regexec.c

硬盘各项性能的测试,如温度容量版本健康度型号

	{
	  assert (pmatch[0].rm_so == start);
	  rval = pmatch[0].rm_eo - start;
	}
      else
	rval = pmatch[0].rm_so;
    }
  re_free (pmatch);
  return rval;
}

static unsigned
re_copy_regs (regs, pmatch, nregs, regs_allocated)
    struct re_registers *regs;
    regmatch_t *pmatch;
    int nregs, regs_allocated;
{
  int rval = REGS_REALLOCATE;
  int i;
  int need_regs = nregs + 1;
  /* We need one extra element beyond `num_regs' for the `-1' marker GNU code
     uses.  */

  /* Have the register data arrays been allocated?  */
  if (regs_allocated == REGS_UNALLOCATED)
    { /* No.  So allocate them with malloc.  */
      regs->start = re_malloc (regoff_t, need_regs);
      if (BE (regs->start == NULL, 0))
	return REGS_UNALLOCATED;
      regs->end = re_malloc (regoff_t, need_regs);
      if (BE (regs->end == NULL, 0))
	{
	  re_free (regs->start);
	  return REGS_UNALLOCATED;
	}
      regs->num_regs = need_regs;
    }
  else if (regs_allocated == REGS_REALLOCATE)
    { /* Yes.  If we need more elements than were already
	 allocated, reallocate them.  If we need fewer, just
	 leave it alone.  */
      if (need_regs > regs->num_regs)
	{
	  regs->start = re_realloc (regs->start, regoff_t, need_regs);
	  if (BE (regs->start == NULL, 0))
	    {
	      if (regs->end != NULL)
		re_free (regs->end);
	      return REGS_UNALLOCATED;
	    }
	  regs->end = re_realloc (regs->end, regoff_t, need_regs);
	  if (BE (regs->end == NULL, 0))
	    {
	      re_free (regs->start);
	      return REGS_UNALLOCATED;
	    }
	  regs->num_regs = need_regs;
	}
    }
  else
    {
      assert (regs_allocated == REGS_FIXED);
      /* This function may not be called with REGS_FIXED and nregs too big.  */
      assert (regs->num_regs >= nregs);
      rval = REGS_FIXED;
    }

  /* Copy the regs.  */
  for (i = 0; i < nregs; ++i)
    {
      regs->start[i] = pmatch[i].rm_so;
      regs->end[i] = pmatch[i].rm_eo;
    }
  for ( ; i < regs->num_regs; ++i)
    regs->start[i] = regs->end[i] = -1;

  return rval;
}

/* Set REGS to hold NUM_REGS registers, storing them in STARTS and
   ENDS.  Subsequent matches using PATTERN_BUFFER and REGS will use
   this memory for recording register information.  STARTS and ENDS
   must be allocated using the malloc library routine, and must each
   be at least NUM_REGS * sizeof (regoff_t) bytes long.

   If NUM_REGS == 0, then subsequent matches should allocate their own
   register data.

   Unless this function is called, the first search or match using
   PATTERN_BUFFER will allocate its own register data, without
   freeing the old data.  */

void
re_set_registers (bufp, regs, num_regs, starts, ends)
    struct re_pattern_buffer *bufp;
    struct re_registers *regs;
    unsigned num_regs;
    regoff_t *starts, *ends;
{
  if (num_regs)
    {
      bufp->regs_allocated = REGS_REALLOCATE;
      regs->num_regs = num_regs;
      regs->start = starts;
      regs->end = ends;
    }
  else
    {
      bufp->regs_allocated = REGS_UNALLOCATED;
      regs->num_regs = 0;
      regs->start = regs->end = (regoff_t *) 0;
    }
}
#ifdef _LIBC
weak_alias (__re_set_registers, re_set_registers)
#endif

/* Entry points compatible with 4.2 BSD regex library.  We don't define
   them unless specifically requested.  */

#if defined _REGEX_RE_COMP || defined _LIBC
int
# ifdef _LIBC
weak_function
# endif
re_exec (s)
     const char *s;
{
  return 0 == regexec (&re_comp_buf, s, 0, NULL, 0);
}
#endif /* _REGEX_RE_COMP */

static re_node_set empty_set;

/* Internal entry point.  */

/* Searches for a compiled pattern PREG in the string STRING, whose
   length is LENGTH.  NMATCH, PMATCH, and EFLAGS have the same
   mingings with regexec.  START, and RANGE have the same meanings
   with re_search.
   Return REG_NOERROR if we find a match, and REG_NOMATCH if not,
   otherwise return the error code.
   Note: We assume front end functions already check ranges.
   (START + RANGE >= 0 && START + RANGE <= LENGTH)  */

static reg_errcode_t
re_search_internal (preg, string, length, start, range, stop, nmatch, pmatch,
		    eflags)
    const regex_t *preg;
    const char *string;
    int length, start, range, stop, eflags;
    size_t nmatch;
    regmatch_t pmatch[];
{
  reg_errcode_t err;
  re_dfa_t *dfa = (re_dfa_t *)preg->buffer;
  re_string_t input;
  int left_lim, right_lim, incr;
  int fl_longest_match, match_first, match_last = -1;
  int fast_translate, sb;
  re_match_context_t mctx;
  char *fastmap = ((preg->fastmap != NULL && preg->fastmap_accurate
		    && range && !preg->can_be_null) ? preg->fastmap : NULL);

