Commit | Line | Data |
---|---|---|
a0d0e21e | 1 | /* regcomp.h |
d6376244 | 2 | * |
4bb101f2 | 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
663f364b | 4 | * 2000, 2001, 2002, 2003, 2005, 2006, 2007, 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 | 9 | */ |
e1d1eefb | 10 | #include "regcharclass.h" |
a687059c | 11 | |
58e23c8d | 12 | /* Convert branch sequences to more efficient trie ops? */ |
07be1b83 | 13 | #define PERL_ENABLE_TRIE_OPTIMISATION 1 |
58e23c8d | 14 | |
486ec47a | 15 | /* Be really aggressive about optimising patterns with trie sequences? */ |
07be1b83 | 16 | #define PERL_ENABLE_EXTENDED_TRIE_OPTIMISATION 1 |
58e23c8d YO |
17 | |
18 | /* Should the optimiser take positive assertions into account? */ | |
68ba3a3f | 19 | #define PERL_ENABLE_POSITIVE_ASSERTION_STUDY 0 |
58e23c8d YO |
20 | |
21 | /* Not for production use: */ | |
07be1b83 | 22 | #define PERL_ENABLE_EXPERIMENTAL_REGEX_OPTIMISATIONS 0 |
58e23c8d | 23 | |
7122b237 YO |
24 | /* Activate offsets code - set to if 1 to enable */ |
25 | #ifdef DEBUGGING | |
26 | #define RE_TRACK_PATTERN_OFFSETS | |
27 | #endif | |
28 | ||
a687059c LW |
29 | /* |
30 | * The "internal use only" fields in regexp.h are present to pass info from | |
31 | * compile to execute that permits the execute phase to run lots faster on | |
32 | * simple cases. They are: | |
33 | * | |
bd61b366 | 34 | * regstart sv that must begin a match; NULL if none obvious |
a687059c LW |
35 | * reganch is the match anchored (at beginning-of-line only)? |
36 | * regmust string (pointer into program) that match must include, or NULL | |
79072805 | 37 | * [regmust changed to SV* for bminstr()--law] |
a687059c LW |
38 | * regmlen length of regmust string |
39 | * [regmlen not used currently] | |
40 | * | |
41 | * Regstart and reganch permit very fast decisions on suitable starting points | |
42 | * for a match, cutting down the work a lot. Regmust permits fast rejection | |
43 | * of lines that cannot possibly match. The regmust tests are costly enough | |
e50aee73 | 44 | * that pregcomp() supplies a regmust only if the r.e. contains something |
a687059c LW |
45 | * potentially expensive (at present, the only such thing detected is * or + |
46 | * at the start of the r.e., which can involve a lot of backup). Regmlen is | |
e50aee73 | 47 | * supplied because the test in pregexec() needs it and pregcomp() is computing |
a687059c LW |
48 | * it anyway. |
49 | * [regmust is now supplied always. The tests that use regmust have a | |
50 | * heuristic that disables the test if it usually matches.] | |
51 | * | |
52 | * [In fact, we now use regmust in many cases to locate where the search | |
53 | * starts in the string, so if regback is >= 0, the regmust search is never | |
54 | * wasted effort. The regback variable says how many characters back from | |
55 | * where regmust matched is the earliest possible start of the match. | |
56 | * For instance, /[a-z].foo/ has a regmust of 'foo' and a regback of 2.] | |
57 | */ | |
58 | ||
59 | /* | |
60 | * Structure for regexp "program". This is essentially a linear encoding | |
61 | * of a nondeterministic finite-state machine (aka syntax charts or | |
62 | * "railroad normal form" in parsing technology). Each node is an opcode | |
63 | * plus a "next" pointer, possibly plus an operand. "Next" pointers of | |
64 | * all nodes except BRANCH implement concatenation; a "next" pointer with | |
65 | * a BRANCH on both ends of it is connecting two alternatives. (Here we | |
66 | * have one of the subtle syntax dependencies: an individual BRANCH (as | |
67 | * opposed to a collection of them) is never concatenated with anything | |
68 | * because of operator precedence.) The operand of some types of node is | |
69 | * a literal string; for others, it is a node leading into a sub-FSM. In | |
70 | * particular, the operand of a BRANCH node is the first node of the branch. | |
71 | * (NB this is *not* a tree structure: the tail of the branch connects | |
a3621e74 YO |
72 | * to the thing following the set of BRANCHes.) The opcodes are defined |
73 | * in regnodes.h which is generated from regcomp.sym by regcomp.pl. | |
a687059c LW |
74 | */ |
75 | ||
a687059c LW |
76 | /* |
77 | * A node is one char of opcode followed by two chars of "next" pointer. | |
78 | * "Next" pointers are stored as two 8-bit pieces, high order first. The | |
79 | * value is a positive offset from the opcode of the node containing it. | |
80 | * An operand, if any, simply follows the node. (Note that much of the | |
81 | * code generation knows about this implicit relationship.) | |
82 | * | |
83 | * Using two bytes for the "next" pointer is vast overkill for most things, | |
84 | * but allows patterns to get big without disasters. | |
85 | * | |
e91177ed | 86 | * [The "next" pointer is always aligned on an even |
a687059c LW |
87 | * boundary, and reads the offset directly as a short. Also, there is no |
88 | * special test to reverse the sign of BACK pointers since the offset is | |
89 | * stored negative.] | |
90 | */ | |
785a26d5 YO |
91 | |
92 | /* This is the stuff that used to live in regexp.h that was truly | |
93 | private to the engine itself. It now lives here. */ | |
94 | ||
28d8d7f4 | 95 | |
785a26d5 YO |
96 | |
97 | typedef struct regexp_internal { | |
1f1031fe | 98 | int name_list_idx; /* Optional data index of an array of paren names */ |
7122b237 YO |
99 | union { |
100 | U32 *offsets; /* offset annotations 20001228 MJD | |
101 | data about mapping the program to the | |
102 | string - | |
103 | offsets[0] is proglen when this is used | |
104 | */ | |
105 | U32 proglen; | |
106 | } u; | |
107 | ||
