regnodes.h

   1 /* -*- buffer-read-only: t -*-
   2    !!!!!!!   DO NOT EDIT THIS FILE   !!!!!!!
   3    This file is built by regen/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             100
  10 #define REGMATCH_STATE_MAX      140
  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 locale word boundary */
  23 #define NBOUND                  11      /* 0x0b Match "" at any word non-boundary */
  24 #define NBOUNDL                 12      /* 0x0c Match "" at any locale 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, 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 using native charset semantics for non-utf8 */
  32 #define ALNUML                  20      /* 0x14 Match any alphanumeric char in locale */
  33 #define ALNUMU                  21      /* 0x15 Match any alphanumeric char using Unicode semantics */
  34 #define NALNUM                  22      /* 0x16 Match any non-alphanumeric character using native charset semantics for non-utf8 */
  35 #define NALNUML                 23      /* 0x17 Match any non-alphanumeric char in locale */
  36 #define NALNUMU                 24      /* 0x18 Match any non-alphanumeric char using Unicode semantics */
  37 #define SPACE                   25      /* 0x19 Match any whitespace character using native charset semantics for non-utf8 */
  38 #define SPACEL                  26      /* 0x1a Match any whitespace char in locale */
  39 #define SPACEU                  27      /* 0x1b Match any whitespace char using Unicode semantics */
  40 #define NSPACE                  28      /* 0x1c Match any non-whitespace character using native charset semantics for non-utf8 */
  41 #define NSPACEL                 29      /* 0x1d Match any non-whitespace char in locale */
  42 #define NSPACEU                 30      /* 0x1e Match any non-whitespace char using Unicode semantics */
  43 #define DIGIT                   31      /* 0x1f Match any numeric character using native charset semantics for non-utf8 */
  44 #define DIGITL                  32      /* 0x20 Match any numeric character in locale */
  45 #define DIGITU                  33      /* 0x21 Match any numeric character using Unicode semantics */
  46 #define NDIGIT                  34      /* 0x22 Match any non-numeric character using native charset semantics for non-utf8 */
  47 #define NDIGITL                 35      /* 0x23 Match any non-numeric character in locale */
  48 #define NDIGITU                 36      /* 0x24 Match any non-numeric character using Unicode semantics */
  49 #define CLUMP                   37      /* 0x25 Match any extended grapheme cluster sequence */
  50 #define BRANCH                  38      /* 0x26 Match this alternative, or the next... */
  51 #define BACK                    39      /* 0x27 Match "", "next" ptr points backward. */
  52 #define EXACT                   40      /* 0x28 Match this string (preceded by length). */
  53 #define EXACTF                  41      /* 0x29 Match this string, folded, native charset semantics for non-utf8 (prec. by length). */
  54 #define EXACTFL                 42      /* 0x2a Match this string, folded in locale (w/len). */
  55 #define EXACTFU                 43      /* 0x2b Match this string, folded, Unicode semantics for non-utf8 (prec. by length). */
  56 #define NOTHING                 44      /* 0x2c Match empty string. */
  57 #define TAIL                    45      /* 0x2d Match empty string. Can jump here from outside. */
  58 #define STAR                    46      /* 0x2e Match this (simple) thing 0 or more times. */
  59 #define PLUS                    47      /* 0x2f Match this (simple) thing 1 or more times. */
  60 #define CURLY                   48      /* 0x30 Match this simple thing {n,m} times. */
  61 #define CURLYN                  49      /* 0x31 Capture next-after-this simple thing */
  62 #define CURLYM                  50      /* 0x32 Capture this medium-complex thing {n,m} times. */
  63 #define CURLYX                  51      /* 0x33 Match this complex thing {n,m} times. */
  64 #define WHILEM                  52      /* 0x34 Do curly processing and see if rest matches. */
  65 #define OPEN                    53      /* 0x35 Mark this point in input as start of */
  66 #define CLOSE                   54      /* 0x36 Analogous to OPEN. */
  67 #define REF                     55      /* 0x37 Match some already matched string */
  68 #define REFF                    56      /* 0x38 Match already matched string, folded using native charset semantics for non-utf8 */
  69 #define REFFL                   57      /* 0x39 Match already matched string, folded in loc. */
  70 #define REFFU                   58      /* 0x3a Match already matched string, folded using unicode semantics for non-utf8 */
  71 #define NREF                    59      /* 0x3b Match some already matched string */
  72 #define NREFF                   60      /* 0x3c Match already matched string, folded using native charset semantics for non-utf8 */
  73 #define NREFFL                  61      /* 0x3d Match already matched string, folded in loc. */
  74 #define NREFFU                  62      /* 0x3e Match already matched string, folded using unicode semantics for non-utf8 */
  75 #define IFMATCH                 63      /* 0x3f Succeeds if the following matches. */
  76 #define UNLESSM                 64      /* 0x40 Fails if the following matches. */
  77 #define SUSPEND                 65      /* 0x41 "Independent" sub-RE. */
  78 #define IFTHEN                  66      /* 0x42 Switch, should be preceded by switcher . */
