pcretest.c

ncbi源码

int rc;
if ((rc = pcre_fullinfo(re, study, option, ptr)) < 0)
  fprintf(outfile, "Error %d from pcre_fullinfo(%d)\n", rc, option);
}




/* Read lines from named file or stdin and write to named file or stdout; lines
consist of a regular expression, in delimiters and optionally followed by
options, followed by a set of test data, terminated by an empty line. */

int main(int argc, char **argv)
{
FILE *infile = stdin;
int options = 0;
int study_options = 0;
int op = 1;
int timeit = 0;
int showinfo = 0;
int showstore = 0;
int size_offsets = 45;
int size_offsets_max;
int *offsets;
#if !defined NOPOSIX
int posix = 0;
#endif
int debug = 0;
int done = 0;
unsigned char buffer[30000];
unsigned char dbuffer[1024];

/* Static so that new_malloc can use it. */

outfile = stdout;

/* Scan options */

while (argc > 1 && argv[op][0] == '-')
  {
  char *endptr;

  if (strcmp(argv[op], "-s") == 0 || strcmp(argv[op], "-m") == 0)
    showstore = 1;
  else if (strcmp(argv[op], "-t") == 0) timeit = 1;
  else if (strcmp(argv[op], "-i") == 0) showinfo = 1;
  else if (strcmp(argv[op], "-d") == 0) showinfo = debug = 1;
  else if (strcmp(argv[op], "-o") == 0 && argc > 2 &&
      ((size_offsets = (int)strtoul(argv[op+1], &endptr, 10)), *endptr == 0))
    {
    op++;
    argc--;
    }
#if !defined NOPOSIX
  else if (strcmp(argv[op], "-p") == 0) posix = 1;
#endif
  else
    {
    printf("** Unknown or malformed option %s\n", argv[op]);
    printf("Usage:   pcretest [-d] [-i] [-o <n>] [-p] [-s] [-t] [<input> [<output>]]\n");
    printf("  -d     debug: show compiled code; implies -i\n"
           "  -i     show information about compiled pattern\n"
           "  -o <n> set size of offsets vector to <n>\n");
#if !defined NOPOSIX
    printf("  -p     use POSIX interface\n");
#endif
    printf("  -s     output store information\n"
           "  -t     time compilation and execution\n");
    return 1;
    }
  op++;
  argc--;
  }

/* Get the store for the offsets vector, and remember what it was */

size_offsets_max = size_offsets;
offsets = malloc(size_offsets_max * sizeof(int));
if (offsets == NULL)
  {
  printf("** Failed to get %d bytes of memory for offsets vector\n",
    size_offsets_max * sizeof(int));
  return 1;
  }

/* Sort out the input and output files */

if (argc > 1)
  {
  infile = fopen(argv[op], "r");
  if (infile == NULL)
    {
    printf("** Failed to open %s\n", argv[op]);
    return 1;
    }
  }

if (argc > 2)
  {
  outfile = fopen(argv[op+1], "wb");
  if (outfile == NULL)
    {
    printf("** Failed to open %s\n", argv[op+1]);
    return 1;
    }
  }

/* Set alternative malloc function */

pcre_malloc = new_malloc;

/* Heading line, then prompt for first regex if stdin */

fprintf(outfile, "PCRE version %s\n\n", pcre_version());

/* Main loop */

while (!done)
  {
  pcre *re = NULL;
  pcre_extra *extra = NULL;

#if !defined NOPOSIX  /* There are still compilers that require no indent */
  regex_t preg;
  int do_posix = 0;
#endif

  const char *error;
  unsigned char *p, *pp, *ppp;
  const unsigned char *tables = NULL;
  int do_study = 0;
  int do_debug = debug;
  int do_G = 0;
  int do_g = 0;
  int do_showinfo = showinfo;
  int do_showrest = 0;
  int utf8 = 0;
  int erroroffset, len, delimiter;

  if (infile == stdin) printf("  re> ");
  if (fgets((char *)buffer, sizeof(buffer), infile) == NULL) break;
  if (infile != stdin) fprintf(outfile, "%s", (char *)buffer);

  p = buffer;
  while (isspace(*p)) p++;
  if (*p == 0) continue;

