This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.sym: Add ANYOFV node
[perl5.git] / regnodes.h
CommitLineData
37442d52
RGS
1/* -*- buffer-read-only: t -*-
2 !!!!!!! DO NOT EDIT THIS FILE !!!!!!!
eb2624c9 3 This file is built by regen/regcomp.pl from regcomp.sym.
d09b2d29
IZ
4 Any changes made here will be lost!
5*/
6
6bda09f9
YO
7/* Regops and State definitions */
8
0e019ad6
KW
9#define REGNODE_MAX 94
10#define REGMATCH_STATE_MAX 134
03363afd 11
f9f4320a
YO
12#define END 0 /* 0000 End of program. */
13#define SUCCEED 1 /* 0x01 Return from a subroutine, basically. */
14#define BOL 2 /* 0x02 Match "" at beginning of line. */
15#define MBOL 3 /* 0x03 Same, assuming multiline. */
16#define SBOL 4 /* 0x04 Same, assuming singleline. */
17#define EOS 5 /* 0x05 Match "" at end of string. */
18#define EOL 6 /* 0x06 Match "" at end of line. */
19#define MEOL 7 /* 0x07 Same, assuming multiline. */
20#define SEOL 8 /* 0x08 Same, assuming singleline. */
21#define BOUND 9 /* 0x09 Match "" at any word boundary */
22#define BOUNDL 10 /* 0x0a Match "" at any word boundary */
23#define NBOUND 11 /* 0x0b Match "" at any word non-boundary */
24#define NBOUNDL 12 /* 0x0c Match "" at any word non-boundary */
25#define GPOS 13 /* 0x0d Matches where last m//g left off. */
26#define REG_ANY 14 /* 0x0e Match any one character (except newline). */
27#define SANY 15 /* 0x0f Match any one character. */
28#define CANY 16 /* 0x10 Match any one byte. */
0e019ad6
KW
29#define ANYOF 17 /* 0x11 Match character in (or not in) this class, single char match only */
30#define ANYOFV 18 /* 0x12 Match character in (or not in) this class, can match-multiple chars */
31#define ALNUM 19 /* 0x13 Match any alphanumeric character */
32#define ALNUML 20 /* 0x14 Match any alphanumeric char in locale */
33#define NALNUM 21 /* 0x15 Match any non-alphanumeric character */
34#define NALNUML 22 /* 0x16 Match any non-alphanumeric char in locale */
35#define SPACE 23 /* 0x17 Match any whitespace character */
36#define SPACEL 24 /* 0x18 Match any whitespace char in locale */
37#define NSPACE 25 /* 0x19 Match any non-whitespace character */
38#define NSPACEL 26 /* 0x1a Match any non-whitespace char in locale */
39#define DIGIT 27 /* 0x1b Match any numeric character */
40#define DIGITL 28 /* 0x1c Match any numeric character in locale */
41#define NDIGIT 29 /* 0x1d Match any non-numeric character */
42#define NDIGITL 30 /* 0x1e Match any non-numeric character in locale */
43#define CLUMP 31 /* 0x1f Match any extended grapheme cluster sequence */
44#define BRANCH 32 /* 0x20 Match this alternative, or the next... */
45#define BACK 33 /* 0x21 Match "", "next" ptr points backward. */
46#define EXACT 34 /* 0x22 Match this string (preceded by length). */
47#define EXACTF 35 /* 0x23 Match this string, folded, native charset semantics for non-utf8 (prec. by length). */
48#define EXACTFL 36 /* 0x24 Match this string, folded in locale (w/len). */
49#define EXACTFU 37 /* 0x25 Match this string, folded, Unicode semantics for non-utf8 (prec. by length). */
50#define NOTHING 38 /* 0x26 Match empty string. */
51#define TAIL 39 /* 0x27 Match empty string. Can jump here from outside. */
52#define STAR 40 /* 0x28 Match this (simple) thing 0 or more times. */
53#define PLUS 41 /* 0x29 Match this (simple) thing 1 or more times. */
54#define CURLY 42 /* 0x2a Match this simple thing {n,m} times. */
