pcregrep.c

PHP v6.0 For Linux 运行环境：Win9X/ WinME/ WinNT/ Win2K/ WinXP

/************* Test stdout for being a terminal when we can't do it **********/

static BOOL
is_stdout_tty(void)
{
return FALSE;
}


#endif



#if ! HAVE_STRERROR
/*************************************************
*     Provide strerror() for non-ANSI libraries  *
*************************************************/

/* Some old-fashioned systems still around (e.g. SunOS4) don't have strerror()
in their libraries, but can provide the same facility by this simple
alternative function. */

extern int   sys_nerr;
extern char *sys_errlist[];

char *
strerror(int n)
{
if (n < 0 || n >= sys_nerr) return "unknown error number";
return sys_errlist[n];
}
#endif /* HAVE_STRERROR */



/*************************************************
*       Print the previous "after" lines         *
*************************************************/

/* This is called if we are about to lose said lines because of buffer filling,
and at the end of the file. The data in the line is written using fwrite() so
that a binary zero does not terminate it.

Arguments:
  lastmatchnumber   the number of the last matching line, plus one
  lastmatchrestart  where we restarted after the last match
  endptr            end of available data
  printname         filename for printing

Returns:            nothing
*/

static void do_after_lines(int lastmatchnumber, char *lastmatchrestart,
  char *endptr, char *printname)
{
if (after_context > 0 && lastmatchnumber > 0)
  {
  int count = 0;
  while (lastmatchrestart < endptr && count++ < after_context)
    {
    char *pp = lastmatchrestart;
    if (printname != NULL) fprintf(stdout, "%s-", printname);
    if (number) fprintf(stdout, "%d-", lastmatchnumber++);
    while (*pp != '\n') pp++;
    fwrite(lastmatchrestart, 1, pp - lastmatchrestart + 1, stdout);
    lastmatchrestart = pp + 1;
    }
  hyphenpending = TRUE;
  }
}



/*************************************************
*            Grep an individual file             *
*************************************************/

/* This is called from grep_or_recurse() below. It uses a buffer that is three
times the value of MBUFTHIRD. The matching point is never allowed to stray into
the top third of the buffer, thus keeping more of the file available for
context printing or for multiline scanning. For large files, the pointer will
be in the middle third most of the time, so the bottom third is available for
"before" context printing.

Arguments:
  in           the fopened FILE stream
  printname    the file name if it is to be printed for each match
               or NULL if the file name is not to be printed
               it cannot be NULL if filenames[_nomatch]_only is set

Returns:       0 if there was at least one match
               1 otherwise (no matches)
*/

static int
pcregrep(FILE *in, char *printname)
{
int rc = 1;
int linenumber = 1;
int lastmatchnumber = 0;
int count = 0;
int offsets[99];
char *lastmatchrestart = NULL;
char buffer[3*MBUFTHIRD];
char *ptr = buffer;
char *endptr;
size_t bufflength;
BOOL endhyphenpending = FALSE;

/* Do the first read into the start of the buffer and set up the pointer to
end of what we have. */

bufflength = fread(buffer, 1, 3*MBUFTHIRD, in);
endptr = buffer + bufflength;

/* Loop while the current pointer is not at the end of the file. For large
files, endptr will be at the end of the buffer when we are in the middle of the
file, but ptr will never get there, because as soon as it gets over 2/3 of the
way, the buffer is shifted left and re-filled. */

while (ptr < endptr)
  {
  int i;
  int mrc = 0;
  BOOL match = FALSE;
  char *t = ptr;
  size_t length, linelength;

  /* At this point, ptr is at the start of a line. We need to find the length
  of the subject string to pass to pcre_exec(). In multiline mode, it is the
  length remainder of the data in the buffer. Otherwise, it is the length of
  the next line. After matching, we always advance by the length of the next
  line. In multiline mode the PCRE_FIRSTLINE option is used for compiling, so
  that any match is constrained to be in the first line. */

  linelength = 0;
  while (t < endptr && *t++ != '\n') linelength++;
  length = multiline? endptr - ptr : linelength;