  /* Check if the DFA haven't been compiled.  */
  if (BE (preg->used == 0 || dfa->init_state == NULL
	  || dfa->init_state_word == NULL || dfa->init_state_nl == NULL
	  || dfa->init_state_begbuf == NULL, 0))
    return REG_NOMATCH;

  re_node_set_init_empty (&empty_set);
  memset (&mctx, '\0', sizeof (re_match_context_t));

  /* We must check the longest matching, if nmatch > 0.  */
  fl_longest_match = (nmatch != 0 || dfa->nbackref);

  err = re_string_allocate (&input, string, length, dfa->nodes_len + 1,
			    preg->translate, preg->syntax & RE_ICASE);
  if (BE (err != REG_NOERROR, 0))
    goto free_return;
  input.stop = stop;

  err = match_ctx_init (&mctx, eflags, &input, dfa->nbackref * 2);
  if (BE (err != REG_NOERROR, 0))
    goto free_return;

  /* We will log all the DFA states through which the dfa pass,
     if nmatch > 1, or this dfa has "multibyte node", which is a
     back-reference or a node which can accept multibyte character or
     multi character collating element.  */
  if (nmatch > 1 || dfa->has_mb_node)
    {
      mctx.state_log = re_malloc (re_dfastate_t *, dfa->nodes_len + 1);
      if (BE (mctx.state_log == NULL, 0))
	{
	  err = REG_ESPACE;
	  goto free_return;
	}
    }
  else
    mctx.state_log = NULL;

#ifdef DEBUG
  /* We assume front-end functions already check them.  */
  assert (start + range >= 0 && start + range <= length);
#endif

  match_first = start;
  input.tip_context = ((eflags & REG_NOTBOL) ? CONTEXT_BEGBUF
		       : CONTEXT_NEWLINE | CONTEXT_BEGBUF);

  /* Check incrementally whether of not the input string match.  */
  incr = (range < 0) ? -1 : 1;
  left_lim = (range < 0) ? start + range : start;
  right_lim = (range < 0) ? start : start + range;
  sb = MB_CUR_MAX == 1;
  fast_translate = sb || !(preg->syntax & RE_ICASE || preg->translate);

  for (;;)
    {
      /* At first get the current byte from input string.  */
      if (fastmap)
	{
	  if (BE (fast_translate, 1))
	    {
	      unsigned RE_TRANSLATE_TYPE t
		= (unsigned RE_TRANSLATE_TYPE) preg->translate;
	      if (BE (range >= 0, 1))
		{
		  if (BE (t != NULL, 0))
		    {
		      while (BE (match_first < right_lim, 1)
			     && !fastmap[t[(unsigned char) string[match_first]]])
			++match_first;
		    }
		  else
		    {
		      while (BE (match_first < right_lim, 1)
			     && !fastmap[(unsigned char) string[match_first]])
			++match_first;
		    }
		  if (BE (match_first == right_lim, 0))
		    {
		      int ch = match_first >= length
			       ? 0 : (unsigned char) string[match_first];
		      if (!fastmap[t ? t[ch] : ch])
			break;
		    }
		}
	      else
		{
		  while (match_first >= left_lim)
		    {
		      int ch = match_first >= length
			       ? 0 : (unsigned char) string[match_first];
		      if (fastmap[t ? t[ch] : ch])
			break;
		      --match_first;
		    }
		  if (match_first < left_lim)
		    break;
		}
	    }
	  else
	    {
	      int ch;

	      do
		{
		  /* In this case, we can't determine easily the current byte,
		     since it might be a component byte of a multibyte
		     character.  Then we use the constructed buffer
		     instead.  */
		  /* If MATCH_FIRST is out of the valid range, reconstruct the
		     buffers.  */
		  if (input.raw_mbs_idx + input.valid_len <= match_first
		      || match_first < input.raw_mbs_idx)
		    {
		      err = re_string_reconstruct (&input, match_first, eflags,
						   preg->newline_anchor);
		      if (BE (err != REG_NOERROR, 0))
			goto free_return;
		    }
		  /* If MATCH_FIRST is out of the buffer, leave it as '\0'.
		     Note that MATCH_FIRST must not be smaller than 0.  */
		  ch = ((match_first >= length) ? 0
		       : re_string_byte_at (&input,
					    match_first - input.raw_mbs_idx));
		  if (fastmap[ch])
		    break;
		  match_first += incr;
		}
	      while (match_first >= left_lim && match_first <= right_lim);
	      if (! fastmap[ch])
		break;
	    }
	}

