This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Add RXf_UNBOUNDED_QUANTIFIER and regexp->maxlen
[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 4 Any changes made here will be lost!
78102347 5 */
d09b2d29 6
6bda09f9
YO
7/* Regops and State definitions */
8
71068078
KW
9#define REGNODE_MAX 93
10#define REGMATCH_STATE_MAX 133
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. */
1e355c70 21#define BOUND 9 /* 0x09 Match "" at any word boundary using native charset semantics for non-utf8 */
f2284805 22#define BOUNDL 10 /* 0x0a Match "" at any locale word boundary */
1e355c70 23#define BOUNDU 11 /* 0x0b Match "" at any word boundary using Unicode semantics */
0c6e81eb
KW
24#define BOUNDA 12 /* 0x0c Match "" at any word boundary using ASCII semantics */
25#define NBOUND 13 /* 0x0d Match "" at any word non-boundary using native charset semantics for non-utf8 */
26#define NBOUNDL 14 /* 0x0e Match "" at any locale word non-boundary */
27#define NBOUNDU 15 /* 0x0f Match "" at any word non-boundary using Unicode semantics */
28#define NBOUNDA 16 /* 0x10 Match "" at any word non-boundary using ASCII semantics */
29#define GPOS 17 /* 0x11 Matches where last m//g left off. */
30#define REG_ANY 18 /* 0x12 Match any one character (except newline). */
31#define SANY 19 /* 0x13 Match any one character. */
32#define CANY 20 /* 0x14 Match any one byte. */
33#define ANYOF 21 /* 0x15 Match character in (or not in) this class, single char match only */
71068078
KW
34#define POSIXD 22 /* 0x16 Some [[:class:]] under /d; the FLAGS field gives which one */
35#define POSIXL 23 /* 0x17 Some [[:class:]] under /l; the FLAGS field gives which one */
36#define POSIXU 24 /* 0x18 Some [[:class:]] under /u; the FLAGS field gives which one */
37#define POSIXA 25 /* 0x19 Some [[:class:]] under /a; the FLAGS field gives which one */
38#define NPOSIXD 26 /* 0x1a complement of POSIXD, [[:^class:]] */
39#define NPOSIXL 27 /* 0x1b complement of POSIXL, [[:^class:]] */
40#define NPOSIXU 28 /* 0x1c complement of POSIXU, [[:^class:]] */
41#define NPOSIXA 29 /* 0x1d complement of POSIXA, [[:^class:]] */
42#define CLUMP 30 /* 0x1e Match any extended grapheme cluster sequence */
43#define BRANCH 31 /* 0x1f Match this alternative, or the next... */
44#define BACK 32 /* 0x20 Match "", "next" ptr points backward. */
45#define EXACT 33 /* 0x21 Match this string (preceded by length). */
46#define EXACTF 34 /* 0x22 Match this non-UTF-8 string (not guaranteed to be folded) using /id rules (w/len). */
47#define EXACTFL 35 /* 0x23 Match this string (not guaranteed to be folded) using /il rules (w/len). */
48#define EXACTFU 36 /* 0x24 Match this string (folded iff in UTF-8, length in folding doesn't change if not in UTF-8) using /iu rules (w/len). */
49#define EXACTFA 37 /* 0x25 Match this string (not guaranteed to be folded) using /iaa rules (w/len). */
50#define EXACTFU_SS 38 /* 0x26 Match this string (folded iff in UTF-8, length in folding may change even if not in UTF-8) using /iu rules (w/len). */
51#define EXACTFA_NO_TRIE 39 /* 0x27 Match this string (which is not trie-able; not guaranteed to be folded) using /iaa rules (w/len). */
52#define NOTHING 40 /* 0x28 Match empty string. */
53#define TAIL 41 /* 0x29 Match empty string. Can jump here from outside. */
54#define STAR 42 /* 0x2a Match this (simple) thing 0 or more times. */
55#define PLUS 43 /* 0x2b Match this (simple) thing 1 or more times. */
56#define CURLY 44 /* 0x2c Match this simple thing {n,m} times. */
57#define CURLYN 45 /* 0x2d Capture next-after-this simple thing */