  79 #define GROUPP                  67      /* 0x43 Whether the group matched. */
  80 #define LONGJMP                 68      /* 0x44 Jump far away. */
  81 #define BRANCHJ                 69      /* 0x45 BRANCH with long offset. */
  82 #define EVAL                    70      /* 0x46 Execute some Perl code. */
  83 #define MINMOD                  71      /* 0x47 Next operator is not greedy. */
  84 #define LOGICAL                 72      /* 0x48 Next opcode should set the flag only. */
  85 #define RENUM                   73      /* 0x49 Group with independently numbered parens. */
  86 #define TRIE                    74      /* 0x4a Match many EXACT(F[LU]?)? at once. flags==type */
  87 #define TRIEC                   75      /* 0x4b Same as TRIE, but with embedded charclass data */
  88 #define AHOCORASICK             76      /* 0x4c Aho Corasick stclass. flags==type */
  89 #define AHOCORASICKC            77      /* 0x4d Same as AHOCORASICK, but with embedded charclass data */
  90 #define GOSUB                   78      /* 0x4e recurse to paren arg1 at (signed) ofs arg2 */
  91 #define GOSTART                 79      /* 0x4f recurse to start of pattern */
  92 #define NGROUPP                 80      /* 0x50 Whether the group matched. */
  93 #define INSUBP                  81      /* 0x51 Whether we are in a specific recurse. */
  94 #define DEFINEP                 82      /* 0x52 Never execute directly. */
  95 #define ENDLIKE                 83      /* 0x53 Used only for the type field of verbs */
  96 #define OPFAIL                  84      /* 0x54 Same as (?!) */
  97 #define ACCEPT                  85      /* 0x55 Accepts the current matched string. */
  98 #define VERB                    86      /* 0x56 Used only for the type field of verbs */
  99 #define PRUNE                   87      /* 0x57 Pattern fails at this startpoint if no-backtracking through this */
 100 #define MARKPOINT               88      /* 0x58 Push the current location for rollback by cut. */
 101 #define SKIP                    89      /* 0x59 On failure skip forward (to the mark) before retrying */
 102 #define COMMIT                  90      /* 0x5a Pattern fails outright if backtracking through this */
 103 #define CUTGROUP                91      /* 0x5b On failure go to the next alternation in the group */
 104 #define KEEPS                   92      /* 0x5c $& begins here. */
 105 #define LNBREAK                 93      /* 0x5d generic newline pattern */
 106 #define VERTWS                  94      /* 0x5e vertical whitespace         (Perl 6) */
 107 #define NVERTWS                 95      /* 0x5f not vertical whitespace     (Perl 6) */
 108 #define HORIZWS                 96      /* 0x60 horizontal whitespace       (Perl 6) */
 109 #define NHORIZWS                97      /* 0x61 not horizontal whitespace   (Perl 6) */
 110 #define FOLDCHAR                98      /* 0x62 codepoint with tricky case folding properties. */
 111 #define OPTIMIZED               99      /* 0x63 Placeholder for dump. */
 112 #define PSEUDO                  100     /* 0x64 Pseudo opcode for internal use. */
 113         /* ------------ States ------------- */
 114 #define TRIE_next               (REGNODE_MAX + 1)       /* state for TRIE */
 115 #define TRIE_next_fail          (REGNODE_MAX + 2)       /* state for TRIE */
 116 #define EVAL_AB                 (REGNODE_MAX + 3)       /* state for EVAL */
 117 #define EVAL_AB_fail            (REGNODE_MAX + 4)       /* state for EVAL */
 118 #define CURLYX_end              (REGNODE_MAX + 5)       /* state for CURLYX */
 119 #define CURLYX_end_fail         (REGNODE_MAX + 6)       /* state for CURLYX */
 120 #define WHILEM_A_pre            (REGNODE_MAX + 7)       /* state for WHILEM */
 121 #define WHILEM_A_pre_fail       (REGNODE_MAX + 8)       /* state for WHILEM */
 122 #define WHILEM_A_min            (REGNODE_MAX + 9)       /* state for WHILEM */
 123 #define WHILEM_A_min_fail       (REGNODE_MAX + 10)      /* state for WHILEM */
 124 #define WHILEM_A_max            (REGNODE_MAX + 11)      /* state for WHILEM */
 125 #define WHILEM_A_max_fail       (REGNODE_MAX + 12)      /* state for WHILEM */
 126 #define WHILEM_B_min            (REGNODE_MAX + 13)      /* state for WHILEM */
 127 #define WHILEM_B_min_fail       (REGNODE_MAX + 14)      /* state for WHILEM */
 128 #define WHILEM_B_max            (REGNODE_MAX + 15)      /* state for WHILEM */
 129 #define WHILEM_B_max_fail       (REGNODE_MAX + 16)      /* state for WHILEM */
 130 #define BRANCH_next             (REGNODE_MAX + 17)      /* state for BRANCH */
 131 #define BRANCH_next_fail        (REGNODE_MAX + 18)      /* state for BRANCH */
 132 #define CURLYM_A                (REGNODE_MAX + 19)      /* state for CURLYM */
 133 #define CURLYM_A_fail           (REGNODE_MAX + 20)      /* state for CURLYM */
 134 #define CURLYM_B                (REGNODE_MAX + 21)      /* state for CURLYM */
 135 #define CURLYM_B_fail           (REGNODE_MAX + 22)      /* state for CURLYM */
 136 #define IFMATCH_A               (REGNODE_MAX + 23)      /* state for IFMATCH */
 137 #define IFMATCH_A_fail          (REGNODE_MAX + 24)      /* state for IFMATCH */
 138 #define CURLY_B_min_known       (REGNODE_MAX + 25)      /* state for CURLY */
 139 #define CURLY_B_min_known_fail  (REGNODE_MAX + 26)      /* state for CURLY */
 140 #define CURLY_B_min             (REGNODE_MAX + 27)      /* state for CURLY */
 141 #define CURLY_B_min_fail        (REGNODE_MAX + 28)      /* state for CURLY */
