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             92
  10 #define REGMATCH_STATE_MAX      132
  11
  12 #define END                     0       /* 0000 End of program. */
  13 #define SUCCEED                 1       /* 0x01 Return from a subroutine, basically. */
  14 #define SBOL                    2       /* 0x02 Match "" at beginning of line: /^/, /\A/ */
  15 #define BOL                     2       /* 0x02 type alias */
  16 #define MBOL                    3       /* 0x03 Same, assuming multiline: /^/m */
  17 #define SEOL                    4       /* 0x04 Match "" at end of line: /$/ */
  18 #define EOL                     4       /* 0x04 type alias */
  19 #define MEOL                    5       /* 0x05 Same, assuming multiline: /$/m */
  20 #define EOS                     6       /* 0x06 Match "" at end of string: /\z/ */
  21 #define GPOS                    7       /* 0x07 Matches where last m//g left off. */
  22 #define BOUND                   8       /* 0x08 Like BOUNDA for non-utf8, otherwise match "" between any Unicode \w\W or \W\w */
  23 #define BOUNDL                  9       /* 0x09 Like BOUND/BOUNDU, but \w and \W are defined by current locale */
  24 #define BOUNDU                  10      /* 0x0a Match "" at any boundary of a given type using Unicode rules */
  25 #define BOUNDA                  11      /* 0x0b Match "" at any boundary between \w\W or \W\w, where \w is [_a-zA-Z0-9] */
  26 #define NBOUND                  12      /* 0x0c Like NBOUNDA for non-utf8, otherwise match "" between any Unicode \w\w or \W\W */
  27 #define NBOUNDL                 13      /* 0x0d Like NBOUND/NBOUNDU, but \w and \W are defined by current locale */
  28 #define NBOUNDU                 14      /* 0x0e Match "" at any non-boundary of a given type using using Unicode rules */
  29 #define NBOUNDA                 15      /* 0x0f Match "" betweeen any \w\w or \W\W, where \w is [_a-zA-Z0-9] */
  30 #define REG_ANY                 16      /* 0x10 Match any one character (except newline). */
  31 #define SANY                    17      /* 0x11 Match any one character. */
  32 #define ANYOF                   18      /* 0x12 Match character in (or not in) this class, single char match only */
  33 #define ANYOFL                  19      /* 0x13 Like ANYOF, but /l is in effect */
  34 #define POSIXD                  20      /* 0x14 Some [[:class:]] under /d; the FLAGS field gives which one */
  35 #define POSIXL                  21      /* 0x15 Some [[:class:]] under /l; the FLAGS field gives which one */
  36 #define POSIXU                  22      /* 0x16 Some [[:class:]] under /u; the FLAGS field gives which one */
  37 #define POSIXA                  23      /* 0x17 Some [[:class:]] under /a; the FLAGS field gives which one */
  38 #define NPOSIXD                 24      /* 0x18 complement of POSIXD, [[:^class:]] */
  39 #define NPOSIXL                 25      /* 0x19 complement of POSIXL, [[:^class:]] */
  40 #define NPOSIXU                 26      /* 0x1a complement of POSIXU, [[:^class:]] */
  41 #define NPOSIXA                 27      /* 0x1b complement of POSIXA, [[:^class:]] */
  42 #define CLUMP                   28      /* 0x1c Match any extended grapheme cluster sequence */
  43 #define BRANCH                  29      /* 0x1d Match this alternative, or the next... */
  44 #define EXACT                   30      /* 0x1e Match this string (preceded by length). */
  45 #define EXACTL                  31      /* 0x1f Like EXACT, but /l is in effect. */
  46 #define EXACTF                  32      /* 0x20 Match this non-UTF-8 string (not guaranteed to be folded) using /id rules (w/len). */
  47 #define EXACTFL                 33      /* 0x21 Match this string (not guaranteed to be folded) using /il rules (w/len). */
  48 #define EXACTFU                 34      /* 0x22 Match this string (folded iff in UTF-8, length in folding doesn't change if not in UTF-8) using /iu rules (w/len). */
  49 #define EXACTFA                 35      /* 0x23 Match this string (not guaranteed to be folded) using /iaa rules (w/len). */
  50 #define EXACTFU_SS              36      /* 0x24 Match this string (folded iff in UTF-8, length in folding may change even if not in UTF-8) using /iu rules (w/len). */
  51 #define EXACTFLU8               37      /* 0x25 Rare cirucmstances: like EXACTFU, but is under /l, UTF-8, folded, and everything in it is above 255. */
  52 #define EXACTFA_NO_TRIE         38      /* 0x26 Match this string (which is not trie-able; not guaranteed to be folded) using /iaa rules (w/len). */
  53 #define NOTHING                 39      /* 0x27 Match empty string. */
  54 #define TAIL                    40      /* 0x28 Match empty string. Can jump here from outside. */
  55 #define STAR                    41      /* 0x29 Match this (simple) thing 0 or more times. */
  56 #define PLUS                    42      /* 0x2a Match this (simple) thing 1 or more times. */
  57 #define CURLY                   43      /* 0x2b Match this simple thing {n,m} times. */
  58 #define CURLYN                  44      /* 0x2c Capture next-after-this simple thing */
  59 #define CURLYM                  45      /* 0x2d Capture this medium-complex thing {n,m} times. */
  60 #define CURLYX                  46      /* 0x2e Match this complex thing {n,m} times. */
  61 #define WHILEM                  47      /* 0x2f Do curly processing and see if rest matches. */
  62 #define OPEN                    48      /* 0x30 Mark this point in input as start of #n. */
  63 #define CLOSE                   49      /* 0x31 Analogous to OPEN. */