58#define CURLYM 46 /* 0x2e Capture this medium-complex thing {n,m} times. */
59#define CURLYX 47 /* 0x2f Match this complex thing {n,m} times. */
60#define WHILEM 48 /* 0x30 Do curly processing and see if rest matches. */
61#define OPEN 49 /* 0x31 Mark this point in input as start of #n. */
62#define CLOSE 50 /* 0x32 Analogous to OPEN. */
63#define REF 51 /* 0x33 Match some already matched string */
64#define REFF 52 /* 0x34 Match already matched string, folded using native charset semantics for non-utf8 */
65#define REFFL 53 /* 0x35 Match already matched string, folded in loc. */
66#define REFFU 54 /* 0x36 Match already matched string, folded using unicode semantics for non-utf8 */
67#define REFFA 55 /* 0x37 Match already matched string, folded using unicode semantics for non-utf8, no mixing ASCII, non-ASCII */
68#define NREF 56 /* 0x38 Match some already matched string */
69#define NREFF 57 /* 0x39 Match already matched string, folded using native charset semantics for non-utf8 */
70#define NREFFL 58 /* 0x3a Match already matched string, folded in loc. */
71#define NREFFU 59 /* 0x3b Match already matched string, folded using unicode semantics for non-utf8 */
72#define NREFFA 60 /* 0x3c Match already matched string, folded using unicode semantics for non-utf8, no mixing ASCII, non-ASCII */
73#define IFMATCH 61 /* 0x3d Succeeds if the following matches. */
74#define UNLESSM 62 /* 0x3e Fails if the following matches. */
75#define SUSPEND 63 /* 0x3f "Independent" sub-RE. */
76#define IFTHEN 64 /* 0x40 Switch, should be preceded by switcher. */
77#define GROUPP 65 /* 0x41 Whether the group matched. */
78#define LONGJMP 66 /* 0x42 Jump far away. */
79#define BRANCHJ 67 /* 0x43 BRANCH with long offset. */
80#define EVAL 68 /* 0x44 Execute some Perl code. */
81#define MINMOD 69 /* 0x45 Next operator is not greedy. */
82#define LOGICAL 70 /* 0x46 Next opcode should set the flag only. */
83#define RENUM 71 /* 0x47 Group with independently numbered parens. */
84#define TRIE 72 /* 0x48 Match many EXACT(F[ALU]?)? at once. flags==type */
85#define TRIEC 73 /* 0x49 Same as TRIE, but with embedded charclass data */
86#define AHOCORASICK 74 /* 0x4a Aho Corasick stclass. flags==type */
87#define AHOCORASICKC 75 /* 0x4b Same as AHOCORASICK, but with embedded charclass data */
88#define GOSUB 76 /* 0x4c recurse to paren arg1 at (signed) ofs arg2 */
89#define GOSTART 77 /* 0x4d recurse to start of pattern */
90#define NGROUPP 78 /* 0x4e Whether the group matched. */
91#define INSUBP 79 /* 0x4f Whether we are in a specific recurse. */
92#define DEFINEP 80 /* 0x50 Never execute directly. */
93#define ENDLIKE 81 /* 0x51 Used only for the type field of verbs */
94#define OPFAIL 82 /* 0x52 Same as (?!) */
95#define ACCEPT 83 /* 0x53 Accepts the current matched string. */
96#define VERB 84 /* 0x54 Used only for the type field of verbs */
97#define PRUNE 85 /* 0x55 Pattern fails at this startpoint if no-backtracking through this */
98#define MARKPOINT 86 /* 0x56 Push the current location for rollback by cut. */
99#define SKIP 87 /* 0x57 On failure skip forward (to the mark) before retrying */
100#define COMMIT 88 /* 0x58 Pattern fails outright if backtracking through this */
101#define CUTGROUP 89 /* 0x59 On failure go to the next alternation in the group */
102#define KEEPS 90 /* 0x5a $& begins here. */
103#define LNBREAK 91 /* 0x5b generic newline pattern */
104#define OPTIMIZED 92 /* 0x5c Placeholder for dump. */
105#define PSEUDO 93 /* 0x5d Pseudo opcode for internal use. */
