement concatenation; a "next" pointer with | |
55 | * a BRANCH on both ends of it is connecting two alternatives. (Here we | |
56 | * have one of the subtle syntax dependencies: an individual BRANCH (as | |
57 | * opposed to a collection of them) is never concatenated with anything | |
58 | * because of operator precedence.) The operand of some types of node is | |
59 | * a literal string; for others, it is a node leading into a sub-FSM. In | |
60 | * particular, the operand of a BRANCH node is the first node of the branch. | |
61 | * (NB this is *not* a tree structure: the tail of the branch connects | |
a3621e74 YO |
62 | * to the thing following the set of BRANCHes.) The opcodes are defined |
63 | * in regnodes.h which is generated from regcomp.sym by regcomp.pl. | |
a687059c LW |
64 | */ |
65 | ||
a687059c LW |
66 | /* |
67 | * A node is one char of opcode followed by two chars of "next" pointer. | |
68 | * "Next" pointers are stored as two 8-bit pieces, high order first. The | |
69 | * value is a positive offset from the opcode of the node containing it. | |
70 | * An operand, if any, simply follows the node. (Note that much of the | |
71 | * code generation knows about this implicit relationship.) | |
72 | * | |
73 | * Using two bytes for the "next" pointer is vast overkill for most things, | |
74 | * but allows patterns to get big without disasters. | |
75 | * | |
e91177ed | 76 | * [The "next" pointer is always aligned on an even |
a687059c LW |
77 | * boundary, and reads the offset directly as a short. Also, there is no |
78 | * special test to reverse the sign of BACK pointers since the offset is | |
79 | * stored negative.] | |
80 | */ | |
81 | ||
c277df42 | 82 | struct regnode_string { |
cd439c50 | 83 | U8 str_len; |
c277df42 IZ |
84 | U8 type; |
85 | U16 next_off; | |
cd439c50 | 86 | char string[1]; |
c277df42 | 87 | }; |
a687059c | 88 | |
6bda09f9 YO |
89 | /* Argument bearing node - workhorse, |
90 | arg1 is often for the data field */ | |
c277df42 IZ |
91 | struct regnode_1 { |
92 | U8 flags; | |
93 | U8 type; | |
94 | U16 next_off; | |
95 | U32 arg1; | |
96 | }; | |
97 | ||
6bda09f9 YO |
98 | /* Similar to a regnode_1 but with an extra signed argument */ |
99 | struct regnode_2L { | |
100 | U8 flags; | |
101 | U8 type; | |
102 | U16 next_off; | |
103 | U32 arg1; | |
104 | I32 arg2; | |
105 | }; | |
106 | ||
107 | /* 'Two field' -- Two 16 bit unsigned args */ | |
c277df42 IZ |
108 | struct regnode_2 { |
109 | U8 flags; | |
110 | U8 type; | |
111 | U16 next_off; | |
112 | U16 arg1; | |
113 | U16 arg2; | |
114 | }; | |
115 | ||
6bda09f9 | 116 | |
936ed897 | 117 | #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ |
ffc61ed2 | 118 | #define ANYOF_CLASSBITMAP_SIZE 4 /* up to 32 (8*4) named classes */ |
936ed897 | 119 | |
786e8c11 | 120 | /* also used by trie */ |
936ed897 IZ |
121 | struct regnode_charclass { |
122 | U8 flags; | |
123 | U8 type; | |
124 | U16 next_off; | |
ffc61ed2 | 125 | U32 arg1; |
19103b91 | 126 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 IZ |
127 | }; |
128 | ||
19103b91 JH |
129 | struct regnode_charclass_class { /* has [[:blah:]] classes */ |
130 | U8 flags; /* should have ANYOF_CLASS here */ | |
936ed897 IZ |
131 | U8 type; |
132 | U16 next_off; | |
ffc61ed2 | 133 | U32 arg1; |
19103b91 JH |
134 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
135 | char classflags[ANYOF_CLASSBITMAP_SIZE]; /* and run-time */ | |
936ed897 IZ |
136 | }; |
137 | ||
9c7e81e8 AD |
138 | /* XXX fix this description. |
139 | Impose a limit of REG_INFTY on various pattern matching operations | |
140 | to limit stack growth and to avoid "infinite" recursions. | |
141 | */ | |
142 | /* The default size for REG_INFTY is I16_MAX, which is the same as | |
143 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits | |
144 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is | |
145 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is | |
146 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a | |
147 | smaller default. | |
148 | --Andy Dougherty 11 June 1998 | |
149 | */ | |
