regexp.h revision 1
fa9e4066f08beec538e775443c5be79dd423fcabahrens/* regexp.h
fa9e4066f08beec538e775443c5be79dd423fcabahrens *
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * by Larry Wall and others
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock *
ea8dc4b6d2251b437950c0056bc626b311c73c27eschrock * You may distribute under the terms of either the GNU General Public
fa9e4066f08beec538e775443c5be79dd423fcabahrens * License or the Artistic License, as specified in the README file.
fa9e4066f08beec538e775443c5be79dd423fcabahrens *
fa9e4066f08beec538e775443c5be79dd423fcabahrens */
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrens/*
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Definitions etc. for regexp(3) routines.
fa9e4066f08beec538e775443c5be79dd423fcabahrens *
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
fa9e4066f08beec538e775443c5be79dd423fcabahrens * not the System V one.
fa9e4066f08beec538e775443c5be79dd423fcabahrens */
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrensstruct regnode {
fa9e4066f08beec538e775443c5be79dd423fcabahrens U8 flags;
fa9e4066f08beec538e775443c5be79dd423fcabahrens U8 type;
97d9e3a676d96208790c885c4766194423e84b24ck U16 next_off;
fa9e4066f08beec538e775443c5be79dd423fcabahrens};
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrenstypedef struct regnode regnode;
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrensstruct reg_substr_data;
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrensstruct reg_data;
fa9e4066f08beec538e775443c5be79dd423fcabahrens
fa9e4066f08beec538e775443c5be79dd423fcabahrenstypedef struct regexp {
fa9e4066f08beec538e775443c5be79dd423fcabahrens I32 *startp;
fa9e4066f08beec538e775443c5be79dd423fcabahrens I32 *endp;
fa9e4066f08beec538e775443c5be79dd423fcabahrens regnode *regstclass;
fa9e4066f08beec538e775443c5be79dd423fcabahrens struct reg_substr_data *substrs;
fa9e4066f08beec538e775443c5be79dd423fcabahrens char *precomp; /* pre-compilation regular expression */
fa9e4066f08beec538e775443c5be79dd423fcabahrens struct reg_data *data; /* Additional data. */
990b4856d0eaada6f8140335733a1b1771ed2746lling char *subbeg; /* saved or original string
990b4856d0eaada6f8140335733a1b1771ed2746lling so \digit works forever. */
fa9e4066f08beec538e775443c5be79dd423fcabahrens U32 *offsets; /* offset annotations 20001228 MJD */
5aba80db367b061758a29154d304977d00d8e4f4ck I32 sublen; /* Length of string pointed by subbeg */
5aba80db367b061758a29154d304977d00d8e4f4ck I32 refcnt;
fa9e4066f08beec538e775443c5be79dd423fcabahrens I32 minlen; /* mininum possible length of $& */
fa9e4066f08beec538e775443c5be79dd423fcabahrens I32 prelen; /* length of precomp */
fa9e4066f08beec538e775443c5be79dd423fcabahrens U32 nparens; /* number of parentheses */
fa9e4066f08beec538e775443c5be79dd423fcabahrens U32 lastparen; /* last paren matched */
91ebeef555ce7f899b6270a3c2df47b51f7ad59aahrens U32 lastcloseparen; /* last paren matched */
fa9e4066f08beec538e775443c5be79dd423fcabahrens U32 reganch; /* Internal use only +
99653d4ee642c6528e88224f12409a5f23060994eschrock Tainted information used by regexec? */
99653d4ee642c6528e88224f12409a5f23060994eschrock regnode program[1]; /* Unwarranted chumminess with compiler. */
99653d4ee642c6528e88224f12409a5f23060994eschrock} regexp;
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_ANCH (ROPT_ANCH_BOL|ROPT_ANCH_MBOL|ROPT_ANCH_GPOS|ROPT_ANCH_SBOL)
fa9e4066f08beec538e775443c5be79dd423fcabahrens#define ROPT_ANCH_SINGLE (ROPT_ANCH_SBOL|ROPT_ANCH_GPOS)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_ANCH_BOL 0x00001
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_ANCH_MBOL 0x00002
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_ANCH_SBOL 0x00004
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_ANCH_GPOS 0x00008
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_SKIP 0x00010
fa9e4066f08beec538e775443c5be79dd423fcabahrens#define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_NOSCAN 0x00040 /* Check-string always at start. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_GPOS_SEEN 0x00080