03363afd 106 /* ------------ States ------------- */
24b23f37
YO
107#define TRIE_next (REGNODE_MAX + 1) /* state for TRIE */
108#define TRIE_next_fail (REGNODE_MAX + 2) /* state for TRIE */
109#define EVAL_AB (REGNODE_MAX + 3) /* state for EVAL */
110#define EVAL_AB_fail (REGNODE_MAX + 4) /* state for EVAL */
111#define CURLYX_end (REGNODE_MAX + 5) /* state for CURLYX */
112#define CURLYX_end_fail (REGNODE_MAX + 6) /* state for CURLYX */
113#define WHILEM_A_pre (REGNODE_MAX + 7) /* state for WHILEM */
114#define WHILEM_A_pre_fail (REGNODE_MAX + 8) /* state for WHILEM */
115#define WHILEM_A_min (REGNODE_MAX + 9) /* state for WHILEM */
116#define WHILEM_A_min_fail (REGNODE_MAX + 10) /* state for WHILEM */
117#define WHILEM_A_max (REGNODE_MAX + 11) /* state for WHILEM */
118#define WHILEM_A_max_fail (REGNODE_MAX + 12) /* state for WHILEM */
119#define WHILEM_B_min (REGNODE_MAX + 13) /* state for WHILEM */
120#define WHILEM_B_min_fail (REGNODE_MAX + 14) /* state for WHILEM */
121#define WHILEM_B_max (REGNODE_MAX + 15) /* state for WHILEM */
122#define WHILEM_B_max_fail (REGNODE_MAX + 16) /* state for WHILEM */
123#define BRANCH_next (REGNODE_MAX + 17) /* state for BRANCH */
124#define BRANCH_next_fail (REGNODE_MAX + 18) /* state for BRANCH */
125#define CURLYM_A (REGNODE_MAX + 19) /* state for CURLYM */
126#define CURLYM_A_fail (REGNODE_MAX + 20) /* state for CURLYM */
127#define CURLYM_B (REGNODE_MAX + 21) /* state for CURLYM */
128#define CURLYM_B_fail (REGNODE_MAX + 22) /* state for CURLYM */
129#define IFMATCH_A (REGNODE_MAX + 23) /* state for IFMATCH */
130#define IFMATCH_A_fail (REGNODE_MAX + 24) /* state for IFMATCH */
131#define CURLY_B_min_known (REGNODE_MAX + 25) /* state for CURLY */
132#define CURLY_B_min_known_fail (REGNODE_MAX + 26) /* state for CURLY */
133#define CURLY_B_min (REGNODE_MAX + 27) /* state for CURLY */
134#define CURLY_B_min_fail (REGNODE_MAX + 28) /* state for CURLY */
135#define CURLY_B_max (REGNODE_MAX + 29) /* state for CURLY */
136#define CURLY_B_max_fail (REGNODE_MAX + 30) /* state for CURLY */
137#define COMMIT_next (REGNODE_MAX + 31) /* state for COMMIT */
138#define COMMIT_next_fail (REGNODE_MAX + 32) /* state for COMMIT */
e2e6a0f1
YO
139#define MARKPOINT_next (REGNODE_MAX + 33) /* state for MARKPOINT */
140#define MARKPOINT_next_fail (REGNODE_MAX + 34) /* state for MARKPOINT */
5d458dd8
YO
141#define SKIP_next (REGNODE_MAX + 35) /* state for SKIP */
142#define SKIP_next_fail (REGNODE_MAX + 36) /* state for SKIP */
143#define CUTGROUP_next (REGNODE_MAX + 37) /* state for CUTGROUP */
144#define CUTGROUP_next_fail (REGNODE_MAX + 38) /* state for CUTGROUP */
ee9b8eae
YO
145#define KEEPS_next (REGNODE_MAX + 39) /* state for KEEPS */
146#define KEEPS_next_fail (REGNODE_MAX + 40) /* state for KEEPS */
03363afd 147
6bda09f9 148/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
149
150#ifndef DOINIT
22c35a8c 151EXTCONST U8 PL_regkind[];
d09b2d29 152#else
22c35a8c 153EXTCONST U8 PL_regkind[] = {
e2e6a0f1
YO
154 END, /* END */
155 END, /* SUCCEED */
156 BOL, /* BOL */
157 BOL, /* MBOL */
158 BOL, /* SBOL */
159 EOL, /* EOS */
160 EOL, /* EOL */
161 EOL, /* MEOL */
162 EOL, /* SEOL */
163 BOUND, /* BOUND */
164 BOUND, /* BOUNDL */
1e355c70 165 BOUND, /* BOUNDU */
0c6e81eb 166 BOUND, /* BOUNDA */
e2e6a0f1
YO
167 NBOUND, /* NBOUND */
168 NBOUND, /* NBOUNDL */
1e355c70 169 NBOUND, /* NBOUNDU */
0c6e81eb 170 NBOUND, /* NBOUNDA */
e2e6a0f1
YO
171 GPOS, /* GPOS */
172 REG_ANY, /* REG_ANY */
173 REG_ANY, /* SANY */
174 REG_ANY, /* CANY */
175 ANYOF, /* ANYOF */
3615ea58
KW
176 POSIXD, /* POSIXD */