150 | #if SHORTSIZE > 2 | |
151 | # ifndef REG_INFTY | |
152 | # define REG_INFTY ((1<<15)-1) | |
153 | # endif | |
154 | #endif | |
155 | ||
156 | #ifndef REG_INFTY | |
83cfe48c IZ |
157 | # define REG_INFTY I16_MAX |
158 | #endif | |
a687059c | 159 | |
e91177ed IZ |
160 | #define ARG_VALUE(arg) (arg) |
161 | #define ARG__SET(arg,val) ((arg) = (val)) | |
c277df42 | 162 | |
e8e3b0fb JH |
163 | #undef ARG |
164 | #undef ARG1 | |
165 | #undef ARG2 | |
166 | ||
c277df42 IZ |
167 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
168 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) | |
169 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) | |
6bda09f9 | 170 | #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) |
786e8c11 | 171 | |
c277df42 IZ |
172 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
173 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) | |
174 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) | |
6bda09f9 | 175 | #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) |
c277df42 | 176 | |
e8e3b0fb JH |
177 | #undef NEXT_OFF |
178 | #undef NODE_ALIGN | |
179 | ||
075abff3 AL |
180 | #define NEXT_OFF(p) ((p)->next_off) |
181 | #define NODE_ALIGN(node) | |
182 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ | |
a687059c | 183 | |
c277df42 IZ |
184 | #define SIZE_ALIGN NODE_ALIGN |
185 | ||
e8e3b0fb JH |
186 | #undef OP |
187 | #undef OPERAND | |
188 | #undef MASK | |
189 | #undef STRING | |
190 | ||
e91177ed IZ |
191 | #define OP(p) ((p)->type) |
192 | #define OPERAND(p) (((struct regnode_string *)p)->string) | |
cd439c50 IZ |
193 | #define MASK(p) ((char*)OPERAND(p)) |
194 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) | |
195 | #define STRING(p) (((struct regnode_string *)p)->string) | |
196 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) | |
197 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) | |
198 | ||
e8e3b0fb JH |
199 | #undef NODE_ALIGN |
200 | #undef ARG_LOC | |
201 | #undef NEXTOPER | |
202 | #undef PREVOPER | |
203 | ||
e91177ed IZ |
204 | #define NODE_ALIGN(node) |
205 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) | |
206 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) | |
207 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) | |
6bda09f9 | 208 | #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) |
786e8c11 | 209 | |
e91177ed IZ |
210 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
211 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) | |
212 | ||
213 | #define NODE_STEP_B 4 | |
c277df42 IZ |
214 | |
215 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) | |
216 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) | |
217 | ||
e91177ed | 218 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 | 219 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed | 220 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 | 221 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c | 222 | |
22c35a8c | 223 | #define REG_MAGIC 0234 |
a687059c | 224 | |
830247a4 | 225 | #define SIZE_ONLY (RExC_emit == &PL_regdummy) |
c277df42 | 226 | |
936ed897 | 227 | /* Flags for node->flags of ANYOF */ |
b8c5462f | 228 | |
19103b91 | 229 | #define ANYOF_CLASS 0x08 /* has [[:blah:]] classes */ |
ffc61ed2 JH |
230 | #define ANYOF_INVERT 0x04 |
231 | #define ANYOF_FOLD 0x02 | |
232 | #define ANYOF_LOCALE 0x01 | |
b8c5462f | 233 | |
653099ff | 234 | /* Used for regstclass only */ |
ffc61ed2 JH |
235 | #define ANYOF_EOS 0x10 /* Can match an empty string too */ |
236 | ||
237 | /* There is a character or a range past 0xff */ | |
238 | #define ANYOF_UNICODE 0x20 | |
1aa99e6b | 239 | #define ANYOF_UNICODE_ALL 0x40 /* Can match any char past 0xff */ |
ffc61ed2 | 240 | |
4f66b38d HS |
241 | /* size of node is large (includes class pointer) */ |
242 | #define ANYOF_LARGE 0x80 | |
243 | ||
ffc61ed2 JH |
244 | /* Are there any runtime flags on in this node? */ |
245 | #define ANYOF_RUNTIME(s) (ANYOF_FLAGS(s) & 0x0f) | |
246 | ||
247 | #define ANYOF_FLAGS_ALL 0xff | |
653099ff | 248 | |
936ed897 | 249 | /* Character classes for node->classflags of ANYOF */ |
