| 1 | /* -*- buffer-read-only: t -*- |
| 2 | !!!!!!! DO NOT EDIT THIS FILE !!!!!!! |
| 3 | This file is built by regcomp.pl from regcomp.sym. |
| 4 | Any changes made here will be lost! |
| 5 | */ |
| 6 | |
| 7 | /* Regops and State definitions */ |
| 8 | |
| 9 | #define REGNODE_MAX 90 |
| 10 | #define REGMATCH_STATE_MAX 130 |
| 11 | |
| 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. */ |
| 29 | #define ANYOF 17 /* 0x11 Match character in (or not in) this class. */ |
| 30 | #define ALNUM 18 /* 0x12 Match any alphanumeric character */ |
| 31 | #define ALNUML 19 /* 0x13 Match any alphanumeric char in locale */ |
| 32 | #define NALNUM 20 /* 0x14 Match any non-alphanumeric character */ |
| 33 | #define NALNUML 21 /* 0x15 Match any non-alphanumeric char in locale */ |
| 34 | #define SPACE 22 /* 0x16 Match any whitespace character */ |
| 35 | #define SPACEL 23 /* 0x17 Match any whitespace char in locale */ |
| 36 | #define NSPACE 24 /* 0x18 Match any non-whitespace character */ |
| 37 | #define NSPACEL 25 /* 0x19 Match any non-whitespace char in locale */ |
| 38 | #define DIGIT 26 /* 0x1a Match any numeric character */ |
| 39 | #define DIGITL 27 /* 0x1b Match any numeric character in locale */ |
| 40 | #define NDIGIT 28 /* 0x1c Match any non-numeric character */ |
| 41 | #define NDIGITL 29 /* 0x1d Match any non-numeric character in locale */ |
| 42 | #define CLUMP 30 /* 0x1e Match any combining character 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 string, folded (prec. by length). */ |
| 47 | #define EXACTFL 35 /* 0x23 Match this string, folded in locale (w/len). */ |
| 48 | #define NOTHING 36 /* 0x24 Match empty string. */ |
| 49 | #define TAIL 37 /* 0x25 Match empty string. Can jump here from outside. */ |
| 50 | #define STAR 38 /* 0x26 Match this (simple) thing 0 or more times. */ |
| 51 | #define PLUS 39 /* 0x27 Match this (simple) thing 1 or more times. */ |
| 52 | #define CURLY 40 /* 0x28 Match this simple thing {n,m} times. */ |
| 53 | #define CURLYN 41 /* 0x29 Capture next-after-this simple thing */ |
| 54 | #define CURLYM 42 /* 0x2a Capture this medium-complex thing {n,m} times. */ |
| 55 | #define CURLYX 43 /* 0x2b Match this complex thing {n,m} times. */ |
| 56 | #define WHILEM 44 /* 0x2c Do curly processing and see if rest matches. */ |
| 57 | #define OPEN 45 /* 0x2d Mark this point in input as start of */ |
| 58 | #define CLOSE 46 /* 0x2e Analogous to OPEN. */ |
| 59 | #define REF 47 /* 0x2f Match some already matched string */ |
| 60 | #define REFF 48 /* 0x30 Match already matched string, folded */ |
| 61 | #define REFFL 49 /* 0x31 Match already matched string, folded in loc. */ |
| 62 | #define IFMATCH 50 /* 0x32 Succeeds if the following matches. */ |
| 63 | #define UNLESSM 51 /* 0x33 Fails if the following matches. */ |
| 64 | #define SUSPEND 52 /* 0x34 "Independent" sub-RE. */ |
| 65 | #define IFTHEN 53 /* 0x35 Switch, should be preceeded by switcher . */ |
| 66 | #define GROUPP 54 /* 0x36 Whether the group matched. */ |
| 67 | #define LONGJMP 55 /* 0x37 Jump far away. */ |
| 68 | #define BRANCHJ 56 /* 0x38 BRANCH with long offset. */ |
| 69 | #define EVAL 57 /* 0x39 Execute some Perl code. */ |
| 70 | #define MINMOD 58 /* 0x3a Next operator is not greedy. */ |
| 71 | #define LOGICAL 59 /* 0x3b Next opcode should set the flag only. */ |
| 72 | #define RENUM 60 /* 0x3c Group with independently numbered parens. */ |
