1 /*- 2 * Copyright (c) 1992, 1993, 1994 Henry Spencer. 3 * Copyright (c) 1992, 1993, 1994 4 * The Regents of the University of California. All rights reserved. 5 * 6 * This code is derived from software contributed to Berkeley by 7 * Henry Spencer of the University of Toronto. 8 * 9 * %sccs.include.redist.c% 10 * 11 * @(#)regexec.c 8.2 (Berkeley) 03/16/94 12 */ 13 14 #if defined(LIBC_SCCS) && !defined(lint) 15 static char sccsid[] = "@(#)regexec.c 8.2 (Berkeley) 03/16/94"; 16 #endif /* LIBC_SCCS and not lint */ 17 18 /* 19 * the outer shell of regexec() 20 * 21 * This file includes engine.c *twice*, after muchos fiddling with the 22 * macros that code uses. This lets the same code operate on two different 23 * representations for state sets. 24 */ 25 #include <sys/types.h> 26 #include <stdio.h> 27 #include <stdlib.h> 28 #include <string.h> 29 #include <limits.h> 30 #include <ctype.h> 31 #include <regex.h> 32 33 #include "utils.h" 34 #include "regex2.h" 35 36 static int nope = 0; /* for use in asserts; shuts lint up */ 37 38 /* macros for manipulating states, small version */ 39 #define states long 40 #define states1 states /* for later use in regexec() decision */ 41 #define CLEAR(v) ((v) = 0) 42 #define SET0(v, n) ((v) &= ~(1 << (n))) 43 #define SET1(v, n) ((v) |= 1 << (n)) 44 #define ISSET(v, n) ((v) & (1 << (n))) 45 #define ASSIGN(d, s) ((d) = (s)) 46 #define EQ(a, b) ((a) == (b)) 47 #define STATEVARS int dummy /* dummy version */ 48 #define STATESETUP(m, n) /* nothing */ 49 #define STATETEARDOWN(m) /* nothing */ 50 #define SETUP(v) ((v) = 0) 51 #define onestate int 52 #define INIT(o, n) ((o) = (unsigned)1 << (n)) 53 #define INC(o) ((o) <<= 1) 54 #define ISSTATEIN(v, o) ((v) & (o)) 55 /* some abbreviations; note that some of these know variable names! */ 56 /* do "if I'm here, I can also be there" etc without branches */ 57 #define FWD(dst, src, n) ((dst) |= ((unsigned)(src)&(here)) << (n)) 58 #define BACK(dst, src, n) ((dst) |= ((unsigned)(src)&(here)) >> (n)) 59 #define ISSETBACK(v, n) ((v) & ((unsigned)here >> (n))) 60 /* function names */ 61 #define SNAMES /* engine.c looks after details */ 62 63 #include "engine.c" 64 65 /* now undo things */ 66 #undef states 67 #undef CLEAR 68 #undef SET0 69 #undef SET1 70 #undef ISSET 71 #undef ASSIGN 72 #undef EQ 73 #undef STATEVARS 74 #undef STATESETUP 75 #undef STATETEARDOWN 76 #undef SETUP 77 #undef onestate 78 #undef INIT 79 #undef INC 80 #undef ISSTATEIN 81 #undef FWD 82 #undef BACK 83 #undef ISSETBACK 84 #undef SNAMES 85 86 /* macros for manipulating states, large version */ 87 #define states char * 88 #define CLEAR(v) memset(v, 0, m->g->nstates) 89 #define SET0(v, n) ((v)[n] = 0) 90 #define SET1(v, n) ((v)[n] = 1) 91 #define ISSET(v, n) ((v)[n]) 92 #define ASSIGN(d, s) memcpy(d, s, m->g->nstates) 93 #define EQ(a, b) (memcmp(a, b, m->g->nstates) == 0) 94 #define STATEVARS int vn; char *space 95 #define STATESETUP(m, nv) { (m)->space = malloc((nv)*(m)->g->nstates); \ 96 if ((m)->space == NULL) return(REG_ESPACE); \ 97 (m)->vn = 0; } 98 #define STATETEARDOWN(m) { free((m)->space); } 99 #define SETUP(v) ((v) = &m->space[m->vn++ * m->g->nstates]) 100 #define onestate int 101 #define INIT(o, n) ((o) = (n)) 102 #define INC(o) ((o)++) 103 #define ISSTATEIN(v, o) ((v)[o]) 104 /* some abbreviations; note that some of these know variable names! */ 105 /* do "if I'm here, I can also be there" etc without branches */ 106 #define FWD(dst, src, n) ((dst)[here+(n)] |= (src)[here]) 107 #define BACK(dst, src, n) ((dst)[here-(n)] |= (src)[here]) 108 #define ISSETBACK(v, n) ((v)[here - (n)]) 109 /* function names */ 110 #define LNAMES /* flag */ 111 112 #include "engine.c" 113 114 /* 115 - regexec - interface for matching 116 = extern int regexec(const regex_t *, const char *, size_t, \ 117 = regmatch_t [], int); 118 = #define REG_NOTBOL 00001 119 = #define REG_NOTEOL 00002 120 = #define REG_STARTEND 00004 121 = #define REG_TRACE 00400 // tracing of execution 122 = #define REG_LARGE 01000 // force large representation 123 = #define REG_BACKR 02000 // force use of backref code 124 * 125 * We put this here so we can exploit knowledge of the state representation 126 * when choosing which matcher to call. Also, by this point the matchers 127 * have been prototyped. 128 */ 129 int /* 0 success, REG_NOMATCH failure */ 130 regexec(preg, string, nmatch, pmatch, eflags) 131 const regex_t *preg; 132 const char *string; 133 size_t nmatch; 134 regmatch_t pmatch[]; 135 int eflags; 136 { 137 register struct re_guts *g = preg->re_g; 138 #ifdef REDEBUG 139 # define GOODFLAGS(f) (f) 140 #else 141 # define GOODFLAGS(f) ((f)&(REG_NOTBOL|REG_NOTEOL|REG_STARTEND)) 142 #endif 143 144 if (preg->re_magic != MAGIC1 || g->magic != MAGIC2) 145 return(REG_BADPAT); 146 assert(!(g->iflags&BAD)); 147 if (g->iflags&BAD) /* backstop for no-debug case */ 148 return(REG_BADPAT); 149 eflags = GOODFLAGS(eflags); 150 151 if (g->nstates <= CHAR_BIT*sizeof(states1) && !(eflags®_LARGE)) 152 return(smatcher(g, (char *)string, nmatch, pmatch, eflags)); 153 else 154 return(lmatcher(g, (char *)string, nmatch, pmatch, eflags)); 155 } 156