This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
regcomp.c: Simply some node calculations
[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
693fefec
KW
9#define REGNODE_MAX 114
10#define REGMATCH_STATE_MAX 154
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 */
34#define ANYOFV 22 /* 0x16 Match character in (or not in) this class, can match-multiple chars */
35#define ALNUM 23 /* 0x17 Match any alphanumeric character using native charset semantics for non-utf8 */
36#define ALNUML 24 /* 0x18 Match any alphanumeric char in locale */
37#define ALNUMU 25 /* 0x19 Match any alphanumeric char using Unicode semantics */
38#define ALNUMA 26 /* 0x1a Match [A-Za-z_0-9] */
39#define NALNUM 27 /* 0x1b Match any non-alphanumeric character using native charset semantics for non-utf8 */
40#define NALNUML 28 /* 0x1c Match any non-alphanumeric char in locale */
41#define NALNUMU 29 /* 0x1d Match any non-alphanumeric char using Unicode semantics */
42#define NALNUMA 30 /* 0x1e Match [^A-Za-z_0-9] */
43#define SPACE 31 /* 0x1f Match any whitespace character using native charset semantics for non-utf8 */
44#define SPACEL 32 /* 0x20 Match any whitespace char in locale */
45#define SPACEU 33 /* 0x21 Match any whitespace char using Unicode semantics */
46#define SPACEA 34 /* 0x22 Match [ \t\n\f\r] */
47#define NSPACE 35 /* 0x23 Match any non-whitespace character using native charset semantics for non-utf8 */
48#define NSPACEL 36 /* 0x24 Match any non-whitespace char in locale */
49#define NSPACEU 37 /* 0x25 Match any non-whitespace char using Unicode semantics */
50#define NSPACEA 38 /* 0x26 Match [^ \t\n\f\r] */
51#define DIGIT 39 /* 0x27 Match any numeric character using native charset semantics for non-utf8 */
52#define DIGITL 40 /* 0x28 Match any numeric character in locale */
693fefec
KW
53#define PLACEHOLDER1 41 /* 0x29 placeholder for missing DIGITU */
54#define DIGITA 42 /* 0x2a Match [0-9] */
55#define NDIGIT 43 /* 0x2b Match any non-numeric character using native charset semantics for non-utf8 */
56#define NDIGITL 44 /* 0x2c Match any non-numeric character in locale */
57#define PLACEHOLDER2 45 /* 0x2d placeholder for missing NDIGITU */
58#define NDIGITA 46 /* 0x2e Match [^0-9] */
59#define CLUMP 47 /* 0x2f Match any extended grapheme cluster sequence */
60#define BRANCH 48 /* 0x30 Match this alternative, or the next... */
61#define BACK 49 /* 0x31 Match "", "next" ptr points backward. */
62#define EXACT 50 /* 0x32 Match this string (preceded by length). */
63#define EXACTF 51 /* 0x33 Match this non-UTF-8 string (not guaranteed to be folded) using /id rules (w/len). */
64#define EXACTFL 52 /* 0x34 Match this string (not guaranteed to be folded) using /il rules (w/len). */
65#define EXACTFU 53 /* 0x35 Match this string (folded iff in UTF-8, length in folding doesn't change if not in UTF-8) using /iu rules (w/len). */
66#define EXACTFA 54 /* 0x36 Match this string (not guaranteed to be folded) using /iaa rules (w/len). */
67#define EXACTFU_SS 55 /* 0x37 Match this string (folded iff in UTF-8, length in folding may change even if not in UTF-8) using /iu rules (w/len). */
68#define EXACTFU_TRICKYFOLD 56 /* 0x38 Match this folded UTF-8 string using /iu rules */
69#define NOTHING 57 /* 0x39 Match empty string. */
70#define TAIL 58 /* 0x3a Match empty string. Can jump here from outside. */
71#define STAR 59 /* 0x3b Match this (simple) thing 0 or more times. */
72#define PLUS 60 /* 0x3c Match this (simple) thing 1 or more times. */
73#define CURLY 61 /* 0x3d Match this simple thing {n,m} times. */
74#define CURLYN 62 /* 0x3e Capture next-after-this simple thing */
75#define CURLYM 63 /* 0x3f Capture this medium-complex thing {n,m} times. */
76#define CURLYX 64 /* 0x40 Match this complex thing {n,m} times. */
77#define WHILEM 65 /* 0x41 Do curly processing and see if rest matches. */