6d5c990f RGS |
108 | regnode *regstclass; /* Optional startclass as identified or constructed |
109 | by the optimiser */ | |
110 | struct reg_data *data; /* Additional miscellaneous data used by the program. | |
111 | Used to make it easier to clone and free arbitrary | |
112 | data that the regops need. Often the ARG field of | |
113 | a regop is an index into this structure */ | |
3d2bd50a DM |
114 | struct reg_code_block *code_blocks;/* positions of literal (?{}) */ |
115 | int num_code_blocks; /* size of code_blocks[] */ | |
6d5c990f RGS |
116 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
117 | } regexp_internal; | |
118 | ||
119 | #define RXi_SET(x,y) (x)->pprivate = (void*)(y) | |
120 | #define RXi_GET(x) ((regexp_internal *)((x)->pprivate)) | |
121 | #define RXi_GET_DECL(r,ri) regexp_internal *ri = RXi_GET(r) | |
785a26d5 YO |
122 | /* |
123 | * Flags stored in regexp->intflags | |
124 | * These are used only internally to the regexp engine | |
125 | * | |
126 | * See regexp.h for flags used externally to the regexp engine | |
127 | */ | |
128 | #define PREGf_SKIP 0x00000001 | |
129 | #define PREGf_IMPLICIT 0x00000002 /* Converted .* to ^.* */ | |
130 | #define PREGf_NAUGHTY 0x00000004 /* how exponential is this pattern? */ | |
131 | #define PREGf_VERBARG_SEEN 0x00000008 | |
132 | #define PREGf_CUTGROUP_SEEN 0x00000010 | |
732caac7 | 133 | #define PREGf_USE_RE_EVAL 0x00000020 /* compiled with "use re 'eval'" */ |
785a26d5 YO |
134 | |
135 | ||
136 | /* this is where the old regcomp.h started */ | |
a687059c | 137 | |
c277df42 | 138 | struct regnode_string { |
cd439c50 | 139 | U8 str_len; |
c277df42 IZ |
140 | U8 type; |
141 | U16 next_off; | |
cd439c50 | 142 | char string[1]; |
c277df42 | 143 | }; |
a687059c | 144 | |
6bda09f9 YO |
145 | /* Argument bearing node - workhorse, |
146 | arg1 is often for the data field */ | |
c277df42 IZ |
147 | struct regnode_1 { |
148 | U8 flags; | |
149 | U8 type; | |
150 | U16 next_off; | |
151 | U32 arg1; | |
152 | }; | |
153 | ||
6bda09f9 YO |
154 | /* Similar to a regnode_1 but with an extra signed argument */ |
155 | struct regnode_2L { | |
156 | U8 flags; | |
157 | U8 type; | |
158 | U16 next_off; | |
159 | U32 arg1; | |
160 | I32 arg2; | |
161 | }; | |
162 | ||
163 | /* 'Two field' -- Two 16 bit unsigned args */ | |
c277df42 IZ |
164 | struct regnode_2 { |
165 | U8 flags; | |
166 | U8 type; | |
167 | U16 next_off; | |
168 | U16 arg1; | |
169 | U16 arg2; | |
170 | }; | |
171 | ||
6bda09f9 | 172 | |
936ed897 | 173 | #define ANYOF_BITMAP_SIZE 32 /* 256 b/(8 b/B) */ |
936ed897 | 174 | |
786e8c11 | 175 | /* also used by trie */ |
936ed897 IZ |
176 | struct regnode_charclass { |
177 | U8 flags; | |
178 | U8 type; | |
179 | U16 next_off; | |
bf3f2d85 | 180 | U32 arg1; /* used as ptr in S_regclass */ |
19103b91 | 181 | char bitmap[ANYOF_BITMAP_SIZE]; /* only compile-time */ |
936ed897 IZ |
182 | }; |
183 | ||
bf3f2d85 KW |
184 | /* has runtime (locale) \d, \w, ..., [:posix:] classes */ |
185 | struct regnode_charclass_class { | |
8efd3f97 | 186 | U8 flags; /* ANYOF_POSIXL bit must go here */ |
936ed897 IZ |
187 | U8 type; |
188 | U16 next_off; | |
bf3f2d85 | 189 | U32 arg1; /* used as ptr in S_regclass */ |
19103b91 | 190 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ |
9cf78fa0 | 191 | U32 classflags; /* and run-time */ |
936ed897 IZ |
192 | }; |
193 | ||
a0dd4231 KW |
194 | /* Synthetic start class; is a regnode_charclass_class plus an SV*. Note that |
195 | * the 'next_off' field is unused, as the SSC stands alone, so there is never a | |
196 | * next node. */ | |
b8f7bb16 | 197 | struct regnode_ssc { |
8efd3f97 | 198 | U8 flags; /* ANYOF_POSIXL bit must go here */ |
b8f7bb16 KW |
199 | U8 type; |
200 | U16 next_off; | |
201 | U32 arg1; /* used as ptr in S_regclass */ | |
202 | char bitmap[ANYOF_BITMAP_SIZE]; /* both compile-time */ | |
203 | U32 classflags; /* and run-time */ | |
204 | SV* invlist; /* list of code points matched */ | |
205 | }; | |
206 | ||
9c7e81e8 AD |
207 | /* XXX fix this description. |
208 | Impose a limit of REG_INFTY on various pattern matching operations | |
209 | to limit stack growth and to avoid "infinite" recursions. | |
210 | */ | |
211 | /* The default size for REG_INFTY is I16_MAX, which is the same as | |
212 | SHORT_MAX (see perl.h). Unfortunately I16 isn't necessarily 16 bits | |
213 | (see handy.h). On the Cray C90, sizeof(short)==4 and hence I16_MAX is | |
214 | ((1<<31)-1), while on the Cray T90, sizeof(short)==8 and I16_MAX is | |
215 | ((1<<63)-1). To limit stack growth to reasonable sizes, supply a | |
216 | smaller default. | |
217 | --Andy Dougherty 11 June 1998 | |
218 | */ | |
219 | #if SHORTSIZE > 2 | |
220 | # ifndef REG_INFTY | |
221 | # define REG_INFTY ((1<<15)-1) | |
222 | # endif | |
223 | #endif | |
224 | ||
225 | #ifndef REG_INFTY | |
83cfe48c IZ |
226 | # define REG_INFTY I16_MAX |
227 | #endif | |
a687059c | 228 | |
e91177ed IZ |
229 | #define ARG_VALUE(arg) (arg) |
230 | #define ARG__SET(arg,val) ((arg) = (val)) | |
c277df42 | 231 | |
e8e3b0fb JH |
232 | #undef ARG |
233 | #undef ARG1 | |
234 | #undef ARG2 | |
235 | ||
c277df42 IZ |
236 | #define ARG(p) ARG_VALUE(ARG_LOC(p)) |
237 | #define ARG1(p) ARG_VALUE(ARG1_LOC(p)) | |
238 | #define ARG2(p) ARG_VALUE(ARG2_LOC(p)) | |
6bda09f9 | 239 | #define ARG2L(p) ARG_VALUE(ARG2L_LOC(p)) |
786e8c11 | 240 | |
c277df42 IZ |
241 | #define ARG_SET(p, val) ARG__SET(ARG_LOC(p), (val)) |
242 | #define ARG1_SET(p, val) ARG__SET(ARG1_LOC(p), (val)) | |
243 | #define ARG2_SET(p, val) ARG__SET(ARG2_LOC(p), (val)) | |
6bda09f9 | 244 | #define ARG2L_SET(p, val) ARG__SET(ARG2L_LOC(p), (val)) |
c277df42 | 245 | |
e8e3b0fb JH |