  64 #define REF                     50      /* 0x32 Match some already matched string */
  65 #define REFF                    51      /* 0x33 Match already matched string, folded using native charset rules for non-utf8 */
  66 #define REFFL                   52      /* 0x34 Match already matched string, folded in loc. */
  67 #define REFFU                   53      /* 0x35 Match already matched string, folded using unicode rules for non-utf8 */
  68 #define REFFA                   54      /* 0x36 Match already matched string, folded using unicode rules for non-utf8, no mixing ASCII, non-ASCII */
  69 #define NREF                    55      /* 0x37 Match some already matched string */
  70 #define NREFF                   56      /* 0x38 Match already matched string, folded using native charset rules for non-utf8 */
  71 #define NREFFL                  57      /* 0x39 Match already matched string, folded in loc. */
  72 #define NREFFU                  58      /* 0x3a Match already matched string, folded using unicode rules for non-utf8 */
  73 #define NREFFA                  59      /* 0x3b Match already matched string, folded using unicode rules for non-utf8, no mixing ASCII, non-ASCII */
  74 #define LONGJMP                 60      /* 0x3c Jump far away. */
  75 #define BRANCHJ                 61      /* 0x3d BRANCH with long offset. */
  76 #define IFMATCH                 62      /* 0x3e Succeeds if the following matches. */
  77 #define UNLESSM                 63      /* 0x3f Fails if the following matches. */
  78 #define SUSPEND                 64      /* 0x40 "Independent" sub-RE. */
  79 #define IFTHEN                  65      /* 0x41 Switch, should be preceded by switcher. */
  80 #define GROUPP                  66      /* 0x42 Whether the group matched. */
  81 #define EVAL                    67      /* 0x43 Execute some Perl code. */
  82 #define MINMOD                  68      /* 0x44 Next operator is not greedy. */
  83 #define LOGICAL                 69      /* 0x45 Next opcode should set the flag only. */
  84 #define RENUM                   70      /* 0x46 Group with independently numbered parens. */
  85 #define TRIE                    71      /* 0x47 Match many EXACT(F[ALU]?)? at once. flags==type */
  86 #define TRIEC                   72      /* 0x48 Same as TRIE, but with embedded charclass data */
  87 #define AHOCORASICK             73      /* 0x49 Aho Corasick stclass. flags==type */
  88 #define AHOCORASICKC            74      /* 0x4a Same as AHOCORASICK, but with embedded charclass data */
  89 #define GOSUB                   75      /* 0x4b recurse to paren arg1 at (signed) ofs arg2 */
  90 #define GOSTART                 76      /* 0x4c recurse to start of pattern */
  91 #define NGROUPP                 77      /* 0x4d Whether the group matched. */
  92 #define INSUBP                  78      /* 0x4e Whether we are in a specific recurse. */
  93 #define DEFINEP                 79      /* 0x4f Never execute directly. */
  94 #define ENDLIKE                 80      /* 0x50 Used only for the type field of verbs */
  95 #define OPFAIL                  81      /* 0x51 Same as (?!) */
  96 #define ACCEPT                  82      /* 0x52 Accepts the current matched string. */
  97 #define VERB                    83      /* 0x53 Used only for the type field of verbs */
  98 #define PRUNE                   84      /* 0x54 Pattern fails at this startpoint if no-backtracking through this */
  99 #define MARKPOINT               85      /* 0x55 Push the current location for rollback by cut. */
 100 #define SKIP                    86      /* 0x56 On failure skip forward (to the mark) before retrying */
 101 #define COMMIT                  87      /* 0x57 Pattern fails outright if backtracking through this */
 102 #define CUTGROUP                88      /* 0x58 On failure go to the next alternation in the group */
 103 #define KEEPS                   89      /* 0x59 $& begins here. */
 104 #define LNBREAK                 90      /* 0x5a generic newline pattern */
 105 #define OPTIMIZED               91      /* 0x5b Placeholder for dump. */
 106 #define PSEUDO                  92      /* 0x5c Pseudo opcode for internal use. */
 107         /* ------------ States ------------- */
 108 #define TRIE_next               (REGNODE_MAX + 1)       /* state for TRIE */
 109 #define TRIE_next_fail          (REGNODE_MAX + 2)       /* state for TRIE */
 110 #define EVAL_AB                 (REGNODE_MAX + 3)       /* state for EVAL */
 111 #define EVAL_AB_fail            (REGNODE_MAX + 4)       /* state for EVAL */
 112 #define CURLYX_end              (REGNODE_MAX + 5)       /* state for CURLYX */
 113 #define CURLYX_end_fail         (REGNODE_MAX + 6)       /* state for CURLYX */
 114 #define WHILEM_A_pre            (REGNODE_MAX + 7)       /* state for WHILEM */
 115 #define WHILEM_A_pre_fail       (REGNODE_MAX + 8)       /* state for WHILEM */
 116 #define WHILEM_A_min            (REGNODE_MAX + 9)       /* state for WHILEM */
 117 #define WHILEM_A_min_fail       (REGNODE_MAX + 10)      /* state for WHILEM */
 118 #define WHILEM_A_max            (REGNODE_MAX + 11)      /* state for WHILEM */
 119 #define WHILEM_A_max_fail       (REGNODE_MAX + 12)      /* state for WHILEM */
 120 #define WHILEM_B_min            (REGNODE_MAX + 13)      /* state for WHILEM */
 121 #define WHILEM_B_min_fail       (REGNODE_MAX + 14)      /* state for WHILEM */
 122 #define WHILEM_B_max            (REGNODE_MAX + 15)      /* state for WHILEM */
 123 #define WHILEM_B_max_fail       (REGNODE_MAX + 16)      /* state for WHILEM */
 124 #define BRANCH_next             (REGNODE_MAX + 17)      /* state for BRANCH */