55#define CURLYN 43 /* 0x2b Capture next-after-this simple thing */
56#define CURLYM 44 /* 0x2c Capture this medium-complex thing {n,m} times. */
57#define CURLYX 45 /* 0x2d Match this complex thing {n,m} times. */
58#define WHILEM 46 /* 0x2e Do curly processing and see if rest matches. */
59#define OPEN 47 /* 0x2f Mark this point in input as start of */
60#define CLOSE 48 /* 0x30 Analogous to OPEN. */
61#define REF 49 /* 0x31 Match some already matched string */
62#define REFF 50 /* 0x32 Match already matched string, folded using native charset semantics for non-utf8 */
63#define REFFL 51 /* 0x33 Match already matched string, folded in loc. */
64#define REFFU 52 /* 0x34 Match already matched string, folded using unicode semantics for non-utf8 */
65#define NREF 53 /* 0x35 Match some already matched string */
66#define NREFF 54 /* 0x36 Match already matched string, folded using native charset semantics for non-utf8 */
67#define NREFFL 55 /* 0x37 Match already matched string, folded in loc. */
68#define NREFFU 56 /* 0x38 Match already matched string, folded using unicode semantics for non-utf8 */
69#define IFMATCH 57 /* 0x39 Succeeds if the following matches. */
70#define UNLESSM 58 /* 0x3a Fails if the following matches. */
71#define SUSPEND 59 /* 0x3b "Independent" sub-RE. */
72#define IFTHEN 60 /* 0x3c Switch, should be preceded by switcher . */
73#define GROUPP 61 /* 0x3d Whether the group matched. */
74#define LONGJMP 62 /* 0x3e Jump far away. */
75#define BRANCHJ 63 /* 0x3f BRANCH with long offset. */
76#define EVAL 64 /* 0x40 Execute some Perl code. */
77#define MINMOD 65 /* 0x41 Next operator is not greedy. */
78#define LOGICAL 66 /* 0x42 Next opcode should set the flag only. */
79#define RENUM 67 /* 0x43 Group with independently numbered parens. */
80#define TRIE 68 /* 0x44 Match many EXACT(F[LU]?)? at once. flags==type */
81#define TRIEC 69 /* 0x45 Same as TRIE, but with embedded charclass data */
82#define AHOCORASICK 70 /* 0x46 Aho Corasick stclass. flags==type */
83#define AHOCORASICKC 71 /* 0x47 Same as AHOCORASICK, but with embedded charclass data */
84#define GOSUB 72 /* 0x48 recurse to paren arg1 at (signed) ofs arg2 */
85#define GOSTART 73 /* 0x49 recurse to start of pattern */
86#define NGROUPP 74 /* 0x4a Whether the group matched. */
87#define INSUBP 75 /* 0x4b Whether we are in a specific recurse. */
88#define DEFINEP 76 /* 0x4c Never execute directly. */
89#define ENDLIKE 77 /* 0x4d Used only for the type field of verbs */
90#define OPFAIL 78 /* 0x4e Same as (?!) */
91#define ACCEPT 79 /* 0x4f Accepts the current matched string. */
92#define VERB 80 /* 0x50 Used only for the type field of verbs */
93#define PRUNE 81 /* 0x51 Pattern fails at this startpoint if no-backtracking through this */
94#define MARKPOINT 82 /* 0x52 Push the current location for rollback by cut. */
95#define SKIP 83 /* 0x53 On failure skip forward (to the mark) before retrying */
96#define COMMIT 84 /* 0x54 Pattern fails outright if backtracking through this */
97#define CUTGROUP 85 /* 0x55 On failure go to the next alternation in the group */
98#define KEEPS 86 /* 0x56 $& begins here. */
99#define LNBREAK 87 /* 0x57 generic newline pattern */
100#define VERTWS 88 /* 0x58 vertical whitespace (Perl 6) */
101#define NVERTWS 89 /* 0x59 not vertical whitespace (Perl 6) */
102#define HORIZWS 90 /* 0x5a horizontal whitespace (Perl 6) */
103#define NHORIZWS 91 /* 0x5b not horizontal whitespace (Perl 6) */