177 POSIXD, /* POSIXL */
178 POSIXD, /* POSIXU */
179 POSIXD, /* POSIXA */
9e84774b
KW
180 NPOSIXD, /* NPOSIXD */
181 NPOSIXD, /* NPOSIXL */
182 NPOSIXD, /* NPOSIXU */
183 NPOSIXD, /* NPOSIXA */
e2e6a0f1
YO
184 CLUMP, /* CLUMP */
185 BRANCH, /* BRANCH */
186 BACK, /* BACK */
187 EXACT, /* EXACT */
188 EXACT, /* EXACTF */
189 EXACT, /* EXACTFL */
01f98ec2 190 EXACT, /* EXACTFU */
8c1182fd 191 EXACT, /* EXACTFA */
3c760661 192 EXACT, /* EXACTFU_SS */
098b07d5 193 EXACT, /* EXACTFA_NO_TRIE */
e2e6a0f1
YO
194 NOTHING, /* NOTHING */
195 NOTHING, /* TAIL */
196 STAR, /* STAR */
197 PLUS, /* PLUS */
198 CURLY, /* CURLY */
199 CURLY, /* CURLYN */
200 CURLY, /* CURLYM */
201 CURLY, /* CURLYX */
202 WHILEM, /* WHILEM */
203 OPEN, /* OPEN */
204 CLOSE, /* CLOSE */
205 REF, /* REF */
206 REF, /* REFF */
207 REF, /* REFFL */
01f98ec2 208 REF, /* REFFU */
781aab5c 209 REF, /* REFFA */
01f98ec2
KW
210 REF, /* NREF */
211 REF, /* NREFF */
212 REF, /* NREFFL */
213 REF, /* NREFFU */
781aab5c 214 REF, /* NREFFA */
e2e6a0f1
YO
215 BRANCHJ, /* IFMATCH */
216 BRANCHJ, /* UNLESSM */
217 BRANCHJ, /* SUSPEND */
218 BRANCHJ, /* IFTHEN */
219 GROUPP, /* GROUPP */
220 LONGJMP, /* LONGJMP */
221 BRANCHJ, /* BRANCHJ */
222 EVAL, /* EVAL */
223 MINMOD, /* MINMOD */
224 LOGICAL, /* LOGICAL */
225 BRANCHJ, /* RENUM */
226 TRIE, /* TRIE */
227 TRIE, /* TRIEC */
228 TRIE, /* AHOCORASICK */
229 TRIE, /* AHOCORASICKC */
230 GOSUB, /* GOSUB */
231 GOSTART, /* GOSTART */
e2e6a0f1
YO
232 NGROUPP, /* NGROUPP */
233 INSUBP, /* INSUBP */
234 DEFINEP, /* DEFINEP */
235 ENDLIKE, /* ENDLIKE */
236 ENDLIKE, /* OPFAIL */
237 ENDLIKE, /* ACCEPT */
238 VERB, /* VERB */
5d458dd8 239 VERB, /* PRUNE */
e2e6a0f1 240 VERB, /* MARKPOINT */
5d458dd8 241 VERB, /* SKIP */
e2e6a0f1 242 VERB, /* COMMIT */
5d458dd8 243 VERB, /* CUTGROUP */
ee9b8eae 244 KEEPS, /* KEEPS */
e1d1eefb 245 LNBREAK, /* LNBREAK */
e2e6a0f1
YO
246 NOTHING, /* OPTIMIZED */
247 PSEUDO, /* PSEUDO */
03363afd 248 /* ------------ States ------------- */
e2e6a0f1
YO
249 TRIE, /* TRIE_next */
250 TRIE, /* TRIE_next_fail */
251 EVAL, /* EVAL_AB */
252 EVAL, /* EVAL_AB_fail */
253 CURLYX, /* CURLYX_end */
254 CURLYX, /* CURLYX_end_fail */
255 WHILEM, /* WHILEM_A_pre */
256 WHILEM, /* WHILEM_A_pre_fail */
257 WHILEM, /* WHILEM_A_min */
258 WHILEM, /* WHILEM_A_min_fail */
259 WHILEM, /* WHILEM_A_max */
260 WHILEM, /* WHILEM_A_max_fail */
261 WHILEM, /* WHILEM_B_min */
262 WHILEM, /* WHILEM_B_min_fail */
263 WHILEM, /* WHILEM_B_max */
264 WHILEM, /* WHILEM_B_max_fail */
265 BRANCH, /* BRANCH_next */
266 BRANCH, /* BRANCH_next_fail */
267 CURLYM, /* CURLYM_A */
268 CURLYM, /* CURLYM_A_fail */
269 CURLYM, /* CURLYM_B */
270 CURLYM, /* CURLYM_B_fail */
271 IFMATCH, /* IFMATCH_A */
272 IFMATCH, /* IFMATCH_A_fail */
273 CURLY, /* CURLY_B_min_known */
274 CURLY, /* CURLY_B_min_known_fail */
275 CURLY, /* CURLY_B_min */
276 CURLY, /* CURLY_B_min_fail */
277 CURLY, /* CURLY_B_max */
278 CURLY, /* CURLY_B_max_fail */
279 COMMIT, /* COMMIT_next */
280 COMMIT, /* COMMIT_next_fail */
281 MARKPOINT, /* MARKPOINT_next */
282 MARKPOINT, /* MARKPOINT_next_fail */
5d458dd8
YO
283 SKIP, /* SKIP_next */
284 SKIP, /* SKIP_next_fail */
285 CUTGROUP, /* CUTGROUP_next */
286 CUTGROUP, /* CUTGROUP_next_fail */
ee9b8eae
YO
287 KEEPS, /* KEEPS_next */
288 KEEPS, /* KEEPS_next_fail */
d09b2d29
IZ
289};
290#endif
291
6bda09f9 292/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
293
294#ifdef REG_COMP_C
29de9391 295static const U8 regarglen[] = {
03363afd
YO
296 0, /* END */
297 0, /* SUCCEED */
298 0, /* BOL */