 125 #define BRANCH_next_fail        (REGNODE_MAX + 18)      /* state for BRANCH */
 126 #define CURLYM_A                (REGNODE_MAX + 19)      /* state for CURLYM */
 127 #define CURLYM_A_fail           (REGNODE_MAX + 20)      /* state for CURLYM */
 128 #define CURLYM_B                (REGNODE_MAX + 21)      /* state for CURLYM */
 129 #define CURLYM_B_fail           (REGNODE_MAX + 22)      /* state for CURLYM */
 130 #define IFMATCH_A               (REGNODE_MAX + 23)      /* state for IFMATCH */
 131 #define IFMATCH_A_fail          (REGNODE_MAX + 24)      /* state for IFMATCH */
 132 #define CURLY_B_min_known       (REGNODE_MAX + 25)      /* state for CURLY */
 133 #define CURLY_B_min_known_fail  (REGNODE_MAX + 26)      /* state for CURLY */
 134 #define CURLY_B_min             (REGNODE_MAX + 27)      /* state for CURLY */
 135 #define CURLY_B_min_fail        (REGNODE_MAX + 28)      /* state for CURLY */
 136 #define CURLY_B_max             (REGNODE_MAX + 29)      /* state for CURLY */
 137 #define CURLY_B_max_fail        (REGNODE_MAX + 30)      /* state for CURLY */
 138 #define COMMIT_next             (REGNODE_MAX + 31)      /* state for COMMIT */
 139 #define COMMIT_next_fail        (REGNODE_MAX + 32)      /* state for COMMIT */
 140 #define MARKPOINT_next          (REGNODE_MAX + 33)      /* state for MARKPOINT */
 141 #define MARKPOINT_next_fail     (REGNODE_MAX + 34)      /* state for MARKPOINT */
 142 #define SKIP_next               (REGNODE_MAX + 35)      /* state for SKIP */
 143 #define SKIP_next_fail          (REGNODE_MAX + 36)      /* state for SKIP */
 144 #define CUTGROUP_next           (REGNODE_MAX + 37)      /* state for CUTGROUP */
 145 #define CUTGROUP_next_fail      (REGNODE_MAX + 38)      /* state for CUTGROUP */
 146 #define KEEPS_next              (REGNODE_MAX + 39)      /* state for KEEPS */
 147 #define KEEPS_next_fail         (REGNODE_MAX + 40)      /* state for KEEPS */
 148
 149 /* PL_regkind[] What type of regop or state is this. */
 150
 151 #ifndef DOINIT
 152 EXTCONST U8 PL_regkind[];
 153 #else
 154 EXTCONST U8 PL_regkind[] = {
 155         END,            /* END                    */
 156         END,            /* SUCCEED                */
 157         BOL,            /* SBOL                   */
 158         BOL,            /* MBOL                   */
 159         EOL,            /* SEOL                   */
 160         EOL,            /* MEOL                   */
 161         EOL,            /* EOS                    */
 162         GPOS,           /* GPOS                   */
 163         BOUND,          /* BOUND                  */
 164         BOUND,          /* BOUNDL                 */
 165         BOUND,          /* BOUNDU                 */
 166         BOUND,          /* BOUNDA                 */
 167         NBOUND,         /* NBOUND                 */
 168         NBOUND,         /* NBOUNDL                */
 169         NBOUND,         /* NBOUNDU                */
 170         NBOUND,         /* NBOUNDA                */
 171         REG_ANY,        /* REG_ANY                */
 172         REG_ANY,        /* SANY                   */
 173         ANYOF,          /* ANYOF                  */
 174         ANYOF,          /* ANYOFL                 */
 175         POSIXD,         /* POSIXD                 */
 176         POSIXD,         /* POSIXL                 */
 177         POSIXD,         /* POSIXU                 */
 178         POSIXD,         /* POSIXA                 */
 179         NPOSIXD,        /* NPOSIXD                */
 180         NPOSIXD,        /* NPOSIXL                */
 181         NPOSIXD,        /* NPOSIXU                */
 182         NPOSIXD,        /* NPOSIXA                */
 183         CLUMP,          /* CLUMP                  */
 184         BRANCH,         /* BRANCH                 */
 185         EXACT,          /* EXACT                  */
 186         EXACT,          /* EXACTL                 */
 187         EXACT,          /* EXACTF                 */
 188         EXACT,          /* EXACTFL                */
 189         EXACT,          /* EXACTFU                */
 190         EXACT,          /* EXACTFA                */
 191         EXACT,          /* EXACTFU_SS             */
 192         EXACT,          /* EXACTFLU8              */
 193         EXACT,          /* EXACTFA_NO_TRIE        */
 194         NOTHING,        /* NOTHING                */
 195         NOTHING,        /* TAIL                   */
 196         STAR,           /* STAR                   */
 197         PLUS,           /* PLUS                   */
 198         CURLY,          /* CURLY                  */
 199         CURLY,          /* CURLYN                 */
 200         CURLY,          /* CURLYM                 */
 201         CURLY,          /* CURLYX                 */
 202         WHILEM,         /* WHILEM                 */
 203         OPEN,           /* OPEN                   */
 204         CLOSE,          /* CLOSE                  */
 205         REF,            /* REF                    */
 206         REF,            /* REFF                   */
 207         REF,            /* REFFL                  */
 208         REF,            /* REFFU                  */
 209         REF,            /* REFFA                  */
 210         REF,            /* NREF                   */
 211         REF,            /* NREFF                  */
 212         REF,            /* NREFFL                 */
 213         REF,            /* NREFFU                 */
 214         REF,            /* NREFFA                 */
 215         LONGJMP,        /* LONGJMP                */
 216         BRANCHJ,        /* BRANCHJ                */
