1198090Srdivacky/*- 2198090Srdivacky * This code is derived from OpenBSD's libc/regex, original license follows: 3198090Srdivacky * 4198090Srdivacky * Copyright (c) 1992, 1993, 1994 Henry Spencer. 5198090Srdivacky * Copyright (c) 1992, 1993, 1994 6198090Srdivacky * The Regents of the University of California. All rights reserved. 7198090Srdivacky * 8198090Srdivacky * This code is derived from software contributed to Berkeley by 9198090Srdivacky * Henry Spencer. 10198090Srdivacky * 11198090Srdivacky * Redistribution and use in source and binary forms, with or without 12198090Srdivacky * modification, are permitted provided that the following conditions 13198090Srdivacky * are met: 14198090Srdivacky * 1. Redistributions of source code must retain the above copyright 15198090Srdivacky * notice, this list of conditions and the following disclaimer. 16198090Srdivacky * 2. Redistributions in binary form must reproduce the above copyright 17198090Srdivacky * notice, this list of conditions and the following disclaimer in the 18198090Srdivacky * documentation and/or other materials provided with the distribution. 19198090Srdivacky * 3. Neither the name of the University nor the names of its contributors 20198090Srdivacky * may be used to endorse or promote products derived from this software 21198090Srdivacky * without specific prior written permission. 22198090Srdivacky * 23198090Srdivacky * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND 24198090Srdivacky * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE 25198090Srdivacky * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE 26198090Srdivacky * ARE DISCLAIMED. IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE 27198090Srdivacky * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL 28198090Srdivacky * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS 29198090Srdivacky * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION) 30198090Srdivacky * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT 31198090Srdivacky * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY 32198090Srdivacky * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF 33198090Srdivacky * SUCH DAMAGE. 34198090Srdivacky * 35198090Srdivacky * @(#)regex2.h 8.4 (Berkeley) 3/20/94 36198090Srdivacky */ 37198090Srdivacky 38198090Srdivacky/* 39198090Srdivacky * internals of regex_t 40198090Srdivacky */ 41198090Srdivacky#define MAGIC1 ((('r'^0200)<<8) | 'e') 42198090Srdivacky 43198090Srdivacky/* 44198090Srdivacky * The internal representation is a *strip*, a sequence of 45198090Srdivacky * operators ending with an endmarker. (Some terminology etc. is a 46198090Srdivacky * historical relic of earlier versions which used multiple strips.) 47198090Srdivacky * Certain oddities in the representation are there to permit running 48198090Srdivacky * the machinery backwards; in particular, any deviation from sequential 49198090Srdivacky * flow must be marked at both its source and its destination. Some 50198090Srdivacky * fine points: 51198090Srdivacky * 52198090Srdivacky * - OPLUS_ and O_PLUS are *inside* the loop they create. 53198090Srdivacky * - OQUEST_ and O_QUEST are *outside* the bypass they create. 54198090Srdivacky * - OCH_ and O_CH are *outside* the multi-way branch they create, while 55198090Srdivacky * OOR1 and OOR2 are respectively the end and the beginning of one of 56198090Srdivacky * the branches. Note that there is an implicit OOR2 following OCH_ 57198090Srdivacky * and an implicit OOR1 preceding O_CH. 58198090Srdivacky * 59198090Srdivacky * In state representations, an operator's bit is on to signify a state 60198090Srdivacky * immediately *preceding* "execution" of that operator. 61198090Srdivacky */ 62198090Srdivackytypedef unsigned long sop; /* strip operator */ 63198090Srdivackytypedef long sopno; 64198090Srdivacky#define OPRMASK 0xf8000000LU 65198090Srdivacky#define OPDMASK 0x07ffffffLU 66198090Srdivacky#define OPSHIFT ((unsigned)27) 67198090Srdivacky#define OP(n) ((n)&OPRMASK) 68198090Srdivacky#define OPND(n) ((n)&OPDMASK) 69198090Srdivacky#define SOP(op, opnd) ((op)|(opnd)) 70198090Srdivacky/* operators meaning operand */ 71198090Srdivacky/* (back, fwd are offsets) */ 72198090Srdivacky#define OEND (1LU<<OPSHIFT) /* endmarker - */ 73198090Srdivacky#define OCHAR (2LU<<OPSHIFT) /* character unsigned char */ 74198090Srdivacky#define OBOL (3LU<<OPSHIFT) /* left anchor - */ 75198090Srdivacky#define OEOL (4LU<<OPSHIFT) /* right anchor - */ 76198090Srdivacky#define OANY (5LU<<OPSHIFT) /* . - */ 77198090Srdivacky#define OANYOF (6LU<<OPSHIFT) /* [...] set number */ 78198090Srdivacky#define OBACK_ (7LU<<OPSHIFT) /* begin \d paren number */ 