653099ff GS |
250 | /* Should be synchronized with a table in regprop() */ |
251 | /* 2n should pair with 2n+1 */ | |
b8c5462f | 252 | |
289d4f09 | 253 | #define ANYOF_ALNUM 0 /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
b8c5462f | 254 | #define ANYOF_NALNUM 1 |
aaa51d5e | 255 | #define ANYOF_SPACE 2 /* \s */ |
b8c5462f JH |
256 | #define ANYOF_NSPACE 3 |
257 | #define ANYOF_DIGIT 4 | |
258 | #define ANYOF_NDIGIT 5 | |
289d4f09 | 259 | #define ANYOF_ALNUMC 6 /* isalnum(3), utf8::IsAlnum, ALNUMC */ |
b8c5462f JH |
260 | #define ANYOF_NALNUMC 7 |
261 | #define ANYOF_ALPHA 8 | |
262 | #define ANYOF_NALPHA 9 | |
263 | #define ANYOF_ASCII 10 | |
264 | #define ANYOF_NASCII 11 | |
265 | #define ANYOF_CNTRL 12 | |
266 | #define ANYOF_NCNTRL 13 | |
267 | #define ANYOF_GRAPH 14 | |
268 | #define ANYOF_NGRAPH 15 | |
269 | #define ANYOF_LOWER 16 | |
270 | #define ANYOF_NLOWER 17 | |
271 | #define ANYOF_PRINT 18 | |
272 | #define ANYOF_NPRINT 19 | |
273 | #define ANYOF_PUNCT 20 | |
274 | #define ANYOF_NPUNCT 21 | |
275 | #define ANYOF_UPPER 22 | |
276 | #define ANYOF_NUPPER 23 | |
277 | #define ANYOF_XDIGIT 24 | |
278 | #define ANYOF_NXDIGIT 25 | |
aaa51d5e JF |
279 | #define ANYOF_PSXSPC 26 /* POSIX space: \s plus the vertical tab */ |
280 | #define ANYOF_NPSXSPC 27 | |
ffc61ed2 | 281 | #define ANYOF_BLANK 28 /* GNU extension: space and tab: non-vertical space */ |
289d4f09 | 282 | #define ANYOF_NBLANK 29 |
b8c5462f | 283 | |
aaa51d5e | 284 | #define ANYOF_MAX 32 |
b8c5462f JH |
285 | |
286 | /* Backward source code compatibility. */ | |
287 | ||
288 | #define ANYOF_ALNUML ANYOF_ALNUM | |
289 | #define ANYOF_NALNUML ANYOF_NALNUM | |
290 | #define ANYOF_SPACEL ANYOF_SPACE | |
291 | #define ANYOF_NSPACEL ANYOF_NSPACE | |
292 | ||
293 | /* Utility macros for the bitmap and classes of ANYOF */ | |
294 | ||
936ed897 IZ |
295 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
296 | #define ANYOF_CLASS_SIZE (sizeof(struct regnode_charclass_class)) | |
b8c5462f | 297 | |
936ed897 | 298 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f JH |
299 | |
300 | #define ANYOF_BIT(c) (1 << ((c) & 7)) | |
301 | ||
936ed897 | 302 | #define ANYOF_CLASS_BYTE(p, c) (((struct regnode_charclass_class*)(p))->classflags[((c) >> 3) & 3]) |
b8c5462f JH |
303 | #define ANYOF_CLASS_SET(p, c) (ANYOF_CLASS_BYTE(p, c) |= ANYOF_BIT(c)) |
304 | #define ANYOF_CLASS_CLEAR(p, c) (ANYOF_CLASS_BYTE(p, c) &= ~ANYOF_BIT(c)) | |
305 | #define ANYOF_CLASS_TEST(p, c) (ANYOF_CLASS_BYTE(p, c) & ANYOF_BIT(c)) | |
306 | ||
936ed897 IZ |
307 | #define ANYOF_CLASS_ZERO(ret) Zero(((struct regnode_charclass_class*)(ret))->classflags, ANYOF_CLASSBITMAP_SIZE, char) |
308 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) | |
309 | ||
310 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) | |
786e8c11 | 311 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f JH |
312 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
313 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) | |
314 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) | |
315 | ||
f8bef550 NC |
316 | #define ANYOF_BITMAP_SETALL(p) \ |
317 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) | |
318 | #define ANYOF_BITMAP_CLEARALL(p) \ | |
319 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) | |
320 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ | |
321 | #define ANYOF_BITMAP_TESTALLSET(p) \ | |
322 | 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) | |
323 | ||
936ed897 IZ |
324 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
325 | #define ANYOF_CLASS_SKIP ((ANYOF_CLASS_SIZE - 1)/sizeof(regnode)) | |
326 | #define ANYOF_CLASS_ADD_SKIP (ANYOF_CLASS_SKIP - ANYOF_SKIP) | |
c277df42 | 327 | |
786e8c11 | 328 | |
a687059c LW |
329 | /* |
330 | * Utility definitions. | |
331 | */ | |
2304df62 | 332 | #ifndef CHARMASK |
075abff3 | 333 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c | 334 | #else |
075abff3 | 335 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c | 336 | #endif |
a687059c | 337 | |
c277df42 IZ |