 217         BRANCHJ,        /* IFMATCH                */
 218         BRANCHJ,        /* UNLESSM                */
 219         BRANCHJ,        /* SUSPEND                */
 220         BRANCHJ,        /* IFTHEN                 */
 221         GROUPP,         /* GROUPP                 */
 222         EVAL,           /* EVAL                   */
 223         MINMOD,         /* MINMOD                 */
 224         LOGICAL,        /* LOGICAL                */
 225         BRANCHJ,        /* RENUM                  */
 226         TRIE,           /* TRIE                   */
 227         TRIE,           /* TRIEC                  */
 228         TRIE,           /* AHOCORASICK            */
 229         TRIE,           /* AHOCORASICKC           */
 230         GOSUB,          /* GOSUB                  */
 231         GOSTART,        /* GOSTART                */
 232         NGROUPP,        /* NGROUPP                */
 233         INSUBP,         /* INSUBP                 */
 234         DEFINEP,        /* DEFINEP                */
 235         ENDLIKE,        /* ENDLIKE                */
 236         ENDLIKE,        /* OPFAIL                 */
 237         ENDLIKE,        /* ACCEPT                 */
 238         VERB,           /* VERB                   */
 239         VERB,           /* PRUNE                  */
 240         VERB,           /* MARKPOINT              */
 241         VERB,           /* SKIP                   */
 242         VERB,           /* COMMIT                 */
 243         VERB,           /* CUTGROUP               */
 244         KEEPS,          /* KEEPS                  */
 245         LNBREAK,        /* LNBREAK                */
 246         NOTHING,        /* OPTIMIZED              */
 247         PSEUDO,         /* PSEUDO                 */
 248         /* ------------ States ------------- */
 249         TRIE,           /* TRIE_next              */
 250         TRIE,           /* TRIE_next_fail         */
 251         EVAL,           /* EVAL_AB                */
 252         EVAL,           /* EVAL_AB_fail           */
 253         CURLYX,         /* CURLYX_end             */
 254         CURLYX,         /* CURLYX_end_fail        */
 255         WHILEM,         /* WHILEM_A_pre           */
 256         WHILEM,         /* WHILEM_A_pre_fail      */
 257         WHILEM,         /* WHILEM_A_min           */
 258         WHILEM,         /* WHILEM_A_min_fail      */
 259         WHILEM,         /* WHILEM_A_max           */
 260         WHILEM,         /* WHILEM_A_max_fail      */
 261         WHILEM,         /* WHILEM_B_min           */
 262         WHILEM,         /* WHILEM_B_min_fail      */
 263         WHILEM,         /* WHILEM_B_max           */
 264         WHILEM,         /* WHILEM_B_max_fail      */
 265         BRANCH,         /* BRANCH_next            */
 266         BRANCH,         /* BRANCH_next_fail       */
 267         CURLYM,         /* CURLYM_A               */
 268         CURLYM,         /* CURLYM_A_fail          */
 269         CURLYM,         /* CURLYM_B               */
 270         CURLYM,         /* CURLYM_B_fail          */
 271         IFMATCH,        /* IFMATCH_A              */
 272         IFMATCH,        /* IFMATCH_A_fail         */
 273         CURLY,          /* CURLY_B_min_known      */
 274         CURLY,          /* CURLY_B_min_known_fail */
 275         CURLY,          /* CURLY_B_min            */
 276         CURLY,          /* CURLY_B_min_fail       */
 277         CURLY,          /* CURLY_B_max            */
 278         CURLY,          /* CURLY_B_max_fail       */
 279         COMMIT,         /* COMMIT_next            */
 280         COMMIT,         /* COMMIT_next_fail       */
 281         MARKPOINT,      /* MARKPOINT_next         */
 282         MARKPOINT,      /* MARKPOINT_next_fail    */
 283         SKIP,           /* SKIP_next              */
 284         SKIP,           /* SKIP_next_fail         */
 285         CUTGROUP,       /* CUTGROUP_next          */
 286         CUTGROUP,       /* CUTGROUP_next_fail     */
 287         KEEPS,          /* KEEPS_next             */
 288         KEEPS,          /* KEEPS_next_fail        */
 289 };
 290 #endif
 291
 292 /* regarglen[] - How large is the argument part of the node (in regnodes) */
 293
 294 #ifdef REG_COMP_C
 295 static const U8 regarglen[] = {
 296         0,                                      /* END          */
 297         0,                                      /* SUCCEED      */
 298         0,                                      /* SBOL         */
 299         0,                                      /* MBOL         */
 300         0,                                      /* SEOL         */
 301         0,                                      /* MEOL         */
 302         0,                                      /* EOS          */
 303         0,                                      /* GPOS         */
 304         0,                                      /* BOUND        */
 305         0,                                      /* BOUNDL       */
 306         0,                                      /* BOUNDU       */
 307         0,                                      /* BOUNDA       */
 308         0,                                      /* NBOUND       */
 309         0,                                      /* NBOUNDL      */
 310         0,                                      /* NBOUNDU      */
 311         0,                                      /* NBOUNDA      */
 312         0,                                      /* REG_ANY      */
 313         0,                                      /* SANY         */