246 | #undef NEXT_OFF |
247 | #undef NODE_ALIGN | |
248 | ||
075abff3 AL |
249 | #define NEXT_OFF(p) ((p)->next_off) |
250 | #define NODE_ALIGN(node) | |
251 | #define NODE_ALIGN_FILL(node) ((node)->flags = 0xde) /* deadbeef */ | |
a687059c | 252 | |
c277df42 IZ |
253 | #define SIZE_ALIGN NODE_ALIGN |
254 | ||
e8e3b0fb JH |
255 | #undef OP |
256 | #undef OPERAND | |
257 | #undef MASK | |
258 | #undef STRING | |
259 | ||
e91177ed | 260 | #define OP(p) ((p)->type) |
b988e673 KW |
261 | #define FLAGS(p) ((p)->flags) /* Caution: Doesn't apply to all \ |
262 | regnode types. For some, it's the \ | |
263 | character set of the regnode */ | |
e91177ed | 264 | #define OPERAND(p) (((struct regnode_string *)p)->string) |
cd439c50 IZ |
265 | #define MASK(p) ((char*)OPERAND(p)) |
266 | #define STR_LEN(p) (((struct regnode_string *)p)->str_len) | |
267 | #define STRING(p) (((struct regnode_string *)p)->string) | |
268 | #define STR_SZ(l) ((l + sizeof(regnode) - 1) / sizeof(regnode)) | |
269 | #define NODE_SZ_STR(p) (STR_SZ(STR_LEN(p))+1) | |
270 | ||
e8e3b0fb JH |
271 | #undef NODE_ALIGN |
272 | #undef ARG_LOC | |
273 | #undef NEXTOPER | |
274 | #undef PREVOPER | |
275 | ||
e91177ed IZ |
276 | #define NODE_ALIGN(node) |
277 | #define ARG_LOC(p) (((struct regnode_1 *)p)->arg1) | |
278 | #define ARG1_LOC(p) (((struct regnode_2 *)p)->arg1) | |
279 | #define ARG2_LOC(p) (((struct regnode_2 *)p)->arg2) | |
6bda09f9 | 280 | #define ARG2L_LOC(p) (((struct regnode_2L *)p)->arg2) |
786e8c11 | 281 | |
e91177ed IZ |
282 | #define NODE_STEP_REGNODE 1 /* sizeof(regnode)/sizeof(regnode) */ |
283 | #define EXTRA_STEP_2ARGS EXTRA_SIZE(struct regnode_2) | |
284 | ||
285 | #define NODE_STEP_B 4 | |
c277df42 IZ |
286 | |
287 | #define NEXTOPER(p) ((p) + NODE_STEP_REGNODE) | |
288 | #define PREVOPER(p) ((p) - NODE_STEP_REGNODE) | |
289 | ||
e91177ed | 290 | #define FILL_ADVANCE_NODE(ptr, op) STMT_START { \ |
c277df42 | 291 | (ptr)->type = op; (ptr)->next_off = 0; (ptr)++; } STMT_END |
e91177ed | 292 | #define FILL_ADVANCE_NODE_ARG(ptr, op, arg) STMT_START { \ |
c277df42 | 293 | ARG_SET(ptr, arg); FILL_ADVANCE_NODE(ptr, op); (ptr) += 1; } STMT_END |
a687059c | 294 | |
22c35a8c | 295 | #define REG_MAGIC 0234 |
a687059c | 296 | |
9a81a976 | 297 | #define SIZE_ONLY (RExC_emit == (regnode *) & RExC_emit_dummy) |
6d15ebf3 KW |
298 | #define PASS1 SIZE_ONLY |
299 | #define PASS2 (! SIZE_ONLY) | |
c277df42 | 300 | |
137165a6 KW |
301 | /* If the bitmap doesn't fully represent what this ANYOF node can match, the |
302 | * ARG is set to this special value (since 0, 1, ... are legal, but will never | |
303 | * reach this high). */ | |
304 | #define ANYOF_NONBITMAP_EMPTY ((U32) -1) | |
305 | ||
306 | /* The information used to be stored as as combination of the ANYOF_UTF8 and | |
307 | * ANYOF_NONBITMAP_NON_UTF8 bits in the flags field, but was moved out of there | |
308 | * to free up a bit for other uses. This tries to hide the change from | |
309 | * existing code as much as possible. Now, the data structure that goes in ARG | |
310 | * is not allocated unless it is needed, and that is what is used to determine | |
311 | * if there is something outside the bitmap. The code now assumes that if | |
312 | * that structure exists, that any UTF-8 encoded string should be tried against | |
313 | * it, but a non-UTF8-encoded string will be tried only if the | |
314 | * ANYOF_NONBITMAP_NON_UTF8 bit is also set. */ | |
315 | #define ANYOF_NONBITMAP(node) (ARG(node) != ANYOF_NONBITMAP_EMPTY) | |
316 | ||
899d20b9 KW |
317 | /* Flags for node->flags of ANYOF. These are in short supply, but there is one |
318 | * currently available. If more than this are needed, the ANYOF_LOCALE and | |
a0dd4231 KW |
319 | * ANYOF_POSIXL bits could be shared, making a space penalty for all locale |
320 | * nodes. Also, the ABOVE_LATIN1_ALL bit could be freed up by resorting to | |
321 | * creating a swash containing everything above 255. This introduces a | |
322 | * performance penalty. Better would be to split it off into a separate node, | |
323 | * which actually would improve performance a bit by allowing regexec.c to test | |
324 | * for a UTF-8 character being above 255 without having to call a function nor | |
9b1eb067 | 325 | * calculate its code point value. However, this solution might need to have a |
a0dd4231 KW |
326 | * second node type, ANYOF_SYNTHETIC_ABOVE_LATIN1_ALL. Several flags are not |
327 | * used in synthetic start class (SSC) nodes, so could be shared should new | |
328 | * flags be needed for SSCs. */ | |
b8c5462f | 329 | |
a4e790c1 | 330 | #define ANYOF_LOCALE 0x01 /* /l modifier */ |
39065660 KW |
331 | |
332 | /* The fold is calculated and stored in the bitmap where possible at compile | |
538b546e KW |
333 | * time. However under locale, the actual folding varies depending on |
334 | * what the locale is at the time of execution, so it has to be deferred until | |
335 | * then */ | |
acf67047 | 336 | #define ANYOF_LOC_FOLD 0x02 |
39065660 KW |
337 | |
338 | #define ANYOF_INVERT 0x04 | |
1984f6a0 | 339 | |
a0dd4231 KW |
340 | /* For the SSC node only, which cannot be inverted, so is shared with that bit. |
341 | * This means "Does this SSC match an empty string?" This is used only during | |
342 | * regex compilation. */ | |
343 | #define ANYOF_EMPTY_STRING ANYOF_INVERT | |
344 | ||
d1d040e5 | 345 | /* Set if this is a regnode_charclass_posixl vs a regnode_charclass. This |
dd58aee1 KW |
346 | * is used for runtime \d, \w, [:posix:], ..., which are used only in locale |
347 | * and the optimizer's synthetic start class. Non-locale \d, etc are resolved | |
9ffd0d41 KW |
