/*############################################################################
  clean: Remove extraneous characters from text files.
  Written by Anastasios Kotsikonas.
  Usage: clean {file}+
*/

# define MAX_BUF_LINES   100
# define MAX_LINE_LENGTH 133

# include <stdio.h>
# include <string.h>

main (argc, argv)
int argc;
char **argv;
{
  FILE *fin, *fout;
  signed char command [80], line [MAX_LINE_LENGTH];
  char *lines [MAX_LINE_LENGTH];
  int i, ch, buf;
  long int obytes, newbytes;

  if (argc == 1)
    fprintf (stderr, "Usage: clean {file}+\nRemove extraneous characters from text files.\n"),
    exit (1);
  --argc;
  while (argc) {
    if ((fin = fopen (argv [argc], "r")) == NULL) {
      printf ("Unable to open input file '%s'\n", argv [argc]);
      fflush (stdout);
      --argc;
      continue;
    }
    if ((fout = fopen ("/tmp/clean.out", "w")) == NULL) {
      printf ("Unable to write to /tmp. File %s not cleaned.\n",
	      argv [argc]);
      fflush (stdout);
      fclose (fin);
      --argc;
      continue;
    }
    obytes = newbytes = buf = 0;
    printf ("Cleaning '%s': ", argv [argc]);
    fflush (stdout);
    while (!feof (fin)) {
      line [0] = '\0';
      fgets (line, MAX_LINE_LENGTH - 1, fin);
      if (strlen (line) > 0) {
	obytes += strlen (line);
	i = 0;
	while ((ch = line [i]) != '\n' && ch != '\0') {
	  if (ch == '\n' || ch == '\r' || ch == '\t' || (ch >= ' ' && ch < 127))
	    ++i;
	  else if (ch >= 0 && ch < 256)
	    sprintf (line + i, "%s", line + i + 1);
	  else {
	    printf ("not a text file; file not cleaned.\n");
	    fflush (stdout);
	    goto abort;
	  }
	}
	newbytes += strlen (line);
	lines[buf] = (char *) malloc ((strlen (line) + 1) * sizeof (char));
	strcpy (lines[buf], line);
	if (++buf > MAX_BUF_LINES - 1) {
	  for (i = 0; i < buf; ++i) {
	    if (fprintf (fout, "%s", lines[i]) < 0) {
	      printf (" unable to write to /tmp anymore; file not cleaned.\n");
	      fflush (stdout);
# ifdef unix
	      unlink ("/tmp/clean.out");
# else
              command[0] = '\0';
              sprintf (command, "del \\tmp\\clean.out");
              system (command);
# endif
	      goto abort;
	    }
	    free (lines[i]);
	    lines[i] = NULL;
	  }
	  buf = 0;
	}
      }
    }
    for (i = 0; i < buf; ++i) {
      if (fprintf (fout, "%s", lines[i]) < 0) {
	printf (" unable to write to /tmp anymore; file not cleaned.\n");
	fflush (stdout);
# ifdef unix
	unlink ("/tmp/clean.out");
# else
        command[0] = '\0';
        sprintf (command, "del \\tmp\\clean.out");
        system (command);
# endif
	goto abort;
      }
      free (lines[i]);
      lines[i] = NULL;
    }
    buf = 0;
    printf ("%ld bytes before --> %ld bytes after", obytes, newbytes);
    if (obytes == newbytes)
      printf (": NO CHANGES");
    printf ("\n");
    fflush (stdout);
    fflush (fout);
    if (obytes != newbytes)
      command[0] = '\0',
# ifdef unix
      sprintf (command, "mv /tmp/clean.out ./%s", argv [argc]),
      system (command);
# else
      sprintf (command, "copy \\tmp\\clean.out .\\%s", argv [argc]),
      system (command);
    command[0] = '\0';
    sprintf (command, "del \\tmp\\clean.out");
    system (command);
# endif
  abort:
    for (i = 0; i < buf; ++i)
      free (lines[i]);
    fclose (fin);
    fclose (fout);
    --argc;
  }
}