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_CHECK_ALL 0x00100
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_LOOKBEHIND_SEEN 0x00200
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_EVAL_SEEN 0x00400
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_CANY_SEEN 0x00800
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock/* 0xf800 of reganch is used by PMf_COMPILETIME */
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_UTF8 0x10000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_NAUGHTY 0x20000 /* how exponential is this pattern? */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_COPY_DONE 0x40000 /* subbeg is a copy of the string */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_TAINTED_SEEN 0x80000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_USE_INTUIT_ML 0x0200000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_AUTORITATIVE_ML 0x0800000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_ONCE_ML 0x2000000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_INTUIT_ONECHAR 0x4000000
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_INTUIT_TAIL 0x8000000
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_USE_INTUIT (RE_USE_INTUIT_NOML|RE_USE_INTUIT_ML)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_AUTORITATIVE (REINT_AUTORITATIVE_NOML|REINT_AUTORITATIVE_ML)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_ONCE (REINT_ONCE_NOML|REINT_ONCE_ML)
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_TAINTED(prog) ((prog)->reganch & ROPT_TAINTED_SEEN)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_TAINTED_on(prog) ((prog)->reganch |= ROPT_TAINTED_SEEN)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_TAINTED_off(prog) ((prog)->reganch &= ~ROPT_TAINTED_SEEN)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_TAINTED_set(prog, t) ((t) \
99653d4ee642c6528e88224f12409a5f23060994eschrock ? RX_MATCH_TAINTED_on(prog) \
99653d4ee642c6528e88224f12409a5f23060994eschrock : RX_MATCH_TAINTED_off(prog))
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_COPIED(prog) ((prog)->reganch & ROPT_COPY_DONE)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_COPIED_on(prog) ((prog)->reganch |= ROPT_COPY_DONE)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_COPIED_off(prog) ((prog)->reganch &= ~ROPT_COPY_DONE)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_COPIED_set(prog,t) ((t) \
99653d4ee642c6528e88224f12409a5f23060994eschrock ? RX_MATCH_COPIED_on(prog) \
99653d4ee642c6528e88224f12409a5f23060994eschrock : RX_MATCH_COPIED_off(prog))
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_UTF8(prog) ((prog)->reganch & ROPT_MATCH_UTF8)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_UTF8_on(prog) ((prog)->reganch |= ROPT_MATCH_UTF8)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_UTF8_off(prog) ((prog)->reganch &= ~ROPT_MATCH_UTF8)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RX_MATCH_UTF8_set(prog, t) ((t) \
99653d4ee642c6528e88224f12409a5f23060994eschrock ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
99653d4ee642c6528e88224f12409a5f23060994eschrock : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_CHECKED 0x02 /* check_substr already checked. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_SCREAM 0x04 /* use scream table. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_ML 0x20 /* $* was set. */
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ReREFCNT_dec(re) CALLREGFREE(aTHX_ re)
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMcf_TAIL_DOLLAR 1
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMcf_TAIL_DOLLARM 2
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMcf_TAIL_Z 4
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMcf_TAIL_z 8
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMcf_TAIL (FBMcf_TAIL_DOLLAR|FBMcf_TAIL_DOLLARM|FBMcf_TAIL_Z|FBMcf_TAIL_z)
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrock#define FBMrf_MULTILINE 1
99653d4ee642c6528e88224f12409a5f23060994eschrock
99653d4ee642c6528e88224f12409a5f23060994eschrockstruct re_scream_pos_data_s;
99653d4ee642c6528e88224f12409a5f23060994eschrock