regexp.h revision 1
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Copyright (C) 1993, 1994, 1996, 1997, 1999, 2000, 2001, 2003,
fa9e4066f08beec538e775443c5be79dd423fcabahrens * by Larry Wall and others
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 * Definitions etc. for regexp(3) routines.
fa9e4066f08beec538e775443c5be79dd423fcabahrens * Caveat: this is V8 regexp(3) [actually, a reimplementation thereof],
fa9e4066f08beec538e775443c5be79dd423fcabahrens * not the System V one.
fa9e4066f08beec538e775443c5be79dd423fcabahrenstypedef struct regexp {
fa9e4066f08beec538e775443c5be79dd423fcabahrens char *precomp; /* pre-compilation regular expression */
990b4856d0eaada6f8140335733a1b1771ed2746lling so \digit works forever. */
99653d4ee642c6528e88224f12409a5f23060994eschrock Tainted information used by regexec? */
99653d4ee642c6528e88224f12409a5f23060994eschrock regnode program[1]; /* Unwarranted chumminess with compiler. */
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)
fa9e4066f08beec538e775443c5be79dd423fcabahrens#define ROPT_IMPLICIT 0x00020 /* Converted .* to ^.* */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_NOSCAN 0x00040 /* Check-string always at start. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ROPT_SANY_SEEN ROPT_CANY_SEEN /* src bckwrd cmpt */
99653d4ee642c6528e88224f12409a5f23060994eschrock/* 0xf800 of reganch is used by PMf_COMPILETIME */
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_MATCH_UTF8 0x10000000 /* subbeg is utf-8 */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define RE_USE_INTUIT_NOML 0x0100000 /* Best to intuit before matching */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_AUTORITATIVE_NOML 0x0400000 /* Can trust a positive answer */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REINT_ONCE_NOML 0x1000000 /* Intuit can succed once only. */
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 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_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_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 ? (RX_MATCH_UTF8_on(prog), (PL_reg_match_utf8 = 1)) \
99653d4ee642c6528e88224f12409a5f23060994eschrock : (RX_MATCH_UTF8_off(prog), (PL_reg_match_utf8 = 0)))
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_COPY_STR 0x01 /* Need to copy the string. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_CHECKED 0x02 /* check_substr already checked. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_IGNOREPOS 0x08 /* \G matches at start. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define REXEC_NOT_FIRST 0x10 /* This is another iteration of //g. */
99653d4ee642c6528e88224f12409a5f23060994eschrock#define ReREFCNT_inc(re) ((void)(re && re->refcnt++), re)