 142 #define CURLY_B_max             (REGNODE_MAX + 29)      /* state for CURLY */
 143 #define CURLY_B_max_fail        (REGNODE_MAX + 30)      /* state for CURLY */
 144 #define COMMIT_next             (REGNODE_MAX + 31)      /* state for COMMIT */
 145 #define COMMIT_next_fail        (REGNODE_MAX + 32)      /* state for COMMIT */
 146 #define MARKPOINT_next          (REGNODE_MAX + 33)      /* state for MARKPOINT */
 147 #define MARKPOINT_next_fail     (REGNODE_MAX + 34)      /* state for MARKPOINT */
 148 #define SKIP_next               (REGNODE_MAX + 35)      /* state for SKIP */
 149 #define SKIP_next_fail          (REGNODE_MAX + 36)      /* state for SKIP */
 150 #define CUTGROUP_next           (REGNODE_MAX + 37)      /* state for CUTGROUP */
 151 #define CUTGROUP_next_fail      (REGNODE_MAX + 38)      /* state for CUTGROUP */
 152 #define KEEPS_next              (REGNODE_MAX + 39)      /* state for KEEPS */
 153 #define KEEPS_next_fail         (REGNODE_MAX + 40)      /* state for KEEPS */
 154
 155 /* PL_regkind[] What type of regop or state is this. */
 156
 157 #ifndef DOINIT
 158 EXTCONST U8 PL_regkind[];
 159 #else
 160 EXTCONST U8 PL_regkind[] = {
 161         END,            /* END                    */
 162         END,            /* SUCCEED                */
 163         BOL,            /* BOL                    */
 164         BOL,            /* MBOL                   */
 165         BOL,            /* SBOL                   */
 166         EOL,            /* EOS                    */
 167         EOL,            /* EOL                    */
 168         EOL,            /* MEOL                   */
 169         EOL,            /* SEOL                   */
 170         BOUND,          /* BOUND                  */
 171         BOUND,          /* BOUNDL                 */
 172         NBOUND,         /* NBOUND                 */
 173         NBOUND,         /* NBOUNDL                */
 174         GPOS,           /* GPOS                   */
 175         REG_ANY,        /* REG_ANY                */
 176         REG_ANY,        /* SANY                   */
 177         REG_ANY,        /* CANY                   */
 178         ANYOF,          /* ANYOF                  */
 179         ANYOF,          /* ANYOFV                 */
 180         ALNUM,          /* ALNUM                  */
 181         ALNUM,          /* ALNUML                 */
 182         ALNUM,          /* ALNUMU                 */
 183         NALNUM,         /* NALNUM                 */
 184         NALNUM,         /* NALNUML                */
 185         NALNUM,         /* NALNUMU                */
 186         SPACE,          /* SPACE                  */
 187         SPACE,          /* SPACEL                 */
 188         SPACE,          /* SPACEU                 */
 189         NSPACE,         /* NSPACE                 */
 190         NSPACE,         /* NSPACEL                */
 191         NSPACE,         /* NSPACEU                */
 192         DIGIT,          /* DIGIT                  */
 193         DIGIT,          /* DIGITL                 */
 194         DIGIT,          /* DIGITU                 */
 195         NDIGIT,         /* NDIGIT                 */
 196         NDIGIT,         /* NDIGITL                */
 197         NDIGIT,         /* NDIGITU                */
 198         CLUMP,          /* CLUMP                  */
 199         BRANCH,         /* BRANCH                 */
 200         BACK,           /* BACK                   */
 201         EXACT,          /* EXACT                  */
 202         EXACT,          /* EXACTF                 */
 203         EXACT,          /* EXACTFL                */
 204         EXACT,          /* EXACTFU                */
 205         NOTHING,        /* NOTHING                */
 206         NOTHING,        /* TAIL                   */
 207         STAR,           /* STAR                   */
 208         PLUS,           /* PLUS                   */
 209         CURLY,          /* CURLY                  */
 210         CURLY,          /* CURLYN                 */
 211         CURLY,          /* CURLYM                 */
 212         CURLY,          /* CURLYX                 */
 213         WHILEM,         /* WHILEM                 */
 214         OPEN,           /* OPEN                   */
 215         CLOSE,          /* CLOSE                  */
 216         REF,            /* REF                    */
 217         REF,            /* REFF                   */
 218         REF,            /* REFFL                  */
 219         REF,            /* REFFU                  */
 220         REF,            /* NREF                   */
 221         REF,            /* NREFF                  */
 222         REF,            /* NREFFL                 */
 223         REF,            /* NREFFU                 */
 224         BRANCHJ,        /* IFMATCH                */
 225         BRANCHJ,        /* UNLESSM                */
 226         BRANCHJ,        /* SUSPEND                */
 227         BRANCHJ,        /* IFTHEN                 */
 228         GROUPP,         /* GROUPP                 */
 229         LONGJMP,        /* LONGJMP                */
 230         BRANCHJ,        /* BRANCHJ                */
 231         EVAL,           /* EVAL                   */
 232         MINMOD,         /* MINMOD                 */
 233         LOGICAL,        /* LOGICAL                */
 234         BRANCHJ,        /* RENUM                  */
 235         TRIE,           /* TRIE                   */
 236         TRIE,           /* TRIEC                  */