348 | * at compile-time. Could be shared with ANYOF_LOCALE, forcing all locale |
349 | * nodes to be large */ | |
8efd3f97 KW |
350 | #define ANYOF_POSIXL 0x08 |
351 | #define ANYOF_CLASS ANYOF_POSIXL | |
352 | #define ANYOF_LARGE ANYOF_POSIXL | |
c8d3cd88 | 353 | |
899d20b9 | 354 | /* Unused: 0x10. When using, be sure to change ANYOF_FLAGS_ALL below */ |
f4244008 | 355 | |
ef87b810 KW |
356 | /* Can match something outside the bitmap that isn't in utf8 */ |
357 | #define ANYOF_NONBITMAP_NON_UTF8 0x20 | |
358 | ||
61ef9872 | 359 | /* Matches every code point 0x100 and above*/ |
9b1eb067 KW |
360 | #define ANYOF_ABOVE_LATIN1_ALL 0x40 |
361 | #define ANYOF_UNICODE_ALL ANYOF_ABOVE_LATIN1_ALL | |
ffc61ed2 | 362 | |
11454c59 KW |
363 | /* Match all Latin1 characters that aren't ASCII when the target string is not |
364 | * in utf8. */ | |
365 | #define ANYOF_NON_UTF8_LATIN1_ALL 0x80 | |
366 | ||
899d20b9 | 367 | #define ANYOF_FLAGS_ALL (0xff & ~0x10) |
653099ff | 368 | |
fb38762f KW |
369 | #define ANYOF_LOCALE_FLAGS (ANYOF_LOCALE \ |
370 | |ANYOF_LOC_FOLD \ | |
371 | |ANYOF_POSIXL) | |
ace6b0e4 | 372 | |
936ed897 | 373 | /* Character classes for node->classflags of ANYOF */ |
653099ff | 374 | /* Should be synchronized with a table in regprop() */ |
575a8a29 | 375 | /* 2n should be the normal one, paired with its complement at 2n+1 */ |
b8c5462f | 376 | |
08ada29e KW |
377 | #define ANYOF_ALPHA ((_CC_ALPHA) * 2) |
378 | #define ANYOF_NALPHA ((ANYOF_ALPHA) + 1) | |
e8d596e0 KW |
379 | #define ANYOF_ALPHANUMERIC ((_CC_ALPHANUMERIC) * 2) /* [[:alnum:]] isalnum(3), utf8::IsAlnum */ |
380 | #define ANYOF_NALPHANUMERIC ((ANYOF_ALPHANUMERIC) + 1) | |
08ada29e KW |
381 | #define ANYOF_ASCII ((_CC_ASCII) * 2) |
382 | #define ANYOF_NASCII ((ANYOF_ASCII) + 1) | |
e8d596e0 KW |
383 | #define ANYOF_BLANK ((_CC_BLANK) * 2) /* GNU extension: space and tab: non-vertical space */ |
384 | #define ANYOF_NBLANK ((ANYOF_BLANK) + 1) | |
b0d691b2 KW |
385 | #define ANYOF_CASED ((_CC_CASED) * 2) /* Pseudo class for [:lower:] or |
386 | [:upper:] under /i */ | |
387 | #define ANYOF_NCASED ((ANYOF_CASED) + 1) | |
08ada29e KW |
388 | #define ANYOF_CNTRL ((_CC_CNTRL) * 2) |
389 | #define ANYOF_NCNTRL ((ANYOF_CNTRL) + 1) | |
e8d596e0 KW |
390 | #define ANYOF_DIGIT ((_CC_DIGIT) * 2) /* \d */ |
391 | #define ANYOF_NDIGIT ((ANYOF_DIGIT) + 1) | |
08ada29e KW |
392 | #define ANYOF_GRAPH ((_CC_GRAPH) * 2) |
393 | #define ANYOF_NGRAPH ((ANYOF_GRAPH) + 1) | |
394 | #define ANYOF_LOWER ((_CC_LOWER) * 2) | |
395 | #define ANYOF_NLOWER ((ANYOF_LOWER) + 1) | |
396 | #define ANYOF_PRINT ((_CC_PRINT) * 2) | |
397 | #define ANYOF_NPRINT ((ANYOF_PRINT) + 1) | |
e8d596e0 KW |
398 | #define ANYOF_PSXSPC ((_CC_PSXSPC) * 2) /* POSIX space: \s plus the vertical tab */ |
399 | #define ANYOF_NPSXSPC ((ANYOF_PSXSPC) + 1) | |
08ada29e KW |
400 | #define ANYOF_PUNCT ((_CC_PUNCT) * 2) |
401 | #define ANYOF_NPUNCT ((ANYOF_PUNCT) + 1) | |
e8d596e0 KW |
402 | #define ANYOF_SPACE ((_CC_SPACE) * 2) /* \s */ |
403 | #define ANYOF_NSPACE ((ANYOF_SPACE) + 1) | |
08ada29e KW |
404 | #define ANYOF_UPPER ((_CC_UPPER) * 2) |
405 | #define ANYOF_NUPPER ((ANYOF_UPPER) + 1) | |
e8d596e0 KW |
406 | #define ANYOF_WORDCHAR ((_CC_WORDCHAR) * 2) /* \w, PL_utf8_alnum, utf8::IsWord, ALNUM */ |
407 | #define ANYOF_NWORDCHAR ((ANYOF_WORDCHAR) + 1) | |
08ada29e KW |
408 | #define ANYOF_XDIGIT ((_CC_XDIGIT) * 2) |
409 | #define ANYOF_NXDIGIT ((ANYOF_XDIGIT) + 1) | |
b8c5462f | 410 | |
87cad4c0 | 411 | /* pseudo classes below this, not stored in the class bitmap, but used as flags |
e1d1eefb YO |
412 | during compilation of char classes */ |
413 | ||
b2ef2e34 KW |
414 | #define ANYOF_VERTWS ((_CC_VERTSPACE) * 2) |
415 | #define ANYOF_NVERTWS ((ANYOF_VERTWS)+1) | |
e1d1eefb | 416 | |
b2ef2e34 KW |
417 | /* It is best if this is the last one, as all above it are stored as bits in a |
418 | * bitmap, and it isn't part of that bitmap */ | |
419 | #if _CC_VERTSPACE != _HIGHEST_REGCOMP_DOT_H_SYNC | |
420 | # error Problem with handy.h _HIGHEST_REGCOMP_DOT_H_SYNC #define | |
421 | #endif | |
422 | ||
8efd3f97 KW |
423 | #define ANYOF_POSIXL_MAX (ANYOF_VERTWS) /* So upper loop limit is written: |
424 | * '< ANYOF_MAX' | |
425 | * Hence doesn't include VERTWS, as that | |
426 | * is a pseudo class */ | |
427 | #define ANYOF_MAX ANYOF_POSIXL_MAX | |
428 | ||
429 | #if (ANYOF_POSIXL_MAX > 32) /* Must fit in 32-bit word */ | |
b2ef2e34 | 430 | # error Problem with handy.h _CC_foo #defines |
a0947d7b KW |
431 | #endif |
432 | ||
8efd3f97 KW |
433 | #define ANYOF_HORIZWS ((ANYOF_POSIXL_MAX)+2) /* = (ANYOF_NVERTWS + 1) */ |
434 | #define ANYOF_NHORIZWS ((ANYOF_POSIXL_MAX)+3) | |
87cad4c0 | 435 | |
8efd3f97 KW |
436 | #define ANYOF_UNIPROP ((ANYOF_POSIXL_MAX)+4) /* Used to indicate a Unicode |
437 | property: \p{} or \P{} */ | |
f39a4dc2 | 438 | |
b8c5462f JH |
439 | /* Backward source code compatibility. */ |
440 | ||
441 | #define ANYOF_ALNUML ANYOF_ALNUM | |
442 | #define ANYOF_NALNUML ANYOF_NALNUM | |
443 | #define ANYOF_SPACEL ANYOF_SPACE | |
444 | #define ANYOF_NSPACEL ANYOF_NSPACE | |
8ebfc29a KW |
445 | #define ANYOF_ALNUM ANYOF_WORDCHAR |
446 | #define ANYOF_NALNUM ANYOF_NWORDCHAR | |
b8c5462f JH |
447 | |
448 | /* Utility macros for the bitmap and classes of ANYOF */ | |
449 | ||
936ed897 | 450 | #define ANYOF_SIZE (sizeof(struct regnode_charclass)) |
d1d040e5 | 451 | #define ANYOF_POSIXL_SIZE (sizeof(regnode_charclass_posixl)) |
8efd3f97 | 452 | #define ANYOF_CLASS_SIZE ANYOF_POSIXL_SIZE |
b8c5462f | 453 | |
936ed897 | 454 | #define ANYOF_FLAGS(p) ((p)->flags) |
b8c5462f JH |
455 | |
456 | #define ANYOF_BIT(c) (1 << ((c) & 7)) | |
457 | ||