299 0, /* MBOL */
300 0, /* SBOL */
301 0, /* EOS */
302 0, /* EOL */
303 0, /* MEOL */
304 0, /* SEOL */
305 0, /* BOUND */
306 0, /* BOUNDL */
1e355c70 307 0, /* BOUNDU */
0c6e81eb 308 0, /* BOUNDA */
03363afd
YO
309 0, /* NBOUND */
310 0, /* NBOUNDL */
1e355c70 311 0, /* NBOUNDU */
0c6e81eb 312 0, /* NBOUNDA */
03363afd
YO
313 0, /* GPOS */
314 0, /* REG_ANY */
315 0, /* SANY */
316 0, /* CANY */
317 0, /* ANYOF */
3615ea58
KW
318 0, /* POSIXD */
319 0, /* POSIXL */
320 0, /* POSIXU */
321 0, /* POSIXA */
322 0, /* NPOSIXD */
323 0, /* NPOSIXL */
324 0, /* NPOSIXU */
325 0, /* NPOSIXA */
03363afd
YO
326 0, /* CLUMP */
327 0, /* BRANCH */
328 0, /* BACK */
329 0, /* EXACT */
330 0, /* EXACTF */
331 0, /* EXACTFL */
01f98ec2 332 0, /* EXACTFU */
8c1182fd 333 0, /* EXACTFA */
3c760661 334 0, /* EXACTFU_SS */
098b07d5 335 0, /* EXACTFA_NO_TRIE */
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 350 EXTRA_SIZE(struct regnode_1), /* REFFU */
781aab5c 351 EXTRA_SIZE(struct regnode_1), /* REFFA */
01f98ec2
KW
352 EXTRA_SIZE(struct regnode_1), /* NREF */
353 EXTRA_SIZE(struct regnode_1), /* NREFF */
354 EXTRA_SIZE(struct regnode_1), /* NREFFL */
355 EXTRA_SIZE(struct regnode_1), /* NREFFU */
781aab5c 356 EXTRA_SIZE(struct regnode_1), /* NREFFA */
03363afd
YO
357 EXTRA_SIZE(struct regnode_1), /* IFMATCH */
358 EXTRA_SIZE(struct regnode_1), /* UNLESSM */
359 EXTRA_SIZE(struct regnode_1), /* SUSPEND */
360 EXTRA_SIZE(struct regnode_1), /* IFTHEN */
361 EXTRA_SIZE(struct regnode_1), /* GROUPP */
362 EXTRA_SIZE(struct regnode_1), /* LONGJMP */
363 EXTRA_SIZE(struct regnode_1), /* BRANCHJ */
364 EXTRA_SIZE(struct regnode_1), /* EVAL */
365 0, /* MINMOD */
366 0, /* LOGICAL */
367 EXTRA_SIZE(struct regnode_1), /* RENUM */
368 EXTRA_SIZE(struct regnode_1), /* TRIE */
369 EXTRA_SIZE(struct regnode_charclass), /* TRIEC */
370 EXTRA_SIZE(struct regnode_1), /* AHOCORASICK */
371 EXTRA_SIZE(struct regnode_charclass), /* AHOCORASICKC */
1a147d38
YO
372 EXTRA_SIZE(struct regnode_2L), /* GOSUB */
373 0, /* GOSTART */
0a4db386 374 EXTRA_SIZE(struct regnode_1), /* NGROUPP */
1a147d38 375 EXTRA_SIZE(struct regnode_1), /* INSUBP */
0a4db386 376 EXTRA_SIZE(struct regnode_1), /* DEFINEP */
e2e6a0f1 377 0, /* ENDLIKE */
7f69552c 378 0, /* OPFAIL */
e2e6a0f1 379 EXTRA_SIZE(struct regnode_1), /* ACCEPT */
20832bc5 380 EXTRA_SIZE(struct regnode_1), /* VERB */
5d458dd8 381 EXTRA_SIZE(struct regnode_1), /* PRUNE */
e2e6a0f1 382 EXTRA_SIZE(struct regnode_1), /* MARKPOINT */
5d458dd8 383 EXTRA_SIZE(struct regnode_1), /* SKIP */
e2e6a0f1 384 EXTRA_SIZE(struct regnode_1), /* COMMIT */
5d458dd8 385 EXTRA_SIZE(struct regnode_1), /* CUTGROUP */
ee9b8eae 386 0, /* KEEPS */
e1d1eefb 387 0, /* LNBREAK */
03363afd
YO
388 0, /* OPTIMIZED */
389 0, /* PSEUDO */
d09b2d29
IZ
390};
391
6bda09f9
YO
392/* reg_off_by_arg[] - Which argument holds the offset to the next node */
393
29de9391 394static const char reg_off_by_arg[] = {
03363afd
YO
395 0, /* END */
396 0, /* SUCCEED */
397 0, /* BOL */
398 0, /* MBOL */
399 0, /* SBOL */
400 0, /* EOS */
401 0, /* EOL */
402 0, /* MEOL */
403 0, /* SEOL */
404 0, /* BOUND */
405 0, /* BOUNDL */
1e355c70 406 0, /* BOUNDU */
0c6e81eb 407 0, /* BOUNDA */
03363afd
YO
408 0, /* NBOUND */
409 0, /* NBOUNDL */
1e355c70 410 0, /* NBOUNDU */
0c6e81eb 411 0, /* NBOUNDA */
03363afd
YO
412 0, /* GPOS */
413 0, /* REG_ANY */
414 0, /* SANY */
415 0, /* CANY */
416 0, /* ANYOF */
3615ea58
KW
417 0, /* POSIXD */
418 0, /* POSIXL */
419 0, /* POSIXU */
420 0, /* POSIXA */
421 0, /* NPOSIXD */
422 0, /* NPOSIXL */