 237         TRIE,           /* AHOCORASICK            */
 238         TRIE,           /* AHOCORASICKC           */
 239         GOSUB,          /* GOSUB                  */
 240         GOSTART,        /* GOSTART                */
 241         NGROUPP,        /* NGROUPP                */
 242         INSUBP,         /* INSUBP                 */
 243         DEFINEP,        /* DEFINEP                */
 244         ENDLIKE,        /* ENDLIKE                */
 245         ENDLIKE,        /* OPFAIL                 */
 246         ENDLIKE,        /* ACCEPT                 */
 247         VERB,           /* VERB                   */
 248         VERB,           /* PRUNE                  */
 249         VERB,           /* MARKPOINT              */
 250         VERB,           /* SKIP                   */
 251         VERB,           /* COMMIT                 */
 252         VERB,           /* CUTGROUP               */
 253         KEEPS,          /* KEEPS                  */
 254         LNBREAK,        /* LNBREAK                */
 255         VERTWS,         /* VERTWS                 */
 256         NVERTWS,        /* NVERTWS                */
 257         HORIZWS,        /* HORIZWS                */
 258         NHORIZWS,       /* NHORIZWS               */
 259         FOLDCHAR,       /* FOLDCHAR               */
 260         NOTHING,        /* OPTIMIZED              */
 261         PSEUDO,         /* PSEUDO                 */
 262         /* ------------ States ------------- */
 263         TRIE,           /* TRIE_next              */
 264         TRIE,           /* TRIE_next_fail         */
 265         EVAL,           /* EVAL_AB                */
 266         EVAL,           /* EVAL_AB_fail           */
 267         CURLYX,         /* CURLYX_end             */
 268         CURLYX,         /* CURLYX_end_fail        */
 269         WHILEM,         /* WHILEM_A_pre           */
 270         WHILEM,         /* WHILEM_A_pre_fail      */
 271         WHILEM,         /* WHILEM_A_min           */
 272         WHILEM,         /* WHILEM_A_min_fail      */
 273         WHILEM,         /* WHILEM_A_max           */
 274         WHILEM,         /* WHILEM_A_max_fail      */
 275         WHILEM,         /* WHILEM_B_min           */
 276         WHILEM,         /* WHILEM_B_min_fail      */
 277         WHILEM,         /* WHILEM_B_max           */
 278         WHILEM,         /* WHILEM_B_max_fail      */
 279         BRANCH,         /* BRANCH_next            */
 280         BRANCH,         /* BRANCH_next_fail       */
 281         CURLYM,         /* CURLYM_A               */
 282         CURLYM,         /* CURLYM_A_fail          */
 283         CURLYM,         /* CURLYM_B               */
 284         CURLYM,         /* CURLYM_B_fail          */
 285         IFMATCH,        /* IFMATCH_A              */
 286         IFMATCH,        /* IFMATCH_A_fail         */
 287         CURLY,          /* CURLY_B_min_known      */
 288         CURLY,          /* CURLY_B_min_known_fail */
 289         CURLY,          /* CURLY_B_min            */
 290         CURLY,          /* CURLY_B_min_fail       */
 291         CURLY,          /* CURLY_B_max            */
 292         CURLY,          /* CURLY_B_max_fail       */
 293         COMMIT,         /* COMMIT_next            */
 294         COMMIT,         /* COMMIT_next_fail       */
 295         MARKPOINT,      /* MARKPOINT_next         */
 296         MARKPOINT,      /* MARKPOINT_next_fail    */
 297         SKIP,           /* SKIP_next              */
 298         SKIP,           /* SKIP_next_fail         */
 299         CUTGROUP,       /* CUTGROUP_next          */
 300         CUTGROUP,       /* CUTGROUP_next_fail     */
 301         KEEPS,          /* KEEPS_next             */
 302         KEEPS,          /* KEEPS_next_fail        */
 303 };
 304 #endif
 305
 306 /* regarglen[] - How large is the argument part of the node (in regnodes) */
 307
 308 #ifdef REG_COMP_C
 309 static const U8 regarglen[] = {
 310         0,                                      /* END          */
 311         0,                                      /* SUCCEED      */
 312         0,                                      /* BOL          */
 313         0,                                      /* MBOL         */
 314         0,                                      /* SBOL         */
 315         0,                                      /* EOS          */
 316         0,                                      /* EOL          */
 317         0,                                      /* MEOL         */
 318         0,                                      /* SEOL         */
 319         0,                                      /* BOUND        */
 320         0,                                      /* BOUNDL       */
 321         0,                                      /* NBOUND       */
 322         0,                                      /* NBOUNDL      */
 323         0,                                      /* GPOS         */
 324         0,                                      /* REG_ANY      */
 325         0,                                      /* SANY         */
 326         0,                                      /* CANY         */
 327         0,                                      /* ANYOF        */
 328         0,                                      /* ANYOFV       */
 329         0,                                      /* ALNUM        */
 330         0,                                      /* ALNUML       */
 331         0,                                      /* ALNUMU       */