      /* Reconstruct the buffers so that the matcher can assume that
	 the matching starts from the begining of the buffer.  */
      err = re_string_reconstruct (&input, match_first, eflags,
				   preg->newline_anchor);
      if (BE (err != REG_NOERROR, 0))
	goto free_return;
#ifdef RE_ENABLE_I18N
     /* Eliminate it when it is a component of a multibyte character
	 and isn't the head of a multibyte character.  */
      if (sb || re_string_first_byte (&input, 0))
#endif
	{
	  /* It seems to be appropriate one, then use the matcher.  */
	  /* We assume that the matching starts from 0.  */
	  mctx.state_log_top = mctx.nbkref_ents = mctx.max_mb_elem_len = 0;
	  match_last = check_matching (preg, &mctx, 0, fl_longest_match);
	  if (match_last != -1)
	    {
	      if (BE (match_last == -2, 0))
		{
		  err = REG_ESPACE;
		  goto free_return;
		}
	      else
		{
		  mctx.match_last = match_last;
		  if ((!preg->no_sub && nmatch > 1) || dfa->nbackref)
		    {
		      re_dfastate_t *pstate = mctx.state_log[match_last];
		      mctx.last_node = check_halt_state_context (preg, pstate,
								 &mctx, match_last);
		    }
		  if ((!preg->no_sub && nmatch > 1 && dfa->has_plural_match)
		      || dfa->nbackref)
		    {
		      err = prune_impossible_nodes (preg, &mctx);
		      if (err == REG_NOERROR)
			break;
		      if (BE (err != REG_NOMATCH, 0))
			goto free_return;
		    }
		  else
		    break; /* We found a matching.  */
		}
	    }
	  match_ctx_clean (&mctx);
	}
      /* Update counter.  */
      match_first += incr;
      if (match_first < left_lim || right_lim < match_first)
	break;
    }

  /* Set pmatch[] if we need.  */
  if (match_last != -1 && nmatch > 0)
    {
      int reg_idx;

      /* Initialize registers.  */
      for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
	pmatch[reg_idx].rm_so = pmatch[reg_idx].rm_eo = -1;

      /* Set the points where matching start/end.  */
      pmatch[0].rm_so = 0;
      pmatch[0].rm_eo = mctx.match_last;

      if (!preg->no_sub && nmatch > 1)
	{
	  err = set_regs (preg, &mctx, nmatch, pmatch,
			  dfa->has_plural_match && dfa->nbackref > 0);
	  if (BE (err != REG_NOERROR, 0))
	    goto free_return;
	}

      /* At last, add the offset to the each registers, since we slided
	 the buffers so that We can assume that the matching starts from 0.  */
      for (reg_idx = 0; reg_idx < nmatch; ++reg_idx)
	if (pmatch[reg_idx].rm_so != -1)
	  {
	    pmatch[reg_idx].rm_so += match_first;
	    pmatch[reg_idx].rm_eo += match_first;
	  }
    }
  err = (match_last == -1) ? REG_NOMATCH : REG_NOERROR;
 free_return:
  re_free (mctx.state_log);
  if (dfa->nbackref)
    match_ctx_free (&mctx);
  re_string_destruct (&input);
  return err;
}

static reg_errcode_t
prune_impossible_nodes (preg, mctx)
     const regex_t *preg;
     re_match_context_t *mctx;
{
  int halt_node, match_last;
  reg_errcode_t ret;
  re_dfa_t *dfa = (re_dfa_t *)preg->buffer;
  re_dfastate_t **sifted_states;
  re_dfastate_t **lim_states = NULL;
  re_sift_context_t sctx;
#ifdef DEBUG
  assert (mctx->state_log != NULL);
#endif
  match_last = mctx->match_last;
  halt_node = mctx->last_node;
  sifted_states = re_malloc (re_dfastate_t *, match_last + 1);
  if (BE (sifted_states == NULL, 0))
    {
      ret = REG_ESPACE;
      goto free_return;
    }
  if (dfa->nbackref)
    {
      lim_states = re_malloc (re_dfastate_t *, match_last + 1);
      if (BE (lim_states == NULL, 0))
	{
	  ret = REG_ESPACE;
	  goto free_return;
	}
      while (1)
	{
	  memset (lim_states, '\0',
		  sizeof (re_dfastate_t *) * (match_last + 1));
	  match_ctx_clear_flag (mctx);
	  sift_ctx_init (&sctx, sifted_states, lim_states, halt_node,
			 match_last, 0);
	  ret = sift_states_backward (preg, mctx, &sctx);
	  re_node_set_free (&sctx.limits);
	  if (BE (ret != REG_NOERROR, 0))