| 1 | /* regexp.h |
| 2 | * |
| 3 | * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, |
| 4 | * by Larry Wall and others |
| 5 | * |
| 6 | * You may distribute under the terms of either the GNU General Public |
| 7 | * License or the Artistic License, as specified in the README file. |
| 8 | * |
| 9 | */ |
| 10 | |
| 11 | /* |
| 12 | * Definitions etc. for regexp(3) routines. |
| 13 | * |
| 14 | * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], |
| 15 | * not the System V one. |
| 16 | */ |
| 17 | |
| 18 | |
| 19 | struct regnode { |
| 20 | U8 flags; |
| 21 | U8 type; |
| 22 | U16 next_off; |
| 23 | }; |
| 24 | |
| 25 | typedef struct regnode regnode; |
| 26 | |
| 27 | struct reg_substr_data; |
| 28 | |
| 29 | struct reg_data; |
| 30 | |
| 31 | typedef struct regexp { |
| 32 | I32 *startp; |
| 33 | I32 *endp; |
| 34 | regnode *regstclass; |
| 35 | struct reg_substr_data *substrs; |
| 36 | char *precomp; /* pre-compilation regular expression */ |
| 37 | struct reg_data *data; /* Additional data. */ |
| 38 | char *subbeg; /* saved or original string |
| 39 | so \digit works forever. */ |
| 40 | U32 *offsets; /* offset annotations 20001228 MJD */ |
| 41 | I32 sublen; /* Length of string pointed by subbeg */ |
| 42 | I32 refcnt; |
| 43 | I32 minlen; /* mininum possible length of $& */ |
| 44 | I32 prelen; /* length of precomp */ |
| 45 | U32 nparens; /* number of parentheses */ |
| 46 | U32 lastparen; /* last paren matched */ |
| 47 | U32 lastcloseparen; /* last paren matched */ |
| 48 | U32 reganch; /* Internal use only + |
| 49 | Tainted information used by regexec? */ |
| 50 | regnode program[1]; /* Unwarranted chumminess with compiler. */ |
| 51 | } regexp; |
| 52 | |
| 53 | #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) |
| 54 | #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) |
| 55 | #define ROPT_ANCH_BOL 0x00001 |
| 56 | #define ROPT_ANCH_MBOL 0x00002 |
| 57 | #define ROPT_ANCH_SBOL 0x00004 |
| 58 | #define ROPT_ANCH_GPOS 0x00008 |
| 59 | #define ROPT_SKIP 0x00010 |
| 60 | #define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */ |
| 61 | #define ROPT_NOSCAN 0x00040 /* Check-string always at start. */ |
| 62 | #define ROPT_GPOS_SEEN 0x00080 |
| 63 | #define ROPT_CHECK_ALL 0x00100 |
| 64 | #define ROPT_LOOKBEHIND_SEEN 0x00200 |
| 65 | #define ROPT_EVAL_SEEN 0x00400 |
| 66 | #define ROPT_CANY_SEEN 0x00800 |
| 67 | #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ |
| 68 | |
| 69 | /* 0xf800 of reganch is used by PMf_COMPILETIME */ |
| 70 | |
| 71 | #define ROPT_UTF8 0x10000 |
| 72 | #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ |
| 73 | #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ |
| 74 | #define ROPT_TAINTED_SEEN 0x80000 |
| 75 | #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ |
| 76 | |
| 77 | #define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */ |
| 78 | #define RE_USE_INTUIT_ML 0x0200000 |
| 79 | #define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */ |
| 80 | #define REINT_AUTORITATIVE_ML 0x0800000 |
| 81 | #define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */ |
| 82 | #define REINT_ONCE_ML 0x2000000 |
| 83 | #define RE_INTUIT_ONECHAR 0x4000000 |
| 84 | #define RE_INTUIT_TAIL 0x8000000 |
| 85 | |
| 86 | #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) |
| 87 | #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) |
| 88 | #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) |
| 89 | |
| 90 | #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) |
| 91 | #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) |
| 92 | #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) |
| 93 | #define RX_MATCH_TAINTED_set(prog, t) ((t) \ |
| 94 | ? RX_MATCH_TAINTED_on(prog) \ |
| 95 | : RX_MATCH_TAINTED_off(prog)) |
| 96 | |
| 97 | #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) |
| 98 | #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) |
| 99 | #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) |
| 100 | #define RX_MATCH_COPIED_set(prog,t) ((t) \ |
| 101 | ? RX_MATCH_COPIED_on(prog) \ |
| 102 | : RX_MATCH_COPIED_off(prog)) |
| 103 | |
| 104 | #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) |
| 105 | #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) |
| 106 | #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) |
| 107 | #define RX_MATCH_UTF8_set(prog, t) ((t) \ |
| 108 | ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ |
| 109 | : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) |
| 110 | |
| 111 | #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ |
| 112 | #define REXEC_CHECKED 0x02 /* check_substr already checked. */ |
| 113 | #define REXEC_SCREAM 0x04 /* use scream table. */ |
| 114 | #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ |
| 115 | #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ |
| 116 | #define REXEC_ML 0x20 /* $* was set. */ |
| 117 | |
| 118 | #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) |
| 119 | #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) |
| 120 | |
| 121 | #define FBMcf_TAIL_DOLLAR 1 |
| 122 | #define FBMcf_TAIL_DOLLARM 2 |
| 123 | #define FBMcf_TAIL_Z 4 |
| 124 | #define FBMcf_TAIL_z 8 |
| 125 | #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) |
| 126 | |
| 127 | #define FBMrf_MULTILINE 1 |
| 128 | |
| 129 | struct re_scream_pos_data_s; |