d1d040e5 | 458 | #define ANYOF_POSIXL_SET(p, c) (((regnode_charclass_posixl*) (p))->classflags |= (1U << (c))) |
8efd3f97 KW |
459 | #define ANYOF_CLASS_SET(p, c) ANYOF_POSIXL_SET((p), (c)) |
460 | ||
d1d040e5 | 461 | #define ANYOF_POSIXL_CLEAR(p, c) (((regnode_charclass_posixl*) (p))->classflags &= ~ (1U <<(c))) |
8efd3f97 KW |
462 | #define ANYOF_CLASS_CLEAR(p, c) ANYOF_POSIXL_CLEAR((p), (c)) |
463 | ||
d1d040e5 | 464 | #define ANYOF_POSIXL_TEST(p, c) (((regnode_charclass_posixl*) (p))->classflags & (1U << (c))) |
8efd3f97 | 465 | #define ANYOF_CLASS_TEST(p, c) ANYOF_POSIXL_TEST((p), (c)) |
b8c5462f | 466 | |
d1d040e5 | 467 | #define ANYOF_POSIXL_ZERO(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = 0; } STMT_END |
8efd3f97 | 468 | #define ANYOF_CLASS_ZERO(ret) ANYOF_POSIXL_ZERO(ret) |
9cf78fa0 KW |
469 | |
470 | /* Shifts a bit to get, eg. 0x4000_0000, then subtracts 1 to get 0x3FFF_FFFF */ | |
d1d040e5 | 471 | #define ANYOF_POSIXL_SETALL(ret) STMT_START { ((regnode_charclass_posixl*) (ret))->classflags = ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1; } STMT_END |
8efd3f97 | 472 | #define ANYOF_CLASS_SETALL(ret) ANYOF_POSIXL_SETALL(ret) |
9cf78fa0 | 473 | |
8efd3f97 KW |
474 | #define ANYOF_POSIXL_TEST_ANY_SET(p) \ |
475 | ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \ | |
d1d040e5 | 476 | && (((regnode_charclass_posixl*)(p))->classflags)) |
8efd3f97 | 477 | #define ANYOF_CLASS_TEST_ANY_SET(p) ANYOF_POSIXL_TEST_ANY_SET(p) |
e867ce45 | 478 | |
fb38762f KW |
479 | #define ANYOF_POSIXL_TEST_ALL_SET(p) \ |
480 | ((ANYOF_FLAGS(p) & ANYOF_POSIXL) \ | |
481 | && ((regnode_charclass_posixl*) (p))->classflags == ((1U << ((ANYOF_POSIXL_MAX) - 1))) - 1) | |
482 | ||
b27fd97a | 483 | #define ANYOF_POSIXL_OR(source, dest) STMT_START { (dest)->classflags |= (source)->classflags ; } STMT_END |
8efd3f97 | 484 | #define ANYOF_CLASS_OR(source, dest) ANYOF_POSIXL_OR((source), (dest)) |
936ed897 | 485 | |
cdd87c1d KW |
486 | #define ANYOF_POSIXL_AND(source, dest) STMT_START { (dest)->classflags &= (source)->classflags ; } STMT_END |
487 | ||
9cf78fa0 | 488 | #define ANYOF_BITMAP_ZERO(ret) Zero(((struct regnode_charclass*)(ret))->bitmap, ANYOF_BITMAP_SIZE, char) |
936ed897 | 489 | #define ANYOF_BITMAP(p) (((struct regnode_charclass*)(p))->bitmap) |
786e8c11 | 490 | #define ANYOF_BITMAP_BYTE(p, c) (ANYOF_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
b8c5462f JH |
491 | #define ANYOF_BITMAP_SET(p, c) (ANYOF_BITMAP_BYTE(p, c) |= ANYOF_BIT(c)) |
492 | #define ANYOF_BITMAP_CLEAR(p,c) (ANYOF_BITMAP_BYTE(p, c) &= ~ANYOF_BIT(c)) | |
493 | #define ANYOF_BITMAP_TEST(p, c) (ANYOF_BITMAP_BYTE(p, c) & ANYOF_BIT(c)) | |
494 | ||
f8bef550 NC |
495 | #define ANYOF_BITMAP_SETALL(p) \ |
496 | memset (ANYOF_BITMAP(p), 255, ANYOF_BITMAP_SIZE) | |
497 | #define ANYOF_BITMAP_CLEARALL(p) \ | |
498 | Zero (ANYOF_BITMAP(p), ANYOF_BITMAP_SIZE) | |
499 | /* Check that all 256 bits are all set. Used in S_cl_is_anything() */ | |
c355c09a | 500 | #define ANYOF_BITMAP_TESTALLSET(p) /* Assumes sizeof(p) == 32 */ \ |
f8bef550 NC |
501 | 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) |
502 | ||
936ed897 | 503 | #define ANYOF_SKIP ((ANYOF_SIZE - 1)/sizeof(regnode)) |
8efd3f97 KW |
504 | #define ANYOF_POSIXL_SKIP ((ANYOF_POSIXL_SIZE - 1)/sizeof(regnode)) |
505 | #define ANYOF_CLASS_SKIP ANYOF_POSIXL_SKIP | |
3a15e693 | 506 | |
a687059c LW |
507 | /* |
508 | * Utility definitions. | |
509 | */ | |
2304df62 | 510 | #ifndef CHARMASK |
075abff3 | 511 | # define UCHARAT(p) ((int)*(const U8*)(p)) |
a687059c | 512 | #else |
075abff3 | 513 | # define UCHARAT(p) ((int)*(p)&CHARMASK) |
a687059c | 514 | #endif |
a687059c | 515 | |
c277df42 IZ |
516 | #define EXTRA_SIZE(guy) ((sizeof(guy)-1)/sizeof(struct regnode)) |
517 | ||
3dab1dad YO |
518 | #define REG_SEEN_ZERO_LEN 0x00000001 |
519 | #define REG_SEEN_LOOKBEHIND 0x00000002 | |
520 | #define REG_SEEN_GPOS 0x00000004 | |
471f5387 | 521 | /* spare */ |
3dab1dad | 522 | #define REG_SEEN_CANY 0x00000010 |
cce850e4 | 523 | #define REG_SEEN_SANY REG_SEEN_CANY /* src bckwrd cmpt */ |
6bda09f9 | 524 | #define REG_SEEN_RECURSE 0x00000020 |
eaf3ca90 | 525 | #define REG_TOP_LEVEL_BRANCHES 0x00000040 |
e2e6a0f1 | 526 | #define REG_SEEN_VERBARG 0x00000080 |
5d458dd8 | 527 | #define REG_SEEN_CUTGROUP 0x00000100 |
bcdf7404 | 528 | #define REG_SEEN_RUN_ON_COMMENT 0x00000200 |
bb914485 | 529 | #define REG_SEEN_EXACTF_SHARP_S 0x00000400 |
d09b2d29 | 530 | |
73c4f7a1 GS |
531 | START_EXTERN_C |
532 | ||
be8e71aa YO |
533 | #ifdef PLUGGABLE_RE_EXTENSION |
534 | #include "re_nodes.h" | |
535 | #else | |
d09b2d29 | 536 | #include "regnodes.h" |
be8e71aa | 537 | #endif |
d09b2d29 | 538 | |
f9f4320a YO |
539 | #ifndef PLUGGABLE_RE_EXTENSION |
540 | #ifndef DOINIT | |
541 | EXTCONST regexp_engine PL_core_reg_engine; | |
542 | #else /* DOINIT */ | |
543 | EXTCONST regexp_engine PL_core_reg_engine = { | |
fe578d7f | 544 | Perl_re_compile, |
2fdbfb4d | 545 | Perl_regexec_flags, |
f9f4320a YO |
546 | Perl_re_intuit_start, |
547 | Perl_re_intuit_string, | |
2fdbfb4d AB |
548 | Perl_regfree_internal, |
549 | Perl_reg_numbered_buff_fetch, | |
550 | Perl_reg_numbered_buff_store, | |
551 | Perl_reg_numbered_buff_length, | |
192b9cd1 AB |
552 | Perl_reg_named_buff, |
553 | Perl_reg_named_buff_iter, | |
49d7dfbc | 554 | Perl_reg_qr_package, |
f9f4320a | 555 | #if defined(USE_ITHREADS) |
3c13cae6 | 556 | Perl_regdupe_internal, |
f9f4320a | 557 | #endif |
3c13cae6 | 558 | Perl_re_op_compile |
f9f4320a YO |
559 | }; |
560 | #endif /* DOINIT */ | |
561 | #endif /* PLUGGABLE_RE_EXTENSION */ | |
562 | ||
563 | ||
73c4f7a1 | 564 | END_EXTERN_C |
cad2e5aa | 565 | |
cad2e5aa | 566 | |
261faec3 MJD |