338 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
339 | ||
3dab1dad YO |
340 | #define REG_SEEN_ZERO_LEN 0x00000001 |
341 | #define REG_SEEN_LOOKBEHIND 0x00000002 | |
342 | #define REG_SEEN_GPOS 0x00000004 | |
343 | #define REG_SEEN_EVAL 0x00000008 | |
344 | #define REG_SEEN_CANY 0x00000010 | |
cce850e4 | 345 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
6bda09f9 | 346 | #define REG_SEEN_RECURSE 0x00000020 |
eaf3ca90 | 347 | #define REG_TOP_LEVEL_BRANCHES 0x00000040 |
d09b2d29 | 348 | |
73c4f7a1 GS |
349 | START_EXTERN_C |
350 | ||
be8e71aa YO |
351 | #ifdef PLUGGABLE_RE_EXTENSION |
352 | #include "re_nodes.h" | |
353 | #else | |
d09b2d29 | 354 | #include "regnodes.h" |
be8e71aa | 355 | #endif |
d09b2d29 | 356 | |
4d61ec05 | 357 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
d09b2d29 | 358 | #ifndef DOINIT |
4d61ec05 | 359 | EXTCONST U8 PL_varies[]; |
d09b2d29 | 360 | #else |
4d61ec05 | 361 | EXTCONST U8 PL_varies[] = { |
0111c4fd | 362 | BRANCH, BACK, STAR, PLUS, CURLY, CURLYX, REF, REFF, REFFL, |
a0ed51b3 | 363 | WHILEM, CURLYM, CURLYN, BRANCHJ, IFTHEN, SUSPEND, CLUMP, 0 |
c277df42 | 364 | }; |
d09b2d29 | 365 | #endif |
c277df42 | 366 | |
4d61ec05 GS |
367 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
368 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ | |
d09b2d29 | 369 | #ifndef DOINIT |
4d61ec05 | 370 | EXTCONST U8 PL_simple[]; |
d09b2d29 | 371 | #else |
4d61ec05 | 372 | EXTCONST U8 PL_simple[] = { |
f33976b4 | 373 | REG_ANY, SANY, CANY, |
ffc61ed2 JH |
374 | ANYOF, |
375 | ALNUM, ALNUML, | |
376 | NALNUM, NALNUML, | |
377 | SPACE, SPACEL, | |
378 | NSPACE, NSPACEL, | |
379 | DIGIT, NDIGIT, | |
380 | 0 | |
c277df42 IZ |
381 | }; |
382 | #endif | |
383 | ||
f9f4320a YO |
384 | #ifndef PLUGGABLE_RE_EXTENSION |
385 | #ifndef DOINIT | |
386 | EXTCONST regexp_engine PL_core_reg_engine; | |
387 | #else /* DOINIT */ | |
388 | EXTCONST regexp_engine PL_core_reg_engine = { | |
389 | Perl_pregcomp, | |
390 | Perl_regexec_flags, | |
391 | Perl_re_intuit_start, | |
392 | Perl_re_intuit_string, | |
393 | Perl_pregfree, | |
394 | #if defined(USE_ITHREADS) | |
395 | Perl_regdupe | |
396 | #endif | |
397 | }; | |
398 | #endif /* DOINIT */ | |
399 | #endif /* PLUGGABLE_RE_EXTENSION */ | |
400 | ||
401 | ||
73c4f7a1 | 402 | END_EXTERN_C |
cad2e5aa | 403 | |
cad2e5aa | 404 | |
261faec3 MJD |
405 | /* .what is a character array with one character for each member of .data |
406 | * The character describes the function of the corresponding .data item: | |
0111c4fd | 407 | * f - start-class data for regstclass optimization |
261faec3 MJD |
408 | * n - Root of op tree for (?{EVAL}) item |
409 | * o - Start op for (?{EVAL}) item | |
3dab1dad | 410 | * p - Pad for (?{EVAL}) item |
9e55ce06 JH |
411 | * s - swash for unicode-style character class, and the multicharacter |
412 | * strings resulting from casefolding the single-character entries | |
413 | * in the character class | |
a3621e74 | 414 | * t - trie struct |
07be1b83 | 415 | * T - aho-trie struct |
261faec3 | 416 | * 20010712 mjd@plover.com |
c3613ed7 | 417 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 | 418 | */ |
cad2e5aa JH |
419 | struct reg_data { |
420 | U32 count; | |
421 | U8 *what; | |
422 | void* data[1]; | |
423 | }; | |
424 | ||
425 | struct reg_substr_datum { | |
426 | I32 min_offset; | |
427 | I32 max_offset; | |
33b8afdf JH |
428 | SV *substr; /* non-utf8 variant */ |
429 | SV *utf8_substr; /* utf8 variant */ | |
1de06328 | 430 | I32 end_shift; |
cad2e5aa JH |
431 | }; |
432 | ||
433 | struct reg_substr_data { | |
434 | struct reg_substr_datum data[3]; /* Actual array */ | |
435 | }; | |
436 | ||
437 | #define anchored_substr substrs->data[0].substr | |
33b8afdf | 438 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa | 439 | #define anchored_offset substrs->data[0].min_offset |
1de06328 YO |
440 | #define anchored_end_shift substrs->data[0].end_shift |
441 | ||
cad2e5aa | 442 | #define float_substr substrs->data[1].substr |
33b8afdf | 443 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa JH |
444 | #define float_min_offset substrs->data[1].min_offset |