| 73 | #define TRIE 61 /* 0x3d Match many EXACT(FL?)? at once. flags==type */ |
| 74 | #define TRIEC 62 /* 0x3e Same as TRIE, but with embedded charclass data */ |
| 75 | #define AHOCORASICK 63 /* 0x3f Aho Corasick stclass. flags==type */ |
| 76 | #define AHOCORASICKC 64 /* 0x40 Same as AHOCORASICK, but with embedded charclass data */ |
| 77 | #define GOSUB 65 /* 0x41 recurse to paren arg1 at (signed) ofs arg2 */ |
| 78 | #define GOSTART 66 /* 0x42 recurse to start of pattern */ |
| 79 | #define NREF 67 /* 0x43 Match some already matched string */ |
| 80 | #define NREFF 68 /* 0x44 Match already matched string, folded */ |
| 81 | #define NREFFL 69 /* 0x45 Match already matched string, folded in loc. */ |
| 82 | #define NGROUPP 70 /* 0x46 Whether the group matched. */ |
| 83 | #define INSUBP 71 /* 0x47 Whether we are in a specific recurse. */ |
| 84 | #define DEFINEP 72 /* 0x48 Never execute directly. */ |
| 85 | #define ENDLIKE 73 /* 0x49 Used only for the type field of verbs */ |
| 86 | #define OPFAIL 74 /* 0x4a Same as (?!) */ |
| 87 | #define ACCEPT 75 /* 0x4b Accepts the current matched string. */ |
| 88 | #define VERB 76 /* 0x4c Used only for the type field of verbs */ |
| 89 | #define PRUNE 77 /* 0x4d Pattern fails at this startpoint if no-backtracking through this */ |
| 90 | #define MARKPOINT 78 /* 0x4e Push the current location for rollback by cut. */ |
| 91 | #define SKIP 79 /* 0x4f On failure skip forward (to the mark) before retrying */ |
| 92 | #define COMMIT 80 /* 0x50 Pattern fails outright if backtracking through this */ |
| 93 | #define CUTGROUP 81 /* 0x51 On failure go to the next alternation in the group */ |
| 94 | #define KEEPS 82 /* 0x52 $& begins here. */ |
| 95 | #define LNBREAK 83 /* 0x53 generic newline pattern */ |
| 96 | #define VERTWS 84 /* 0x54 vertical whitespace (Perl 6) */ |
| 97 | #define NVERTWS 85 /* 0x55 not vertical whitespace (Perl 6) */ |
| 98 | #define HORIZWS 86 /* 0x56 horizontal whitespace (Perl 6) */ |
| 99 | #define NHORIZWS 87 /* 0x57 not horizontal whitespace (Perl 6) */ |
| 100 | #define FOLDCHAR 88 /* 0x58 codepoint with tricky case folding properties. */ |
| 101 | #define OPTIMIZED 89 /* 0x59 Placeholder for dump. */ |
| 102 | #define PSEUDO 90 /* 0x5a Pseudo opcode for internal use. */ |
| 103 | /* ------------ States ------------- */ |
| 104 | #define TRIE_next (REGNODE_MAX + 1) /* state for TRIE */ |
| 105 | #define TRIE_next_fail (REGNODE_MAX + 2) /* state for TRIE */ |
| 106 | #define EVAL_AB (REGNODE_MAX + 3) /* state for EVAL */ |
| 107 | #define EVAL_AB_fail (REGNODE_MAX + 4) /* state for EVAL */ |
| 108 | #define CURLYX_end (REGNODE_MAX + 5) /* state for CURLYX */ |
| 109 | #define CURLYX_end_fail (REGNODE_MAX + 6) /* state for CURLYX */ |
| 110 | #define WHILEM_A_pre (REGNODE_MAX + 7) /* state for WHILEM */ |
| 111 | #define WHILEM_A_pre_fail (REGNODE_MAX + 8) /* state for WHILEM */ |
| 112 | #define WHILEM_A_min (REGNODE_MAX + 9) /* state for WHILEM */ |
| 113 | #define WHILEM_A_min_fail (REGNODE_MAX + 10) /* state for WHILEM */ |
| 114 | #define WHILEM_A_max (REGNODE_MAX + 11) /* state for WHILEM */ |
| 115 | #define WHILEM_A_max_fail (REGNODE_MAX + 12) /* state for WHILEM */ |
| 116 | #define WHILEM_B_min (REGNODE_MAX + 13) /* state for WHILEM */ |
| 117 | #define WHILEM_B_min_fail (REGNODE_MAX + 14) /* state for WHILEM */ |
| 118 | #define WHILEM_B_max (REGNODE_MAX + 15) /* state for WHILEM */ |