 314         EXTRA_SIZE(struct regnode_1),           /* ANYOF        */
 315         EXTRA_SIZE(struct regnode_1),           /* ANYOFL       */
 316         0,                                      /* POSIXD       */
 317         0,                                      /* POSIXL       */
 318         0,                                      /* POSIXU       */
 319         0,                                      /* POSIXA       */
 320         0,                                      /* NPOSIXD      */
 321         0,                                      /* NPOSIXL      */
 322         0,                                      /* NPOSIXU      */
 323         0,                                      /* NPOSIXA      */
 324         0,                                      /* CLUMP        */
 325         0,                                      /* BRANCH       */
 326         0,                                      /* EXACT        */
 327         0,                                      /* EXACTL       */
 328         0,                                      /* EXACTF       */
 329         0,                                      /* EXACTFL      */
 330         0,                                      /* EXACTFU      */
 331         0,                                      /* EXACTFA      */
 332         0,                                      /* EXACTFU_SS   */
 333         0,                                      /* EXACTFLU8    */
 334         0,                                      /* EXACTFA_NO_TRIE */
 335         0,                                      /* NOTHING      */
 336         0,                                      /* TAIL         */
 337         0,                                      /* STAR         */
 338         0,                                      /* PLUS         */
 339         EXTRA_SIZE(struct regnode_2),           /* CURLY        */
 340         EXTRA_SIZE(struct regnode_2),           /* CURLYN       */
 341         EXTRA_SIZE(struct regnode_2),           /* CURLYM       */
 342         EXTRA_SIZE(struct regnode_2),           /* CURLYX       */
 343         0,                                      /* WHILEM       */
 344         EXTRA_SIZE(struct regnode_1),           /* OPEN         */
 345         EXTRA_SIZE(struct regnode_1),           /* CLOSE        */
 346         EXTRA_SIZE(struct regnode_1),           /* REF          */
 347         EXTRA_SIZE(struct regnode_1),           /* REFF         */
 348         EXTRA_SIZE(struct regnode_1),           /* REFFL        */
 349         EXTRA_SIZE(struct regnode_1),           /* REFFU        */
 350         EXTRA_SIZE(struct regnode_1),           /* REFFA        */
 351         EXTRA_SIZE(struct regnode_1),           /* NREF         */
 352         EXTRA_SIZE(struct regnode_1),           /* NREFF        */
 353         EXTRA_SIZE(struct regnode_1),           /* NREFFL       */
 354         EXTRA_SIZE(struct regnode_1),           /* NREFFU       */
 355         EXTRA_SIZE(struct regnode_1),           /* NREFFA       */
 356         EXTRA_SIZE(struct regnode_1),           /* LONGJMP      */
 357         EXTRA_SIZE(struct regnode_1),           /* BRANCHJ      */
 358         EXTRA_SIZE(struct regnode_1),           /* IFMATCH      */
 359         EXTRA_SIZE(struct regnode_1),           /* UNLESSM      */
 360         EXTRA_SIZE(struct regnode_1),           /* SUSPEND      */
 361         EXTRA_SIZE(struct regnode_1),           /* IFTHEN       */
 362         EXTRA_SIZE(struct regnode_1),           /* GROUPP       */
 363         EXTRA_SIZE(struct regnode_2L),          /* EVAL         */
 364         0,                                      /* MINMOD       */
 365         0,                                      /* LOGICAL      */
 366         EXTRA_SIZE(struct regnode_1),           /* RENUM        */
 367         EXTRA_SIZE(struct regnode_1),           /* TRIE         */
 368         EXTRA_SIZE(struct regnode_charclass),   /* TRIEC        */
 369         EXTRA_SIZE(struct regnode_1),           /* AHOCORASICK  */
 370         EXTRA_SIZE(struct regnode_charclass),   /* AHOCORASICKC */
 371         EXTRA_SIZE(struct regnode_2L),          /* GOSUB        */
 372         0,                                      /* GOSTART      */
 373         EXTRA_SIZE(struct regnode_1),           /* NGROUPP      */
 374         EXTRA_SIZE(struct regnode_1),           /* INSUBP       */
 375         EXTRA_SIZE(struct regnode_1),           /* DEFINEP      */
 376         0,                                      /* ENDLIKE      */
 377         0,                                      /* OPFAIL       */
 378         EXTRA_SIZE(struct regnode_1),           /* ACCEPT       */
 379         EXTRA_SIZE(struct regnode_1),           /* VERB         */
 380         EXTRA_SIZE(struct regnode_1),           /* PRUNE        */
 381         EXTRA_SIZE(struct regnode_1),           /* MARKPOINT    */
 382         EXTRA_SIZE(struct regnode_1),           /* SKIP         */
 383         EXTRA_SIZE(struct regnode_1),           /* COMMIT       */
 384         EXTRA_SIZE(struct regnode_1),           /* CUTGROUP     */
 385         0,                                      /* KEEPS        */
 386         0,                                      /* LNBREAK      */
 387         0,                                      /* OPTIMIZED    */
 388         0,                                      /* PSEUDO       */
 389 };
 390
 391 /* reg_off_by_arg[] - Which argument holds the offset to the next node */
 392
 393 static const char reg_off_by_arg[] = {
 394         0,      /* END          */
 395         0,      /* SUCCEED      */
 396         0,      /* SBOL         */