445 | #define float_max_offset substrs->data[1].max_offset | |
1de06328 YO |
446 | #define float_end_shift substrs->data[1].end_shift |
447 | ||
cad2e5aa | 448 | #define check_substr substrs->data[2].substr |
33b8afdf | 449 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa JH |
450 | #define check_offset_min substrs->data[2].min_offset |
451 | #define check_offset_max substrs->data[2].max_offset | |
1de06328 | 452 | #define check_end_shift substrs->data[2].end_shift |
a3621e74 YO |
453 | |
454 | ||
455 | ||
456 | /* trie related stuff */ | |
5d9a96ca | 457 | |
a3621e74 YO |
458 | /* a transition record for the state machine. the |
459 | check field determines which state "owns" the | |
460 | transition. the char the transition is for is | |
461 | determined by offset from the owning states base | |
462 | field. the next field determines which state | |
463 | is to be transitioned to if any. | |
464 | */ | |
465 | struct _reg_trie_trans { | |
466 | U32 next; | |
467 | U32 check; | |
468 | }; | |
469 | ||
470 | /* a transition list element for the list based representation */ | |
471 | struct _reg_trie_trans_list_elem { | |
472 | U16 forid; | |
473 | U32 newstate; | |
474 | }; | |
475 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; | |
476 | ||
477 | /* a state for compressed nodes. base is an offset | |
478 | into an array of reg_trie_trans array. If wordnum is | |
479 | nonzero the state is accepting. if base is zero then | |
480 | the state has no children (and will be accepting) | |
481 | */ | |
482 | struct _reg_trie_state { | |
483 | U16 wordnum; | |
484 | union { | |
485 | U32 base; | |
486 | reg_trie_trans_le* list; | |
487 | } trans; | |
488 | }; | |
489 | ||
490 | ||
491 | ||
a3621e74 YO |
492 | typedef struct _reg_trie_state reg_trie_state; |
493 | typedef struct _reg_trie_trans reg_trie_trans; | |
494 | ||
495 | ||
496 | /* anything in here that needs to be freed later | |
8e11feef | 497 | should be dealt with in pregfree */ |
a3621e74 | 498 | struct _reg_trie_data { |
786e8c11 YO |
499 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
500 | U32 lasttrans; /* last valid transition element */ | |
501 | U16 *charmap; /* byte to charid lookup array */ | |
502 | HV *widecharmap; /* code points > 255 to charid */ | |
503 | reg_trie_state *states; /* state data */ | |
504 | reg_trie_trans *trans; /* array of transition elements */ | |
505 | char *bitmap; /* stclass bitmap */ | |
506 | U32 refcount; /* number of times this trie is referenced */ | |
507 | U32 startstate; /* initial state - used for common prefix optimisation */ | |
508 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ | |
509 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ | |
510 | U32 *wordlen; /* array of lengths of words */ | |
511 | U16 *jump; /* optional 1 indexed array of offsets before tail | |
512 | for the node following a given word. */ | |
513 | U16 *nextword; /* optional 1 indexed array to support linked list | |
514 | of duplicate wordnums */ | |
515 | U32 laststate; /* Build only */ | |
516 | U32 wordcount; /* Build only */ | |
a3621e74 | 517 | #ifdef DEBUGGING |
786e8c11 YO |
518 | STRLEN charcount; /* Build only */ |
519 | AV *words; /* Array of words contained in trie, for dumping */ | |
520 | AV *revcharmap; /* Map of each charid back to its character representation */ | |
a3621e74 YO |
521 | #endif |
522 | }; | |
a3621e74 YO |
523 | typedef struct _reg_trie_data reg_trie_data; |
524 | ||
07be1b83 YO |
525 | struct _reg_ac_data { |
526 | U32 *fail; | |
527 | reg_trie_state *states; | |
528 | reg_trie_data *trie; | |
529 | U32 refcount; | |
530 | }; | |
531 | typedef struct _reg_ac_data reg_ac_data; | |
532 | ||
3dab1dad YO |
533 | /* ANY_BIT doesnt use the structure, so we can borrow it here. |
534 | This is simpler than refactoring all of it as wed end up with | |
535 | three different sets... */ | |
536 | ||
537 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) | |
786e8c11 | 538 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f YO |