| 119 | #define WHILEM_B_max_fail (REGNODE_MAX + 16) /* state for WHILEM */ |
| 120 | #define BRANCH_next (REGNODE_MAX + 17) /* state for BRANCH */ |
| 121 | #define BRANCH_next_fail (REGNODE_MAX + 18) /* state for BRANCH */ |
| 122 | #define CURLYM_A (REGNODE_MAX + 19) /* state for CURLYM */ |
| 123 | #define CURLYM_A_fail (REGNODE_MAX + 20) /* state for CURLYM */ |
| 124 | #define CURLYM_B (REGNODE_MAX + 21) /* state for CURLYM */ |
| 125 | #define CURLYM_B_fail (REGNODE_MAX + 22) /* state for CURLYM */ |
| 126 | #define IFMATCH_A (REGNODE_MAX + 23) /* state for IFMATCH */ |
| 127 | #define IFMATCH_A_fail (REGNODE_MAX + 24) /* state for IFMATCH */ |
| 128 | #define CURLY_B_min_known (REGNODE_MAX + 25) /* state for CURLY */ |
| 129 | #define CURLY_B_min_known_fail (REGNODE_MAX + 26) /* state for CURLY */ |
| 130 | #define CURLY_B_min (REGNODE_MAX + 27) /* state for CURLY */ |
| 131 | #define CURLY_B_min_fail (REGNODE_MAX + 28) /* state for CURLY */ |
| 132 | #define CURLY_B_max (REGNODE_MAX + 29) /* state for CURLY */ |
| 133 | #define CURLY_B_max_fail (REGNODE_MAX + 30) /* state for CURLY */ |
| 134 | #define COMMIT_next (REGNODE_MAX + 31) /* state for COMMIT */ |
| 135 | #define COMMIT_next_fail (REGNODE_MAX + 32) /* state for COMMIT */ |
| 136 | #define MARKPOINT_next (REGNODE_MAX + 33) /* state for MARKPOINT */ |
| 137 | #define MARKPOINT_next_fail (REGNODE_MAX + 34) /* state for MARKPOINT */ |
| 138 | #define SKIP_next (REGNODE_MAX + 35) /* state for SKIP */ |
| 139 | #define SKIP_next_fail (REGNODE_MAX + 36) /* state for SKIP */ |
| 140 | #define CUTGROUP_next (REGNODE_MAX + 37) /* state for CUTGROUP */ |
| 141 | #define CUTGROUP_next_fail (REGNODE_MAX + 38) /* state for CUTGROUP */ |
| 142 | #define KEEPS_next (REGNODE_MAX + 39) /* state for KEEPS */ |
| 143 | #define KEEPS_next_fail (REGNODE_MAX + 40) /* state for KEEPS */ |
| 144 | |
| 145 | /* PL_regkind[] What type of regop or state is this. */ |
| 146 | |
| 147 | #ifndef DOINIT |
| 148 | EXTCONST U8 PL_regkind[]; |
| 149 | #else |
| 150 | EXTCONST U8 PL_regkind[] = { |
| 151 | END, /* END */ |
| 152 | END, /* SUCCEED */ |
| 153 | BOL, /* BOL */ |
| 154 | BOL, /* MBOL */ |
| 155 | BOL, /* SBOL */ |
| 156 | EOL, /* EOS */ |
| 157 | EOL, /* EOL */ |
| 158 | EOL, /* MEOL */ |
| 159 | EOL, /* SEOL */ |
| 160 | BOUND, /* BOUND */ |
| 161 | BOUND, /* BOUNDL */ |
| 162 | NBOUND, /* NBOUND */ |
| 163 | NBOUND, /* NBOUNDL */ |
| 164 | GPOS, /* GPOS */ |
| 165 | REG_ANY, /* REG_ANY */ |
| 166 | REG_ANY, /* SANY */ |
| 167 | REG_ANY, /* CANY */ |
| 168 | ANYOF, /* ANYOF */ |
| 169 | ALNUM, /* ALNUM */ |
| 170 | ALNUM, /* ALNUML */ |
| 171 | NALNUM, /* NALNUM */ |
| 172 | NALNUM, /* NALNUML */ |
| 173 | SPACE, /* SPACE */ |
| 174 | SPACE, /* SPACEL */ |
| 175 | NSPACE, /* NSPACE */ |
| 176 | NSPACE, /* NSPACEL */ |
| 177 | DIGIT, /* DIGIT */ |
| 178 | DIGIT, /* DIGITL */ |
| 179 | NDIGIT, /* NDIGIT */ |
| 180 | NDIGIT, /* NDIGITL */ |
| 181 | CLUMP, /* CLUMP */ |
| 182 | BRANCH, /* BRANCH */ |
| 183 | BACK, /* BACK */ |
| 184 | EXACT, /* EXACT */ |
| 185 | EXACT, /* EXACTF */ |
| 186 | EXACT, /* EXACTFL */ |
| 187 | NOTHING, /* NOTHING */ |
| 188 | NOTHING, /* TAIL */ |
| 189 | STAR, /* STAR */ |
| 190 | PLUS, /* PLUS */ |
| 191 | CURLY, /* CURLY */ |
| 192 | CURLY, /* CURLYN */ |
| 193 | CURLY, /* CURLYM */ |
| 194 | CURLY, /* CURLYX */ |
| 195 | WHILEM, /* WHILEM */ |
| 196 | OPEN, /* OPEN */ |