78#define OPEN 66 /* 0x42 Mark this point in input as start of */
79#define CLOSE 67 /* 0x43 Analogous to OPEN. */
80#define REF 68 /* 0x44 Match some already matched string */
81#define REFF 69 /* 0x45 Match already matched string, folded using native charset semantics for non-utf8 */
82#define REFFL 70 /* 0x46 Match already matched string, folded in loc. */
83#define REFFU 71 /* 0x47 Match already matched string, folded using unicode semantics for non-utf8 */
84#define REFFA 72 /* 0x48 Match already matched string, folded using unicode semantics for non-utf8, no mixing ASCII, non-ASCII */
85#define NREF 73 /* 0x49 Match some already matched string */
86#define NREFF 74 /* 0x4a Match already matched string, folded using native charset semantics for non-utf8 */
87#define NREFFL 75 /* 0x4b Match already matched string, folded in loc. */
88#define NREFFU 76 /* 0x4c Match already matched string, folded using unicode semantics for non-utf8 */
89#define NREFFA 77 /* 0x4d Match already matched string, folded using unicode semantics for non-utf8, no mixing ASCII, non-ASCII */
90#define IFMATCH 78 /* 0x4e Succeeds if the following matches. */
91#define UNLESSM 79 /* 0x4f Fails if the following matches. */
92#define SUSPEND 80 /* 0x50 "Independent" sub-RE. */
93#define IFTHEN 81 /* 0x51 Switch, should be preceded by switcher . */
94#define GROUPP 82 /* 0x52 Whether the group matched. */
95#define LONGJMP 83 /* 0x53 Jump far away. */
96#define BRANCHJ 84 /* 0x54 BRANCH with long offset. */
97#define EVAL 85 /* 0x55 Execute some Perl code. */
98#define MINMOD 86 /* 0x56 Next operator is not greedy. */
99#define LOGICAL 87 /* 0x57 Next opcode should set the flag only. */
100#define RENUM 88 /* 0x58 Group with independently numbered parens. */
101#define TRIE 89 /* 0x59 Match many EXACT(F[ALU]?)? at once. flags==type */
102#define TRIEC 90 /* 0x5a Same as TRIE, but with embedded charclass data */
103#define AHOCORASICK 91 /* 0x5b Aho Corasick stclass. flags==type */
104#define AHOCORASICKC 92 /* 0x5c Same as AHOCORASICK, but with embedded charclass data */
105#define GOSUB 93 /* 0x5d recurse to paren arg1 at (signed) ofs arg2 */
106#define GOSTART 94 /* 0x5e recurse to start of pattern */
107#define NGROUPP 95 /* 0x5f Whether the group matched. */
108#define INSUBP 96 /* 0x60 Whether we are in a specific recurse. */
109#define DEFINEP 97 /* 0x61 Never execute directly. */
110#define ENDLIKE 98 /* 0x62 Used only for the type field of verbs */
111#define OPFAIL 99 /* 0x63 Same as (?!) */
112#define ACCEPT 100 /* 0x64 Accepts the current matched string. */
113#define VERB 101 /* 0x65 Used only for the type field of verbs */
114#define PRUNE 102 /* 0x66 Pattern fails at this startpoint if no-backtracking through this */
115#define MARKPOINT 103 /* 0x67 Push the current location for rollback by cut. */
116#define SKIP 104 /* 0x68 On failure skip forward (to the mark) before retrying */
117#define COMMIT 105 /* 0x69 Pattern fails outright if backtracking through this */
118#define CUTGROUP 106 /* 0x6a On failure go to the next alternation in the group */
119#define KEEPS 107 /* 0x6b $& begins here. */
120#define LNBREAK 108 /* 0x6c generic newline pattern */
121#define VERTWS 109 /* 0x6d vertical whitespace (Perl 6) */
122#define NVERTWS 110 /* 0x6e not vertical whitespace (Perl 6) */
123#define HORIZWS 111 /* 0x6f horizontal whitespace (Perl 6) */
124#define NHORIZWS 112 /* 0x70 not horizontal whitespace (Perl 6) */
125#define OPTIMIZED 113 /* 0x71 Placeholder for dump. */
126#define PSEUDO 114 /* 0x72 Pseudo opcode for internal use. */
03363afd 127 /* ------------ States ------------- */
24b23f37
YO
128#define TRIE_next (REGNODE_MAX + 1) /* state for TRIE */
129#define TRIE_next_fail (REGNODE_MAX + 2) /* state for TRIE */
130#define EVAL_AB (REGNODE_MAX + 3) /* state for EVAL */