 332         0,                                      /* NALNUM       */
 333         0,                                      /* NALNUML      */
 334         0,                                      /* NALNUMU      */
 335         0,                                      /* SPACE        */
 336         0,                                      /* SPACEL       */
 337         0,                                      /* SPACEU       */
 338         0,                                      /* NSPACE       */
 339         0,                                      /* NSPACEL      */
 340         0,                                      /* NSPACEU      */
 341         0,                                      /* DIGIT        */
 342         0,                                      /* DIGITL       */
 343         0,                                      /* DIGITU       */
 344         0,                                      /* NDIGIT       */
 345         0,                                      /* NDIGITL      */
 346         0,                                      /* NDIGITU      */
 347         0,                                      /* CLUMP        */
 348         0,                                      /* BRANCH       */
 349         0,                                      /* BACK         */
 350         0,                                      /* EXACT        */
 351         0,                                      /* EXACTF       */
 352         0,                                      /* EXACTFL      */
 353         0,                                      /* EXACTFU      */
 354         0,                                      /* NOTHING      */
 355         0,                                      /* TAIL         */
 356         0,                                      /* STAR         */
 357         0,                                      /* PLUS         */
 358         EXTRA_SIZE(struct regnode_2),           /* CURLY        */
 359         EXTRA_SIZE(struct regnode_2),           /* CURLYN       */
 360         EXTRA_SIZE(struct regnode_2),           /* CURLYM       */
 361         EXTRA_SIZE(struct regnode_2),           /* CURLYX       */
 362         0,                                      /* WHILEM       */
 363         EXTRA_SIZE(struct regnode_1),           /* OPEN         */
 364         EXTRA_SIZE(struct regnode_1),           /* CLOSE        */
 365         EXTRA_SIZE(struct regnode_1),           /* REF          */
 366         EXTRA_SIZE(struct regnode_1),           /* REFF         */
 367         EXTRA_SIZE(struct regnode_1),           /* REFFL        */
 368         EXTRA_SIZE(struct regnode_1),           /* REFFU        */
 369         EXTRA_SIZE(struct regnode_1),           /* NREF         */
 370         EXTRA_SIZE(struct regnode_1),           /* NREFF        */
 371         EXTRA_SIZE(struct regnode_1),           /* NREFFL       */
 372         EXTRA_SIZE(struct regnode_1),           /* NREFFU       */
 373         EXTRA_SIZE(struct regnode_1),           /* IFMATCH      */
 374         EXTRA_SIZE(struct regnode_1),           /* UNLESSM      */
 375         EXTRA_SIZE(struct regnode_1),           /* SUSPEND      */
 376         EXTRA_SIZE(struct regnode_1),           /* IFTHEN       */
 377         EXTRA_SIZE(struct regnode_1),           /* GROUPP       */
 378         EXTRA_SIZE(struct regnode_1),           /* LONGJMP      */
 379         EXTRA_SIZE(struct regnode_1),           /* BRANCHJ      */
 380         EXTRA_SIZE(struct regnode_1),           /* EVAL         */
 381         0,                                      /* MINMOD       */
 382         0,                                      /* LOGICAL      */
 383         EXTRA_SIZE(struct regnode_1),           /* RENUM        */
 384         EXTRA_SIZE(struct regnode_1),           /* TRIE         */
 385         EXTRA_SIZE(struct regnode_charclass),   /* TRIEC        */
 386         EXTRA_SIZE(struct regnode_1),           /* AHOCORASICK  */
 387         EXTRA_SIZE(struct regnode_charclass),   /* AHOCORASICKC */
 388         EXTRA_SIZE(struct regnode_2L),          /* GOSUB        */
 389         0,                                      /* GOSTART      */
 390         EXTRA_SIZE(struct regnode_1),           /* NGROUPP      */
 391         EXTRA_SIZE(struct regnode_1),           /* INSUBP       */
 392         EXTRA_SIZE(struct regnode_1),           /* DEFINEP      */
 393         0,                                      /* ENDLIKE      */
 394         0,                                      /* OPFAIL       */
 395         EXTRA_SIZE(struct regnode_1),           /* ACCEPT       */
 396         EXTRA_SIZE(struct regnode_1),           /* VERB         */
 397         EXTRA_SIZE(struct regnode_1),           /* PRUNE        */
 398         EXTRA_SIZE(struct regnode_1),           /* MARKPOINT    */
 399         EXTRA_SIZE(struct regnode_1),           /* SKIP         */
 400         EXTRA_SIZE(struct regnode_1),           /* COMMIT       */
 401         EXTRA_SIZE(struct regnode_1),           /* CUTGROUP     */
 402         0,                                      /* KEEPS        */
 403         0,                                      /* LNBREAK      */
 404         0,                                      /* VERTWS       */
 405         0,                                      /* NVERTWS      */
 406         0,                                      /* HORIZWS      */
 407         0,                                      /* NHORIZWS     */
 408         EXTRA_SIZE(struct regnode_1),           /* FOLDCHAR     */