104#define FOLDCHAR 92 /* 0x5c codepoint with tricky case folding properties. */
105#define OPTIMIZED 93 /* 0x5d Placeholder for dump. */
106#define PSEUDO 94 /* 0x5e Pseudo opcode for internal use. */
03363afd 107 /* ------------ States ------------- */
24b23f37
YO
108#define TRIE_next (REGNODE_MAX + 1) /* state for TRIE */
109#define TRIE_next_fail (REGNODE_MAX + 2) /* state for TRIE */
110#define EVAL_AB (REGNODE_MAX + 3) /* state for EVAL */
111#define EVAL_AB_fail (REGNODE_MAX + 4) /* state for EVAL */
112#define CURLYX_end (REGNODE_MAX + 5) /* state for CURLYX */
113#define CURLYX_end_fail (REGNODE_MAX + 6) /* state for CURLYX */
114#define WHILEM_A_pre (REGNODE_MAX + 7) /* state for WHILEM */
115#define WHILEM_A_pre_fail (REGNODE_MAX + 8) /* state for WHILEM */
116#define WHILEM_A_min (REGNODE_MAX + 9) /* state for WHILEM */
117#define WHILEM_A_min_fail (REGNODE_MAX + 10) /* state for WHILEM */
118#define WHILEM_A_max (REGNODE_MAX + 11) /* state for WHILEM */
119#define WHILEM_A_max_fail (REGNODE_MAX + 12) /* state for WHILEM */
120#define WHILEM_B_min (REGNODE_MAX + 13) /* state for WHILEM */
121#define WHILEM_B_min_fail (REGNODE_MAX + 14) /* state for WHILEM */
122#define WHILEM_B_max (REGNODE_MAX + 15) /* state for WHILEM */
123#define WHILEM_B_max_fail (REGNODE_MAX + 16) /* state for WHILEM */
124#define BRANCH_next (REGNODE_MAX + 17) /* state for BRANCH */
125#define BRANCH_next_fail (REGNODE_MAX + 18) /* state for BRANCH */
126#define CURLYM_A (REGNODE_MAX + 19) /* state for CURLYM */
127#define CURLYM_A_fail (REGNODE_MAX + 20) /* state for CURLYM */
128#define CURLYM_B (REGNODE_MAX + 21) /* state for CURLYM */
129#define CURLYM_B_fail (REGNODE_MAX + 22) /* state for CURLYM */
130#define IFMATCH_A (REGNODE_MAX + 23) /* state for IFMATCH */
131#define IFMATCH_A_fail (REGNODE_MAX + 24) /* state for IFMATCH */
132#define CURLY_B_min_known (REGNODE_MAX + 25) /* state for CURLY */
133#define CURLY_B_min_known_fail (REGNODE_MAX + 26) /* state for CURLY */
134#define CURLY_B_min (REGNODE_MAX + 27) /* state for CURLY */
135#define CURLY_B_min_fail (REGNODE_MAX + 28) /* state for CURLY */
136#define CURLY_B_max (REGNODE_MAX + 29) /* state for CURLY */
137#define CURLY_B_max_fail (REGNODE_MAX + 30) /* state for CURLY */
138#define COMMIT_next (REGNODE_MAX + 31) /* state for COMMIT */
139#define COMMIT_next_fail (REGNODE_MAX + 32) /* state for COMMIT */
e2e6a0f1
YO
140#define MARKPOINT_next (REGNODE_MAX + 33) /* state for MARKPOINT */
141#define MARKPOINT_next_fail (REGNODE_MAX + 34) /* state for MARKPOINT */
5d458dd8
YO
142#define SKIP_next (REGNODE_MAX + 35) /* state for SKIP */
143#define SKIP_next_fail (REGNODE_MAX + 36) /* state for SKIP */
144#define CUTGROUP_next (REGNODE_MAX + 37) /* state for CUTGROUP */
145#define CUTGROUP_next_fail (REGNODE_MAX + 38) /* state for CUTGROUP */
ee9b8eae
YO
146#define KEEPS_next (REGNODE_MAX + 39) /* state for KEEPS */
147#define KEEPS_next_fail (REGNODE_MAX + 40) /* state for KEEPS */
03363afd 148
6bda09f9 149/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
150
151#ifndef DOINIT
22c35a8c 152EXTCONST U8 PL_regkind[];
d09b2d29 153#else
22c35a8c 154EXTCONST U8 PL_regkind[] = {
e2e6a0f1
YO
155 END, /* END */
156 END, /* SUCCEED */
157 BOL, /* BOL */
158 BOL, /* MBOL */
159 BOL, /* SBOL */
160 EOL, /* EOS */
161 EOL, /* EOL */
162 EOL, /* MEOL */
163 EOL, /* SEOL */
164 BOUND, /* BOUND */