131#define EVAL_AB_fail (REGNODE_MAX + 4) /* state for EVAL */
132#define CURLYX_end (REGNODE_MAX + 5) /* state for CURLYX */
133#define CURLYX_end_fail (REGNODE_MAX + 6) /* state for CURLYX */
134#define WHILEM_A_pre (REGNODE_MAX + 7) /* state for WHILEM */
135#define WHILEM_A_pre_fail (REGNODE_MAX + 8) /* state for WHILEM */
136#define WHILEM_A_min (REGNODE_MAX + 9) /* state for WHILEM */
137#define WHILEM_A_min_fail (REGNODE_MAX + 10) /* state for WHILEM */
138#define WHILEM_A_max (REGNODE_MAX + 11) /* state for WHILEM */
139#define WHILEM_A_max_fail (REGNODE_MAX + 12) /* state for WHILEM */
140#define WHILEM_B_min (REGNODE_MAX + 13) /* state for WHILEM */
141#define WHILEM_B_min_fail (REGNODE_MAX + 14) /* state for WHILEM */
142#define WHILEM_B_max (REGNODE_MAX + 15) /* state for WHILEM */
143#define WHILEM_B_max_fail (REGNODE_MAX + 16) /* state for WHILEM */
144#define BRANCH_next (REGNODE_MAX + 17) /* state for BRANCH */
145#define BRANCH_next_fail (REGNODE_MAX + 18) /* state for BRANCH */
146#define CURLYM_A (REGNODE_MAX + 19) /* state for CURLYM */
147#define CURLYM_A_fail (REGNODE_MAX + 20) /* state for CURLYM */
148#define CURLYM_B (REGNODE_MAX + 21) /* state for CURLYM */
149#define CURLYM_B_fail (REGNODE_MAX + 22) /* state for CURLYM */
150#define IFMATCH_A (REGNODE_MAX + 23) /* state for IFMATCH */
151#define IFMATCH_A_fail (REGNODE_MAX + 24) /* state for IFMATCH */
152#define CURLY_B_min_known (REGNODE_MAX + 25) /* state for CURLY */
153#define CURLY_B_min_known_fail (REGNODE_MAX + 26) /* state for CURLY */
154#define CURLY_B_min (REGNODE_MAX + 27) /* state for CURLY */
155#define CURLY_B_min_fail (REGNODE_MAX + 28) /* state for CURLY */
156#define CURLY_B_max (REGNODE_MAX + 29) /* state for CURLY */
157#define CURLY_B_max_fail (REGNODE_MAX + 30) /* state for CURLY */
158#define COMMIT_next (REGNODE_MAX + 31) /* state for COMMIT */
159#define COMMIT_next_fail (REGNODE_MAX + 32) /* state for COMMIT */
e2e6a0f1
YO
160#define MARKPOINT_next (REGNODE_MAX + 33) /* state for MARKPOINT */
161#define MARKPOINT_next_fail (REGNODE_MAX + 34) /* state for MARKPOINT */
5d458dd8
YO
162#define SKIP_next (REGNODE_MAX + 35) /* state for SKIP */
163#define SKIP_next_fail (REGNODE_MAX + 36) /* state for SKIP */
164#define CUTGROUP_next (REGNODE_MAX + 37) /* state for CUTGROUP */
165#define CUTGROUP_next_fail (REGNODE_MAX + 38) /* state for CUTGROUP */
ee9b8eae
YO
166#define KEEPS_next (REGNODE_MAX + 39) /* state for KEEPS */
167#define KEEPS_next_fail (REGNODE_MAX + 40) /* state for KEEPS */
03363afd 168
6bda09f9 169/* PL_regkind[] What type of regop or state is this. */
d09b2d29
IZ
170
171#ifndef DOINIT
22c35a8c 172EXTCONST U8 PL_regkind[];
d09b2d29 173#else
22c35a8c 174EXTCONST U8 PL_regkind[] = {
e2e6a0f1
YO
175 END, /* END */
176 END, /* SUCCEED */
177 BOL, /* BOL */
178 BOL, /* MBOL */
179 BOL, /* SBOL */
180 EOL, /* EOS */
181 EOL, /* EOL */
182 EOL, /* MEOL */
183 EOL, /* SEOL */
184 BOUND, /* BOUND */
185 BOUND, /* BOUNDL */
1e355c70 186 BOUND, /* BOUNDU */
0c6e81eb 187 BOUND, /* BOUNDA */
e2e6a0f1
YO
188 NBOUND, /* NBOUND */
189 NBOUND, /* NBOUNDL */
1e355c70 190 NBOUND, /* NBOUNDU */
0c6e81eb 191 NBOUND, /* NBOUNDA */
e2e6a0f1
YO
192 GPOS, /* GPOS */
193 REG_ANY, /* REG_ANY */
194 REG_ANY, /* SANY */
195 REG_ANY, /* CANY */
196 ANYOF, /* ANYOF */
0e019ad6 197 ANYOF, /* ANYOFV */
e2e6a0f1
YO
198 ALNUM, /* ALNUM */
199 ALNUM, /* ALNUML */
fdf48794 200 ALNUM, /* ALNUMU */
0c6e81eb 201 ALNUM, /* ALNUMA */
e2e6a0f1
YO
202 NALNUM, /* NALNUM */
203 NALNUM, /* NALNUML */
fdf48794 204 NALNUM, /* NALNUMU */
0c6e81eb 205 NALNUM, /* NALNUMA */
e2e6a0f1
YO
206 SPACE, /* SPACE */
207 SPACE, /* SPACEL */
fdf48794 208 SPACE, /* SPACEU */
0c6e81eb 209 SPACE, /* SPACEA */
e2e6a0f1
YO
210 NSPACE, /* NSPACE */
211 NSPACE, /* NSPACEL */
fdf48794 212 NSPACE, /* NSPACEU */
0c6e81eb 213 NSPACE, /* NSPACEA */