 409         0,                                      /* OPTIMIZED    */
 410         0,                                      /* PSEUDO       */
 411 };
 412
 413 /* reg_off_by_arg[] - Which argument holds the offset to the next node */
 414
 415 static const char reg_off_by_arg[] = {
 416         0,      /* END          */
 417         0,      /* SUCCEED      */
 418         0,      /* BOL          */
 419         0,      /* MBOL         */
 420         0,      /* SBOL         */
 421         0,      /* EOS          */
 422         0,      /* EOL          */
 423         0,      /* MEOL         */
 424         0,      /* SEOL         */
 425         0,      /* BOUND        */
 426         0,      /* BOUNDL       */
 427         0,      /* NBOUND       */
 428         0,      /* NBOUNDL      */
 429         0,      /* GPOS         */
 430         0,      /* REG_ANY      */
 431         0,      /* SANY         */
 432         0,      /* CANY         */
 433         0,      /* ANYOF        */
 434         0,      /* ANYOFV       */
 435         0,      /* ALNUM        */
 436         0,      /* ALNUML       */
 437         0,      /* ALNUMU       */
 438         0,      /* NALNUM       */
 439         0,      /* NALNUML      */
 440         0,      /* NALNUMU      */
 441         0,      /* SPACE        */
 442         0,      /* SPACEL       */
 443         0,      /* SPACEU       */
 444         0,      /* NSPACE       */
 445         0,      /* NSPACEL      */
 446         0,      /* NSPACEU      */
 447         0,      /* DIGIT        */
 448         0,      /* DIGITL       */
 449         0,      /* DIGITU       */
 450         0,      /* NDIGIT       */
 451         0,      /* NDIGITL      */
 452         0,      /* NDIGITU      */
 453         0,      /* CLUMP        */
 454         0,      /* BRANCH       */
 455         0,      /* BACK         */
 456         0,      /* EXACT        */
 457         0,      /* EXACTF       */
 458         0,      /* EXACTFL      */
 459         0,      /* EXACTFU      */
 460         0,      /* NOTHING      */
 461         0,      /* TAIL         */
 462         0,      /* STAR         */
 463         0,      /* PLUS         */
 464         0,      /* CURLY        */
 465         0,      /* CURLYN       */
 466         0,      /* CURLYM       */
 467         0,      /* CURLYX       */
 468         0,      /* WHILEM       */
 469         0,      /* OPEN         */
 470         0,      /* CLOSE        */
 471         0,      /* REF          */
 472         0,      /* REFF         */
 473         0,      /* REFFL        */
 474         0,      /* REFFU        */
 475         0,      /* NREF         */
 476         0,      /* NREFF        */
 477         0,      /* NREFFL       */
 478         0,      /* NREFFU       */
 479         2,      /* IFMATCH      */
 480         2,      /* UNLESSM      */
 481         1,      /* SUSPEND      */
 482         1,      /* IFTHEN       */
 483         0,      /* GROUPP       */
 484         1,      /* LONGJMP      */
 485         1,      /* BRANCHJ      */
 486         0,      /* EVAL         */
 487         0,      /* MINMOD       */
 488         0,      /* LOGICAL      */
 489         1,      /* RENUM        */
 490         0,      /* TRIE         */
 491         0,      /* TRIEC        */
 492         0,      /* AHOCORASICK  */
 493         0,      /* AHOCORASICKC */
 494         0,      /* GOSUB        */
 495         0,      /* GOSTART      */
 496         0,      /* NGROUPP      */
 497         0,      /* INSUBP       */
 498         0,      /* DEFINEP      */
 499         0,      /* ENDLIKE      */
 500         0,      /* OPFAIL       */
 501         0,      /* ACCEPT       */
 502         0,      /* VERB         */
 503         0,      /* PRUNE        */
 504         0,      /* MARKPOINT    */
 505         0,      /* SKIP         */
 506         0,      /* COMMIT       */
 507         0,      /* CUTGROUP     */
 508         0,      /* KEEPS        */
 509         0,      /* LNBREAK      */
 510         0,      /* VERTWS       */
 511         0,      /* NVERTWS      */
 512         0,      /* HORIZWS      */
 513         0,      /* NHORIZWS     */
 514         0,      /* FOLDCHAR     */
 515         0,      /* OPTIMIZED    */
 516         0,      /* PSEUDO       */
 517 };
 518
 519 #endif /* REG_COMP_C */
 520
 521 /* reg_name[] - Opcode/state names in string form, for debugging */
 522
 523 #ifndef DOINIT
 524 EXTCONST char * PL_reg_name[];
 525 #else
 526 EXTCONST char * const PL_reg_name[] = {
 527         "END",                          /* 0000 */
 528         "SUCCEED",                      /* 0x01 */
 529         "BOL",                          /* 0x02 */
 530         "MBOL",                         /* 0x03 */
 531         "SBOL",                         /* 0x04 */
 532         "EOS",                          /* 0x05 */
 533         "EOL",                          /* 0x06 */
 534         "MEOL",                         /* 0x07 */
 535         "SEOL",                         /* 0x08 */
 536         "BOUND",                        /* 0x09 */
 537         "BOUNDL",                       /* 0x0a */
 538         "NBOUND",                       /* 0x0b */
 539         "NBOUNDL",                      /* 0x0c */
 540         "GPOS",                         /* 0x0d */
 541         "REG_ANY",                      /* 0x0e */
 542         "SANY",                         /* 0x0f */
 543         "CANY",                         /* 0x10 */
 544         "ANYOF",                        /* 0x11 */