  /* Get the delimiter and seek the end of the pattern; if is isn't
  complete, read more. */

  delimiter = *p++;

  if (isalnum(delimiter) || delimiter == '\\')
    {
    fprintf(outfile, "** Delimiter must not be alphameric or \\\n");
    goto SKIP_DATA;
    }

  pp = p;

  for(;;)
    {
    while (*pp != 0)
      {
      if (*pp == '\\' && pp[1] != 0) pp++;
        else if (*pp == delimiter) break;
      pp++;
      }
    if (*pp != 0) break;

    len = sizeof(buffer) - (pp - buffer);
    if (len < 256)
      {
      fprintf(outfile, "** Expression too long - missing delimiter?\n");
      goto SKIP_DATA;
      }

    if (infile == stdin) printf("    > ");
    if (fgets((char *)pp, len, infile) == NULL)
      {
      fprintf(outfile, "** Unexpected EOF\n");
      done = 1;
      goto CONTINUE;
      }
    if (infile != stdin) fprintf(outfile, "%s", (char *)pp);
    }

  /* If the first character after the delimiter is backslash, make
  the pattern end with backslash. This is purely to provide a way
  of testing for the error message when a pattern ends with backslash. */

  if (pp[1] == '\\') *pp++ = '\\';

  /* Terminate the pattern at the delimiter */

  *pp++ = 0;

  /* Look for options after final delimiter */

  options = 0;
  study_options = 0;
  log_store = showstore;  /* default from command line */

  while (*pp != 0)
    {
    switch (*pp++)
      {
      case 'g': do_g = 1; break;
      case 'i': options |= PCRE_CASELESS; break;
      case 'm': options |= PCRE_MULTILINE; break;
      case 's': options |= PCRE_DOTALL; break;
      case 'x': options |= PCRE_EXTENDED; break;

      case '+': do_showrest = 1; break;
      case 'A': options |= PCRE_ANCHORED; break;
      case 'D': do_debug = do_showinfo = 1; break;
      case 'E': options |= PCRE_DOLLAR_ENDONLY; break;
      case 'G': do_G = 1; break;
      case 'I': do_showinfo = 1; break;
      case 'M': log_store = 1; break;

#if !defined NOPOSIX
      case 'P': do_posix = 1; break;
#endif

      case 'S': do_study = 1; break;
      case 'U': options |= PCRE_UNGREEDY; break;
      case 'X': options |= PCRE_EXTRA; break;
      case '8': options |= PCRE_UTF8; utf8 = 1; break;

      case 'L':
      ppp = pp;
      while (*ppp != '\n' && *ppp != ' ') ppp++;
      *ppp = 0;
      if (setlocale(LC_CTYPE, (const char *)pp) == NULL)
        {
        fprintf(outfile, "** Failed to set locale \"%s\"\n", pp);
        goto SKIP_DATA;
        }
      tables = pcre_maketables();
      pp = ppp;
      break;

      case '\n': case ' ': break;
      default:
      fprintf(outfile, "** Unknown option '%c'\n", pp[-1]);
      goto SKIP_DATA;
      }
    }

  /* Handle compiling via the POSIX interface, which doesn't support the
  timing, showing, or debugging options, nor the ability to pass over
  local character tables. */

#if !defined NOPOSIX
  if (posix || do_posix)
    {
    int rc;
    int cflags = 0;
    if ((options & PCRE_CASELESS) != 0) cflags |= REG_ICASE;
    if ((options & PCRE_MULTILINE) != 0) cflags |= REG_NEWLINE;
    rc = regcomp(&preg, (char *)p, cflags);

    /* Compilation failed; go back for another re, skipping to blank line
    if non-interactive. */

    if (rc != 0)
      {
      (void)regerror(rc, &preg, (char *)buffer, sizeof(buffer));
      fprintf(outfile, "Failed: POSIX code %d: %s\n", rc, buffer);
      goto SKIP_DATA;
      }
    }

  /* Handle compiling via the native interface */

  else
#endif  /* !defined NOPOSIX */

    {
    if (timeit)
      {
      register int i;
      clock_t time_taken;
      clock_t start_time = clock();
      for (i = 0; i < LOOPREPEAT; i++)
        {
        re = pcre_compile((char *)p, options, &error, &erroroffset, tables);
        if (re != NULL) free(re);
        }
      time_taken = clock() - start_time;
      fprintf(outfile, "Compile time %.3f milliseconds\n",
        ((double)time_taken * 1000.0) /
        ((double)LOOPREPEAT * (double)CLOCKS_PER_SEC));
      }

    re = pcre_compile((char *)p, options, &error, &erroroffset, tables);