e2e6a0f1
YO
214 DIGIT, /* DIGIT */
215 DIGIT, /* DIGITL */
693fefec 216 NOTHING, /* PLACEHOLDER1 */
0c6e81eb 217 DIGIT, /* DIGITA */
e2e6a0f1
YO
218 NDIGIT, /* NDIGIT */
219 NDIGIT, /* NDIGITL */
693fefec 220 NOTHING, /* PLACEHOLDER2 */
0c6e81eb 221 NDIGIT, /* NDIGITA */
e2e6a0f1
YO
222 CLUMP, /* CLUMP */
223 BRANCH, /* BRANCH */
224 BACK, /* BACK */
225 EXACT, /* EXACT */
226 EXACT, /* EXACTF */
227 EXACT, /* EXACTFL */
01f98ec2 228 EXACT, /* EXACTFU */
8c1182fd 229 EXACT, /* EXACTFA */
3c760661 230 EXACT, /* EXACTFU_SS */
fab2782b 231 EXACT, /* EXACTFU_TRICKYFOLD */
e2e6a0f1
YO
232 NOTHING, /* NOTHING */
233 NOTHING, /* TAIL */
234 STAR, /* STAR */
235 PLUS, /* PLUS */
236 CURLY, /* CURLY */
237 CURLY, /* CURLYN */
238 CURLY, /* CURLYM */
239 CURLY, /* CURLYX */
240 WHILEM, /* WHILEM */
241 OPEN, /* OPEN */
242 CLOSE, /* CLOSE */
243 REF, /* REF */
244 REF, /* REFF */
245 REF, /* REFFL */
01f98ec2 246 REF, /* REFFU */
781aab5c 247 REF, /* REFFA */
01f98ec2
KW
248 REF, /* NREF */
249 REF, /* NREFF */
250 REF, /* NREFFL */
251 REF, /* NREFFU */
781aab5c 252 REF, /* NREFFA */
e2e6a0f1
YO
253 BRANCHJ, /* IFMATCH */
254 BRANCHJ, /* UNLESSM */
255 BRANCHJ, /* SUSPEND */
256 BRANCHJ, /* IFTHEN */
257 GROUPP, /* GROUPP */
258 LONGJMP, /* LONGJMP */
259 BRANCHJ, /* BRANCHJ */
260 EVAL, /* EVAL */
261 MINMOD, /* MINMOD */
262 LOGICAL, /* LOGICAL */
263 BRANCHJ, /* RENUM */
264 TRIE, /* TRIE */
265 TRIE, /* TRIEC */
266 TRIE, /* AHOCORASICK */
267 TRIE, /* AHOCORASICKC */
268 GOSUB, /* GOSUB */
269 GOSTART, /* GOSTART */
e2e6a0f1
YO
270 NGROUPP, /* NGROUPP */
271 INSUBP, /* INSUBP */
272 DEFINEP, /* DEFINEP */
273 ENDLIKE, /* ENDLIKE */
274 ENDLIKE, /* OPFAIL */
275 ENDLIKE, /* ACCEPT */
276 VERB, /* VERB */
5d458dd8 277 VERB, /* PRUNE */
e2e6a0f1 278 VERB, /* MARKPOINT */
5d458dd8 279 VERB, /* SKIP */
e2e6a0f1 280 VERB, /* COMMIT */
5d458dd8 281 VERB, /* CUTGROUP */
ee9b8eae 282 KEEPS, /* KEEPS */
e1d1eefb
YO
283 LNBREAK, /* LNBREAK */
284 VERTWS, /* VERTWS */
285 NVERTWS, /* NVERTWS */
286 HORIZWS, /* HORIZWS */
287 NHORIZWS, /* NHORIZWS */
e2e6a0f1
YO
288 NOTHING, /* OPTIMIZED */
289 PSEUDO, /* PSEUDO */
03363afd 290 /* ------------ States ------------- */
e2e6a0f1
YO
291 TRIE, /* TRIE_next */
292 TRIE, /* TRIE_next_fail */
293 EVAL, /* EVAL_AB */
294 EVAL, /* EVAL_AB_fail */
295 CURLYX, /* CURLYX_end */
296 CURLYX, /* CURLYX_end_fail */
297 WHILEM, /* WHILEM_A_pre */
298 WHILEM, /* WHILEM_A_pre_fail */
299 WHILEM, /* WHILEM_A_min */
300 WHILEM, /* WHILEM_A_min_fail */
301 WHILEM, /* WHILEM_A_max */
302 WHILEM, /* WHILEM_A_max_fail */
303 WHILEM, /* WHILEM_B_min */
304 WHILEM, /* WHILEM_B_min_fail */
305 WHILEM, /* WHILEM_B_max */
306 WHILEM, /* WHILEM_B_max_fail */
307 BRANCH, /* BRANCH_next */
308 BRANCH, /* BRANCH_next_fail */
309 CURLYM, /* CURLYM_A */
310 CURLYM, /* CURLYM_A_fail */
311 CURLYM, /* CURLYM_B */
312 CURLYM, /* CURLYM_B_fail */
313 IFMATCH, /* IFMATCH_A */
314 IFMATCH, /* IFMATCH_A_fail */
315 CURLY, /* CURLY_B_min_known */
316 CURLY, /* CURLY_B_min_known_fail */
317 CURLY, /* CURLY_B_min */
318 CURLY, /* CURLY_B_min_fail */
319 CURLY, /* CURLY_B_max */
320 CURLY, /* CURLY_B_max_fail */
321 COMMIT, /* COMMIT_next */
322 COMMIT, /* COMMIT_next_fail */
323 MARKPOINT, /* MARKPOINT_next */
324 MARKPOINT, /* MARKPOINT_next_fail */
5d458dd8
YO
325 SKIP, /* SKIP_next */
326 SKIP, /* SKIP_next_fail */
327 CUTGROUP, /* CUTGROUP_next */
328 CUTGROUP, /* CUTGROUP_next_fail */
ee9b8eae
YO
329 KEEPS, /* KEEPS_next */
330 KEEPS, /* KEEPS_next_fail */
d09b2d29
IZ
331};
332#endif
333
6bda09f9 334/* regarglen[] - How large is the argument part of the node (in regnodes) */
d09b2d29
IZ
335
336#ifdef REG_COMP_C
29de9391 337static const U8 regarglen[] = {
03363afd
YO
338 0, /* END */
339 0, /* SUCCEED */
340 0, /* BOL */
341 0, /* MBOL */
342 0, /* SBOL */
343 0, /* EOS */