 545         "ANYOFV",                       /* 0x12 */
 546         "ALNUM",                        /* 0x13 */
 547         "ALNUML",                       /* 0x14 */
 548         "ALNUMU",                       /* 0x15 */
 549         "NALNUM",                       /* 0x16 */
 550         "NALNUML",                      /* 0x17 */
 551         "NALNUMU",                      /* 0x18 */
 552         "SPACE",                        /* 0x19 */
 553         "SPACEL",                       /* 0x1a */
 554         "SPACEU",                       /* 0x1b */
 555         "NSPACE",                       /* 0x1c */
 556         "NSPACEL",                      /* 0x1d */
 557         "NSPACEU",                      /* 0x1e */
 558         "DIGIT",                        /* 0x1f */
 559         "DIGITL",                       /* 0x20 */
 560         "DIGITU",                       /* 0x21 */
 561         "NDIGIT",                       /* 0x22 */
 562         "NDIGITL",                      /* 0x23 */
 563         "NDIGITU",                      /* 0x24 */
 564         "CLUMP",                        /* 0x25 */
 565         "BRANCH",                       /* 0x26 */
 566         "BACK",                         /* 0x27 */
 567         "EXACT",                        /* 0x28 */
 568         "EXACTF",                       /* 0x29 */
 569         "EXACTFL",                      /* 0x2a */
 570         "EXACTFU",                      /* 0x2b */
 571         "NOTHING",                      /* 0x2c */
 572         "TAIL",                         /* 0x2d */
 573         "STAR",                         /* 0x2e */
 574         "PLUS",                         /* 0x2f */
 575         "CURLY",                        /* 0x30 */
 576         "CURLYN",                       /* 0x31 */
 577         "CURLYM",                       /* 0x32 */
 578         "CURLYX",                       /* 0x33 */
 579         "WHILEM",                       /* 0x34 */
 580         "OPEN",                         /* 0x35 */
 581         "CLOSE",                        /* 0x36 */
 582         "REF",                          /* 0x37 */
 583         "REFF",                         /* 0x38 */
 584         "REFFL",                        /* 0x39 */
 585         "REFFU",                        /* 0x3a */
 586         "NREF",                         /* 0x3b */
 587         "NREFF",                        /* 0x3c */
 588         "NREFFL",                       /* 0x3d */
 589         "NREFFU",                       /* 0x3e */
 590         "IFMATCH",                      /* 0x3f */
 591         "UNLESSM",                      /* 0x40 */
 592         "SUSPEND",                      /* 0x41 */
 593         "IFTHEN",                       /* 0x42 */
 594         "GROUPP",                       /* 0x43 */
 595         "LONGJMP",                      /* 0x44 */
 596         "BRANCHJ",                      /* 0x45 */
 597         "EVAL",                         /* 0x46 */
 598         "MINMOD",                       /* 0x47 */
 599         "LOGICAL",                      /* 0x48 */
 600         "RENUM",                        /* 0x49 */
 601         "TRIE",                         /* 0x4a */
 602         "TRIEC",                        /* 0x4b */
 603         "AHOCORASICK",                  /* 0x4c */
 604         "AHOCORASICKC",                 /* 0x4d */
 605         "GOSUB",                        /* 0x4e */
 606         "GOSTART",                      /* 0x4f */
 607         "NGROUPP",                      /* 0x50 */
 608         "INSUBP",                       /* 0x51 */
 609         "DEFINEP",                      /* 0x52 */
 610         "ENDLIKE",                      /* 0x53 */
 611         "OPFAIL",                       /* 0x54 */
 612         "ACCEPT",                       /* 0x55 */
 613         "VERB",                         /* 0x56 */
 614         "PRUNE",                        /* 0x57 */
 615         "MARKPOINT",                    /* 0x58 */
 616         "SKIP",                         /* 0x59 */
 617         "COMMIT",                       /* 0x5a */
 618         "CUTGROUP",                     /* 0x5b */
 619         "KEEPS",                        /* 0x5c */
 620         "LNBREAK",                      /* 0x5d */
 621         "VERTWS",                       /* 0x5e */
 622         "NVERTWS",                      /* 0x5f */
 623         "HORIZWS",                      /* 0x60 */
 624         "NHORIZWS",                     /* 0x61 */
 625         "FOLDCHAR",                     /* 0x62 */
 626         "OPTIMIZED",                    /* 0x63 */
 627         "PSEUDO",                       /* 0x64 */
 628         /* ------------ States ------------- */
 629         "TRIE_next",                    /* REGNODE_MAX +0x01 */
 630         "TRIE_next_fail",               /* REGNODE_MAX +0x02 */
 631         "EVAL_AB",                      /* REGNODE_MAX +0x03 */
 632         "EVAL_AB_fail",                 /* REGNODE_MAX +0x04 */
 633         "CURLYX_end",                   /* REGNODE_MAX +0x05 */
 634         "CURLYX_end_fail",              /* REGNODE_MAX +0x06 */
 635         "WHILEM_A_pre",                 /* REGNODE_MAX +0x07 */
 636         "WHILEM_A_pre_fail",            /* REGNODE_MAX +0x08 */
 637         "WHILEM_A_min",                 /* REGNODE_MAX +0x09 */
 638         "WHILEM_A_min_fail",            /* REGNODE_MAX +0x0a */
 639         "WHILEM_A_max",                 /* REGNODE_MAX +0x0b */
 640         "WHILEM_A_max_fail",            /* REGNODE_MAX +0x0c */
 641         "WHILEM_B_min",                 /* REGNODE_MAX +0x0d */