165 BOUND, /* BOUNDL */
166 NBOUND, /* NBOUND */
167 NBOUND, /* NBOUNDL */
168 GPOS, /* GPOS */
169 REG_ANY, /* REG_ANY */
170 REG_ANY, /* SANY */
171 REG_ANY, /* CANY */
172 ANYOF, /* ANYOF */
0e019ad6 173 ANYOF, /* ANYOFV */
e2e6a0f1
YO
174 ALNUM, /* ALNUM */
175 ALNUM, /* ALNUML */
176 NALNUM, /* NALNUM */
177 NALNUM, /* NALNUML */
178 SPACE, /* SPACE */
179 SPACE, /* SPACEL */
180 NSPACE, /* NSPACE */
181 NSPACE, /* NSPACEL */
182 DIGIT, /* DIGIT */
183 DIGIT, /* DIGITL */
184 NDIGIT, /* NDIGIT */
185 NDIGIT, /* NDIGITL */
186 CLUMP, /* CLUMP */
187 BRANCH, /* BRANCH */
188 BACK, /* BACK */
189 EXACT, /* EXACT */
190 EXACT, /* EXACTF */
191 EXACT, /* EXACTFL */
01f98ec2 192 EXACT, /* EXACTFU */
e2e6a0f1
YO
193 NOTHING, /* NOTHING */
194 NOTHING, /* TAIL */
195 STAR, /* STAR */
196 PLUS, /* PLUS */
197 CURLY, /* CURLY */
198 CURLY, /* CURLYN */
199 CURLY, /* CURLYM */
200 CURLY, /* CURLYX */
201 WHILEM, /* WHILEM */
202 OPEN, /* OPEN */
203 CLOSE, /* CLOSE */
204 REF, /* REF */
205 REF, /* REFF */
206 REF, /* REFFL */
01f98ec2
KW
207 REF, /* REFFU */
208 REF, /* NREF */
209 REF, /* NREFF */
210 REF, /* NREFFL */
211 REF, /* NREFFU */
e2e6a0f1
YO
212 BRANCHJ, /* IFMATCH */
213 BRANCHJ, /* UNLESSM */
214 BRANCHJ, /* SUSPEND */
215 BRANCHJ, /* IFTHEN */
216 GROUPP, /* GROUPP */
217 LONGJMP, /* LONGJMP */
218 BRANCHJ, /* BRANCHJ */
219 EVAL, /* EVAL */
220 MINMOD, /* MINMOD */
221 LOGICAL, /* LOGICAL */
222 BRANCHJ, /* RENUM */
223 TRIE, /* TRIE */
224 TRIE, /* TRIEC */
225 TRIE, /* AHOCORASICK */
226 TRIE, /* AHOCORASICKC */
227 GOSUB, /* GOSUB */
228 GOSTART, /* GOSTART */
e2e6a0f1
YO
229 NGROUPP, /* NGROUPP */
230 INSUBP, /* INSUBP */
231 DEFINEP, /* DEFINEP */
232 ENDLIKE, /* ENDLIKE */
233 ENDLIKE, /* OPFAIL */
234 ENDLIKE, /* ACCEPT */
235 VERB, /* VERB */
5d458dd8 236 VERB, /* PRUNE */
e2e6a0f1 237 VERB, /* MARKPOINT */
5d458dd8 238 VERB, /* SKIP */
e2e6a0f1 239 VERB, /* COMMIT */
5d458dd8 240 VERB, /* CUTGROUP */
ee9b8eae 241 KEEPS, /* KEEPS */
e1d1eefb
YO
242 LNBREAK, /* LNBREAK */
243 VERTWS, /* VERTWS */
244 NVERTWS, /* NVERTWS */
245 HORIZWS, /* HORIZWS */
246 NHORIZWS, /* NHORIZWS */
32e6a07c 247 FOLDCHAR, /* FOLDCHAR */
e2e6a0f1
YO
248 NOTHING, /* OPTIMIZED */
249 PSEUDO, /* PSEUDO */
03363afd 250 /* ------------ States ------------- */
e2e6a0f1
YO
251 TRIE, /* TRIE_next */
252 TRIE, /* TRIE_next_fail */
253 EVAL, /* EVAL_AB */
254 EVAL, /* EVAL_AB_fail */
255 CURLYX, /* CURLYX_end */
256 CURLYX, /* CURLYX_end_fail */
257 WHILEM, /* WHILEM_A_pre */
258 WHILEM, /* WHILEM_A_pre_fail */
259 WHILEM, /* WHILEM_A_min */
260 WHILEM, /* WHILEM_A_min_fail */
261 WHILEM, /* WHILEM_A_max */
262 WHILEM, /* WHILEM_A_max_fail */
263 WHILEM, /* WHILEM_B_min */
264 WHILEM, /* WHILEM_B_min_fail */
265 WHILEM, /* WHILEM_B_max */
266 WHILEM, /* WHILEM_B_max_fail */
267 BRANCH, /* BRANCH_next */
268 BRANCH, /* BRANCH_next_fail */
269 CURLYM, /* CURLYM_A */
270 CURLYM, /* CURLYM_A_fail */
271 CURLYM, /* CURLYM_B */
272 CURLYM, /* CURLYM_B_fail */
273 IFMATCH, /* IFMATCH_A */
274 IFMATCH, /* IFMATCH_A_fail */
275 CURLY, /* CURLY_B_min_known */
276 CURLY, /* CURLY_B_min_known_fail */
277 CURLY, /* CURLY_B_min */
278 CURLY, /* CURLY_B_min_fail */
279 CURLY, /* CURLY_B_max */
280 CURLY, /* CURLY_B_max_fail */
281 COMMIT, /* COMMIT_next */
282 COMMIT, /* COMMIT_next_fail */
283 MARKPOINT, /* MARKPOINT_next */
284 MARKPOINT, /* MARKPOINT_next_fail */
5d458dd8
YO
285 SKIP, /* SKIP_next */