344 0, /* EOL */
345 0, /* MEOL */
346 0, /* SEOL */
347 0, /* BOUND */
348 0, /* BOUNDL */
1e355c70 349 0, /* BOUNDU */
0c6e81eb 350 0, /* BOUNDA */
03363afd
YO
351 0, /* NBOUND */
352 0, /* NBOUNDL */
1e355c70 353 0, /* NBOUNDU */
0c6e81eb 354 0, /* NBOUNDA */
03363afd
YO
355 0, /* GPOS */
356 0, /* REG_ANY */
357 0, /* SANY */
358 0, /* CANY */
359 0, /* ANYOF */
0e019ad6 360 0, /* ANYOFV */
03363afd
YO
361 0, /* ALNUM */
362 0, /* ALNUML */
fdf48794 363 0, /* ALNUMU */
0c6e81eb 364 0, /* ALNUMA */
03363afd
YO
365 0, /* NALNUM */
366 0, /* NALNUML */
fdf48794 367 0, /* NALNUMU */
0c6e81eb 368 0, /* NALNUMA */
03363afd
YO
369 0, /* SPACE */
370 0, /* SPACEL */
fdf48794 371 0, /* SPACEU */
0c6e81eb 372 0, /* SPACEA */
03363afd
YO
373 0, /* NSPACE */
374 0, /* NSPACEL */
fdf48794 375 0, /* NSPACEU */
0c6e81eb 376 0, /* NSPACEA */
03363afd
YO
377 0, /* DIGIT */
378 0, /* DIGITL */
693fefec 379 0, /* PLACEHOLDER1 */
0c6e81eb 380 0, /* DIGITA */
03363afd
YO
381 0, /* NDIGIT */
382 0, /* NDIGITL */
693fefec 383 0, /* PLACEHOLDER2 */
0c6e81eb 384 0, /* NDIGITA */
03363afd
YO
385 0, /* CLUMP */
386 0, /* BRANCH */
387 0, /* BACK */
388 0, /* EXACT */
389 0, /* EXACTF */
390 0, /* EXACTFL */
01f98ec2 391 0, /* EXACTFU */
8c1182fd 392 0, /* EXACTFA */
3c760661 393 0, /* EXACTFU_SS */
fab2782b 394 0, /* EXACTFU_TRICKYFOLD */
03363afd
YO
395 0, /* NOTHING */
396 0, /* TAIL */
397 0, /* STAR */
398 0, /* PLUS */
399 EXTRA_SIZE(struct regnode_2), /* CURLY */
400 EXTRA_SIZE(struct regnode_2), /* CURLYN */
401 EXTRA_SIZE(struct regnode_2), /* CURLYM */
402 EXTRA_SIZE(struct regnode_2), /* CURLYX */
403 0, /* WHILEM */
404 EXTRA_SIZE(struct regnode_1), /* OPEN */
405 EXTRA_SIZE(struct regnode_1), /* CLOSE */
406 EXTRA_SIZE(struct regnode_1), /* REF */
407 EXTRA_SIZE(struct regnode_1), /* REFF */
408 EXTRA_SIZE(struct regnode_1), /* REFFL */
01f98ec2 409 EXTRA_SIZE(struct regnode_1), /* REFFU */
781aab5c 410 EXTRA_SIZE(struct regnode_1), /* REFFA */
01f98ec2
KW
411 EXTRA_SIZE(struct regnode_1), /* NREF */
412 EXTRA_SIZE(struct regnode_1), /* NREFF */
413 EXTRA_SIZE(struct regnode_1), /* NREFFL */
414 EXTRA_SIZE(struct regnode_1), /* NREFFU */
781aab5c 415 EXTRA_SIZE(struct regnode_1), /* NREFFA */
03363afd
YO
416 EXTRA_SIZE(struct regnode_1), /* IFMATCH */
417 EXTRA_SIZE(struct regnode_1), /* UNLESSM */
418 EXTRA_SIZE(struct regnode_1), /* SUSPEND */
419 EXTRA_SIZE(struct regnode_1), /* IFTHEN */
420 EXTRA_SIZE(struct regnode_1), /* GROUPP */
421 EXTRA_SIZE(struct regnode_1), /* LONGJMP */
422 EXTRA_SIZE(struct regnode_1), /* BRANCHJ */
423 EXTRA_SIZE(struct regnode_1), /* EVAL */
424 0, /* MINMOD */
425 0, /* LOGICAL */
426 EXTRA_SIZE(struct regnode_1), /* RENUM */
427 EXTRA_SIZE(struct regnode_1), /* TRIE */
428 EXTRA_SIZE(struct regnode_charclass), /* TRIEC */
429 EXTRA_SIZE(struct regnode_1), /* AHOCORASICK */
430 EXTRA_SIZE(struct regnode_charclass), /* AHOCORASICKC */
1a147d38
YO
431 EXTRA_SIZE(struct regnode_2L), /* GOSUB */
432 0, /* GOSTART */
0a4db386 433 EXTRA_SIZE(struct regnode_1), /* NGROUPP */
1a147d38 434 EXTRA_SIZE(struct regnode_1), /* INSUBP */
0a4db386 435 EXTRA_SIZE(struct regnode_1), /* DEFINEP */
e2e6a0f1 436 0, /* ENDLIKE */
7f69552c 437 0, /* OPFAIL */
e2e6a0f1 438 EXTRA_SIZE(struct regnode_1), /* ACCEPT */
20832bc5 439 EXTRA_SIZE(struct regnode_1), /* VERB */
5d458dd8 440 EXTRA_SIZE(struct regnode_1), /* PRUNE */
e2e6a0f1 441 EXTRA_SIZE(struct regnode_1), /* MARKPOINT */
5d458dd8 442 EXTRA_SIZE(struct regnode_1), /* SKIP */
e2e6a0f1 443 EXTRA_SIZE(struct regnode_1), /* COMMIT */
5d458dd8 444 EXTRA_SIZE(struct regnode_1), /* CUTGROUP */
ee9b8eae 445 0, /* KEEPS */
e1d1eefb
YO
446 0, /* LNBREAK */
447 0, /* VERTWS */
448 0, /* NVERTWS */
449 0, /* HORIZWS */
450 0, /* NHORIZWS */
03363afd
YO
451 0, /* OPTIMIZED */
452 0, /* PSEUDO */
d09b2d29
IZ
453};
454
