src/regex/regexec.c

   1 /*
   2  * the outer shell of regexec()
   3  *
   4  * This file includes engine.c *twice*, after muchos fiddling with the
   5  * macros that code uses.  This lets the same code operate on two different
   6  * representations for state sets.
   7  */
   8 #include <sys/types.h>
   9 #include <stdio.h>
  10 #include <stdlib.h>
  11 #include <string.h>
  12 #include <limits.h>
  13 #include <ctype.h>
  14 #include <regex.h>
  15
  16 #include "utils.h"
  17 #include "regex2.h"
  18
  19 /* macros for manipulating states, small version */
  20 #define states  unsigned
  21 #define states1 unsigned        /* for later use in regexec() decision */
  22 #define CLEAR(v)        ((v) = 0)
  23 #define SET0(v, n)      ((v) &= ~((unsigned)1 << (n)))
  24 #define SET1(v, n)      ((v) |= (unsigned)1 << (n))
  25 #define ISSET(v, n)     ((v) & ((unsigned)1 << (n)))
  26 #define ASSIGN(d, s)    ((d) = (s))
  27 #define EQ(a, b)        ((a) == (b))
  28 #define STATEVARS       int dummy       /* dummy version */
  29 #define STATESETUP(m, n)        /* nothing */
  30 #define STATETEARDOWN(m)        /* nothing */
  31 #define SETUP(v)        ((v) = 0)
  32 #define onestate        unsigned
  33 #define INIT(o, n)      ((o) = (unsigned)1 << (n))
  34 #define INC(o)  ((o) <<= 1)
  35 #define ISSTATEIN(v, o) ((v) & (o))
  36 /* some abbreviations; note that some of these know variable names! */
  37 /* do "if I'm here, I can also be there" etc without branches */
  38 #define FWD(dst, src, n)        ((dst) |= ((unsigned)(src)&(here)) << (n))
  39 #define BACK(dst, src, n)       ((dst) |= ((unsigned)(src)&(here)) >> (n))
  40 #define ISSETBACK(v, n) ((v) & ((unsigned)here >> (n)))
  41 /* function names */
  42 #define SNAMES                  /* engine.c looks after details */
  43
  44 #include "engine.c"
  45
  46 /* now undo things */
  47 #undef  states
  48 #undef  CLEAR
  49 #undef  SET0
  50 #undef  SET1
  51 #undef  ISSET
  52 #undef  ASSIGN
  53 #undef  EQ
  54 #undef  STATEVARS
  55 #undef  STATESETUP
  56 #undef  STATETEARDOWN
  57 #undef  SETUP
  58 #undef  onestate
  59 #undef  INIT
  60 #undef  INC
  61 #undef  ISSTATEIN
  62 #undef  FWD
  63 #undef  BACK
  64 #undef  ISSETBACK
  65 #undef  SNAMES
  66
  67 /* macros for manipulating states, large version */
  68 #define states  char *
  69 #define CLEAR(v)        memset(v, 0, m->g->nstates)
  70 #define SET0(v, n)      ((v)[n] = 0)
  71 #define SET1(v, n)      ((v)[n] = 1)
  72 #define ISSET(v, n)     ((v)[n])
  73 #define ASSIGN(d, s)    memcpy(d, s, m->g->nstates)
  74 #define EQ(a, b)        (memcmp(a, b, m->g->nstates) == 0)
  75 #define STATEVARS       int vn; char *space
  76 #define STATESETUP(m, nv)       { (m)->space = malloc((nv)*(m)->g->nstates); \
  77                                 if ((m)->space == NULL) return(REG_ESPACE); \
  78                                 (m)->vn = 0; }
  79 #define STATETEARDOWN(m)        { free((m)->space); }
  80 #define SETUP(v)        ((v) = &m->space[m->vn++ * m->g->nstates])
  81 #define onestate        int
  82 #define INIT(o, n)      ((o) = (n))
  83 #define INC(o)  ((o)++)
  84 #define ISSTATEIN(v, o) ((v)[o])
  85 /* some abbreviations; note that some of these know variable names! */
  86 /* do "if I'm here, I can also be there" etc without branches */
  87 #define FWD(dst, src, n)        ((dst)[here+(n)] |= (src)[here])
  88 #define BACK(dst, src, n)       ((dst)[here-(n)] |= (src)[here])
  89 #define ISSETBACK(v, n) ((v)[here - (n)])
  90 /* function names */
  91 #define LNAMES                  /* flag */
  92
  93 #include "engine.c"
  94
  95 /*
  96  - regexec - interface for matching
  97  = extern int regexec(const regex_t *, const char *, size_t, \
  98  =                                      regmatch_t [], int);
  99  = #define      REG_NOTBOL      00001
 100  = #define      REG_NOTEOL      00002
 101  = #define      REG_STARTEND    00004
 102  = #define      REG_TRACE       00400   // tracing of execution
 103  = #define      REG_LARGE       01000   // force large representation
 104  = #define      REG_BACKR       02000   // force use of backref code
 105  *
 106  * We put this here so we can exploit knowledge of the state representation
 107  * when choosing which matcher to call.  Also, by this point the matchers
 108  * have been prototyped.
 109  */
 110 int                             /* 0 success, REG_NOMATCH failure */
 111 regexec(preg, string, nmatch, pmatch, eflags)
 112 const regex_t *preg;
 113 const char *string;
 114 size_t nmatch;
 115 regmatch_t pmatch[];
 116 int eflags;
 117 {
 118         register struct re_guts *g = preg->re_g;
 119 #ifdef REDEBUG
 120 #       define  GOODFLAGS(f)    (f)
 121 #else
 122 #       define  GOODFLAGS(f)    ((f)&(REG_NOTBOL|REG_NOTEOL|REG_STARTEND))
 123 #endif
 124
 125         if (preg->re_magic != MAGIC1 || g->magic != MAGIC2)
 126                 return(REG_BADPAT);
 127         assert(!(g->iflags&BAD));
 128         if (g->iflags&BAD)              /* backstop for no-debug case */
 129                 return(REG_BADPAT);
 130         eflags = GOODFLAGS(eflags);
 131
 132         if (g->nstates <= CHAR_BIT*sizeof(states1) && !(eflags&REG_LARGE))
 133                 return(smatcher(g, (char *)string, nmatch, pmatch, eflags));
 134         else
 135                 return(lmatcher(g, (char *)string, nmatch, pmatch, eflags));
 136 }