| 197 | CLOSE, /* CLOSE */ |
| 198 | REF, /* REF */ |
| 199 | REF, /* REFF */ |
| 200 | REF, /* REFFL */ |
| 201 | BRANCHJ, /* IFMATCH */ |
| 202 | BRANCHJ, /* UNLESSM */ |
| 203 | BRANCHJ, /* SUSPEND */ |
| 204 | BRANCHJ, /* IFTHEN */ |
| 205 | GROUPP, /* GROUPP */ |
| 206 | LONGJMP, /* LONGJMP */ |
| 207 | BRANCHJ, /* BRANCHJ */ |
| 208 | EVAL, /* EVAL */ |
| 209 | MINMOD, /* MINMOD */ |
| 210 | LOGICAL, /* LOGICAL */ |
| 211 | BRANCHJ, /* RENUM */ |
| 212 | TRIE, /* TRIE */ |
| 213 | TRIE, /* TRIEC */ |
| 214 | TRIE, /* AHOCORASICK */ |
| 215 | TRIE, /* AHOCORASICKC */ |
| 216 | GOSUB, /* GOSUB */ |
| 217 | GOSTART, /* GOSTART */ |
| 218 | REF, /* NREF */ |
| 219 | REF, /* NREFF */ |
| 220 | REF, /* NREFFL */ |
| 221 | NGROUPP, /* NGROUPP */ |
| 222 | INSUBP, /* INSUBP */ |
| 223 | DEFINEP, /* DEFINEP */ |
| 224 | ENDLIKE, /* ENDLIKE */ |
| 225 | ENDLIKE, /* OPFAIL */ |
| 226 | ENDLIKE, /* ACCEPT */ |
| 227 | VERB, /* VERB */ |
| 228 | VERB, /* PRUNE */ |
| 229 | VERB, /* MARKPOINT */ |
| 230 | VERB, /* SKIP */ |
| 231 | VERB, /* COMMIT */ |
| 232 | VERB, /* CUTGROUP */ |
| 233 | KEEPS, /* KEEPS */ |
| 234 | LNBREAK, /* LNBREAK */ |
| 235 | VERTWS, /* VERTWS */ |
| 236 | NVERTWS, /* NVERTWS */ |
| 237 | HORIZWS, /* HORIZWS */ |
| 238 | NHORIZWS, /* NHORIZWS */ |
| 239 | FOLDCHAR, /* FOLDCHAR */ |
| 240 | NOTHING, /* OPTIMIZED */ |
| 241 | PSEUDO, /* PSEUDO */ |
| 242 | /* ------------ States ------------- */ |
| 243 | TRIE, /* TRIE_next */ |
| 244 | TRIE, /* TRIE_next_fail */ |
| 245 | EVAL, /* EVAL_AB */ |
| 246 | EVAL, /* EVAL_AB_fail */ |
| 247 | CURLYX, /* CURLYX_end */ |
| 248 | CURLYX, /* CURLYX_end_fail */ |
| 249 | WHILEM, /* WHILEM_A_pre */ |
| 250 | WHILEM, /* WHILEM_A_pre_fail */ |
| 251 | WHILEM, /* WHILEM_A_min */ |
| 252 | WHILEM, /* WHILEM_A_min_fail */ |
| 253 | WHILEM, /* WHILEM_A_max */ |
| 254 | WHILEM, /* WHILEM_A_max_fail */ |
| 255 | WHILEM, /* WHILEM_B_min */ |
| 256 | WHILEM, /* WHILEM_B_min_fail */ |
| 257 | WHILEM, /* WHILEM_B_max */ |
| 258 | WHILEM, /* WHILEM_B_max_fail */ |
| 259 | BRANCH, /* BRANCH_next */ |
| 260 | BRANCH, /* BRANCH_next_fail */ |
| 261 | CURLYM, /* CURLYM_A */ |
| 262 | CURLYM, /* CURLYM_A_fail */ |
| 263 | CURLYM, /* CURLYM_B */ |
| 264 | CURLYM, /* CURLYM_B_fail */ |
| 265 | IFMATCH, /* IFMATCH_A */ |
| 266 | IFMATCH, /* IFMATCH_A_fail */ |
| 267 | CURLY, /* CURLY_B_min_known */ |
| 268 | CURLY, /* CURLY_B_min_known_fail */ |
| 269 | CURLY, /* CURLY_B_min */ |
| 270 | CURLY, /* CURLY_B_min_fail */ |
| 271 | CURLY, /* CURLY_B_max */ |
| 272 | CURLY, /* CURLY_B_max_fail */ |
| 273 | COMMIT, /* COMMIT_next */ |
| 274 | COMMIT, /* COMMIT_next_fail */ |
| 275 | MARKPOINT, /* MARKPOINT_next */ |
| 276 | MARKPOINT, /* MARKPOINT_next_fail */ |
| 277 | SKIP, /* SKIP_next */ |
| 278 | SKIP, /* SKIP_next_fail */ |
| 279 | CUTGROUP, /* CUTGROUP_next */ |
| 280 | CUTGROUP, /* CUTGROUP_next_fail */ |
| 281 | KEEPS, /* KEEPS_next */ |
| 282 | KEEPS, /* KEEPS_next_fail */ |
| 283 | }; |
| 284 | #endif |
| 285 | |
| 286 | /* regarglen[] - How large is the argument part of the node (in regnodes) */ |
| 287 | |
| 288 | #ifdef REG_COMP_C |
| 289 | static const U8 regarglen[] = { |
| 290 | 0, /* END */ |
| 291 | 0, /* SUCCEED */ |
| 292 | 0, /* BOL */ |
| 293 | 0, /* MBOL */ |
| 294 | 0, /* SBOL */ |
| 295 | 0, /* EOS */ |
| 296 | 0, /* EOL */ |
| 297 | 0, /* MEOL */ |
| 298 | 0, /* SEOL */ |
| 299 | 0, /* BOUND */ |