 397         0,      /* MBOL         */
 398         0,      /* SEOL         */
 399         0,      /* MEOL         */
 400         0,      /* EOS          */
 401         0,      /* GPOS         */
 402         0,      /* BOUND        */
 403         0,      /* BOUNDL       */
 404         0,      /* BOUNDU       */
 405         0,      /* BOUNDA       */
 406         0,      /* NBOUND       */
 407         0,      /* NBOUNDL      */
 408         0,      /* NBOUNDU      */
 409         0,      /* NBOUNDA      */
 410         0,      /* REG_ANY      */
 411         0,      /* SANY         */
 412         0,      /* ANYOF        */
 413         0,      /* ANYOFL       */
 414         0,      /* POSIXD       */
 415         0,      /* POSIXL       */
 416         0,      /* POSIXU       */
 417         0,      /* POSIXA       */
 418         0,      /* NPOSIXD      */
 419         0,      /* NPOSIXL      */
 420         0,      /* NPOSIXU      */
 421         0,      /* NPOSIXA      */
 422         0,      /* CLUMP        */
 423         0,      /* BRANCH       */
 424         0,      /* EXACT        */
 425         0,      /* EXACTL       */
 426         0,      /* EXACTF       */
 427         0,      /* EXACTFL      */
 428         0,      /* EXACTFU      */
 429         0,      /* EXACTFA      */
 430         0,      /* EXACTFU_SS   */
 431         0,      /* EXACTFLU8    */
 432         0,      /* EXACTFA_NO_TRIE */
 433         0,      /* NOTHING      */
 434         0,      /* TAIL         */
 435         0,      /* STAR         */
 436         0,      /* PLUS         */
 437         0,      /* CURLY        */
 438         0,      /* CURLYN       */
 439         0,      /* CURLYM       */
 440         0,      /* CURLYX       */
 441         0,      /* WHILEM       */
 442         0,      /* OPEN         */
 443         0,      /* CLOSE        */
 444         0,      /* REF          */
 445         0,      /* REFF         */
 446         0,      /* REFFL        */
 447         0,      /* REFFU        */
 448         0,      /* REFFA        */
 449         0,      /* NREF         */
 450         0,      /* NREFF        */
 451         0,      /* NREFFL       */
 452         0,      /* NREFFU       */
 453         0,      /* NREFFA       */
 454         1,      /* LONGJMP      */
 455         1,      /* BRANCHJ      */
 456         2,      /* IFMATCH      */
 457         2,      /* UNLESSM      */
 458         1,      /* SUSPEND      */
 459         1,      /* IFTHEN       */
 460         0,      /* GROUPP       */
 461         0,      /* EVAL         */
 462         0,      /* MINMOD       */
 463         0,      /* LOGICAL      */
 464         1,      /* RENUM        */
 465         0,      /* TRIE         */
 466         0,      /* TRIEC        */
 467         0,      /* AHOCORASICK  */
 468         0,      /* AHOCORASICKC */
 469         0,      /* GOSUB        */
 470         0,      /* GOSTART      */
 471         0,      /* NGROUPP      */
 472         0,      /* INSUBP       */
 473         0,      /* DEFINEP      */
 474         0,      /* ENDLIKE      */
 475         0,      /* OPFAIL       */
 476         0,      /* ACCEPT       */
 477         0,      /* VERB         */
 478         0,      /* PRUNE        */
 479         0,      /* MARKPOINT    */
 480         0,      /* SKIP         */
 481         0,      /* COMMIT       */
 482         0,      /* CUTGROUP     */
 483         0,      /* KEEPS        */
 484         0,      /* LNBREAK      */
 485         0,      /* OPTIMIZED    */
 486         0,      /* PSEUDO       */
 487 };
 488
 489 #endif /* REG_COMP_C */
 490
 491 /* reg_name[] - Opcode/state names in string form, for debugging */
 492
 493 #ifndef DOINIT
 494 EXTCONST char * PL_reg_name[];
 495 #else
 496 EXTCONST char * const PL_reg_name[] = {
 497         "END",                          /* 0000 */
 498         "SUCCEED",                      /* 0x01 */
 499         "SBOL",                         /* 0x02 */
 500         "MBOL",                         /* 0x03 */
 501         "SEOL",                         /* 0x04 */
 502         "MEOL",                         /* 0x05 */
 503         "EOS",                          /* 0x06 */
 504         "GPOS",                         /* 0x07 */
 505         "BOUND",                        /* 0x08 */
 506         "BOUNDL",                       /* 0x09 */
 507         "BOUNDU",                       /* 0x0a */
 508         "BOUNDA",                       /* 0x0b */
 509         "NBOUND",                       /* 0x0c */
 510         "NBOUNDL",                      /* 0x0d */
 511         "NBOUNDU",                      /* 0x0e */
 512         "NBOUNDA",                      /* 0x0f */
 513         "REG_ANY",                      /* 0x10 */
 514         "SANY",                         /* 0x11 */
 515         "ANYOF",                        /* 0x12 */
 516         "ANYOFL",                       /* 0x13 */
 517         "POSIXD",                       /* 0x14 */
 518         "POSIXL",                       /* 0x15 */
 519         "POSIXU",                       /* 0x16 */
 520         "POSIXA",                       /* 0x17 */
 521         "NPOSIXD",                      /* 0x18 */
 522         "NPOSIXL",                      /* 0x19 */
 523         "NPOSIXU",                      /* 0x1a */
 524         "NPOSIXA",                      /* 0x1b */
 525         "CLUMP",                        /* 0x1c */
 526         "BRANCH",                       /* 0x1d */
 527         "EXACT",                        /* 0x1e */
 528         "EXACTL",                       /* 0x1f */
 529         "EXACTF",                       /* 0x20 */