6bda09f9
YO
455/* reg_off_by_arg[] - Which argument holds the offset to the next node */
456
29de9391 457static const char reg_off_by_arg[] = {
03363afd
YO
458 0, /* END */
459 0, /* SUCCEED */
460 0, /* BOL */
461 0, /* MBOL */
462 0, /* SBOL */
463 0, /* EOS */
464 0, /* EOL */
465 0, /* MEOL */
466 0, /* SEOL */
467 0, /* BOUND */
468 0, /* BOUNDL */
1e355c70 469 0, /* BOUNDU */
0c6e81eb 470 0, /* BOUNDA */
03363afd
YO
471 0, /* NBOUND */
472 0, /* NBOUNDL */
1e355c70 473 0, /* NBOUNDU */
0c6e81eb 474 0, /* NBOUNDA */
03363afd
YO
475 0, /* GPOS */
476 0, /* REG_ANY */
477 0, /* SANY */
478 0, /* CANY */
479 0, /* ANYOF */
0e019ad6 480 0, /* ANYOFV */
03363afd
YO
481 0, /* ALNUM */
482 0, /* ALNUML */
fdf48794 483 0, /* ALNUMU */
0c6e81eb 484 0, /* ALNUMA */
03363afd
YO
485 0, /* NALNUM */
486 0, /* NALNUML */
fdf48794 487 0, /* NALNUMU */
0c6e81eb 488 0, /* NALNUMA */
03363afd
YO
489 0, /* SPACE */
490 0, /* SPACEL */
fdf48794 491 0, /* SPACEU */
0c6e81eb 492 0, /* SPACEA */
03363afd
YO
493 0, /* NSPACE */
494 0, /* NSPACEL */
fdf48794 495 0, /* NSPACEU */
0c6e81eb 496 0, /* NSPACEA */
03363afd
YO
497 0, /* DIGIT */
498 0, /* DIGITL */
693fefec 499 0, /* PLACEHOLDER1 */
0c6e81eb 500 0, /* DIGITA */
03363afd
YO
501 0, /* NDIGIT */
502 0, /* NDIGITL */
693fefec 503 0, /* PLACEHOLDER2 */
0c6e81eb 504 0, /* NDIGITA */
03363afd
YO
505 0, /* CLUMP */
506 0, /* BRANCH */
507 0, /* BACK */
508 0, /* EXACT */
509 0, /* EXACTF */
510 0, /* EXACTFL */
01f98ec2 511 0, /* EXACTFU */
8c1182fd 512 0, /* EXACTFA */
3c760661 513 0, /* EXACTFU_SS */
fab2782b 514 0, /* EXACTFU_TRICKYFOLD */
03363afd
YO
515 0, /* NOTHING */
516 0, /* TAIL */
517 0, /* STAR */
518 0, /* PLUS */
519 0, /* CURLY */
520 0, /* CURLYN */
521 0, /* CURLYM */
522 0, /* CURLYX */
523 0, /* WHILEM */
524 0, /* OPEN */
525 0, /* CLOSE */
526 0, /* REF */
527 0, /* REFF */
528 0, /* REFFL */
01f98ec2 529 0, /* REFFU */
781aab5c 530 0, /* REFFA */
01f98ec2
KW
531 0, /* NREF */
532 0, /* NREFF */
533 0, /* NREFFL */
534 0, /* NREFFU */
781aab5c 535 0, /* NREFFA */
03363afd
YO
536 2, /* IFMATCH */
537 2, /* UNLESSM */
538 1, /* SUSPEND */
539 1, /* IFTHEN */
540 0, /* GROUPP */
541 1, /* LONGJMP */
542 1, /* BRANCHJ */
543 0, /* EVAL */
544 0, /* MINMOD */
545 0, /* LOGICAL */
546 1, /* RENUM */
547 0, /* TRIE */
548 0, /* TRIEC */
549 0, /* AHOCORASICK */
550 0, /* AHOCORASICKC */
1a147d38
YO
551 0, /* GOSUB */
552 0, /* GOSTART */
0a4db386 553 0, /* NGROUPP */
1a147d38 554 0, /* INSUBP */
0a4db386 555 0, /* DEFINEP */
e2e6a0f1 556 0, /* ENDLIKE */
7f69552c 557 0, /* OPFAIL */
e2e6a0f1
YO
558 0, /* ACCEPT */
559 0, /* VERB */
5d458dd8 560 0, /* PRUNE */
e2e6a0f1 561 0, /* MARKPOINT */
5d458dd8 562 0, /* SKIP */
e2e6a0f1 563 0, /* COMMIT */
5d458dd8 564 0, /* CUTGROUP */
ee9b8eae 565 0, /* KEEPS */
e1d1eefb
YO
566 0, /* LNBREAK */
567 0, /* VERTWS */
568 0, /* NVERTWS */
569 0, /* HORIZWS */
570 0, /* NHORIZWS */
03363afd
YO
571 0, /* OPTIMIZED */
572 0, /* PSEUDO */
d09b2d29 573};
885f9e59 574
13d6edb4
NC
575#endif /* REG_COMP_C */
576
6bda09f9
YO
577/* reg_name[] - Opcode/state names in string form, for debugging */
578
22429478 579#ifndef DOINIT
13d6edb4 580EXTCONST char * PL_reg_name[];
22429478 581#else
4764e399 582EXTCONST char * const PL_reg_name[] = {
03363afd
YO
583 "END", /* 0000 */
584 "SUCCEED", /* 0x01 */
585 "BOL", /* 0x02 */
586 "MBOL", /* 0x03 */
587 "SBOL", /* 0x04 */
588 "EOS", /* 0x05 */
589 "EOL", /* 0x06 */
590 "MEOL", /* 0x07 */
591 "SEOL", /* 0x08 */
592 "BOUND", /* 0x09 */
593 "BOUNDL", /* 0x0a */
1e355c70 594 "BOUNDU", /* 0x0b */
0c6e81eb
KW
595 "BOUNDA", /* 0x0c */
596 "NBOUND", /* 0x0d */
597 "NBOUNDL", /* 0x0e */
598 "NBOUNDU", /* 0x0f */
599 "NBOUNDA", /* 0x10 */
600 "GPOS", /* 0x11 */
601 "REG_ANY", /* 0x12 */
602 "SANY", /* 0x13 */
603 "CANY", /* 0x14 */
604 "ANYOF", /* 0x15 */
605 "ANYOFV", /* 0x16 */