539 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
540 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) | |
541 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 542 | |
1de06328 YO |
543 | #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) |
544 | #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) | |
545 | ||
546 | ||
786e8c11 YO |
547 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
548 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 549 | |
a3621e74 YO |
550 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
551 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) | |
552 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) | |
553 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) | |
554 | ||
3dab1dad | 555 | #ifdef DEBUGGING |
3dab1dad | 556 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad YO |
557 | #define TRIE_REVCHARMAP(trie) ((trie)->revcharmap) |
558 | #else | |
3dab1dad | 559 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad YO |
560 | #define TRIE_REVCHARMAP(trie) (trie_revcharmap) |
561 | #endif | |
a3621e74 | 562 | |
0111c4fd RGS |
563 | #define RE_TRIE_MAXBUF_INIT 65536 |
564 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" | |
a3621e74 YO |
565 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
566 | ||
be8e71aa YO |
567 | /* |
568 | ||
569 | RE_DEBUG_FLAGS is used to control what debug output is emitted | |
570 | its divided into three groups of options, some of which interact. | |
571 | The three groups are: Compile, Execute, Extra. There is room for a | |
572 | further group, as currently only the low three bytes are used. | |
573 | ||
574 | Compile Options: | |
07be1b83 | 575 | |
be8e71aa YO |
576 | PARSE |
577 | PEEP | |
578 | TRIE | |
579 | PROGRAM | |
580 | OFFSETS | |
a3621e74 | 581 | |
be8e71aa | 582 | Execute Options: |
a3621e74 | 583 | |
be8e71aa YO |
584 | INTUIT |
585 | MATCH | |
586 | TRIE | |
a3621e74 | 587 | |
be8e71aa | 588 | Extra Options |
a3621e74 | 589 | |
be8e71aa YO |
590 | TRIE |
591 | OFFSETS | |
592 | ||
593 | If you modify any of these make sure you make corresponding changes to | |
594 | re.pm, especially to the documentation. | |
a3621e74 | 595 | |
be8e71aa YO |
596 | */ |
597 | ||
598 | ||
599 | /* Compile */ | |
600 | #define RE_DEBUG_COMPILE_MASK 0x0000FF | |
601 | #define RE_DEBUG_COMPILE_PARSE 0x000001 | |
602 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 | |
603 | #define RE_DEBUG_COMPILE_TRIE 0x000004 | |
604 | #define RE_DEBUG_COMPILE_DUMP 0x000008 | |
be8e71aa YO |
605 | |
606 | /* Execute */ | |
607 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 | |
608 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 | |
609 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 | |
610 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 | |
611 | ||
612 | /* Extra */ | |
613 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 | |
614 | #define RE_DEBUG_EXTRA_TRIE 0x010000 | |
615 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 | |
a5ca303d YO |
616 | #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 |
617 | #define RE_DEBUG_EXTRA_STATE 0x080000 | |
618 | #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 | |
be8e71aa | 619 | |
e68ec53f | 620 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa YO |
621 | /* Compile */ |
622 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ | |
e68ec53f | 623 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa | 624 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 625 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 626 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f | 627 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa | 628 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 629 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 630 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f | 631 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa | 632 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f | 633 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
be8e71aa YO |
634 | |
635 | /* Execute */ | |