286 SKIP, /* SKIP_next_fail */
287 CUTGROUP, /* CUTGROUP_next */
288 CUTGROUP, /* CUTGROUP_next_fail */
ee9b8eae
YO
289 KEEPS, /* KEEPS_next */
290 KEEPS, /* KEEPS_next_fail */
d09b2d29
IZ
291};
292#endif
293
6bda09f9 294/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
295
296#ifdef REG_COMP_C
29de9391 297static const U8 regarglen[] = {
03363afd
YO
298 0, /* END */
299 0, /* SUCCEED */
300 0, /* BOL */
301 0, /* MBOL */
302 0, /* SBOL */
303 0, /* EOS */
304 0, /* EOL */
305 0, /* MEOL */
306 0, /* SEOL */
307 0, /* BOUND */
308 0, /* BOUNDL */
309 0, /* NBOUND */
310 0, /* NBOUNDL */
311 0, /* GPOS */
312 0, /* REG_ANY */
313 0, /* SANY */
314 0, /* CANY */
315 0, /* ANYOF */
0e019ad6 316 0, /* ANYOFV */
03363afd
YO
317 0, /* ALNUM */
318 0, /* ALNUML */
319 0, /* NALNUM */
320 0, /* NALNUML */
321 0, /* SPACE */
322 0, /* SPACEL */
323 0, /* NSPACE */
324 0, /* NSPACEL */
325 0, /* DIGIT */
326 0, /* DIGITL */
327 0, /* NDIGIT */
328 0, /* NDIGITL */
329 0, /* CLUMP */
330 0, /* BRANCH */
331 0, /* BACK */
332 0, /* EXACT */
333 0, /* EXACTF */
334 0, /* EXACTFL */
01f98ec2 335 0, /* EXACTFU */
03363afd
YO
336 0, /* NOTHING */
337 0, /* TAIL */
338 0, /* STAR */
339 0, /* PLUS */
340 EXTRA_SIZE(struct regnode_2), /* CURLY */
341 EXTRA_SIZE(struct regnode_2), /* CURLYN */
342 EXTRA_SIZE(struct regnode_2), /* CURLYM */
343 EXTRA_SIZE(struct regnode_2), /* CURLYX */
344 0, /* WHILEM */
345 EXTRA_SIZE(struct regnode_1), /* OPEN */
346 EXTRA_SIZE(struct regnode_1), /* CLOSE */
347 EXTRA_SIZE(struct regnode_1), /* REF */
348 EXTRA_SIZE(struct regnode_1), /* REFF */
349 EXTRA_SIZE(struct regnode_1), /* REFFL */
01f98ec2
KW
350 EXTRA_SIZE(struct regnode_1), /* REFFU */
351 EXTRA_SIZE(struct regnode_1), /* NREF */
352 EXTRA_SIZE(struct regnode_1), /* NREFF */
353 EXTRA_SIZE(struct regnode_1), /* NREFFL */
354 EXTRA_SIZE(struct regnode_1), /* NREFFU */
03363afd
YO
355 EXTRA_SIZE(struct regnode_1), /* IFMATCH */
356 EXTRA_SIZE(struct regnode_1), /* UNLESSM */
357 EXTRA_SIZE(struct regnode_1), /* SUSPEND */
358 EXTRA_SIZE(struct regnode_1), /* IFTHEN */
359 EXTRA_SIZE(struct regnode_1), /* GROUPP */
360 EXTRA_SIZE(struct regnode_1), /* LONGJMP */
361 EXTRA_SIZE(struct regnode_1), /* BRANCHJ */
362 EXTRA_SIZE(struct regnode_1), /* EVAL */
363 0, /* MINMOD */
364 0, /* LOGICAL */
365 EXTRA_SIZE(struct regnode_1), /* RENUM */
366 EXTRA_SIZE(struct regnode_1), /* TRIE */
367 EXTRA_SIZE(struct regnode_charclass), /* TRIEC */
368 EXTRA_SIZE(struct regnode_1), /* AHOCORASICK */
369 EXTRA_SIZE(struct regnode_charclass), /* AHOCORASICKC */
1a147d38
YO
370 EXTRA_SIZE(struct regnode_2L), /* GOSUB */
371 0, /* GOSTART */
0a4db386 372 EXTRA_SIZE(struct regnode_1), /* NGROUPP */
1a147d38 373 EXTRA_SIZE(struct regnode_1), /* INSUBP */
0a4db386 374 EXTRA_SIZE(struct regnode_1), /* DEFINEP */
e2e6a0f1 375 0, /* ENDLIKE */
7f69552c 376 0, /* OPFAIL */
e2e6a0f1 377 EXTRA_SIZE(struct regnode_1), /* ACCEPT */
20832bc5 378 EXTRA_SIZE(struct regnode_1), /* VERB */
5d458dd8 379 EXTRA_SIZE(struct regnode_1), /* PRUNE */
e2e6a0f1 380 EXTRA_SIZE(struct regnode_1), /* MARKPOINT */
5d458dd8 381 EXTRA_SIZE(struct regnode_1), /* SKIP */
e2e6a0f1 382 EXTRA_SIZE(struct regnode_1), /* COMMIT */
5d458dd8 383 EXTRA_SIZE(struct regnode_1), /* CUTGROUP */
ee9b8eae 384 0, /* KEEPS */
e1d1eefb
YO
385 0, /* LNBREAK */
386 0, /* VERTWS */
387 0, /* NVERTWS */
388 0, /* HORIZWS */