  /* Extra processing for Jeffrey Friedl's debugging. */

#ifdef JFRIEDL_DEBUG
  if (jfriedl_XT || jfriedl_XR)
  {
      #include <sys/time.h>
      #include <time.h>
      struct timeval start_time, end_time;
      struct timezone dummy;

      if (jfriedl_XT)
      {
          unsigned long newlen = length * jfriedl_XT + strlen(jfriedl_prefix) + strlen(jfriedl_postfix);
          const char *orig = ptr;
          ptr = malloc(newlen + 1);
          if (!ptr) {
                  printf("out of memory");
                  exit(2);
          }
          endptr = ptr;
          strcpy(endptr, jfriedl_prefix); endptr += strlen(jfriedl_prefix);
          for (i = 0; i < jfriedl_XT; i++) {
                  strncpy(endptr, orig,  length);
                  endptr += length;
          }
          strcpy(endptr, jfriedl_postfix); endptr += strlen(jfriedl_postfix);
          length = newlen;
      }

      if (gettimeofday(&start_time, &dummy) != 0)
              perror("bad gettimeofday");


      for (i = 0; i < jfriedl_XR; i++)
          match = (pcre_exec(pattern_list[0], hints_list[0], ptr, length, 0, 0, offsets, 99) >= 0);

      if (gettimeofday(&end_time, &dummy) != 0)
              perror("bad gettimeofday");

      double delta = ((end_time.tv_sec + (end_time.tv_usec / 1000000.0))
                      -
                      (start_time.tv_sec + (start_time.tv_usec / 1000000.0)));

      printf("%s TIMER[%.4f]\n", match ? "MATCH" : "FAIL", delta);
      return 0;
  }
#endif


  /* Run through all the patterns until one matches. Note that we don't include
  the final newline in the subject string. */

  for (i = 0; i < pattern_count; i++)
    {
    mrc = pcre_exec(pattern_list[i], hints_list[i], ptr, length, 0, 0,
      offsets, 99);
    if (mrc >= 0) { match = TRUE; break; }
    if (mrc != PCRE_ERROR_NOMATCH)
      {
      fprintf(stderr, "pcregrep: pcre_exec() error %d while matching ", mrc);
      if (pattern_count > 1) fprintf(stderr, "pattern number %d to ", i+1);
      fprintf(stderr, "this line:\n");
      fwrite(ptr, 1, linelength, stderr);   /* In case binary zero included */
      fprintf(stderr, "\n");
      if (error_count == 0 &&
          (mrc == PCRE_ERROR_MATCHLIMIT || mrc == PCRE_ERROR_RECURSIONLIMIT))
        {
        fprintf(stderr, "pcregrep: error %d means that a resource limit "
          "was exceeded\n", mrc);
        fprintf(stderr, "pcregrep: check your regex for nested unlimited loops\n");
        }
      if (error_count++ > 20)
        {
        fprintf(stderr, "pcregrep: too many errors - abandoned\n");
        exit(2);
        }
      match = invert;    /* No more matching; don't show the line again */
      break;
      }
    }

  /* If it's a match or a not-match (as required), do what's wanted. */

  if (match != invert)
    {
    BOOL hyphenprinted = FALSE;

    /* We've failed if we want a file that doesn't have any matches. */

    if (filenames == FN_NOMATCH_ONLY) return 1;

    /* Just count if just counting is wanted. */

    if (count_only) count++;

    /* If all we want is a file name, there is no need to scan any more lines
    in the file. */

    else if (filenames == FN_ONLY)
      {
      fprintf(stdout, "%s\n", printname);
      return 0;
      }

    /* Likewise, if all we want is a yes/no answer. */

    else if (quiet) return 0;

    /* The --only-matching option prints just the substring that matched, and
    does not pring any context. */

    else if (only_matching)
      {
      if (printname != NULL) fprintf(stdout, "%s:", printname);
      if (number) fprintf(stdout, "%d:", linenumber);
      fwrite(ptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
      fprintf(stdout, "\n");
      }

    /* This is the default case when none of the above options is set. We print
    the matching lines(s), possibly preceded and/or followed by other lines of
    context. */

    else
      {
      /* See if there is a requirement to print some "after" lines from a
      previous match. We never print any overlaps. */

      if (after_context > 0 && lastmatchnumber > 0)
        {
        int linecount = 0;
        char *p = lastmatchrestart;

        while (p < ptr && linecount < after_context)
          {
          while (*p != '\n') p++;
          p++;
          linecount++;
          }

        /* It is important to advance lastmatchrestart during this printing so
        that it interacts correctly with any "before" printing below. Print
        each line's data using fwrite() in case there are binary zeroes. */

        while (lastmatchrestart < p)
          {
          char *pp = lastmatchrestart;
          if (printname != NULL) fprintf(stdout, "%s-", printname);
          if (number) fprintf(stdout, "%d-", lastmatchnumber++);
          while (*pp != '\n') pp++;
          fwrite(lastmatchrestart, 1, pp - lastmatchrestart + 1, stdout);
          lastmatchrestart = pp + 1;
          }
        if (lastmatchrestart != ptr) hyphenpending = TRUE;
        }