 642         "WHILEM_B_min_fail",            /* REGNODE_MAX +0x0e */
 643         "WHILEM_B_max",                 /* REGNODE_MAX +0x0f */
 644         "WHILEM_B_max_fail",            /* REGNODE_MAX +0x10 */
 645         "BRANCH_next",                  /* REGNODE_MAX +0x11 */
 646         "BRANCH_next_fail",             /* REGNODE_MAX +0x12 */
 647         "CURLYM_A",                     /* REGNODE_MAX +0x13 */
 648         "CURLYM_A_fail",                /* REGNODE_MAX +0x14 */
 649         "CURLYM_B",                     /* REGNODE_MAX +0x15 */
 650         "CURLYM_B_fail",                /* REGNODE_MAX +0x16 */
 651         "IFMATCH_A",                    /* REGNODE_MAX +0x17 */
 652         "IFMATCH_A_fail",               /* REGNODE_MAX +0x18 */
 653         "CURLY_B_min_known",            /* REGNODE_MAX +0x19 */
 654         "CURLY_B_min_known_fail",       /* REGNODE_MAX +0x1a */
 655         "CURLY_B_min",                  /* REGNODE_MAX +0x1b */
 656         "CURLY_B_min_fail",             /* REGNODE_MAX +0x1c */
 657         "CURLY_B_max",                  /* REGNODE_MAX +0x1d */
 658         "CURLY_B_max_fail",             /* REGNODE_MAX +0x1e */
 659         "COMMIT_next",                  /* REGNODE_MAX +0x1f */
 660         "COMMIT_next_fail",             /* REGNODE_MAX +0x20 */
 661         "MARKPOINT_next",               /* REGNODE_MAX +0x21 */
 662         "MARKPOINT_next_fail",          /* REGNODE_MAX +0x22 */
 663         "SKIP_next",                    /* REGNODE_MAX +0x23 */
 664         "SKIP_next_fail",               /* REGNODE_MAX +0x24 */
 665         "CUTGROUP_next",                /* REGNODE_MAX +0x25 */
 666         "CUTGROUP_next_fail",           /* REGNODE_MAX +0x26 */
 667         "KEEPS_next",                   /* REGNODE_MAX +0x27 */
 668         "KEEPS_next_fail",              /* REGNODE_MAX +0x28 */
 669 };
 670 #endif /* DOINIT */
 671
 672 /* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
 673
 674 #ifndef DOINIT
 675 EXTCONST char * PL_reg_extflags_name[];
 676 #else
 677 EXTCONST char * const PL_reg_extflags_name[] = {
 678         /* Bits in extflags defined: 11111111111111111111111001111111 */
 679         "MULTILINE",        /* 0x00000001 */
 680         "SINGLELINE",       /* 0x00000002 */
 681         "FOLD",             /* 0x00000004 */
 682         "EXTENDED",         /* 0x00000008 */
 683         "KEEPCOPY",         /* 0x00000010 */
 684         "CHARSET",          /* 0x00000060 */
 685         "CHARSET",          /* 0x00000060 */
 686         "UNUSED_BIT_7",     /* 0x00000080 */
 687         "UNUSED_BIT_8",     /* 0x00000100 */
 688         "ANCH_BOL",         /* 0x00000200 */
 689         "ANCH_MBOL",        /* 0x00000400 */
 690         "ANCH_SBOL",        /* 0x00000800 */
 691         "ANCH_GPOS",        /* 0x00001000 */
 692         "GPOS_SEEN",        /* 0x00002000 */
 693         "GPOS_FLOAT",       /* 0x00004000 */
 694         "LOOKBEHIND_SEEN",  /* 0x00008000 */
 695         "EVAL_SEEN",        /* 0x00010000 */
 696         "CANY_SEEN",        /* 0x00020000 */
 697         "NOSCAN",           /* 0x00040000 */
 698         "CHECK_ALL",        /* 0x00080000 */
 699         "MATCH_UTF8",       /* 0x00100000 */
 700         "USE_INTUIT_NOML",  /* 0x00200000 */
 701         "USE_INTUIT_ML",    /* 0x00400000 */
 702         "INTUIT_TAIL",      /* 0x00800000 */
 703         "SPLIT",            /* 0x01000000 */
 704         "COPY_DONE",        /* 0x02000000 */
 705         "TAINTED_SEEN",     /* 0x04000000 */
 706         "TAINTED",          /* 0x08000000 */
 707         "START_ONLY",       /* 0x10000000 */
 708         "SKIPWHITE",        /* 0x20000000 */
 709         "WHITE",            /* 0x40000000 */
 710         "NULL",             /* 0x80000000 */
 711 };
 712 #endif /* DOINIT */
 713
 714 /* The following have no fixed length. U8 so we can do strchr() on it. */
 715 #define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 716
 717 #ifndef DOINIT
 718 EXTCONST U8 PL_varies[] __attribute__deprecated__;
 719 #else
 720 EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
 721     ANYOFV, CLUMP, BRANCH, BACK, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX,
 722     WHILEM, REF, REFF, REFFL, REFFU, NREF, NREFF, NREFFL, NREFFU, SUSPEND,
 723     IFTHEN, BRANCHJ,
 724     0
 725 };
 726 #endif /* DOINIT */
 727
 728 #ifndef DOINIT
 729 EXTCONST U8 PL_varies_bitmask[];
 730 #else
 731 EXTCONST U8 PL_varies_bitmask[] = {
 732     0x00, 0x00, 0x04, 0x00, 0xE0, 0xC0, 0x9F, 0x7F, 0x26, 0x00, 0x00, 0x00, 0x00
 733 };
 734 #endif /* DOINIT */
 735
 736 /* The following always have a length of 1. U8 we can do strchr() on it. */
 737 /* (Note that length 1 means "one character" under UTF8, not "one octet".) */
 738 #define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 739
 740 #ifndef DOINIT
 741 EXTCONST U8 PL_simple[] __attribute__deprecated__;
 742 #else
 743 EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
 744     REG_ANY, SANY, CANY, ANYOF, ALNUM, ALNUML, ALNUMU, NALNUM, NALNUML,
 745     NALNUMU, SPACE, SPACEL, SPACEU, NSPACE, NSPACEL, NSPACEU, DIGIT,
 746     DIGITL, DIGITU, NDIGIT, NDIGITL, NDIGITU, VERTWS, NVERTWS, HORIZWS,
 747     NHORIZWS,
 748     0
 749 };
 750 #endif /* DOINIT */
 751
 752 #ifndef DOINIT
 753 EXTCONST U8 PL_simple_bitmask[];
 754 #else
 755 EXTCONST U8 PL_simple_bitmask[] = {
 756     0x00, 0xC0, 0xFB, 0xFF, 0x1F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0xC0, 0x03
 757 };
 758 #endif /* DOINIT */
 759
 760 /* ex: set ro: */