567 | /* .what is a character array with one character for each member of .data |
568 | * The character describes the function of the corresponding .data item: | |
af534a04 | 569 | * a - AV for paren_name_list under DEBUGGING |
0111c4fd | 570 | * f - start-class data for regstclass optimization |
68e2671b | 571 | * l - start op for literal (?{EVAL}) item |
d63c20f2 | 572 | * L - start op for literal (?{EVAL}) item, with separate CV (qr//) |
b30fcab9 | 573 | * r - pointer to an embedded code-containing qr, e.g. /ab$qr/ |
38a44b82 | 574 | * s - swash for Unicode-style character class, and the multicharacter |
9e55ce06 JH |
575 | * strings resulting from casefolding the single-character entries |
576 | * in the character class | |
a3621e74 | 577 | * t - trie struct |
55eed653 | 578 | * u - trie struct's widecharmap (a HV, so can't share, must dup) |
2b8b4781 | 579 | * also used for revcharmap and words under DEBUGGING |
07be1b83 | 580 | * T - aho-trie struct |
81714fb9 | 581 | * S - sv for named capture lookup |
261faec3 | 582 | * 20010712 mjd@plover.com |
c3613ed7 | 583 | * (Remember to update re_dup() and pregfree() if you add any items.) |
261faec3 | 584 | */ |
cad2e5aa JH |
585 | struct reg_data { |
586 | U32 count; | |
587 | U8 *what; | |
588 | void* data[1]; | |
589 | }; | |
590 | ||
a1cac82e NC |
591 | /* Code in S_to_utf8_substr() and S_to_byte_substr() in regexec.c accesses |
592 | anchored* and float* via array indexes 0 and 1. */ | |
cad2e5aa | 593 | #define anchored_substr substrs->data[0].substr |
33b8afdf | 594 | #define anchored_utf8 substrs->data[0].utf8_substr |
cad2e5aa | 595 | #define anchored_offset substrs->data[0].min_offset |
1de06328 YO |
596 | #define anchored_end_shift substrs->data[0].end_shift |
597 | ||
cad2e5aa | 598 | #define float_substr substrs->data[1].substr |
33b8afdf | 599 | #define float_utf8 substrs->data[1].utf8_substr |
cad2e5aa JH |
600 | #define float_min_offset substrs->data[1].min_offset |
601 | #define float_max_offset substrs->data[1].max_offset | |
1de06328 YO |
602 | #define float_end_shift substrs->data[1].end_shift |
603 | ||
cad2e5aa | 604 | #define check_substr substrs->data[2].substr |
33b8afdf | 605 | #define check_utf8 substrs->data[2].utf8_substr |
cad2e5aa JH |
606 | #define check_offset_min substrs->data[2].min_offset |
607 | #define check_offset_max substrs->data[2].max_offset | |
1de06328 | 608 | #define check_end_shift substrs->data[2].end_shift |
a3621e74 | 609 | |
8d919b0a FC |
610 | #define RX_ANCHORED_SUBSTR(rx) (ReANY(rx)->anchored_substr) |
611 | #define RX_ANCHORED_UTF8(rx) (ReANY(rx)->anchored_utf8) | |
612 | #define RX_FLOAT_SUBSTR(rx) (ReANY(rx)->float_substr) | |
613 | #define RX_FLOAT_UTF8(rx) (ReANY(rx)->float_utf8) | |
a3621e74 YO |
614 | |
615 | /* trie related stuff */ | |
5d9a96ca | 616 | |
a3621e74 YO |
617 | /* a transition record for the state machine. the |
618 | check field determines which state "owns" the | |
619 | transition. the char the transition is for is | |
620 | determined by offset from the owning states base | |
621 | field. the next field determines which state | |
622 | is to be transitioned to if any. | |
623 | */ | |
624 | struct _reg_trie_trans { | |
625 | U32 next; | |
626 | U32 check; | |
627 | }; | |
628 | ||
629 | /* a transition list element for the list based representation */ | |
630 | struct _reg_trie_trans_list_elem { | |
631 | U16 forid; | |
632 | U32 newstate; | |
633 | }; | |
634 | typedef struct _reg_trie_trans_list_elem reg_trie_trans_le; | |
635 | ||
636 | /* a state for compressed nodes. base is an offset | |
637 | into an array of reg_trie_trans array. If wordnum is | |
638 | nonzero the state is accepting. if base is zero then | |
639 | the state has no children (and will be accepting) | |
640 | */ | |
641 | struct _reg_trie_state { | |
642 | U16 wordnum; | |
643 | union { | |
644 | U32 base; | |
645 | reg_trie_trans_le* list; | |
646 | } trans; | |
647 | }; | |
648 | ||
2e64971a DM |
649 | /* info per word; indexed by wordnum */ |
650 | typedef struct { | |
651 | U16 prev; /* previous word in acceptance chain; eg in | |
652 | * zzz|abc|ab/ after matching the chars abc, the | |
653 | * accepted word is #2, and the previous accepted | |
654 | * word is #3 */ | |
655 | U32 len; /* how many chars long is this word? */ | |
656 | U32 accept; /* accept state for this word */ | |
657 | } reg_trie_wordinfo; | |
a3621e74 YO |
658 | |
659 | ||
a3621e74 YO |
660 | typedef struct _reg_trie_state reg_trie_state; |
661 | typedef struct _reg_trie_trans reg_trie_trans; | |
662 | ||
663 | ||
664 | /* anything in here that needs to be freed later | |
23eab42c NC |
665 | should be dealt with in pregfree. |
666 | refcount is first in both this and _reg_ac_data to allow a space | |
667 | optimisation in Perl_regdupe. */ | |
a3621e74 | 668 | struct _reg_trie_data { |
23eab42c | 669 | U32 refcount; /* number of times this trie is referenced */ |
786e8c11 YO |
670 | U32 lasttrans; /* last valid transition element */ |
671 | U16 *charmap; /* byte to charid lookup array */ | |
786e8c11 YO |
672 | reg_trie_state *states; /* state data */ |
673 | reg_trie_trans *trans; /* array of transition elements */ | |
674 | char *bitmap; /* stclass bitmap */ | |
786e8c11 YO |
675 | U16 *jump; /* optional 1 indexed array of offsets before tail |
676 | for the node following a given word. */ | |
2e64971a | 677 | reg_trie_wordinfo *wordinfo; /* array of info per word */ |
1b5c067c NC |
678 | U16 uniquecharcount; /* unique chars in trie (width of trans table) */ |
679 | U32 startstate; /* initial state - used for common prefix optimisation */ | |
680 | STRLEN minlen; /* minimum length of words in trie - build/opt only? */ | |