423 0, /* NPOSIXU */
424 0, /* NPOSIXA */
03363afd
YO
425 0, /* CLUMP */
426 0, /* BRANCH */
427 0, /* BACK */
428 0, /* EXACT */
429 0, /* EXACTF */
430 0, /* EXACTFL */
01f98ec2 431 0, /* EXACTFU */
8c1182fd 432 0, /* EXACTFA */
3c760661 433 0, /* EXACTFU_SS */
098b07d5 434 0, /* EXACTFA_NO_TRIE */
03363afd
YO
435 0, /* NOTHING */
436 0, /* TAIL */
437 0, /* STAR */
438 0, /* PLUS */
439 0, /* CURLY */
440 0, /* CURLYN */
441 0, /* CURLYM */
442 0, /* CURLYX */
443 0, /* WHILEM */
444 0, /* OPEN */
445 0, /* CLOSE */
446 0, /* REF */
447 0, /* REFF */
448 0, /* REFFL */
01f98ec2 449 0, /* REFFU */
781aab5c 450 0, /* REFFA */
01f98ec2
KW
451 0, /* NREF */
452 0, /* NREFF */
453 0, /* NREFFL */
454 0, /* NREFFU */
781aab5c 455 0, /* NREFFA */
03363afd
YO
456 2, /* IFMATCH */
457 2, /* UNLESSM */
458 1, /* SUSPEND */
459 1, /* IFTHEN */
460 0, /* GROUPP */
461 1, /* LONGJMP */
462 1, /* BRANCHJ */
463 0, /* EVAL */
464 0, /* MINMOD */
465 0, /* LOGICAL */
466 1, /* RENUM */
467 0, /* TRIE */
468 0, /* TRIEC */
469 0, /* AHOCORASICK */
470 0, /* AHOCORASICKC */
1a147d38
YO
471 0, /* GOSUB */
472 0, /* GOSTART */
0a4db386 473 0, /* NGROUPP */
1a147d38 474 0, /* INSUBP */
0a4db386 475 0, /* DEFINEP */
e2e6a0f1 476 0, /* ENDLIKE */
7f69552c 477 0, /* OPFAIL */
e2e6a0f1
YO
478 0, /* ACCEPT */
479 0, /* VERB */
5d458dd8 480 0, /* PRUNE */
e2e6a0f1 481 0, /* MARKPOINT */
5d458dd8 482 0, /* SKIP */
e2e6a0f1 483 0, /* COMMIT */
5d458dd8 484 0, /* CUTGROUP */
ee9b8eae 485 0, /* KEEPS */
e1d1eefb 486 0, /* LNBREAK */
03363afd
YO
487 0, /* OPTIMIZED */
488 0, /* PSEUDO */
d09b2d29 489};
885f9e59 490
13d6edb4
NC
491#endif /* REG_COMP_C */
492
6bda09f9
YO
493/* reg_name[] - Opcode/state names in string form, for debugging */
494
22429478 495#ifndef DOINIT
13d6edb4 496EXTCONST char * PL_reg_name[];
22429478 497#else
4764e399 498EXTCONST char * const PL_reg_name[] = {
03363afd
YO
499 "END", /* 0000 */
500 "SUCCEED", /* 0x01 */
501 "BOL", /* 0x02 */
502 "MBOL", /* 0x03 */
503 "SBOL", /* 0x04 */
504 "EOS", /* 0x05 */
505 "EOL", /* 0x06 */
506 "MEOL", /* 0x07 */
507 "SEOL", /* 0x08 */
508 "BOUND", /* 0x09 */
509 "BOUNDL", /* 0x0a */
1e355c70 510 "BOUNDU", /* 0x0b */
0c6e81eb
KW
511 "BOUNDA", /* 0x0c */
512 "NBOUND", /* 0x0d */
513 "NBOUNDL", /* 0x0e */
514 "NBOUNDU", /* 0x0f */
515 "NBOUNDA", /* 0x10 */
516 "GPOS", /* 0x11 */
517 "REG_ANY", /* 0x12 */
518 "SANY", /* 0x13 */
519 "CANY", /* 0x14 */
520 "ANYOF", /* 0x15 */
71068078
KW
521 "POSIXD", /* 0x16 */
522 "POSIXL", /* 0x17 */
523 "POSIXU", /* 0x18 */
524 "POSIXA", /* 0x19 */
525 "NPOSIXD", /* 0x1a */
526 "NPOSIXL", /* 0x1b */
527 "NPOSIXU", /* 0x1c */
528 "NPOSIXA", /* 0x1d */
529 "CLUMP", /* 0x1e */
530 "BRANCH", /* 0x1f */
531 "BACK", /* 0x20 */
532 "EXACT", /* 0x21 */
533 "EXACTF", /* 0x22 */
534 "EXACTFL", /* 0x23 */
535 "EXACTFU", /* 0x24 */
536 "EXACTFA", /* 0x25 */
537 "EXACTFU_SS", /* 0x26 */
538 "EXACTFA_NO_TRIE", /* 0x27 */
539 "NOTHING", /* 0x28 */
540 "TAIL", /* 0x29 */
541 "STAR", /* 0x2a */
542 "PLUS", /* 0x2b */
543 "CURLY", /* 0x2c */
544 "CURLYN", /* 0x2d */
545 "CURLYM", /* 0x2e */
546 "CURLYX", /* 0x2f */
547 "WHILEM", /* 0x30 */
548 "OPEN", /* 0x31 */
549 "CLOSE", /* 0x32 */
550 "REF", /* 0x33 */
551 "REFF", /* 0x34 */
552 "REFFL", /* 0x35 */
553 "REFFU", /* 0x36 */
554 "REFFA", /* 0x37 */
555 "NREF", /* 0x38 */
556 "NREFF", /* 0x39 */
557 "NREFFL", /* 0x3a */
558 "NREFFU", /* 0x3b */
559 "NREFFA", /* 0x3c */
560 "IFMATCH", /* 0x3d */
561 "UNLESSM", /* 0x3e */
562 "SUSPEND", /* 0x3f */
563 "IFTHEN", /* 0x40 */
564 "GROUPP", /* 0x41 */