389 0, /* NHORIZWS */
32e6a07c 390 EXTRA_SIZE(struct regnode_1), /* FOLDCHAR */
03363afd
YO
391 0, /* OPTIMIZED */
392 0, /* PSEUDO */
d09b2d29
IZ
393};
394
6bda09f9
YO
395/* reg_off_by_arg[] - Which argument holds the offset to the next node */
396
29de9391 397static const char reg_off_by_arg[] = {
03363afd
YO
398 0, /* END */
399 0, /* SUCCEED */
400 0, /* BOL */
401 0, /* MBOL */
402 0, /* SBOL */
403 0, /* EOS */
404 0, /* EOL */
405 0, /* MEOL */
406 0, /* SEOL */
407 0, /* BOUND */
408 0, /* BOUNDL */
409 0, /* NBOUND */
410 0, /* NBOUNDL */
411 0, /* GPOS */
412 0, /* REG_ANY */
413 0, /* SANY */
414 0, /* CANY */
415 0, /* ANYOF */
0e019ad6 416 0, /* ANYOFV */
03363afd
YO
417 0, /* ALNUM */
418 0, /* ALNUML */
419 0, /* NALNUM */
420 0, /* NALNUML */
421 0, /* SPACE */
422 0, /* SPACEL */
423 0, /* NSPACE */
424 0, /* NSPACEL */
425 0, /* DIGIT */
426 0, /* DIGITL */
427 0, /* NDIGIT */
428 0, /* NDIGITL */
429 0, /* CLUMP */
430 0, /* BRANCH */
431 0, /* BACK */
432 0, /* EXACT */
433 0, /* EXACTF */
434 0, /* EXACTFL */
01f98ec2 435 0, /* EXACTFU */
03363afd
YO
436 0, /* NOTHING */
437 0, /* TAIL */
438 0, /* STAR */
439 0, /* PLUS */
440 0, /* CURLY */
441 0, /* CURLYN */
442 0, /* CURLYM */
443 0, /* CURLYX */
444 0, /* WHILEM */
445 0, /* OPEN */
446 0, /* CLOSE */
447 0, /* REF */
448 0, /* REFF */
449 0, /* REFFL */
01f98ec2
KW
450 0, /* REFFU */
451 0, /* NREF */
452 0, /* NREFF */
453 0, /* NREFFL */
454 0, /* NREFFU */
03363afd
YO
455 2, /* IFMATCH */
456 2, /* UNLESSM */
457 1, /* SUSPEND */
458 1, /* IFTHEN */
459 0, /* GROUPP */
460 1, /* LONGJMP */
461 1, /* BRANCHJ */
462 0, /* EVAL */
463 0, /* MINMOD */
464 0, /* LOGICAL */
465 1, /* RENUM */
466 0, /* TRIE */
467 0, /* TRIEC */
468 0, /* AHOCORASICK */
469 0, /* AHOCORASICKC */
1a147d38
YO
470 0, /* GOSUB */
471 0, /* GOSTART */
0a4db386 472 0, /* NGROUPP */
1a147d38 473 0, /* INSUBP */
0a4db386 474 0, /* DEFINEP */
e2e6a0f1 475 0, /* ENDLIKE */
7f69552c 476 0, /* OPFAIL */
e2e6a0f1
YO
477 0, /* ACCEPT */
478 0, /* VERB */
5d458dd8 479 0, /* PRUNE */
e2e6a0f1 480 0, /* MARKPOINT */
5d458dd8 481 0, /* SKIP */
e2e6a0f1 482 0, /* COMMIT */
5d458dd8 483 0, /* CUTGROUP */
ee9b8eae 484 0, /* KEEPS */
e1d1eefb
YO
485 0, /* LNBREAK */
486 0, /* VERTWS */
487 0, /* NVERTWS */
488 0, /* HORIZWS */
489 0, /* NHORIZWS */
32e6a07c 490 0, /* FOLDCHAR */
03363afd
YO
491 0, /* OPTIMIZED */
492 0, /* PSEUDO */
d09b2d29 493};
885f9e59 494
13d6edb4
NC
495#endif /* REG_COMP_C */
496
6bda09f9
YO
497/* reg_name[] - Opcode/state names in string form, for debugging */
498
22429478 499#ifndef DOINIT
13d6edb4 500EXTCONST char * PL_reg_name[];
22429478 501#else
4764e399 502EXTCONST char * const PL_reg_name[] = {
03363afd
YO
503 "END", /* 0000 */
504 "SUCCEED", /* 0x01 */
505 "BOL", /* 0x02 */
506 "MBOL", /* 0x03 */
507 "SBOL", /* 0x04 */
508 "EOS", /* 0x05 */
509 "EOL", /* 0x06 */
510 "MEOL", /* 0x07 */
511 "SEOL", /* 0x08 */
512 "BOUND", /* 0x09 */
513 "BOUNDL", /* 0x0a */
514 "NBOUND", /* 0x0b */
515 "NBOUNDL", /* 0x0c */
516 "GPOS", /* 0x0d */
517 "REG_ANY", /* 0x0e */
518 "SANY", /* 0x0f */
519 "CANY", /* 0x10 */
520 "ANYOF", /* 0x11 */
0e019ad6
KW
521 "ANYOFV", /* 0x12 */
522 "ALNUM", /* 0x13 */
523 "ALNUML", /* 0x14 */
524 "NALNUM", /* 0x15 */
525 "NALNUML", /* 0x16 */
526 "SPACE", /* 0x17 */
527 "SPACEL", /* 0x18 */
528 "NSPACE", /* 0x19 */
529 "NSPACEL", /* 0x1a */
530 "DIGIT", /* 0x1b */
531 "DIGITL", /* 0x1c */