| 300 | 0, /* BOUNDL */ |
| 301 | 0, /* NBOUND */ |
| 302 | 0, /* NBOUNDL */ |
| 303 | 0, /* GPOS */ |
| 304 | 0, /* REG_ANY */ |
| 305 | 0, /* SANY */ |
| 306 | 0, /* CANY */ |
| 307 | 0, /* ANYOF */ |
| 308 | 0, /* ALNUM */ |
| 309 | 0, /* ALNUML */ |
| 310 | 0, /* NALNUM */ |
| 311 | 0, /* NALNUML */ |
| 312 | 0, /* SPACE */ |
| 313 | 0, /* SPACEL */ |
| 314 | 0, /* NSPACE */ |
| 315 | 0, /* NSPACEL */ |
| 316 | 0, /* DIGIT */ |
| 317 | 0, /* DIGITL */ |
| 318 | 0, /* NDIGIT */ |
| 319 | 0, /* NDIGITL */ |
| 320 | 0, /* CLUMP */ |
| 321 | 0, /* BRANCH */ |
| 322 | 0, /* BACK */ |
| 323 | 0, /* EXACT */ |
| 324 | 0, /* EXACTF */ |
| 325 | 0, /* EXACTFL */ |
| 326 | 0, /* NOTHING */ |
| 327 | 0, /* TAIL */ |
| 328 | 0, /* STAR */ |
| 329 | 0, /* PLUS */ |
| 330 | EXTRA_SIZE(struct regnode_2), /* CURLY */ |
| 331 | EXTRA_SIZE(struct regnode_2), /* CURLYN */ |
| 332 | EXTRA_SIZE(struct regnode_2), /* CURLYM */ |
| 333 | EXTRA_SIZE(struct regnode_2), /* CURLYX */ |
| 334 | 0, /* WHILEM */ |
| 335 | EXTRA_SIZE(struct regnode_1), /* OPEN */ |
| 336 | EXTRA_SIZE(struct regnode_1), /* CLOSE */ |
| 337 | EXTRA_SIZE(struct regnode_1), /* REF */ |
| 338 | EXTRA_SIZE(struct regnode_1), /* REFF */ |
| 339 | EXTRA_SIZE(struct regnode_1), /* REFFL */ |
| 340 | EXTRA_SIZE(struct regnode_1), /* IFMATCH */ |
| 341 | EXTRA_SIZE(struct regnode_1), /* UNLESSM */ |
| 342 | EXTRA_SIZE(struct regnode_1), /* SUSPEND */ |
| 343 | EXTRA_SIZE(struct regnode_1), /* IFTHEN */ |
| 344 | EXTRA_SIZE(struct regnode_1), /* GROUPP */ |
| 345 | EXTRA_SIZE(struct regnode_1), /* LONGJMP */ |
| 346 | EXTRA_SIZE(struct regnode_1), /* BRANCHJ */ |
| 347 | EXTRA_SIZE(struct regnode_1), /* EVAL */ |
| 348 | 0, /* MINMOD */ |
| 349 | 0, /* LOGICAL */ |
| 350 | EXTRA_SIZE(struct regnode_1), /* RENUM */ |
| 351 | EXTRA_SIZE(struct regnode_1), /* TRIE */ |
| 352 | EXTRA_SIZE(struct regnode_charclass), /* TRIEC */ |
| 353 | EXTRA_SIZE(struct regnode_1), /* AHOCORASICK */ |
| 354 | EXTRA_SIZE(struct regnode_charclass), /* AHOCORASICKC */ |
| 355 | EXTRA_SIZE(struct regnode_2L), /* GOSUB */ |
| 356 | 0, /* GOSTART */ |
| 357 | EXTRA_SIZE(struct regnode_1), /* NREF */ |
| 358 | EXTRA_SIZE(struct regnode_1), /* NREFF */ |
| 359 | EXTRA_SIZE(struct regnode_1), /* NREFFL */ |
| 360 | EXTRA_SIZE(struct regnode_1), /* NGROUPP */ |
| 361 | EXTRA_SIZE(struct regnode_1), /* INSUBP */ |
| 362 | EXTRA_SIZE(struct regnode_1), /* DEFINEP */ |
| 363 | 0, /* ENDLIKE */ |
| 364 | 0, /* OPFAIL */ |
| 365 | EXTRA_SIZE(struct regnode_1), /* ACCEPT */ |
| 366 | EXTRA_SIZE(struct regnode_1), /* VERB */ |
| 367 | EXTRA_SIZE(struct regnode_1), /* PRUNE */ |
| 368 | EXTRA_SIZE(struct regnode_1), /* MARKPOINT */ |
| 369 | EXTRA_SIZE(struct regnode_1), /* SKIP */ |
| 370 | EXTRA_SIZE(struct regnode_1), /* COMMIT */ |
| 371 | EXTRA_SIZE(struct regnode_1), /* CUTGROUP */ |
| 372 | 0, /* KEEPS */ |
| 373 | 0, /* LNBREAK */ |
| 374 | 0, /* VERTWS */ |
| 375 | 0, /* NVERTWS */ |
| 376 | 0, /* HORIZWS */ |
| 377 | 0, /* NHORIZWS */ |
| 378 | EXTRA_SIZE(struct regnode_1), /* FOLDCHAR */ |
| 379 | 0, /* OPTIMIZED */ |
| 380 | 0, /* PSEUDO */ |
| 381 | }; |
| 382 | |
| 383 | /* reg_off_by_arg[] - Which argument holds the offset to the next node */ |
| 384 | |
| 385 | static const char reg_off_by_arg[] = { |
| 386 | 0, /* END */ |
| 387 | 0, /* SUCCEED */ |
| 388 | 0, /* BOL */ |
| 389 | 0, /* MBOL */ |
| 390 | 0, /* SBOL */ |
| 391 | 0, /* EOS */ |