 530         "EXACTFL",                      /* 0x21 */
 531         "EXACTFU",                      /* 0x22 */
 532         "EXACTFA",                      /* 0x23 */
 533         "EXACTFU_SS",                   /* 0x24 */
 534         "EXACTFLU8",                    /* 0x25 */
 535         "EXACTFA_NO_TRIE",              /* 0x26 */
 536         "NOTHING",                      /* 0x27 */
 537         "TAIL",                         /* 0x28 */
 538         "STAR",                         /* 0x29 */
 539         "PLUS",                         /* 0x2a */
 540         "CURLY",                        /* 0x2b */
 541         "CURLYN",                       /* 0x2c */
 542         "CURLYM",                       /* 0x2d */
 543         "CURLYX",                       /* 0x2e */
 544         "WHILEM",                       /* 0x2f */
 545         "OPEN",                         /* 0x30 */
 546         "CLOSE",                        /* 0x31 */
 547         "REF",                          /* 0x32 */
 548         "REFF",                         /* 0x33 */
 549         "REFFL",                        /* 0x34 */
 550         "REFFU",                        /* 0x35 */
 551         "REFFA",                        /* 0x36 */
 552         "NREF",                         /* 0x37 */
 553         "NREFF",                        /* 0x38 */
 554         "NREFFL",                       /* 0x39 */
 555         "NREFFU",                       /* 0x3a */
 556         "NREFFA",                       /* 0x3b */
 557         "LONGJMP",                      /* 0x3c */
 558         "BRANCHJ",                      /* 0x3d */
 559         "IFMATCH",                      /* 0x3e */
 560         "UNLESSM",                      /* 0x3f */
 561         "SUSPEND",                      /* 0x40 */
 562         "IFTHEN",                       /* 0x41 */
 563         "GROUPP",                       /* 0x42 */
 564         "EVAL",                         /* 0x43 */
 565         "MINMOD",                       /* 0x44 */
 566         "LOGICAL",                      /* 0x45 */
 567         "RENUM",                        /* 0x46 */
 568         "TRIE",                         /* 0x47 */
 569         "TRIEC",                        /* 0x48 */
 570         "AHOCORASICK",                  /* 0x49 */
 571         "AHOCORASICKC",                 /* 0x4a */
 572         "GOSUB",                        /* 0x4b */
 573         "GOSTART",                      /* 0x4c */
 574         "NGROUPP",                      /* 0x4d */
 575         "INSUBP",                       /* 0x4e */
 576         "DEFINEP",                      /* 0x4f */
 577         "ENDLIKE",                      /* 0x50 */
 578         "OPFAIL",                       /* 0x51 */
 579         "ACCEPT",                       /* 0x52 */
 580         "VERB",                         /* 0x53 */
 581         "PRUNE",                        /* 0x54 */
 582         "MARKPOINT",                    /* 0x55 */
 583         "SKIP",                         /* 0x56 */
 584         "COMMIT",                       /* 0x57 */
 585         "CUTGROUP",                     /* 0x58 */
 586         "KEEPS",                        /* 0x59 */
 587         "LNBREAK",                      /* 0x5a */
 588         "OPTIMIZED",                    /* 0x5b */
 589         "PSEUDO",                       /* 0x5c */
 590         /* ------------ States ------------- */
 591         "TRIE_next",                    /* REGNODE_MAX +0x01 */
 592         "TRIE_next_fail",               /* REGNODE_MAX +0x02 */
 593         "EVAL_AB",                      /* REGNODE_MAX +0x03 */
 594         "EVAL_AB_fail",                 /* REGNODE_MAX +0x04 */
 595         "CURLYX_end",                   /* REGNODE_MAX +0x05 */
 596         "CURLYX_end_fail",              /* REGNODE_MAX +0x06 */
 597         "WHILEM_A_pre",                 /* REGNODE_MAX +0x07 */
 598         "WHILEM_A_pre_fail",            /* REGNODE_MAX +0x08 */
 599         "WHILEM_A_min",                 /* REGNODE_MAX +0x09 */
 600         "WHILEM_A_min_fail",            /* REGNODE_MAX +0x0a */
 601         "WHILEM_A_max",                 /* REGNODE_MAX +0x0b */
 602         "WHILEM_A_max_fail",            /* REGNODE_MAX +0x0c */
 603         "WHILEM_B_min",                 /* REGNODE_MAX +0x0d */
 604         "WHILEM_B_min_fail",            /* REGNODE_MAX +0x0e */
 605         "WHILEM_B_max",                 /* REGNODE_MAX +0x0f */
 606         "WHILEM_B_max_fail",            /* REGNODE_MAX +0x10 */
 607         "BRANCH_next",                  /* REGNODE_MAX +0x11 */
 608         "BRANCH_next_fail",             /* REGNODE_MAX +0x12 */
 609         "CURLYM_A",                     /* REGNODE_MAX +0x13 */
 610         "CURLYM_A_fail",                /* REGNODE_MAX +0x14 */
 611         "CURLYM_B",                     /* REGNODE_MAX +0x15 */
 612         "CURLYM_B_fail",                /* REGNODE_MAX +0x16 */
 613         "IFMATCH_A",                    /* REGNODE_MAX +0x17 */
 614         "IFMATCH_A_fail",               /* REGNODE_MAX +0x18 */
 615         "CURLY_B_min_known",            /* REGNODE_MAX +0x19 */
 616         "CURLY_B_min_known_fail",       /* REGNODE_MAX +0x1a */
 617         "CURLY_B_min",                  /* REGNODE_MAX +0x1b */
 618         "CURLY_B_min_fail",             /* REGNODE_MAX +0x1c */
 619         "CURLY_B_max",                  /* REGNODE_MAX +0x1d */
 620         "CURLY_B_max_fail",             /* REGNODE_MAX +0x1e */
 621         "COMMIT_next",                  /* REGNODE_MAX +0x1f */
 622         "COMMIT_next_fail",             /* REGNODE_MAX +0x20 */
