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