681 | STRLEN maxlen; /* maximum length of words in trie - build/opt only? */ | |
2e64971a | 682 | U32 prefixlen; /* #chars in common prefix */ |
1e2e3d02 YO |
683 | U32 statecount; /* Build only - number of states in the states array |
684 | (including the unused zero state) */ | |
786e8c11 | 685 | U32 wordcount; /* Build only */ |
a3621e74 | 686 | #ifdef DEBUGGING |
786e8c11 | 687 | STRLEN charcount; /* Build only */ |
a3621e74 YO |
688 | #endif |
689 | }; | |
2b8b4781 NC |
690 | /* There is one (3 under DEBUGGING) pointers that logically belong in this |
691 | structure, but are held outside as they need duplication on thread cloning, | |
692 | whereas the rest of the structure can be read only: | |
693 | HV *widecharmap; code points > 255 to charid | |
694 | #ifdef DEBUGGING | |
695 | AV *words; Array of words contained in trie, for dumping | |
696 | AV *revcharmap; Map of each charid back to its character representation | |
697 | #endif | |
698 | */ | |
699 | ||
700 | #define TRIE_WORDS_OFFSET 2 | |
701 | ||
a3621e74 YO |
702 | typedef struct _reg_trie_data reg_trie_data; |
703 | ||
23eab42c NC |
704 | /* refcount is first in both this and _reg_trie_data to allow a space |
705 | optimisation in Perl_regdupe. */ | |
07be1b83 | 706 | struct _reg_ac_data { |
23eab42c | 707 | U32 refcount; |
1b5c067c | 708 | U32 trie; |
07be1b83 YO |
709 | U32 *fail; |
710 | reg_trie_state *states; | |
07be1b83 YO |
711 | }; |
712 | typedef struct _reg_ac_data reg_ac_data; | |
713 | ||
486ec47a | 714 | /* ANY_BIT doesn't use the structure, so we can borrow it here. |
3dab1dad YO |
715 | This is simpler than refactoring all of it as wed end up with |
716 | three different sets... */ | |
717 | ||
718 | #define TRIE_BITMAP(p) (((reg_trie_data *)(p))->bitmap) | |
786e8c11 | 719 | #define TRIE_BITMAP_BYTE(p, c) (TRIE_BITMAP(p)[(((U8)(c)) >> 3) & 31]) |
5dfe062f YO |
720 | #define TRIE_BITMAP_SET(p, c) (TRIE_BITMAP_BYTE(p, c) |= ANYOF_BIT((U8)c)) |
721 | #define TRIE_BITMAP_CLEAR(p,c) (TRIE_BITMAP_BYTE(p, c) &= ~ANYOF_BIT((U8)c)) | |
722 | #define TRIE_BITMAP_TEST(p, c) (TRIE_BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 723 | |
1de06328 YO |
724 | #define IS_ANYOF_TRIE(op) ((op)==TRIEC || (op)==AHOCORASICKC) |
725 | #define IS_TRIE_AC(op) ((op)>=AHOCORASICK) | |
726 | ||
727 | ||
786e8c11 YO |
728 | #define BITMAP_BYTE(p, c) (((U8*)p)[(((U8)(c)) >> 3) & 31]) |
729 | #define BITMAP_TEST(p, c) (BITMAP_BYTE(p, c) & ANYOF_BIT((U8)c)) | |
3dab1dad | 730 | |
a3621e74 YO |
731 | /* these defines assume uniquecharcount is the correct variable, and state may be evaluated twice */ |
732 | #define TRIE_NODENUM(state) (((state)-1)/(trie->uniquecharcount)+1) | |
733 | #define SAFE_TRIE_NODENUM(state) ((state) ? (((state)-1)/(trie->uniquecharcount)+1) : (state)) | |
734 | #define TRIE_NODEIDX(state) ((state) ? (((state)-1)*(trie->uniquecharcount)+1) : (state)) | |
735 | ||
3dab1dad | 736 | #ifdef DEBUGGING |
3dab1dad | 737 | #define TRIE_CHARCOUNT(trie) ((trie)->charcount) |
3dab1dad | 738 | #else |
3dab1dad | 739 | #define TRIE_CHARCOUNT(trie) (trie_charcount) |
3dab1dad | 740 | #endif |
a3621e74 | 741 | |
0111c4fd RGS |
742 | #define RE_TRIE_MAXBUF_INIT 65536 |
743 | #define RE_TRIE_MAXBUF_NAME "\022E_TRIE_MAXBUF" | |
a3621e74 YO |
744 | #define RE_DEBUG_FLAGS "\022E_DEBUG_FLAGS" |
745 | ||
be8e71aa YO |
746 | /* |
747 | ||
748 | RE_DEBUG_FLAGS is used to control what debug output is emitted | |
749 | its divided into three groups of options, some of which interact. | |
750 | The three groups are: Compile, Execute, Extra. There is room for a | |
751 | further group, as currently only the low three bytes are used. | |
752 | ||
753 | Compile Options: | |
07be1b83 | 754 | |
be8e71aa YO |
755 | PARSE |
756 | PEEP | |
757 | TRIE | |
758 | PROGRAM | |
759 | OFFSETS | |
a3621e74 | 760 | |
be8e71aa | 761 | Execute Options: |
a3621e74 | 762 | |
be8e71aa YO |
763 | INTUIT |
764 | MATCH | |
765 | TRIE | |
a3621e74 | 766 | |
be8e71aa | 767 | Extra Options |
a3621e74 | 768 | |
be8e71aa YO |
769 | TRIE |
770 | OFFSETS | |
771 | ||
772 | If you modify any of these make sure you make corresponding changes to | |
773 | re.pm, especially to the documentation. | |
a3621e74 | 774 | |
be8e71aa YO |
775 | */ |
776 | ||
777 | ||
778 | /* Compile */ | |
779 | #define RE_DEBUG_COMPILE_MASK 0x0000FF | |
780 | #define RE_DEBUG_COMPILE_PARSE 0x000001 | |
781 | #define RE_DEBUG_COMPILE_OPTIMISE 0x000002 | |
782 | #define RE_DEBUG_COMPILE_TRIE 0x000004 | |
783 | #define RE_DEBUG_COMPILE_DUMP 0x000008 | |
f7819f85 | 784 | #define RE_DEBUG_COMPILE_FLAGS 0x000010 |
be8e71aa YO |
785 | |
786 | /* Execute */ | |
787 | #define RE_DEBUG_EXECUTE_MASK 0x00FF00 | |
788 | #define RE_DEBUG_EXECUTE_INTUIT 0x000100 | |
789 | #define RE_DEBUG_EXECUTE_MATCH 0x000200 | |
790 | #define RE_DEBUG_EXECUTE_TRIE 0x000400 | |
791 | ||
792 | /* Extra */ | |
793 | #define RE_DEBUG_EXTRA_MASK 0xFF0000 | |
794 | #define RE_DEBUG_EXTRA_TRIE 0x010000 | |
795 | #define RE_DEBUG_EXTRA_OFFSETS 0x020000 | |
a5ca303d YO |
796 | #define RE_DEBUG_EXTRA_OFFDEBUG 0x040000 |
797 | #define RE_DEBUG_EXTRA_STATE 0x080000 | |
798 | #define RE_DEBUG_EXTRA_OPTIMISE 0x100000 | |
e7707071 | 799 | #define RE_DEBUG_EXTRA_BUFFERS 0x400000 |
2c296965 | 800 | #define RE_DEBUG_EXTRA_GPOS 0x800000 |
24b23f37 YO |
801 | /* combined */ |
802 | #define RE_DEBUG_EXTRA_STACK 0x280000 | |
be8e71aa | 803 | |
e68ec53f | 804 | #define RE_DEBUG_FLAG(x) (re_debug_flags & x) |
be8e71aa YO |
805 | /* Compile */ |
806 | #define DEBUG_COMPILE_r(x) DEBUG_r( \ | |
e68ec53f | 807 | if (re_debug_flags & RE_DEBUG_COMPILE_MASK) x ) |