 623         "MARKPOINT_next",               /* REGNODE_MAX +0x21 */
 624         "MARKPOINT_next_fail",          /* REGNODE_MAX +0x22 */
 625         "SKIP_next",                    /* REGNODE_MAX +0x23 */
 626         "SKIP_next_fail",               /* REGNODE_MAX +0x24 */
 627         "CUTGROUP_next",                /* REGNODE_MAX +0x25 */
 628         "CUTGROUP_next_fail",           /* REGNODE_MAX +0x26 */
 629         "KEEPS_next",                   /* REGNODE_MAX +0x27 */
 630         "KEEPS_next_fail",              /* REGNODE_MAX +0x28 */
 631 };
 632 #endif /* DOINIT */
 633
 634 /* PL_reg_extflags_name[] - Opcode/state names in string form, for debugging */
 635
 636 #ifndef DOINIT
 637 EXTCONST char * PL_reg_extflags_name[];
 638 #else
 639 EXTCONST char * const PL_reg_extflags_name[] = {
 640         /* Bits in extflags defined: 11111111111111110000111111111111 */
 641         "MULTILINE",        /* 0x00000001 */
 642         "SINGLELINE",       /* 0x00000002 */
 643         "FOLD",             /* 0x00000004 */
 644         "EXTENDED",         /* 0x00000008 */
 645         "EXTENDED_MORE",    /* 0x00000010 */
 646         "NOCAPTURE",        /* 0x00000020 */
 647         "KEEPCOPY",         /* 0x00000040 */
 648         "CHARSET0",         /* 0x00000080 : "CHARSET" - 0x00000380 */
 649         "CHARSET1",         /* 0x00000100 : "CHARSET" - 0x00000380 */
 650         "CHARSET2",         /* 0x00000200 : "CHARSET" - 0x00000380 */
 651         "STRICT",           /* 0x00000400 */
 652         "SPLIT",            /* 0x00000800 */
 653         "UNUSED_BIT_12",    /* 0x00001000 */
 654         "UNUSED_BIT_13",    /* 0x00002000 */
 655         "UNUSED_BIT_14",    /* 0x00004000 */
 656         "UNUSED_BIT_15",    /* 0x00008000 */
 657         "NO_INPLACE_SUBST", /* 0x00010000 */
 658         "EVAL_SEEN",        /* 0x00020000 */
 659         "UNBOUNDED_QUANTIFIER_SEEN",/* 0x00040000 */
 660         "CHECK_ALL",        /* 0x00080000 */
 661         "MATCH_UTF8",       /* 0x00100000 */
 662         "USE_INTUIT_NOML",  /* 0x00200000 */
 663         "USE_INTUIT_ML",    /* 0x00400000 */
 664         "INTUIT_TAIL",      /* 0x00800000 */
 665         "IS_ANCHORED",      /* 0x01000000 */
 666         "COPY_DONE",        /* 0x02000000 */
 667         "TAINTED_SEEN",     /* 0x04000000 */
 668         "TAINTED",          /* 0x08000000 */
 669         "START_ONLY",       /* 0x10000000 */
 670         "SKIPWHITE",        /* 0x20000000 */
 671         "WHITE",            /* 0x40000000 */
 672         "NULL",             /* 0x80000000 */
 673 };
 674 #endif /* DOINIT */
 675
 676 #ifdef DEBUGGING
 677 #  define REG_EXTFLAGS_NAME_SIZE 32
 678 #endif
 679
 680 /* PL_reg_intflags_name[] - Opcode/state names in string form, for debugging */
 681
 682 #ifndef DOINIT
 683 EXTCONST char * PL_reg_intflags_name[];
 684 #else
 685 EXTCONST char * const PL_reg_intflags_name[] = {
 686         "SKIP",                       /* 0x00000001 - PREGf_SKIP */
 687         "IMPLICIT",                   /* 0x00000002 - PREGf_IMPLICIT -  Converted .* to ^.*  */
 688         "NAUGHTY",                    /* 0x00000004 - PREGf_NAUGHTY -  how exponential is this pattern?  */
 689         "VERBARG_SEEN",               /* 0x00000008 - PREGf_VERBARG_SEEN */
 690         "CUTGROUP_SEEN",              /* 0x00000010 - PREGf_CUTGROUP_SEEN */
 691         "USE_RE_EVAL",                /* 0x00000020 - PREGf_USE_RE_EVAL -  compiled with "use re 'eval'"  */
 692         "NOSCAN",                     /* 0x00000040 - PREGf_NOSCAN */
 693         "GPOS_SEEN",                  /* 0x00000100 - PREGf_GPOS_SEEN */
 694         "GPOS_FLOAT",                 /* 0x00000200 - PREGf_GPOS_FLOAT */
 695         "ANCH_MBOL",                  /* 0x00000400 - PREGf_ANCH_MBOL */
 696         "ANCH_SBOL",                  /* 0x00000800 - PREGf_ANCH_SBOL */
 697         "ANCH_GPOS",                  /* 0x00001000 - PREGf_ANCH_GPOS */
 698 };
 699 #endif /* DOINIT */
 700
 701 #ifdef DEBUGGING
 702 #  define REG_INTFLAGS_NAME_SIZE 12
 703 #endif
 704
 705 /* The following have no fixed length. U8 so we can do strchr() on it. */
 706 #define REGNODE_VARIES(node) (PL_varies_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 707
 708 #ifndef DOINIT
 709 EXTCONST U8 PL_varies[] __attribute__deprecated__;
 710 #else
 711 EXTCONST U8 PL_varies[] __attribute__deprecated__ = {
 712     CLUMP, BRANCH, STAR, PLUS, CURLY, CURLYN, CURLYM, CURLYX, WHILEM, REF,
 713     REFF, REFFL, REFFU, REFFA, NREF, NREFF, NREFFL, NREFFU, NREFFA,
 714     BRANCHJ, SUSPEND, IFTHEN,
 715     0
 716 };
 717 #endif /* DOINIT */
 718
 719 #ifndef DOINIT
 720 EXTCONST U8 PL_varies_bitmask[];
 721 #else
 722 EXTCONST U8 PL_varies_bitmask[] = {
 723     0x00, 0x00, 0x00, 0x30, 0x00, 0xFE, 0xFC, 0x2F, 0x03, 0x00, 0x00, 0x00
 724 };
 725 #endif /* DOINIT */
 726
 727 /* The following always have a length of 1. U8 we can do strchr() on it. */
 728 /* (Note that length 1 means "one character" under UTF8, not "one octet".) */
 729 #define REGNODE_SIMPLE(node) (PL_simple_bitmask[(node) >> 3] & (1 << ((node) & 7)))
 730
 731 #ifndef DOINIT
 732 EXTCONST U8 PL_simple[] __attribute__deprecated__;
 733 #else
 734 EXTCONST U8 PL_simple[] __attribute__deprecated__ = {
 735     REG_ANY, SANY, ANYOF, ANYOFL, POSIXD, POSIXL, POSIXU, POSIXA, NPOSIXD,
 736     NPOSIXL, NPOSIXU, NPOSIXA,
 737     0
 738 };
 739 #endif /* DOINIT */
 740
 741 #ifndef DOINIT
 742 EXTCONST U8 PL_simple_bitmask[];
 743 #else
 744 EXTCONST U8 PL_simple_bitmask[] = {
 745     0x00, 0x00, 0xFF, 0x0F, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00, 0x00
 746 };
 747 #endif /* DOINIT */
 748
 749 /* ex: set ro: */