| 392 | 0, /* EOL */ |
| 393 | 0, /* MEOL */ |
| 394 | 0, /* SEOL */ |
| 395 | 0, /* BOUND */ |
| 396 | 0, /* BOUNDL */ |
| 397 | 0, /* NBOUND */ |
| 398 | 0, /* NBOUNDL */ |
| 399 | 0, /* GPOS */ |
| 400 | 0, /* REG_ANY */ |
| 401 | 0, /* SANY */ |
| 402 | 0, /* CANY */ |
| 403 | 0, /* ANYOF */ |
| 404 | 0, /* ALNUM */ |
| 405 | 0, /* ALNUML */ |
| 406 | 0, /* NALNUM */ |
| 407 | 0, /* NALNUML */ |
| 408 | 0, /* SPACE */ |
| 409 | 0, /* SPACEL */ |
| 410 | 0, /* NSPACE */ |
| 411 | 0, /* NSPACEL */ |
| 412 | 0, /* DIGIT */ |
| 413 | 0, /* DIGITL */ |
| 414 | 0, /* NDIGIT */ |
| 415 | 0, /* NDIGITL */ |
| 416 | 0, /* CLUMP */ |
| 417 | 0, /* BRANCH */ |
| 418 | 0, /* BACK */ |
| 419 | 0, /* EXACT */ |
| 420 | 0, /* EXACTF */ |
| 421 | 0, /* EXACTFL */ |
| 422 | 0, /* NOTHING */ |
| 423 | 0, /* TAIL */ |
| 424 | 0, /* STAR */ |
| 425 | 0, /* PLUS */ |
| 426 | 0, /* CURLY */ |
| 427 | 0, /* CURLYN */ |
| 428 | 0, /* CURLYM */ |
| 429 | 0, /* CURLYX */ |
| 430 | 0, /* WHILEM */ |
| 431 | 0, /* OPEN */ |
| 432 | 0, /* CLOSE */ |
| 433 | 0, /* REF */ |
| 434 | 0, /* REFF */ |
| 435 | 0, /* REFFL */ |
| 436 | 2, /* IFMATCH */ |
| 437 | 2, /* UNLESSM */ |
| 438 | 1, /* SUSPEND */ |
| 439 | 1, /* IFTHEN */ |
| 440 | 0, /* GROUPP */ |
| 441 | 1, /* LONGJMP */ |
| 442 | 1, /* BRANCHJ */ |
| 443 | 0, /* EVAL */ |
| 444 | 0, /* MINMOD */ |
| 445 | 0, /* LOGICAL */ |
| 446 | 1, /* RENUM */ |
| 447 | 0, /* TRIE */ |
| 448 | 0, /* TRIEC */ |
| 449 | 0, /* AHOCORASICK */ |
| 450 | 0, /* AHOCORASICKC */ |
| 451 | 0, /* GOSUB */ |
| 452 | 0, /* GOSTART */ |
| 453 | 0, /* NREF */ |
| 454 | 0, /* NREFF */ |
| 455 | 0, /* NREFFL */ |
| 456 | 0, /* NGROUPP */ |
| 457 | 0, /* INSUBP */ |
| 458 | 0, /* DEFINEP */ |
| 459 | 0, /* ENDLIKE */ |
| 460 | 0, /* OPFAIL */ |
| 461 | 0, /* ACCEPT */ |
| 462 | 0, /* VERB */ |
| 463 | 0, /* PRUNE */ |
| 464 | 0, /* MARKPOINT */ |
| 465 | 0, /* SKIP */ |
| 466 | 0, /* COMMIT */ |
| 467 | 0, /* CUTGROUP */ |
| 468 | 0, /* KEEPS */ |
| 469 | 0, /* LNBREAK */ |
| 470 | 0, /* VERTWS */ |
| 471 | 0, /* NVERTWS */ |
| 472 | 0, /* HORIZWS */ |
| 473 | 0, /* NHORIZWS */ |
| 474 | 0, /* FOLDCHAR */ |
| 475 | 0, /* OPTIMIZED */ |
| 476 | 0, /* PSEUDO */ |
| 477 | }; |
| 478 | |
| 479 | #endif /* REG_COMP_C */ |
| 480 | |
| 481 | /* reg_name[] - Opcode/state names in string form, for debugging */ |
| 482 | |
| 483 | #ifndef DOINIT |
| 484 | EXTCONST char * PL_reg_name[]; |
| 485 | #else |
| 486 | EXTCONST char * const PL_reg_name[] = { |
| 487 | "END", /* 0000 */ |
| 488 | "SUCCEED", /* 0x01 */ |
| 489 | "BOL", /* 0x02 */ |
| 490 | "MBOL", /* 0x03 */ |
| 491 | "SBOL", /* 0x04 */ |
| 492 | "EOS", /* 0x05 */ |
| 493 | "EOL", /* 0x06 */ |
| 494 | "MEOL", /* 0x07 */ |
| 495 | "SEOL", /* 0x08 */ |
| 496 | "BOUND", /* 0x09 */ |
| 497 | "BOUNDL", /* 0x0a */ |
| 498 | "NBOUND", /* 0x0b */ |
| 499 | "NBOUNDL", /* 0x0c */ |
| 500 | "GPOS", /* 0x0d */ |
| 501 | "REG_ANY", /* 0x0e */ |
| 502 | "SANY", /* 0x0f */ |
| 503 | "CANY", /* 0x10 */ |
| 504 | "ANYOF", /* 0x11 */ |
| 505 | "ALNUM", /* 0x12 */ |
| 506 | "ALNUML", /* 0x13 */ |
| 507 | "NALNUM", /* 0x14 */ |
| 508 | "NALNUML", /* 0x15 */ |
| 509 | "SPACE", /* 0x16 */ |
| 510 | "SPACEL", /* 0x17 */ |
| 511 | "NSPACE", /* 0x18 */ |
| 512 | "NSPACEL", /* 0x19 */ |
| 513 | "DIGIT", /* 0x1a */ |
| 514 | "DIGITL", /* 0x1b */ |
| 515 | "NDIGIT", /* 0x1c */ |
| 516 | "NDIGITL", /* 0x1d */ |