532 "NDIGIT", /* 0x1d */
533 "NDIGITL", /* 0x1e */
534 "CLUMP", /* 0x1f */
535 "BRANCH", /* 0x20 */
536 "BACK", /* 0x21 */
537 "EXACT", /* 0x22 */
538 "EXACTF", /* 0x23 */
539 "EXACTFL", /* 0x24 */
540 "EXACTFU", /* 0x25 */
541 "NOTHING", /* 0x26 */
542 "TAIL", /* 0x27 */
543 "STAR", /* 0x28 */
544 "PLUS", /* 0x29 */
545 "CURLY", /* 0x2a */
546 "CURLYN", /* 0x2b */
547 "CURLYM", /* 0x2c */
548 "CURLYX", /* 0x2d */
549 "WHILEM", /* 0x2e */
550 "OPEN", /* 0x2f */
551 "CLOSE", /* 0x30 */
552 "REF", /* 0x31 */
553 "REFF", /* 0x32 */
554 "REFFL", /* 0x33 */
555 "REFFU", /* 0x34 */
556 "NREF", /* 0x35 */
557 "NREFF", /* 0x36 */
558 "NREFFL", /* 0x37 */
559 "NREFFU", /* 0x38 */
560 "IFMATCH", /* 0x39 */
561 "UNLESSM", /* 0x3a */
562 "SUSPEND", /* 0x3b */
563 "IFTHEN", /* 0x3c */
564 "GROUPP", /* 0x3d */
565 "LONGJMP", /* 0x3e */
566 "BRANCHJ", /* 0x3f */
567 "EVAL", /* 0x40 */
568 "MINMOD", /* 0x41 */
569 "LOGICAL", /* 0x42 */
570 "RENUM", /* 0x43 */
571 "TRIE", /* 0x44 */
572 "TRIEC", /* 0x45 */
573 "AHOCORASICK", /* 0x46 */
574 "AHOCORASICKC", /* 0x47 */
575 "GOSUB", /* 0x48 */
576 "GOSTART", /* 0x49 */
577 "NGROUPP", /* 0x4a */
578 "INSUBP", /* 0x4b */
579 "DEFINEP", /* 0x4c */
580 "ENDLIKE", /* 0x4d */
581 "OPFAIL", /* 0x4e */
582 "ACCEPT", /* 0x4f */
583 "VERB", /* 0x50 */
584 "PRUNE", /* 0x51 */
585 "MARKPOINT", /* 0x52 */
586 "SKIP", /* 0x53 */
587 "COMMIT", /* 0x54 */
588 "CUTGROUP", /* 0x55 */
589 "KEEPS", /* 0x56 */
590 "LNBREAK", /* 0x57 */
591 "VERTWS", /* 0x58 */
592 "NVERTWS", /* 0x59 */
593 "HORIZWS", /* 0x5a */
594 "NHORIZWS", /* 0x5b */
595 "FOLDCHAR", /* 0x5c */
596 "OPTIMIZED", /* 0x5d */
597 "PSEUDO", /* 0x5e */
03363afd 598 /* ------------ States ------------- */
24b23f37
YO
599 "TRIE_next", /* REGNODE_MAX +0x01 */
600 "TRIE_next_fail", /* REGNODE_MAX +0x02 */
601 "EVAL_AB", /* REGNODE_MAX +0x03 */
602 "EVAL_AB_fail", /* REGNODE_MAX +0x04 */
603 "CURLYX_end", /* REGNODE_MAX +0x05 */
604 "CURLYX_end_fail", /* REGNODE_MAX +0x06 */
605 "WHILEM_A_pre", /* REGNODE_MAX +0x07 */
606 "WHILEM_A_pre_fail", /* REGNODE_MAX +0x08 */
607 "WHILEM_A_min", /* REGNODE_MAX +0x09 */
608 "WHILEM_A_min_fail", /* REGNODE_MAX +0x0a */
609 "WHILEM_A_max", /* REGNODE_MAX +0x0b */
610 "WHILEM_A_max_fail", /* REGNODE_MAX +0x0c */
611 "WHILEM_B_min", /* REGNODE_MAX +0x0d */
612 "WHILEM_B_min_fail", /* REGNODE_MAX +0x0e */
613 "WHILEM_B_max", /* REGNODE_MAX +0x0f */
614 "WHILEM_B_max_fail", /* REGNODE_MAX +0x10 */
615 "BRANCH_next", /* REGNODE_MAX +0x11 */
616 "BRANCH_next_fail", /* REGNODE_MAX +0x12 */
617 "CURLYM_A", /* REGNODE_MAX +0x13 */
618 "CURLYM_A_fail", /* REGNODE_MAX +0x14 */
619 "CURLYM_B", /* REGNODE_MAX +0x15 */
620 "CURLYM_B_fail", /* REGNODE_MAX +0x16 */
621 "IFMATCH_A", /* REGNODE_MAX +0x17 */
622 "IFMATCH_A_fail", /* REGNODE_MAX +0x18 */
623 "CURLY_B_min_known", /* REGNODE_MAX +0x19 */
624 "CURLY_B_min_known_fail", /* REGNODE_MAX +0x1a */
625 "CURLY_B_min", /* REGNODE_MAX +0x1b */
626 "CURLY_B_min_fail", /* REGNODE_MAX +0x1c */
627 "CURLY_B_max", /* REGNODE_MAX +0x1d */
628 "CURLY_B_max_fail", /* REGNODE_MAX +0x1e */
629 "COMMIT_next", /* REGNODE_MAX +0x1f */
630 "COMMIT_next_fail", /* REGNODE_MAX +0x20 */
e2e6a0f1
YO
631 "MARKPOINT_next", /* REGNODE_MAX +0x21 */
632 "MARKPOINT_next_fail", /* REGNODE_MAX +0x22 */
5d458dd8
YO
633 "SKIP_next", /* REGNODE_MAX +0x23 */
634 "SKIP_next_fail", /* REGNODE_MAX +0x24 */
635 "CUTGROUP_next", /* REGNODE_MAX +0x25 */
