1*0Sstevel@tonic-gate /* regexp.h 2*0Sstevel@tonic-gate * 3*0Sstevel@tonic-gate * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003, 4*0Sstevel@tonic-gate * by Larry Wall and others 5*0Sstevel@tonic-gate * 6*0Sstevel@tonic-gate * You may distribute under the terms of either the GNU General Public 7*0Sstevel@tonic-gate * License or the Artistic License, as specified in the README file. 8*0Sstevel@tonic-gate * 9*0Sstevel@tonic-gate */ 10*0Sstevel@tonic-gate 11*0Sstevel@tonic-gate /* 12*0Sstevel@tonic-gate * Definitions etc. for regexp(3) routines. 13*0Sstevel@tonic-gate * 14*0Sstevel@tonic-gate * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof], 15*0Sstevel@tonic-gate * not the System V one. 16*0Sstevel@tonic-gate */ 17*0Sstevel@tonic-gate 18*0Sstevel@tonic-gate 19*0Sstevel@tonic-gate struct regnode { 20*0Sstevel@tonic-gate U8 flags; 21*0Sstevel@tonic-gate U8 type; 22*0Sstevel@tonic-gate U16 next_off; 23*0Sstevel@tonic-gate }; 24*0Sstevel@tonic-gate 25*0Sstevel@tonic-gate typedef struct regnode regnode; 26*0Sstevel@tonic-gate 27*0Sstevel@tonic-gate struct reg_substr_data; 28*0Sstevel@tonic-gate 29*0Sstevel@tonic-gate struct reg_data; 30*0Sstevel@tonic-gate 31*0Sstevel@tonic-gate typedef struct regexp { 32*0Sstevel@tonic-gate I32 *startp; 33*0Sstevel@tonic-gate I32 *endp; 34*0Sstevel@tonic-gate regnode *regstclass; 35*0Sstevel@tonic-gate struct reg_substr_data *substrs; 36*0Sstevel@tonic-gate char *precomp; /* pre-compilation regular expression */ 37*0Sstevel@tonic-gate struct reg_data *data; /* Additional data. */ 38*0Sstevel@tonic-gate char *subbeg; /* saved or original string 39*0Sstevel@tonic-gate so \digit works forever. */ 40*0Sstevel@tonic-gate U32 *offsets; /* offset annotations 20001228 MJD */ 41*0Sstevel@tonic-gate I32 sublen; /* Length of string pointed by subbeg */ 42*0Sstevel@tonic-gate I32 refcnt; 43*0Sstevel@tonic-gate I32 minlen; /* mininum possible length of $& */ 44*0Sstevel@tonic-gate I32 prelen; /* length of precomp */ 45*0Sstevel@tonic-gate U32 nparens; /* number of parentheses */ 46*0Sstevel@tonic-gate U32 lastparen; /* last paren matched */ 47*0Sstevel@tonic-gate U32 lastcloseparen; /* last paren matched */ 48*0Sstevel@tonic-gate U32 reganch; /* Internal use only + 49*0Sstevel@tonic-gate Tainted information used by regexec? */ 50*0Sstevel@tonic-gate regnode program[1]; /* Unwarranted chumminess with compiler. */ 51*0Sstevel@tonic-gate } regexp; 52*0Sstevel@tonic-gate 53*0Sstevel@tonic-gate #define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL) 54*0Sstevel@tonic-gate #define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS) 55*0Sstevel@tonic-gate #define ROPT_ANCH_BOL 0x00001 56*0Sstevel@tonic-gate #define ROPT_ANCH_MBOL 0x00002 57*0Sstevel@tonic-gate #define ROPT_ANCH_SBOL 0x00004 58*0Sstevel@tonic-gate #define ROPT_ANCH_GPOS 0x00008 59*0Sstevel@tonic-gate #define ROPT_SKIP 0x00010 60*0Sstevel@tonic-gate #define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */ 61*0Sstevel@tonic-gate #define ROPT_NOSCAN 0x00040 /* Check-string always at start. */ 62*0Sstevel@tonic-gate #define ROPT_GPOS_SEEN 0x00080 63*0Sstevel@tonic-gate #define ROPT_CHECK_ALL 0x00100 64*0Sstevel@tonic-gate #define ROPT_LOOKBEHIND_SEEN 0x00200 65*0Sstevel@tonic-gate #define ROPT_EVAL_SEEN 0x00400 66*0Sstevel@tonic-gate #define ROPT_CANY_SEEN 0x00800 67*0Sstevel@tonic-gate #define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */ 68*0Sstevel@tonic-gate 69*0Sstevel@tonic-gate /* 0xf800 of reganch is used by PMf_COMPILETIME */ 70*0Sstevel@tonic-gate 71*0Sstevel@tonic-gate #define ROPT_UTF8 0x10000 72*0Sstevel@tonic-gate #define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */ 73*0Sstevel@tonic-gate #define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */ 74*0Sstevel@tonic-gate #define ROPT_TAINTED_SEEN 0x80000 75*0Sstevel@tonic-gate #define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */ 76*0Sstevel@tonic-gate 77*0Sstevel@tonic-gate #define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */ 78*0Sstevel@tonic-gate #define RE_USE_INTUIT_ML 0x0200000 79*0Sstevel@tonic-gate #define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */ 80*0Sstevel@tonic-gate #define REINT_AUTORITATIVE_ML 0x0800000 81*0Sstevel@tonic-gate #define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */ 82*0Sstevel@tonic-gate #define REINT_ONCE_ML 0x2000000 83*0Sstevel@tonic-gate #define RE_INTUIT_ONECHAR 0x4000000 84*0Sstevel@tonic-gate #define RE_INTUIT_TAIL 0x8000000 85*0Sstevel@tonic-gate 86*0Sstevel@tonic-gate #define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML) 87*0Sstevel@tonic-gate #define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML) 88*0Sstevel@tonic-gate #define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML) 89*0Sstevel@tonic-gate 90*0Sstevel@tonic-gate #define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN) 91*0Sstevel@tonic-gate #define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN) 92*0Sstevel@tonic-gate #define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN) 93*0Sstevel@tonic-gate #define RX_MATCH_TAINTED_set(prog, t) ((t) \ 94*0Sstevel@tonic-gate ? RX_MATCH_TAINTED_on(prog) \ 95*0Sstevel@tonic-gate : RX_MATCH_TAINTED_off(prog)) 96*0Sstevel@tonic-gate 97*0Sstevel@tonic-gate #define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE) 98*0Sstevel@tonic-gate #define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE) 99*0Sstevel@tonic-gate #define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE) 100*0Sstevel@tonic-gate #define RX_MATCH_COPIED_set(prog,t) ((t) \ 101*0Sstevel@tonic-gate ? RX_MATCH_COPIED_on(prog) \ 102*0Sstevel@tonic-gate : RX_MATCH_COPIED_off(prog)) 103*0Sstevel@tonic-gate 104*0Sstevel@tonic-gate #define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8) 105*0Sstevel@tonic-gate #define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8) 106*0Sstevel@tonic-gate #define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8) 107*0Sstevel@tonic-gate #define RX_MATCH_UTF8_set(prog, t) ((t) \ 108*0Sstevel@tonic-gate ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \ 109*0Sstevel@tonic-gate : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0))) 110*0Sstevel@tonic-gate 111*0Sstevel@tonic-gate #define REXEC_COPY_STR 0x01 /* Need to copy the string. */ 112*0Sstevel@tonic-gate #define REXEC_CHECKED 0x02 /* check_substr already checked. */ 113*0Sstevel@tonic-gate #define REXEC_SCREAM 0x04 /* use scream table. */ 114*0Sstevel@tonic-gate #define REXEC_IGNOREPOS 0x08 /* \G matches at start. */ 115*0Sstevel@tonic-gate #define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */ 116*0Sstevel@tonic-gate #define REXEC_ML 0x20 /* $* was set. */ 117*0Sstevel@tonic-gate 118*0Sstevel@tonic-gate #define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re) 119*0Sstevel@tonic-gate #define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re) 120*0Sstevel@tonic-gate 121*0Sstevel@tonic-gate #define FBMcf_TAIL_DOLLAR 1 122*0Sstevel@tonic-gate #define FBMcf_TAIL_DOLLARM 2 123*0Sstevel@tonic-gate #define FBMcf_TAIL_Z 4 124*0Sstevel@tonic-gate #define FBMcf_TAIL_z 8 125*0Sstevel@tonic-gate #define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z) 126*0Sstevel@tonic-gate 127*0Sstevel@tonic-gate #define FBMrf_MULTILINE 1 128*0Sstevel@tonic-gate 129*0Sstevel@tonic-gate struct re_scream_pos_data_s; 130