| 517 | "CLUMP", /* 0x1e */ |
| 518 | "BRANCH", /* 0x1f */ |
| 519 | "BACK", /* 0x20 */ |
| 520 | "EXACT", /* 0x21 */ |
| 521 | "EXACTF", /* 0x22 */ |
| 522 | "EXACTFL", /* 0x23 */ |
| 523 | "NOTHING", /* 0x24 */ |
| 524 | "TAIL", /* 0x25 */ |
| 525 | "STAR", /* 0x26 */ |
| 526 | "PLUS", /* 0x27 */ |
| 527 | "CURLY", /* 0x28 */ |
| 528 | "CURLYN", /* 0x29 */ |
| 529 | "CURLYM", /* 0x2a */ |
| 530 | "CURLYX", /* 0x2b */ |
| 531 | "WHILEM", /* 0x2c */ |
| 532 | "OPEN", /* 0x2d */ |
| 533 | "CLOSE", /* 0x2e */ |
| 534 | "REF", /* 0x2f */ |
| 535 | "REFF", /* 0x30 */ |
| 536 | "REFFL", /* 0x31 */ |
| 537 | "IFMATCH", /* 0x32 */ |
| 538 | "UNLESSM", /* 0x33 */ |
| 539 | "SUSPEND", /* 0x34 */ |
| 540 | "IFTHEN", /* 0x35 */ |
| 541 | "GROUPP", /* 0x36 */ |
| 542 | "LONGJMP", /* 0x37 */ |
| 543 | "BRANCHJ", /* 0x38 */ |
| 544 | "EVAL", /* 0x39 */ |
| 545 | "MINMOD", /* 0x3a */ |
| 546 | "LOGICAL", /* 0x3b */ |
| 547 | "RENUM", /* 0x3c */ |
| 548 | "TRIE", /* 0x3d */ |
| 549 | "TRIEC", /* 0x3e */ |
| 550 | "AHOCORASICK", /* 0x3f */ |
| 551 | "AHOCORASICKC", /* 0x40 */ |
| 552 | "GOSUB", /* 0x41 */ |
| 553 | "GOSTART", /* 0x42 */ |
| 554 | "NREF", /* 0x43 */ |
| 555 | "NREFF", /* 0x44 */ |
| 556 | "NREFFL", /* 0x45 */ |
| 557 | "NGROUPP", /* 0x46 */ |
| 558 | "INSUBP", /* 0x47 */ |
| 559 | "DEFINEP", /* 0x48 */ |
| 560 | "ENDLIKE", /* 0x49 */ |
| 561 | "OPFAIL", /* 0x4a */ |
| 562 | "ACCEPT", /* 0x4b */ |
| 563 | "VERB", /* 0x4c */ |
| 564 | "PRUNE", /* 0x4d */ |
| 565 | "MARKPOINT", /* 0x4e */ |
| 566 | "SKIP", /* 0x4f */ |
| 567 | "COMMIT", /* 0x50 */ |
| 568 | "CUTGROUP", /* 0x51 */ |
| 569 | "KEEPS", /* 0x52 */ |
| 570 | "LNBREAK", /* 0x53 */ |
| 571 | "VERTWS", /* 0x54 */ |
| 572 | "NVERTWS", /* 0x55 */ |
| 573 | "HORIZWS", /* 0x56 */ |
| 574 | "NHORIZWS", /* 0x57 */ |
| 575 | "FOLDCHAR", /* 0x58 */ |
| 576 | "OPTIMIZED", /* 0x59 */ |
| 577 | "PSEUDO", /* 0x5a */ |
| 578 | /* ------------ States ------------- */ |
| 579 | "TRIE_next", /* REGNODE_MAX +0x01 */ |
| 580 | "TRIE_next_fail", /* REGNODE_MAX +0x02 */ |
| 581 | "EVAL_AB", /* REGNODE_MAX +0x03 */ |
| 582 | "EVAL_AB_fail", /* REGNODE_MAX +0x04 */ |
| 583 | "CURLYX_end", /* REGNODE_MAX +0x05 */ |
| 584 | "CURLYX_end_fail", /* REGNODE_MAX +0x06 */ |
| 585 | "WHILEM_A_pre", /* REGNODE_MAX +0x07 */ |
| 586 | "WHILEM_A_pre_fail", /* REGNODE_MAX +0x08 */ |
| 587 | "WHILEM_A_min", /* REGNODE_MAX +0x09 */ |
| 588 | "WHILEM_A_min_fail", /* REGNODE_MAX +0x0a */ |
| 589 | "WHILEM_A_max", /* REGNODE_MAX +0x0b */ |
| 590 | "WHILEM_A_max_fail", /* REGNODE_MAX +0x0c */ |
| 591 | "WHILEM_B_min", /* REGNODE_MAX +0x0d */ |
| 592 | "WHILEM_B_min_fail", /* REGNODE_MAX +0x0e */ |
| 593 | "WHILEM_B_max", /* REGNODE_MAX +0x0f */ |
| 594 | "WHILEM_B_max_fail", /* REGNODE_MAX +0x10 */ |
| 595 | "BRANCH_next", /* REGNODE_MAX +0x11 */ |
| 596 | "BRANCH_next_fail", /* REGNODE_MAX +0x12 */ |
| 597 | "CURLYM_A", /* REGNODE_MAX +0x13 */ |
| 598 | "CURLYM_A_fail", /* REGNODE_MAX +0x14 */ |
| 599 | "CURLYM_B", /* REGNODE_MAX +0x15 */ |
| 600 | "CURLYM_B_fail", /* REGNODE_MAX +0x16 */ |
| 601 | "IFMATCH_A", /* REGNODE_MAX +0x17 */ |
| 602 | "IFMATCH_A_fail", /* REGNODE_MAX +0x18 */ |
| 603 | "CURLY_B_min_known", /* REGNODE_MAX +0x19 */ |
| 604 | "CURLY_B_min_known_fail", /* REGNODE_MAX +0x1a */ |
| 605 | "CURLY_B_min", /* REGNODE_MAX +0x1b */ |
| 606 | "CURLY_B_min_fail", /* REGNODE_MAX +0x1c */ |
| 607 | "CURLY_B_max", /* REGNODE_MAX +0x1d */ |