606 "ALNUM", /* 0x17 */
607 "ALNUML", /* 0x18 */
608 "ALNUMU", /* 0x19 */
609 "ALNUMA", /* 0x1a */
610 "NALNUM", /* 0x1b */
611 "NALNUML", /* 0x1c */
612 "NALNUMU", /* 0x1d */
613 "NALNUMA", /* 0x1e */
614 "SPACE", /* 0x1f */
615 "SPACEL", /* 0x20 */
616 "SPACEU", /* 0x21 */
617 "SPACEA", /* 0x22 */
618 "NSPACE", /* 0x23 */
619 "NSPACEL", /* 0x24 */
620 "NSPACEU", /* 0x25 */
621 "NSPACEA", /* 0x26 */
622 "DIGIT", /* 0x27 */
623 "DIGITL", /* 0x28 */
693fefec
KW
624 "PLACEHOLDER1", /* 0x29 */
625 "DIGITA", /* 0x2a */
626 "NDIGIT", /* 0x2b */
627 "NDIGITL", /* 0x2c */
628 "PLACEHOLDER2", /* 0x2d */
629 "NDIGITA", /* 0x2e */
630 "CLUMP", /* 0x2f */
631 "BRANCH", /* 0x30 */
632 "BACK", /* 0x31 */
633 "EXACT", /* 0x32 */
634 "EXACTF", /* 0x33 */
635 "EXACTFL", /* 0x34 */
636 "EXACTFU", /* 0x35 */
637 "EXACTFA", /* 0x36 */
638 "EXACTFU_SS", /* 0x37 */
639 "EXACTFU_TRICKYFOLD", /* 0x38 */
640 "NOTHING", /* 0x39 */
641 "TAIL", /* 0x3a */
642 "STAR", /* 0x3b */
643 "PLUS", /* 0x3c */
644 "CURLY", /* 0x3d */
645 "CURLYN", /* 0x3e */
646 "CURLYM", /* 0x3f */
647 "CURLYX", /* 0x40 */
648 "WHILEM", /* 0x41 */
649 "OPEN", /* 0x42 */
650 "CLOSE", /* 0x43 */
651 "REF", /* 0x44 */
652 "REFF", /* 0x45 */
653 "REFFL", /* 0x46 */
654 "REFFU", /* 0x47 */
655 "REFFA", /* 0x48 */
656 "NREF", /* 0x49 */
657 "NREFF", /* 0x4a */
658 "NREFFL", /* 0x4b */
659 "NREFFU", /* 0x4c */
660 "NREFFA", /* 0x4d */
661 "IFMATCH", /* 0x4e */
662 "UNLESSM", /* 0x4f */
663 "SUSPEND", /* 0x50 */
664 "IFTHEN", /* 0x51 */
665 "GROUPP", /* 0x52 */
666 "LONGJMP", /* 0x53 */
667 "BRANCHJ", /* 0x54 */
668 "EVAL", /* 0x55 */
669 "MINMOD", /* 0x56 */
670 "LOGICAL", /* 0x57 */
671 "RENUM", /* 0x58 */
672 "TRIE", /* 0x59 */
673 "TRIEC", /* 0x5a */
674 "AHOCORASICK", /* 0x5b */
675 "AHOCORASICKC", /* 0x5c */
676 "GOSUB", /* 0x5d */
677 "GOSTART", /* 0x5e */
678 "NGROUPP", /* 0x5f */
679 "INSUBP", /* 0x60 */
680 "DEFINEP", /* 0x61 */
681 "ENDLIKE", /* 0x62 */
682 "OPFAIL", /* 0x63 */
683 "ACCEPT", /* 0x64 */
684 "VERB", /* 0x65 */
685 "PRUNE", /* 0x66 */
686 "MARKPOINT", /* 0x67 */
687 "SKIP", /* 0x68 */
688 "COMMIT", /* 0x69 */
689 "CUTGROUP", /* 0x6a */
690 "KEEPS", /* 0x6b */
691 "LNBREAK", /* 0x6c */
692 "VERTWS", /* 0x6d */
693 "NVERTWS", /* 0x6e */
694 "HORIZWS", /* 0x6f */
695 "NHORIZWS", /* 0x70 */
696 "OPTIMIZED", /* 0x71 */
697 "PSEUDO", /* 0x72 */
03363afd 698 /* ------------ States ------------- */
24b23f37
YO
699 "TRIE_next", /* REGNODE_MAX +0x01 */
700 "TRIE_next_fail", /* REGNODE_MAX +0x02 */
701 "EVAL_AB", /* REGNODE_MAX +0x03 */
702 "EVAL_AB_fail", /* REGNODE_MAX +0x04 */
703 "CURLYX_end", /* REGNODE_MAX +0x05 */
704 "CURLYX_end_fail", /* REGNODE_MAX +0x06 */
705 "WHILEM_A_pre", /* REGNODE_MAX +0x07 */
706 "WHILEM_A_pre_fail", /* REGNODE_MAX +0x08 */
707 "WHILEM_A_min", /* REGNODE_MAX +0x09 */
708 "WHILEM_A_min_fail", /* REGNODE_MAX +0x0a */
709 "WHILEM_A_max", /* REGNODE_MAX +0x0b */
710 "WHILEM_A_max_fail", /* REGNODE_MAX +0x0c */
711 "WHILEM_B_min", /* REGNODE_MAX +0x0d */
712 "WHILEM_B_min_fail", /* REGNODE_MAX +0x0e */
713 "WHILEM_B_max", /* REGNODE_MAX +0x0f */
714 "WHILEM_B_max_fail", /* REGNODE_MAX +0x10 */
715 "BRANCH_next", /* REGNODE_MAX +0x11 */
716 "BRANCH_next_fail", /* REGNODE_MAX +0x12 */
717 "CURLYM_A", /* REGNODE_MAX +0x13 */
718 "CURLYM_A_fail", /* REGNODE_MAX +0x14 */
719 "CURLYM_B", /* REGNODE_MAX +0x15 */
720 "CURLYM_B_fail", /* REGNODE_MAX +0x16 */
721 "IFMATCH_A", /* REGNODE_MAX +0x17 */
722 "IFMATCH_A_fail", /* REGNODE_MAX +0x18 */
723 "CURLY_B_min_known", /* REGNODE_MAX +0x19 */
724 "CURLY_B_min_known_fail", /* REGNODE_MAX +0x1a */
725 "CURLY_B_min", /* REGNODE_MAX +0x1b */
726 "CURLY_B_min_fail", /* REGNODE_MAX +0x1c */
727 "CURLY_B_max", /* REGNODE_MAX +0x1d */
728 "CURLY_B_max_fail", /* REGNODE_MAX +0x1e */
729 "COMMIT_next", /* REGNODE_MAX +0x1f */
730 "COMMIT_next_fail", /* REGNODE_MAX +0x20 */