565 "LONGJMP", /* 0x42 */
566 "BRANCHJ", /* 0x43 */
567 "EVAL", /* 0x44 */
568 "MINMOD", /* 0x45 */
569 "LOGICAL", /* 0x46 */
570 "RENUM", /* 0x47 */
571 "TRIE", /* 0x48 */
572 "TRIEC", /* 0x49 */
573 "AHOCORASICK", /* 0x4a */
574 "AHOCORASICKC", /* 0x4b */
575 "GOSUB", /* 0x4c */
576 "GOSTART", /* 0x4d */
577 "NGROUPP", /* 0x4e */
578 "INSUBP", /* 0x4f */
579 "DEFINEP", /* 0x50 */
580 "ENDLIKE", /* 0x51 */
581 "OPFAIL", /* 0x52 */
582 "ACCEPT", /* 0x53 */
583 "VERB", /* 0x54 */
584 "PRUNE", /* 0x55 */
585 "MARKPOINT", /* 0x56 */
586 "SKIP", /* 0x57 */
587 "COMMIT", /* 0x58 */
588 "CUTGROUP", /* 0x59 */
589 "KEEPS", /* 0x5a */
590 "LNBREAK", /* 0x5b */
591 "OPTIMIZED", /* 0x5c */
592 "PSEUDO", /* 0x5d */
03363afd 593 /* ------------ States ------------- */
24b23f37
YO
594 "TRIE_next", /* REGNODE_MAX +0x01 */
595 "TRIE_next_fail", /* REGNODE_MAX +0x02 */
596 "EVAL_AB", /* REGNODE_MAX +0x03 */
597 "EVAL_AB_fail", /* REGNODE_MAX +0x04 */
598 "CURLYX_end", /* REGNODE_MAX +0x05 */
599 "CURLYX_end_fail", /* REGNODE_MAX +0x06 */
600 "WHILEM_A_pre", /* REGNODE_MAX +0x07 */
601 "WHILEM_A_pre_fail", /* REGNODE_MAX +0x08 */
602 "WHILEM_A_min", /* REGNODE_MAX +0x09 */
603 "WHILEM_A_min_fail", /* REGNODE_MAX +0x0a */
604 "WHILEM_A_max", /* REGNODE_MAX +0x0b */
605 "WHILEM_A_max_fail", /* REGNODE_MAX +0x0c */
606 "WHILEM_B_min", /* REGNODE_MAX +0x0d */
607 "WHILEM_B_min_fail", /* REGNODE_MAX +0x0e */
608 "WHILEM_B_max", /* REGNODE_MAX +0x0f */
609 "WHILEM_B_max_fail", /* REGNODE_MAX +0x10 */
610 "BRANCH_next", /* REGNODE_MAX +0x11 */
611 "BRANCH_next_fail", /* REGNODE_MAX +0x12 */
612 "CURLYM_A", /* REGNODE_MAX +0x13 */
613 "CURLYM_A_fail", /* REGNODE_MAX +0x14 */
614 "CURLYM_B", /* REGNODE_MAX +0x15 */
615 "CURLYM_B_fail", /* REGNODE_MAX +0x16 */
616 "IFMATCH_A", /* REGNODE_MAX +0x17 */
617 "IFMATCH_A_fail", /* REGNODE_MAX +0x18 */
618 "CURLY_B_min_known", /* REGNODE_MAX +0x19 */
619 "CURLY_B_min_known_fail", /* REGNODE_MAX +0x1a */
620 "CURLY_B_min", /* REGNODE_MAX +0x1b */
621 "CURLY_B_min_fail", /* REGNODE_MAX +0x1c */
622 "CURLY_B_max", /* REGNODE_MAX +0x1d */
623 "CURLY_B_max_fail", /* REGNODE_MAX +0x1e */
624 "COMMIT_next", /* REGNODE_MAX +0x1f */
625 "COMMIT_next_fail", /* REGNODE_MAX +0x20 */
e2e6a0f1
YO
626 "MARKPOINT_next", /* REGNODE_MAX +0x21 */
627 "MARKPOINT_next_fail", /* REGNODE_MAX +0x22 */
5d458dd8
YO
628 "SKIP_next", /* REGNODE_MAX +0x23 */
629 "SKIP_next_fail", /* REGNODE_MAX +0x24 */
630 "CUTGROUP_next", /* REGNODE_MAX +0x25 */
631 "CUTGROUP_next_fail", /* REGNODE_MAX +0x26 */
ee9b8eae
YO
632 "KEEPS_next", /* REGNODE_MAX +0x27 */
633 "KEEPS_next_fail", /* REGNODE_MAX +0x28 */
885f9e59 634};
22429478 635#endif /* DOINIT */
d09b2d29 636
f7819f85
A
637/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
638
639#ifndef DOINIT
640EXTCONST char * PL_reg_extflags_name[];
641#else
642EXTCONST char * const PL_reg_extflags_name[] = {
dbc200c5 643 /* Bits in extflags defined: 11111110111111111111111111111111 */
52d81aa8
NC
644 "MULTILINE", /* 0x00000001 */
645 "SINGLELINE", /* 0x00000002 */
646 "FOLD", /* 0x00000004 */
647 "EXTENDED", /* 0x00000008 */
648 "KEEPCOPY", /* 0x00000010 */
6976c986
YO
649 "CHARSET0", /* 0x00000020 : "CHARSET" - 0x000000e0 */
650 "CHARSET1", /* 0x00000040 : "CHARSET" - 0x000000e0 */
651 "CHARSET2", /* 0x00000080 : "CHARSET" - 0x000000e0 */
dbc200c5 652 "SPLIT", /* 0x00000100 */
8e1490ee
YO
653 "IS_ANCHORED", /* 0x00000200 */
654 "UNUSED1", /* 0x00000400 */
655 "UNUSED2", /* 0x00000800 */
656 "UNUSED3", /* 0x00001000 */
657 "UNUSED4", /* 0x00002000 */
