gnu1987/gnucmds/tr.c

/* -*- Mode: C -*- tr -- translates characters of streams
   Reads from standard-input, outputs to standard-output
   Arguments: [ -cds ] [character-set-1 [ character-set-2 ] ]
    -d means output all characters -not- in character-set-1
       If -d is specified, then character-set-2 may not be specified,
       (as there is no real translation occurring)
    -c means to interpret character-set-1 as its complement
       (ie as if it were the contents of a [^...] regexp rather than [...])
    -s means to only output one character if several successive
       charactes translated into the same character.
       eg "tr -s -c a-z _" on "abc!!ddd ef&$  x" would produce "abc_ddd_ef_x"
    Character-set-1 and character-set-2 are like regexp [...] ranges
     except that "\" is a quoting character, which may be followed by either
     by another "\" or 1 to 3 octal characters.  (\n, \f etc are not supported)

  This version of "tr" does not have the unix tr's bug of being unable
   to deal with \000
 */

/* mly 850320 */

#include <stdio.h>

int	dflag	= 0;
int	sflag	= 0;
int	cflag	= 0;
int	last;
char	xlate[256];		/* translation table */
char	nuke[256];		/* if -d, anything in this table is not
				 *   output at all,
				 * else this is the table of possible
				 *  translated characters, for the use
				 *  of -s
				 */
#define SRS_INIT 0
#define SRS_READ 1
#define SRS_RANGE 2
#define SRS_EOF 3

struct string_range_stream
  {
    int state;			/* 0 initial, 1 reading, 2 range, 3 eof */
    int lower;			/* lower bound of range we're in, or -1 */
    int upper;			/* upper bound of above */
    char *p;			/* last char looked at */
  }
  string1, string2;


char *progname;

main (argc,argv)
     int argc;
     char **argv;
{
  register i;
  register c, d;

  progname = argv[0];
  if (--argc > 0)
    /* inhale flags */
    { argv++;
      while (argc > 2 && argv[0][0] == '-' && argv[0][1] != 0)
	{ while (*++argv[0])
	    switch (argv[0][0])
	      { case 'c':
	          cflag++;
		  break;
		case 'd':
		  dflag++;
		  break;
		case 's':
		  sflag++;
		  break;
		default:
		  barf ("unknown flag \"-%c\":  Expected \"c\" \"d\", or \"s\"",
			argv[0][0]);
		  break; }
	  argc--;
	  argv++; } }


  string1.p = argv[0]; string1.state = SRS_INIT; string2.state = SRS_EOF;
  if (argc == 0)
    string1.state = SRS_EOF;
  if (argc == 2)
    { if (dflag)
	barf ("-d takes only one argument");
      else
	string2.p = argv[1]; string2.state = SRS_INIT; }
  else if (argc > 2)
    barf ("too many arguments (%d)", argc);

  if (cflag)
    { if (argc < 2)
	barf ("-c must have two arguments given");
      for (i = 0 ; i < 256; i++)
	{ xlate[i] = 0; nuke[i] = 0; }
      while ((c = snarf (&string1)) != EOF)
	xlate[c] = 1;
      for (i = 0 ; i < 256 ; i++)
	{ if ((d = snarf (&string2)) == EOF)
	    d = last;
	  else {last = d; nuke[d] = 1; }
	  xlate[i] = xlate[i] ? i : d; } }
  else
    { for (i = 0 ; i < 256 ; i++)
	{ xlate[i] = i; nuke[i] = 0; }
      last = EOF;
      while ((c = snarf (&string1)) != EOF)
	{ if (dflag)
	    nuke[c] = 1;
	  else { if ((d = snarf (&string2)) == EOF)
		   d = last;
		 else {last = d; nuke[d] = 1; }
		 xlate[c] = d; } } }

  last = EOF;
  while ((c = getchar ()) != EOF )
    if (dflag ? !nuke[c]
	      : ((c = xlate[c]) != last || !sflag || !nuke[c]))
      putchar (last = c);
  exit (0);
}

int
snarf (s)
     struct string_range_stream *s;
{ register c;
  switch (s->state)
    { case SRS_EOF:
        c = EOF;
	break;
      case SRS_RANGE:
	c = ++s->lower;
	if (c == s->upper) s->state = SRS_INIT;
	break;
      case SRS_INIT:
      case SRS_READ:
	if (!(c = *s->p++))
	  { s->state = SRS_EOF;
	    c = EOF; }
	else if (c == '-' && s->state == SRS_READ)
	  { register d;
	    if ((d = snarfchar (s)) == EOF) d = 255;
	    s->state = (((c = ++s->lower) == (s->upper = d))
			? SRS_INIT : SRS_RANGE); }
	else { s->state = SRS_READ;
	       if (c == '\\')
		 c = snarfquoted (s); }
	s->lower = c; }
  return (c); }

int
snarfchar (s)
     struct string_range_stream *s;
{ register c;
  if (!(c = *s->p++))
    { s->state = SRS_EOF;
      return (EOF); }
  else if (c == '\\')
    return (snarfquoted (s));
  else return (c); }

int
snarfquoted (s)
     struct string_range_stream *s;
{ register c;
  if (!(c = *s->p++))
    { s->state = SRS_EOF;
      return (EOF); }
  else if ('0' <= c && c <= '7')
    { register d = c - '0';
      register i;
      for (i = 0; i < 2; i++)
	{ if (!(c = *s->p++))
	    { s->state = SRS_EOF;
	      break; }
	  else if ('0' <= c && c <= '7')
	    d = d * 8 + c - '0';
	  else { s->p--;
		 break; } }
      return (d); }
  else return (c); }

/* print error message and die with exit code 1 */
barf (string, arg)
     char *string, *arg;
{
  fprintf (stderr,"%s: ", progname);
  fprintf (stderr, string, arg);
  fprintf (stderr,"\n");
  exit (1);
}