be8e71aa | 808 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 809 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 810 | #define DEBUG_OPTIMISE_r(x) DEBUG_r( \ |
e68ec53f | 811 | if (re_debug_flags & RE_DEBUG_COMPILE_OPTIMISE) x ) |
be8e71aa | 812 | #define DEBUG_PARSE_r(x) DEBUG_r( \ |
e68ec53f | 813 | if (re_debug_flags & RE_DEBUG_COMPILE_PARSE) x ) |
be8e71aa | 814 | #define DEBUG_DUMP_r(x) DEBUG_r( \ |
e68ec53f | 815 | if (re_debug_flags & RE_DEBUG_COMPILE_DUMP) x ) |
be8e71aa | 816 | #define DEBUG_TRIE_COMPILE_r(x) DEBUG_r( \ |
e68ec53f | 817 | if (re_debug_flags & RE_DEBUG_COMPILE_TRIE) x ) |
f7819f85 A |
818 | #define DEBUG_FLAGS_r(x) DEBUG_r( \ |
819 | if (re_debug_flags & RE_DEBUG_COMPILE_FLAGS) x ) | |
be8e71aa YO |
820 | /* Execute */ |
821 | #define DEBUG_EXECUTE_r(x) DEBUG_r( \ | |
e68ec53f | 822 | if (re_debug_flags & RE_DEBUG_EXECUTE_MASK) x ) |
be8e71aa | 823 | #define DEBUG_INTUIT_r(x) DEBUG_r( \ |
e68ec53f | 824 | if (re_debug_flags & RE_DEBUG_EXECUTE_INTUIT) x ) |
be8e71aa | 825 | #define DEBUG_MATCH_r(x) DEBUG_r( \ |
e68ec53f | 826 | if (re_debug_flags & RE_DEBUG_EXECUTE_MATCH) x ) |
be8e71aa | 827 | #define DEBUG_TRIE_EXECUTE_r(x) DEBUG_r( \ |
e68ec53f | 828 | if (re_debug_flags & RE_DEBUG_EXECUTE_TRIE) x ) |
be8e71aa YO |
829 | |
830 | /* Extra */ | |
831 | #define DEBUG_EXTRA_r(x) DEBUG_r( \ | |
e68ec53f | 832 | if (re_debug_flags & RE_DEBUG_EXTRA_MASK) x ) |
a5ca303d YO |
833 | #define DEBUG_OFFSETS_r(x) DEBUG_r( \ |
834 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFSETS) x ) | |
ab3bbdeb YO |
835 | #define DEBUG_STATE_r(x) DEBUG_r( \ |
836 | if (re_debug_flags & RE_DEBUG_EXTRA_STATE) x ) | |
24b23f37 YO |
837 | #define DEBUG_STACK_r(x) DEBUG_r( \ |
838 | if (re_debug_flags & RE_DEBUG_EXTRA_STACK) x ) | |
e7707071 YO |
839 | #define DEBUG_BUFFERS_r(x) DEBUG_r( \ |
840 | if (re_debug_flags & RE_DEBUG_EXTRA_BUFFERS) x ) | |
841 | ||
a5ca303d YO |
842 | #define DEBUG_OPTIMISE_MORE_r(x) DEBUG_r( \ |
843 | if ((RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE) == \ | |
844 | (re_debug_flags & (RE_DEBUG_EXTRA_OPTIMISE|RE_DEBUG_COMPILE_OPTIMISE)) ) x ) | |
07be1b83 | 845 | #define MJD_OFFSET_DEBUG(x) DEBUG_r( \ |
a5ca303d | 846 | if (re_debug_flags & RE_DEBUG_EXTRA_OFFDEBUG) \ |
be8e71aa YO |
847 | Perl_warn_nocontext x ) |
848 | #define DEBUG_TRIE_COMPILE_MORE_r(x) DEBUG_TRIE_COMPILE_r( \ | |
e68ec53f | 849 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa | 850 | #define DEBUG_TRIE_EXECUTE_MORE_r(x) DEBUG_TRIE_EXECUTE_r( \ |
e68ec53f | 851 | if (re_debug_flags & RE_DEBUG_EXTRA_TRIE) x ) |
be8e71aa YO |
852 | |
853 | #define DEBUG_TRIE_r(x) DEBUG_r( \ | |
e68ec53f | 854 | if (re_debug_flags & (RE_DEBUG_COMPILE_TRIE \ |
be8e71aa | 855 | | RE_DEBUG_EXECUTE_TRIE )) x ) |
2c296965 YO |
856 | #define DEBUG_GPOS_r(x) DEBUG_r( \ |
857 | if (re_debug_flags & RE_DEBUG_EXTRA_GPOS) x ) | |
be8e71aa YO |
858 | |
859 | /* initialization */ | |
d2d8c711 | 860 | /* get_sv() can return NULL during global destruction. */ |
e68ec53f YO |
861 | #define GET_RE_DEBUG_FLAGS DEBUG_r({ \ |
862 | SV * re_debug_flags_sv = NULL; \ | |
863 | re_debug_flags_sv = get_sv(RE_DEBUG_FLAGS, 1); \ | |
864 | if (re_debug_flags_sv) { \ | |
8e9a8a48 | 865 | if (!SvIOK(re_debug_flags_sv)) \ |
e68ec53f | 866 | sv_setuv(re_debug_flags_sv, RE_DEBUG_COMPILE_DUMP | RE_DEBUG_EXECUTE_MASK ); \ |
8e9a8a48 | 867 | re_debug_flags=SvIV(re_debug_flags_sv); \ |
e68ec53f YO |
868 | }\ |
869 | }) | |
a3621e74 YO |
870 | |
871 | #ifdef DEBUGGING | |
ab3bbdeb | 872 | |
cb65b687 DM |
873 | #define GET_RE_DEBUG_FLAGS_DECL VOL IV re_debug_flags = 0; \ |
874 | PERL_UNUSED_VAR(re_debug_flags); GET_RE_DEBUG_FLAGS; | |
ab3bbdeb YO |
875 | |
876 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) \ | |
877 | const char * const rpv = \ | |
ddc5bc0f | 878 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb | 879 | PL_colors[(c1)],PL_colors[(c2)], \ |
c89df6cf | 880 | PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ); \ |
0df25f3d | 881 | const int rlen = SvCUR(dsv) |
ab3bbdeb YO |
882 | |
883 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) \ | |
884 | const char * const rpv = \ | |
ddc5bc0f | 885 | pv_pretty((dsv), (SvPV_nolen_const(sv)), (SvCUR(sv)), (m), \ |
ab3bbdeb | 886 | PL_colors[(c1)],PL_colors[(c2)], \ |
c89df6cf | 887 | PERL_PV_ESCAPE_RE|PERL_PV_ESCAPE_NONASCII |((isuni) ? PERL_PV_ESCAPE_UNI : 0) ) |
ab3bbdeb YO |
888 | |
889 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) \ | |
890 | const char * const rpv = \ | |
ddc5bc0f | 891 | pv_pretty((dsv), (pv), (l), (m), \ |
ab3bbdeb | 892 | PL_colors[0], PL_colors[1], \ |
c89df6cf | 893 | ( PERL_PV_PRETTY_QUOTE | PERL_PV_ESCAPE_RE | PERL_PV_ESCAPE_NONASCII | PERL_PV_PRETTY_ELLIPSES | \ |
ab3bbdeb | 894 | ((isuni) ? PERL_PV_ESCAPE_UNI : 0)) \ |
0a0e9afc | 895 | ) |
ab3bbdeb YO |
896 | |
897 | #define RE_SV_DUMPLEN(ItEm) (SvCUR(ItEm) - (SvTAIL(ItEm)!=0)) | |
898 | #define RE_SV_TAIL(ItEm) (SvTAIL(ItEm) ? "$" : "") | |
899 | ||
900 | #else /* if not DEBUGGING */ | |
901 | ||
a3621e74 | 902 | #define GET_RE_DEBUG_FLAGS_DECL |
ab3bbdeb YO |
903 | #define RE_PV_COLOR_DECL(rpv,rlen,isuni,dsv,pv,l,m,c1,c2) |
904 | #define RE_SV_ESCAPE(rpv,isuni,dsv,sv,m) | |
905 | #define RE_PV_QUOTED_DECL(rpv,isuni,dsv,pv,l,m) | |
906 | #define RE_SV_DUMPLEN(ItEm) | |
907 | #define RE_SV_TAIL(ItEm) | |
908 | ||
909 | #endif /* DEBUG RELATED DEFINES */ | |
a3621e74 | 910 | |
e9a8c099 MHM |
911 | /* |
912 | * Local variables: | |
913 | * c-indentation-style: bsd | |
914 | * c-basic-offset: 4 | |
14d04a33 | 915 | * indent-tabs-mode: nil |
e9a8c099 MHM |
916 | * End: |
917 | * | |
14d04a33 | 918 | * ex: set ts=8 sts=4 sw=4 et: |
e9a8c099 | 919 | */ |