658 "UNUSED5", /* 0x00004000 */
dbc200c5 659 "NO_INPLACE_SUBST", /* 0x00008000 */
e795e964 660 "EVAL_SEEN", /* 0x00010000 */
8e1490ee 661 "UNUSED8", /* 0x00020000 */
ee273784 662 "UNBOUNDED_QUANTIFIER_SEEN",/* 0x00040000 */
e795e964 663 "CHECK_ALL", /* 0x00080000 */
52d81aa8
NC
664 "MATCH_UTF8", /* 0x00100000 */
665 "USE_INTUIT_NOML", /* 0x00200000 */
666 "USE_INTUIT_ML", /* 0x00400000 */
667 "INTUIT_TAIL", /* 0x00800000 */
dbc200c5 668 "UNUSED_BIT_24", /* 0x01000000 */
52d81aa8
NC
669 "COPY_DONE", /* 0x02000000 */
670 "TAINTED_SEEN", /* 0x04000000 */
671 "TAINTED", /* 0x08000000 */
672 "START_ONLY", /* 0x10000000 */
dbc200c5 673 "SKIPWHITE", /* 0x20000000 */
52d81aa8
NC
674 "WHITE", /* 0x40000000 */
675 "NULL", /* 0x80000000 */
f7819f85
A
676};
677#endif /* DOINIT */
678
337ff307
YO
679/* PL_reg_intflags_name[] - Opcode/state names in string form, for debugging */
680
681#ifndef DOINIT
682EXTCONST char * PL_reg_intflags_name[];
683#else
684EXTCONST char * const PL_reg_intflags_name[] = {
b8f6efdd
YO
685 "SKIP", /* 0x00000001 - PREGf_SKIP */
686 "IMPLICIT", /* 0x00000002 - PREGf_IMPLICIT - Converted .* to ^.* */
687 "NAUGHTY", /* 0x00000004 - PREGf_NAUGHTY - how exponential is this pattern? */
688 "VERBARG_SEEN", /* 0x00000008 - PREGf_VERBARG_SEEN */
689 "CUTGROUP_SEEN", /* 0x00000010 - PREGf_CUTGROUP_SEEN */
690 "USE_RE_EVAL", /* 0x00000020 - PREGf_USE_RE_EVAL - compiled with "use re 'eval'" */
58430ea8
YO
691 "NOSCAN", /* 0x00000040 - PREGf_NOSCAN */
692 "CANY_SEEN", /* 0x00000080 - PREGf_CANY_SEEN */
693 "GPOS_SEEN", /* 0x00000100 - PREGf_GPOS_SEEN */
694 "GPOS_FLOAT", /* 0x00000200 - PREGf_GPOS_FLOAT */
8e1490ee
YO
695 "ANCH_BOL", /* 0x00000400 - PREGf_ANCH_BOL */
696 "ANCH_MBOL", /* 0x00000800 - PREGf_ANCH_MBOL */
697 "ANCH_SBOL", /* 0x00001000 - PREGf_ANCH_SBOL */
698 "ANCH_GPOS", /* 0x00002000 - PREGf_ANCH_GPOS */
337ff307
YO
699};
700#endif /* DOINIT */
701
f9ef50a7 702/* The following have no fixed length. U8 so we can do strchr() on it. */
ded4dd2a 703#define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 704
f9ef50a7 705#ifndef DOINIT
ded4dd2a 706EXTCONST U8 PL_varies[] __attribute__deprecated__;
f9ef50a7 707#else
ded4dd2a 708EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
e0193e47
KW
709 CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX, WHILEM,
710 REF, REFF, REFFL, REFFU, REFFA, NREF, NREFF, NREFFL, NREFFU, NREFFA,
711 SUSPEND, IFTHEN, BRANCHJ,
f9ef50a7
NC
712 0
713};
714#endif /* DOINIT */
715
ded4dd2a
NC
716#ifndef DOINIT
717EXTCONST U8 PL_varies_bitmask[];
718#else
719EXTCONST U8 PL_varies_bitmask[] = {
71068078 720 0x00, 0x00, 0x00, 0xC0, 0x01, 0xFC, 0xF9, 0x9F, 0x09, 0x00, 0x00, 0x00
ded4dd2a
NC
721};
722#endif /* DOINIT */
723
f9ef50a7
NC
724/* The following always have a length of 1. U8 we can do strchr() on it. */
725/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
ded4dd2a 726#define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 727
f9ef50a7 728#ifndef DOINIT
ded4dd2a 729EXTCONST U8 PL_simple[] __attribute__deprecated__;
f9ef50a7 730#else
ded4dd2a 731EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
71068078
KW
732 REG_ANY, SANY, CANY, ANYOF, POSIXD, POSIXL, POSIXU, POSIXA, NPOSIXD,
733 NPOSIXL, NPOSIXU, NPOSIXA,
f9ef50a7
NC
734 0
735};
736#endif /* DOINIT */
737
ded4dd2a
NC
738#ifndef DOINIT
739EXTCONST U8 PL_simple_bitmask[];
740#else
741EXTCONST U8 PL_simple_bitmask[] = {
71068078 742 0x00, 0x00, 0xFC, 0x3F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
ded4dd2a
NC
743};
744#endif /* DOINIT */
745
37442d52 746/* ex: set ro: */