79198090Srdivacky#define O_BACK (8LU<<OPSHIFT) /* end \d paren number */ 80198090Srdivacky#define OPLUS_ (9LU<<OPSHIFT) /* + prefix fwd to suffix */ 81198090Srdivacky#define O_PLUS (10LU<<OPSHIFT) /* + suffix back to prefix */ 82198090Srdivacky#define OQUEST_ (11LU<<OPSHIFT) /* ? prefix fwd to suffix */ 83198090Srdivacky#define O_QUEST (12LU<<OPSHIFT) /* ? suffix back to prefix */ 84198090Srdivacky#define OLPAREN (13LU<<OPSHIFT) /* ( fwd to ) */ 85198090Srdivacky#define ORPAREN (14LU<<OPSHIFT) /* ) back to ( */ 86198090Srdivacky#define OCH_ (15LU<<OPSHIFT) /* begin choice fwd to OOR2 */ 87198090Srdivacky#define OOR1 (16LU<<OPSHIFT) /* | pt. 1 back to OOR1 or OCH_ */ 88198090Srdivacky#define OOR2 (17LU<<OPSHIFT) /* | pt. 2 fwd to OOR2 or O_CH */ 89198090Srdivacky#define O_CH (18LU<<OPSHIFT) /* end choice back to OOR1 */ 90198090Srdivacky#define OBOW (19LU<<OPSHIFT) /* begin word - */ 91198090Srdivacky#define OEOW (20LU<<OPSHIFT) /* end word - */ 92198090Srdivacky 93198090Srdivacky/* 94198090Srdivacky * Structure for [] character-set representation. Character sets are 95198090Srdivacky * done as bit vectors, grouped 8 to a byte vector for compactness. 96198090Srdivacky * The individual set therefore has both a pointer to the byte vector 97198090Srdivacky * and a mask to pick out the relevant bit of each byte. A hash code 98198090Srdivacky * simplifies testing whether two sets could be identical. 99198090Srdivacky * 100198090Srdivacky * This will get trickier for multicharacter collating elements. As 101198090Srdivacky * preliminary hooks for dealing with such things, we also carry along 102198090Srdivacky * a string of multi-character elements, and decide the size of the 103198090Srdivacky * vectors at run time. 104198090Srdivacky */ 105198090Srdivackytypedef struct { 106198090Srdivacky uch *ptr; /* -> uch [csetsize] */ 107198090Srdivacky uch mask; /* bit within array */ 108198090Srdivacky uch hash; /* hash code */ 109198090Srdivacky size_t smultis; 110198090Srdivacky char *multis; /* -> char[smulti] ab\0cd\0ef\0\0 */ 111198090Srdivacky} cset; 112198090Srdivacky/* note that CHadd and CHsub are unsafe, and CHIN doesn't yield 0/1 */ 113198090Srdivacky#define CHadd(cs, c) ((cs)->ptr[(uch)(c)] |= (cs)->mask, (cs)->hash += (c)) 114198090Srdivacky#define CHsub(cs, c) ((cs)->ptr[(uch)(c)] &= ~(cs)->mask, (cs)->hash -= (c)) 115198090Srdivacky#define CHIN(cs, c) ((cs)->ptr[(uch)(c)] & (cs)->mask) 116198090Srdivacky#define MCadd(p, cs, cp) mcadd(p, cs, cp) /* llvm_regcomp() internal fns */ 117198090Srdivacky#define MCsub(p, cs, cp) mcsub(p, cs, cp) 118198090Srdivacky#define MCin(p, cs, cp) mcin(p, cs, cp) 119198090Srdivacky 120198090Srdivacky/* stuff for character categories */ 121198090Srdivackytypedef unsigned char cat_t; 122198090Srdivacky 123198090Srdivacky/* 124198090Srdivacky * main compiled-expression structure 125198090Srdivacky */ 126198090Srdivackystruct re_guts { 127198090Srdivacky int magic; 128198090Srdivacky# define MAGIC2 ((('R'^0200)<<8)|'E') 129198090Srdivacky sop *strip; /* malloced area for strip */ 130198090Srdivacky int csetsize; /* number of bits in a cset vector */ 131198090Srdivacky int ncsets; /* number of csets in use */ 132198090Srdivacky cset *sets; /* -> cset [ncsets] */ 133198090Srdivacky uch *setbits; /* -> uch[csetsize][ncsets/CHAR_BIT] */ 134198090Srdivacky int cflags; /* copy of llvm_regcomp() cflags argument */ 135198090Srdivacky sopno nstates; /* = number of sops */ 136198090Srdivacky sopno firststate; /* the initial OEND (normally 0) */ 137198090Srdivacky sopno laststate; /* the final OEND */ 138198090Srdivacky int iflags; /* internal flags */ 139198090Srdivacky# define USEBOL 01 /* used ^ */ 140198090Srdivacky# define USEEOL 02 /* used $ */ 141198090Srdivacky# define REGEX_BAD 04 /* something wrong */ 142198090Srdivacky int nbol; /* number of ^ used */ 143198090Srdivacky int neol; /* number of $ used */ 144198090Srdivacky int ncategories; /* how many character categories */ 145198090Srdivacky cat_t *categories; /* ->catspace[-CHAR_MIN] */ 146198090Srdivacky char *must; /* match must contain this string */ 147198090Srdivacky int mlen; /* length of must */ 148198090Srdivacky size_t nsub; /* copy of re_nsub */ 149198090Srdivacky int backrefs; /* does it use back references? */ 150198090Srdivacky sopno nplus; /* how deep does it nest +s? */ 151198090Srdivacky /* catspace must be last */ 152198090Srdivacky cat_t catspace[1]; /* actually [NC] */ 153198090Srdivacky}; 154198090Srdivacky 155198090Srdivacky/* misc utilities */ 156198090Srdivacky#define OUT (CHAR_MAX+1) /* a non-character value */ 157198090Srdivacky#define ISWORD(c) (isalnum(c&0xff) || (c) == '_') 158