| 608 | "CURLY_B_max_fail", /* REGNODE_MAX +0x1e */ |
| 609 | "COMMIT_next", /* REGNODE_MAX +0x1f */ |
| 610 | "COMMIT_next_fail", /* REGNODE_MAX +0x20 */ |
| 611 | "MARKPOINT_next", /* REGNODE_MAX +0x21 */ |
| 612 | "MARKPOINT_next_fail", /* REGNODE_MAX +0x22 */ |
| 613 | "SKIP_next", /* REGNODE_MAX +0x23 */ |
| 614 | "SKIP_next_fail", /* REGNODE_MAX +0x24 */ |
| 615 | "CUTGROUP_next", /* REGNODE_MAX +0x25 */ |
| 616 | "CUTGROUP_next_fail", /* REGNODE_MAX +0x26 */ |
| 617 | "KEEPS_next", /* REGNODE_MAX +0x27 */ |
| 618 | "KEEPS_next_fail", /* REGNODE_MAX +0x28 */ |
| 619 | }; |
| 620 | #endif /* DOINIT */ |
| 621 | |
| 622 | /* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */ |
| 623 | |
| 624 | #ifndef DOINIT |
| 625 | EXTCONST char * PL_reg_extflags_name[]; |
| 626 | #else |
| 627 | EXTCONST char * const PL_reg_extflags_name[] = { |
| 628 | /* Bits in extflags defined: 11111111111101111111111100111111 */ |
| 629 | "MULTILINE", /* 0x00000001 */ |
| 630 | "SINGLELINE", /* 0x00000002 */ |
| 631 | "FOLD", /* 0x00000004 */ |
| 632 | "EXTENDED", /* 0x00000008 */ |
| 633 | "KEEPCOPY", /* 0x00000010 */ |
| 634 | "LOCALE", /* 0x00000020 */ |
| 635 | "UNUSED_BIT_6", /* 0x00000040 */ |
| 636 | "UNUSED_BIT_7", /* 0x00000080 */ |
| 637 | "ANCH_BOL", /* 0x00000100 */ |
| 638 | "ANCH_MBOL", /* 0x00000200 */ |
| 639 | "ANCH_SBOL", /* 0x00000400 */ |
| 640 | "ANCH_GPOS", /* 0x00000800 */ |
| 641 | "GPOS_SEEN", /* 0x00001000 */ |
| 642 | "GPOS_FLOAT", /* 0x00002000 */ |
| 643 | "LOOKBEHIND_SEEN", /* 0x00004000 */ |
| 644 | "EVAL_SEEN", /* 0x00008000 */ |
| 645 | "CANY_SEEN", /* 0x00010000 */ |
| 646 | "NOSCAN", /* 0x00020000 */ |
| 647 | "CHECK_ALL", /* 0x00040000 */ |
| 648 | "UNUSED_BIT_19", /* 0x00080000 */ |
| 649 | "MATCH_UTF8", /* 0x00100000 */ |
| 650 | "USE_INTUIT_NOML", /* 0x00200000 */ |
| 651 | "USE_INTUIT_ML", /* 0x00400000 */ |
| 652 | "INTUIT_TAIL", /* 0x00800000 */ |
| 653 | "SPLIT", /* 0x01000000 */ |
| 654 | "COPY_DONE", /* 0x02000000 */ |
| 655 | "TAINTED_SEEN", /* 0x04000000 */ |
| 656 | "TAINTED", /* 0x08000000 */ |
| 657 | "START_ONLY", /* 0x10000000 */ |
| 658 | "SKIPWHITE", /* 0x20000000 */ |
| 659 | "WHITE", /* 0x40000000 */ |
| 660 | "NULL", /* 0x80000000 */ |
| 661 | }; |
| 662 | #endif /* DOINIT */ |
| 663 | |
| 664 | /* The following have no fixed length. U8 so we can do strchr() on it. */ |
| 665 | #define REGNODE_VARIES(node) strchr((const char *)PL_varies, (node)) |
| 666 | |
| 667 | #ifndef DOINIT |
| 668 | EXTCONST U8 PL_varies[]; |
| 669 | #else |
| 670 | EXTCONST U8 PL_varies[] = { |
| 671 | CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX, WHILEM, |
| 672 | REF, REFF, REFFL, SUSPEND, IFTHEN, BRANCHJ, NREF, NREFF, NREFFL, |
| 673 | 0 |
| 674 | }; |
| 675 | #endif /* DOINIT */ |
| 676 | |
| 677 | /* The following always have a length of 1. U8 we can do strchr() on it. */ |
| 678 | /* (Note that length 1 means "one character" under UTF8, not "one octet".) */ |
| 679 | #define REGNODE_SIMPLE(node) strchr((const char *)PL_simple, (node)) |
| 680 | |
| 681 | #ifndef DOINIT |
| 682 | EXTCONST U8 PL_simple[]; |
| 683 | #else |
| 684 | EXTCONST U8 PL_simple[] = { |
| 685 | REG_ANY, SANY, CANY, ANYOF, ALNUM, ALNUML, NALNUM, NALNUML, SPACE, |
| 686 | SPACEL, NSPACE, NSPACEL, DIGIT, NDIGIT, VERTWS, NVERTWS, HORIZWS, |
| 687 | NHORIZWS, |
| 688 | 0 |
| 689 | }; |
| 690 | #endif /* DOINIT */ |
| 691 | |
| 692 | /* ex: set ro: */ |