636 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ | |
e68ec53f | 637 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa | 638 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f | 639 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa | 640 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f | 641 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa | 642 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f | 643 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa YO |
644 | |
645 | /* Extra */ | |
646 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ | |
e68ec53f | 647 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
a5ca303d YO |
648 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
649 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) | |
ab3bbdeb YO |
650 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
651 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) | |
a5ca303d YO |
652 | #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ |
653 | if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ | |
654 | (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) | |
07be1b83 | 655 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
a5ca303d | 656 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ |
be8e71aa YO |
657 | Perl_warn_nocontext x ) |
658 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ | |
e68ec53f | 659 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa | 660 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f | 661 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa YO |
662 | |
663 | #define DEBUG_TRIE_r(x) DEBUG_r( \ | |
e68ec53f | 664 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa YO |
665 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
666 | ||
667 | /* initialization */ | |
668 | /* get_sv() can return NULL during global destruction. */ | |
e68ec53f YO |
669 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
670 | SV * re_debug_flags_sv = NULL; \ | |
671 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ | |
672 | if (re_debug_flags_sv) { \ | |
8e9a8a48 | 673 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f | 674 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 | 675 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f YO |
676 | }\ |
677 | }) | |
a3621e74 YO |
678 | |
679 | #ifdef DEBUGGING | |
ab3bbdeb | 680 | |
8e9a8a48 | 681 | #define GET_RE_DEBUG_FLAGS_DECL IV re_debug_flags = 0; GET_RE_DEBUG_FLAGS; |
ab3bbdeb YO |
682 | |
683 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ | |
684 | const char * const rpv = \ | |
ddc5bc0f | 685 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb YO |
686 | PL_colors[(c1)],PL_colors[(c2)], \ |
687 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ | |
0df25f3d | 688 | const int rlen = SvCUR(dsv) |
ab3bbdeb YO |
689 | |
690 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ | |
691 | const char * const rpv = \ | |
ddc5bc0f | 692 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb YO |
693 | PL_colors[(c1)],PL_colors[(c2)], \ |
694 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) | |
695 | ||
696 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ | |
697 | const char * const rpv = \ | |
ddc5bc0f | 698 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb YO |
699 | PL_colors[0], PL_colors[1], \ |
700 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_PRETTY_ELIPSES | \ | |
701 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ | |
702 | ) | |
703 | ||
704 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) | |
705 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") | |
706 | ||
707 | #else /* if not DEBUGGING */ | |
708 | ||
a3621e74 | 709 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb YO |
710 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
711 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) | |
712 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) | |
713 | #define RE_SV_DUMPLEN(ItEm) | |
714 | #define RE_SV_TAIL(ItEm) | |
715 | ||
716 | #endif /* DEBUG RELATED DEFINES */ | |
a3621e74 | 717 |