636 "CUTGROUP_next_fail", /* REGNODE_MAX +0x26 */
ee9b8eae
YO
637 "KEEPS_next", /* REGNODE_MAX +0x27 */
638 "KEEPS_next_fail", /* REGNODE_MAX +0x28 */
885f9e59 639};
22429478 640#endif /* DOINIT */
d09b2d29 641
f7819f85
A
642/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
643
644#ifndef DOINIT
645EXTCONST char * PL_reg_extflags_name[];
646#else
647EXTCONST char * const PL_reg_extflags_name[] = {
9de15fec 648 /* Bits in extflags defined: 11111111111111111111111001111111 */
52d81aa8
NC
649 "MULTILINE", /* 0x00000001 */
650 "SINGLELINE", /* 0x00000002 */
651 "FOLD", /* 0x00000004 */
652 "EXTENDED", /* 0x00000008 */
653 "KEEPCOPY", /* 0x00000010 */
654 "LOCALE", /* 0x00000020 */
9de15fec 655 "UNICODE", /* 0x00000040 */
f7819f85 656 "UNUSED_BIT_7", /* 0x00000080 */
e795e964
KW
657 "UNUSED_BIT_8", /* 0x00000100 */
658 "ANCH_BOL", /* 0x00000200 */
659 "ANCH_MBOL", /* 0x00000400 */
660 "ANCH_SBOL", /* 0x00000800 */
661 "ANCH_GPOS", /* 0x00001000 */
662 "GPOS_SEEN", /* 0x00002000 */
663 "GPOS_FLOAT", /* 0x00004000 */
664 "LOOKBEHIND_SEEN", /* 0x00008000 */
665 "EVAL_SEEN", /* 0x00010000 */
666 "CANY_SEEN", /* 0x00020000 */
667 "NOSCAN", /* 0x00040000 */
668 "CHECK_ALL", /* 0x00080000 */
52d81aa8
NC
669 "MATCH_UTF8", /* 0x00100000 */
670 "USE_INTUIT_NOML", /* 0x00200000 */
671 "USE_INTUIT_ML", /* 0x00400000 */
672 "INTUIT_TAIL", /* 0x00800000 */
673 "SPLIT", /* 0x01000000 */
674 "COPY_DONE", /* 0x02000000 */
675 "TAINTED_SEEN", /* 0x04000000 */
676 "TAINTED", /* 0x08000000 */
677 "START_ONLY", /* 0x10000000 */
678 "SKIPWHITE", /* 0x20000000 */
679 "WHITE", /* 0x40000000 */
680 "NULL", /* 0x80000000 */
f7819f85
A
681};
682#endif /* DOINIT */
683
f9ef50a7 684/* The following have no fixed length. U8 so we can do strchr() on it. */
ded4dd2a 685#define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 686
f9ef50a7 687#ifndef DOINIT
ded4dd2a 688EXTCONST U8 PL_varies[] __attribute__deprecated__;
f9ef50a7 689#else
ded4dd2a 690EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
0e019ad6
KW
691 ANYOFV, CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX,
692 WHILEM, REF, REFF, REFFL, REFFU, NREF, NREFF, NREFFL, NREFFU, SUSPEND,
693 IFTHEN, BRANCHJ,
f9ef50a7
NC
694 0
695};
696#endif /* DOINIT */
697
ded4dd2a
NC
698#ifndef DOINIT
699EXTCONST U8 PL_varies_bitmask[];
700#else
701EXTCONST U8 PL_varies_bitmask[] = {
0e019ad6 702 0x00, 0x00, 0x04, 0x80, 0x03, 0x7F, 0xFE, 0x99, 0x00, 0x00, 0x00, 0x00
ded4dd2a
NC
703};
704#endif /* DOINIT */
705
f9ef50a7
NC
706/* The following always have a length of 1. U8 we can do strchr() on it. */
707/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
ded4dd2a 708#define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 709
f9ef50a7 710#ifndef DOINIT
ded4dd2a 711EXTCONST U8 PL_simple[] __attribute__deprecated__;
f9ef50a7 712#else
ded4dd2a 713EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
f9ef50a7 714 REG_ANY, SANY, CANY, ANYOF, ALNUM, ALNUML, NALNUM, NALNUML, SPACE,
28b5d7bf
KW
715 SPACEL, NSPACE, NSPACEL, DIGIT, DIGITL, NDIGIT, NDIGITL, VERTWS,
716 NVERTWS, HORIZWS, NHORIZWS,
f9ef50a7
NC
717 0
718};
719#endif /* DOINIT */
720
ded4dd2a
NC
721#ifndef DOINIT
722EXTCONST U8 PL_simple_bitmask[];
723#else
724EXTCONST U8 PL_simple_bitmask[] = {
0e019ad6 725 0x00, 0xC0, 0xFB, 0x7F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x0F
ded4dd2a
NC
726};
727#endif /* DOINIT */
728
37442d52 729/* ex: set ro: */