      /* If there were non-contiguous lines printed above, insert hyphens. */

      if (hyphenpending)
        {
        fprintf(stdout, "--\n");
        hyphenpending = FALSE;
        hyphenprinted = TRUE;
        }

      /* See if there is a requirement to print some "before" lines for this
      match. Again, don't print overlaps. */

      if (before_context > 0)
        {
        int linecount = 0;
        char *p = ptr;

        while (p > buffer && (lastmatchnumber == 0 || p > lastmatchrestart) &&
               linecount < before_context)
          {
          linecount++;
          p--;
          while (p > buffer && p[-1] != '\n') p--;
          }

        if (lastmatchnumber > 0 && p > lastmatchrestart && !hyphenprinted)
          fprintf(stdout, "--\n");

        while (p < ptr)
          {
          char *pp = p;
          if (printname != NULL) fprintf(stdout, "%s-", printname);
          if (number) fprintf(stdout, "%d-", linenumber - linecount--);
          while (*pp != '\n') pp++;
          fwrite(p, 1, pp - p + 1, stdout);   /* In case binary zero */
          p = pp + 1;
          }
        }

      /* Now print the matching line(s); ensure we set hyphenpending at the end
      of the file if any context lines are being output. */

      if (after_context > 0 || before_context > 0)
        endhyphenpending = TRUE;

      if (printname != NULL) fprintf(stdout, "%s:", printname);
      if (number) fprintf(stdout, "%d:", linenumber);

      /* In multiline mode, we want to print to the end of the line in which
      the end of the matched string is found, so we adjust linelength and the
      line number appropriately. Because the PCRE_FIRSTLINE option is set, the
      start of the match will always be before the first \n character. */

      if (multiline)
        {
        char *endmatch = ptr + offsets[1];
        t = ptr;
        while (t < endmatch) { if (*t++ == '\n') linenumber++; }
        while (endmatch < endptr && *endmatch != '\n') endmatch++;
        linelength = endmatch - ptr;
        }

      /*** NOTE: Use only fwrite() to output the data line, so that binary
      zeroes are treated as just another data character. */

      /* This extra option, for Jeffrey Friedl's debugging requirements,
      replaces the matched string, or a specific captured string if it exists,
      with X. When this happens, colouring is ignored. */

#ifdef JFRIEDL_DEBUG
      if (S_arg >= 0 && S_arg < mrc)
        {
        int first = S_arg * 2;
        int last  = first + 1;
        fwrite(ptr, 1, offsets[first], stdout);
        fprintf(stdout, "X");
        fwrite(ptr + offsets[last], 1, linelength - offsets[last], stdout);
        }
      else
#endif

      /* We have to split the line(s) up if colouring. */

      if (do_colour)
        {
        fwrite(ptr, 1, offsets[0], stdout);
        fprintf(stdout, "%c[%sm", 0x1b, colour_string);
        fwrite(ptr + offsets[0], 1, offsets[1] - offsets[0], stdout);
        fprintf(stdout, "%c[00m", 0x1b);
        fwrite(ptr + offsets[1], 1, linelength - offsets[1], stdout);
        }
      else fwrite(ptr, 1, linelength, stdout);

      fprintf(stdout, "\n");
      }

    /* End of doing what has to be done for a match */

    rc = 0;    /* Had some success */

    /* Remember where the last match happened for after_context. We remember
    where we are about to restart, and that line's number. */

    lastmatchrestart = ptr + linelength + 1;
    lastmatchnumber = linenumber + 1;
    }

  /* Advance to after the newline and increment the line number. */

  ptr += linelength + 1;
  linenumber++;

  /* If we haven't yet reached the end of the file (the buffer is full), and
  the current point is in the top 1/3 of the buffer, slide the buffer down by
  1/3 and refill it. Before we do this, if some unprinted "after" lines are
  about to be lost, print them. */

  if (bufflength >= sizeof(buffer) && ptr > buffer + 2*MBUFTHIRD)
    {
    if (after_context > 0 &&
        lastmatchnumber > 0 &&
        lastmatchrestart < buffer + MBUFTHIRD)
      {
      do_after_lines(lastmatchnumber, lastmatchrestart, endptr, printname);
      lastmatchnumber = 0;
      }

    /* Now do the shuffle */

    memmove(buffer, buffer + MBUFTHIRD, 2*MBUFTHIRD);
    ptr -= MBUFTHIRD;
    bufflength = 2*MBUFTHIRD + fread(buffer + 2*MBUFTHIRD, 1, MBUFTHIRD, in);