    /* Compilation failed; go back for another re, skipping to blank line
    if non-interactive. */

    if (re == NULL)
      {
      fprintf(outfile, "Failed: %s at offset %d\n", error, erroroffset);
      SKIP_DATA:
      if (infile != stdin)
        {
        for (;;)
          {
          if (fgets((char *)buffer, sizeof(buffer), infile) == NULL)
            {
            done = 1;
            goto CONTINUE;
            }
          len = (int)strlen((char *)buffer);
          while (len > 0 && isspace(buffer[len-1])) len--;
          if (len == 0) break;
          }
        fprintf(outfile, "\n");
        }
      goto CONTINUE;
      }

    /* Compilation succeeded; print data if required. There are now two
    info-returning functions. The old one has a limited interface and
    returns only limited data. Check that it agrees with the newer one. */

    if (do_showinfo)
      {
      unsigned long int get_options;
      int old_first_char, old_options, old_count;
      int count, backrefmax, first_char, need_char;
      size_t size;

      if (do_debug) print_internals(re);

      new_info(re, NULL, PCRE_INFO_OPTIONS, &get_options);
      new_info(re, NULL, PCRE_INFO_SIZE, &size);
      new_info(re, NULL, PCRE_INFO_CAPTURECOUNT, &count);
      new_info(re, NULL, PCRE_INFO_BACKREFMAX, &backrefmax);
      new_info(re, NULL, PCRE_INFO_FIRSTCHAR, &first_char);
      new_info(re, NULL, PCRE_INFO_LASTLITERAL, &need_char);

      old_count = pcre_info(re, &old_options, &old_first_char);
      if (count < 0) fprintf(outfile,
        "Error %d from pcre_info()\n", count);
      else
        {
        if (old_count != count) fprintf(outfile,
          "Count disagreement: pcre_fullinfo=%d pcre_info=%d\n", count,
            old_count);

        if (old_first_char != first_char) fprintf(outfile,
          "First char disagreement: pcre_fullinfo=%d pcre_info=%d\n",
            first_char, old_first_char);

        if (old_options != (int)get_options) fprintf(outfile,
          "Options disagreement: pcre_fullinfo=%ld pcre_info=%d\n",
            get_options, old_options);
        }

      if (size != gotten_store) fprintf(outfile,
        "Size disagreement: pcre_fullinfo=%d call to malloc for %d\n",
        size, gotten_store);

      fprintf(outfile, "Capturing subpattern count = %d\n", count);
      if (backrefmax > 0)
        fprintf(outfile, "Max back reference = %d\n", backrefmax);
      if (get_options == 0) fprintf(outfile, "No options\n");
        else fprintf(outfile, "Options:%s%s%s%s%s%s%s%s%s\n",
          ((get_options & PCRE_ANCHORED) != 0)? " anchored" : "",
          ((get_options & PCRE_CASELESS) != 0)? " caseless" : "",
          ((get_options & PCRE_EXTENDED) != 0)? " extended" : "",
          ((get_options & PCRE_MULTILINE) != 0)? " multiline" : "",
          ((get_options & PCRE_DOTALL) != 0)? " dotall" : "",
          ((get_options & PCRE_DOLLAR_ENDONLY) != 0)? " dollar_endonly" : "",
          ((get_options & PCRE_EXTRA) != 0)? " extra" : "",
          ((get_options & PCRE_UNGREEDY) != 0)? " ungreedy" : "",
          ((get_options & PCRE_UTF8) != 0)? " utf8" : "");

      if (((((real_pcre *)re)->options) & PCRE_ICHANGED) != 0)
        fprintf(outfile, "Case state changes\n");

      if (first_char == -1)
        {
        fprintf(outfile, "First char at start or follows \\n\n");
        }
      else if (first_char < 0)
        {
        fprintf(outfile, "No first char\n");
        }
      else
        {
        if (isprint(first_char))
          fprintf(outfile, "First char = \'%c\'\n", first_char);
        else
          fprintf(outfile, "First char = %d\n", first_char);
        }

      if (need_char < 0)
        {
        fprintf(outfile, "No need char\n");
        }
      else
        {
        if (isprint(need_char))
          fprintf(outfile, "Need char = \'%c\'\n", need_char);
        else
          fprintf(outfile, "Need char = %d\n", need_char);
        }
      }

    /* If /S was present, study the regexp to generate additional info to
    help with the matching. */

    if (do_study)
      {
      if (timeit)
        {
        register int i;
        clock_t time_taken;
        clock_t start_time = clock();
        for (i = 0; i < LOOPREPEAT; i++)
          extra = pcre_study(re, study_options, &error);