e2e6a0f1
YO
731 "MARKPOINT_next", /* REGNODE_MAX +0x21 */
732 "MARKPOINT_next_fail", /* REGNODE_MAX +0x22 */
5d458dd8
YO
733 "SKIP_next", /* REGNODE_MAX +0x23 */
734 "SKIP_next_fail", /* REGNODE_MAX +0x24 */
735 "CUTGROUP_next", /* REGNODE_MAX +0x25 */
736 "CUTGROUP_next_fail", /* REGNODE_MAX +0x26 */
ee9b8eae
YO
737 "KEEPS_next", /* REGNODE_MAX +0x27 */
738 "KEEPS_next_fail", /* REGNODE_MAX +0x28 */
885f9e59 739};
22429478 740#endif /* DOINIT */
d09b2d29 741
f7819f85
A
742/* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
743
744#ifndef DOINIT
745EXTCONST char * PL_reg_extflags_name[];
746#else
747EXTCONST char * const PL_reg_extflags_name[] = {
df7a8460 748 /* Bits in extflags defined: 11111111111111111111111011111111 */
52d81aa8
NC
749 "MULTILINE", /* 0x00000001 */
750 "SINGLELINE", /* 0x00000002 */
751 "FOLD", /* 0x00000004 */
752 "EXTENDED", /* 0x00000008 */
753 "KEEPCOPY", /* 0x00000010 */
df7a8460
KW
754 "CHARSET", /* 0x000000e0 */
755 "CHARSET", /* 0x000000e0 */
756 "CHARSET", /* 0x000000e0 */
e795e964
KW
757 "UNUSED_BIT_8", /* 0x00000100 */
758 "ANCH_BOL", /* 0x00000200 */
759 "ANCH_MBOL", /* 0x00000400 */
760 "ANCH_SBOL", /* 0x00000800 */
761 "ANCH_GPOS", /* 0x00001000 */
762 "GPOS_SEEN", /* 0x00002000 */
763 "GPOS_FLOAT", /* 0x00004000 */
764 "LOOKBEHIND_SEEN", /* 0x00008000 */
765 "EVAL_SEEN", /* 0x00010000 */
766 "CANY_SEEN", /* 0x00020000 */
767 "NOSCAN", /* 0x00040000 */
768 "CHECK_ALL", /* 0x00080000 */
52d81aa8
NC
769 "MATCH_UTF8", /* 0x00100000 */
770 "USE_INTUIT_NOML", /* 0x00200000 */
771 "USE_INTUIT_ML", /* 0x00400000 */
772 "INTUIT_TAIL", /* 0x00800000 */
773 "SPLIT", /* 0x01000000 */
774 "COPY_DONE", /* 0x02000000 */
775 "TAINTED_SEEN", /* 0x04000000 */
776 "TAINTED", /* 0x08000000 */
777 "START_ONLY", /* 0x10000000 */
778 "SKIPWHITE", /* 0x20000000 */
779 "WHITE", /* 0x40000000 */
780 "NULL", /* 0x80000000 */
f7819f85
A
781};
782#endif /* DOINIT */
783
f9ef50a7 784/* The following have no fixed length. U8 so we can do strchr() on it. */
ded4dd2a 785#define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 786
f9ef50a7 787#ifndef DOINIT
ded4dd2a 788EXTCONST U8 PL_varies[] __attribute__deprecated__;
f9ef50a7 789#else
ded4dd2a 790EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
0e019ad6 791 ANYOFV, CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX,
781aab5c
KW
792 WHILEM, REF, REFF, REFFL, REFFU, REFFA, NREF, NREFF, NREFFL, NREFFU,
793 NREFFA, SUSPEND, IFTHEN, BRANCHJ,
f9ef50a7
NC
794 0
795};
796#endif /* DOINIT */
797
ded4dd2a
NC
798#ifndef DOINIT
799EXTCONST U8 PL_varies_bitmask[];
800#else
801EXTCONST U8 PL_varies_bitmask[] = {
693fefec 802 0x00, 0x00, 0x40, 0x00, 0x00, 0x80, 0x03, 0xF8, 0xF3, 0x3F, 0x13, 0x00, 0x00, 0x00, 0x00
ded4dd2a
NC
803};
804#endif /* DOINIT */
805
f9ef50a7
NC
806/* The following always have a length of 1. U8 we can do strchr() on it. */
807/* (Note that length 1 means "one character" under UTF8, not "one octet".) */
ded4dd2a 808#define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
e52fc539 809
f9ef50a7 810#ifndef DOINIT
ded4dd2a 811EXTCONST U8 PL_simple[] __attribute__deprecated__;
f9ef50a7 812#else
ded4dd2a 813EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
0c6e81eb
KW
814 REG_ANY, SANY, CANY, ANYOF, ALNUM, ALNUML, ALNUMU, ALNUMA, NALNUM,
815 NALNUML, NALNUMU, NALNUMA, SPACE, SPACEL, SPACEU, SPACEA, NSPACE,
816 NSPACEL, NSPACEU, NSPACEA, DIGIT, DIGITL, DIGITA, NDIGIT, NDIGITL,
817 NDIGITA, VERTWS, NVERTWS, HORIZWS, NHORIZWS,
f9ef50a7
NC
818 0
819};
820#endif /* DOINIT */
821
ded4dd2a
NC
822#ifndef DOINIT
823EXTCONST U8 PL_simple_bitmask[];
824#else
825EXTCONST U8 PL_simple_bitmask[] = {
693fefec 826 0x00, 0x00, 0xBC, 0xFF, 0xFF, 0x5D, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xE0, 0x01
ded4dd2a
NC
827};
828#endif /* DOINIT */
829
37442d52 830/* ex: set ro: */