/*
 * SORTLEX.C
 *
 * Sort a file by lines.
 * Ignore case and punctuation in comparison.
 * Optionally skip leading chars in comparison.
 *
 * This is a modification of the ALPH program written by Leor Zolman
 *
 * 	The sort is alphameric, ignoring case and punctuation.
 *
 * 	The "-l" parameter is for the maximum number of lines of text
 * to be sorted (default = MAXLINES).  This allocates an array of
 * pointers to text.
 *
 * 	The "-s" parameter is to specify a number of LEADING characters
 * in each line to be ignored in determining order in the sort.
 *
 * Usage:
 *
 *   sortlex <inputfile> <outputfile> [-l<maxlines>] [-s<skip chars>]
 *
 * Written to use BDS C by H.Moran
 * Last Update: 10/9/83 (hrm)
 */

#include "bdscio.h"

#define REV_DATE "10/9/83"

#define PROC int	/* PROCedure i.e. function returning no value */
#define TRIAD int	/* -1, 0, or +1 */
#define NATURAL unsigned	/* Natural or counting numbers */

#define MAXLINES   500		/* default max number of lines */

/*
 * forward declared functions
 */

	TRIAD cmpstr();

/*
 * cmpstr() locals made global for speed
 */

	char *s1, *t1;
	int i;

/*
 * Globals
 */

  FILE iobuf;	/* used for both input and output file */
  char **ltab;
  int lcount;
  char *allocp;
  NATURAL maxlines;
  NATURAL skipchars;

PROC
main(argc, argv)
  char **argv;
{
	char linbuf[300];
	int maxwidth, width, wideline;
	int i;

	skipchars = maxwidth = 0;
	maxlines = MAXLINES;
	if ( argc < 3 || argc >5 )
	  abort("incorrect number of arguments");
	if ( fopen(argv[1], &iobuf) == ERROR ) {
	  printf("Can't open %s\n", argv[1]);
	  exit(1);
	  }
	for ( i = 3; i < argc; ++i ) { /* decode options */
	  if ( argv[i][0] != '-' )
	    abort("expecting '-' option");
	  else { /* option flag detected */
	    if ( tolower(argv[i][1]) == 'l' )
	      maxlines = atoi(&argv[i][2]);
	    else if ( tolower(argv[i][1]) == 's' )
	      skipchars = atoi(&argv[i][2]);
	    else /* invalid option */
	      abort("unrecognized option");
	    }
	  }
	if ( (ltab = sbrk(2*(maxlines+1))) == ERROR ) {
	  printf("Can't allocate space for <%d> lines\n", maxlines);
	  exit(1);
	  }
	lcount = 0;
	for ( i = 0; i < maxlines; i++ ) {
	  if ( ! fgets(linbuf, &iobuf) )
	    break;
	  if ( (width = strlen(linbuf)) > maxwidth ) {
	    maxwidth = width;
	    wideline = i;
	    }
	  if ( (allocp = sbrk(width + 1)) == ERROR ) {
	    puts("Out of allocation space\n");
	    exit(1);
	    }
	  ltab[lcount++] = allocp;
	  strcpy(allocp, linbuf);
	  }
	fclose(&iobuf);
	if ( i == maxlines ) {
	  puts("Too many lines. Change MAXLINES or use -L option\n");
	  exit(1);
	  }
	printf("%d lines of a possible %d were used\n", i, maxlines);
	printf("Line %d is the longest line (%d characters)\n",
		wideline, maxwidth);
	qsort(ltab, lcount, 2, &cmpstr);
	if ( fcreat(argv[2], &iobuf) == ERROR ) {
	  printf("Can't create %s\n", argv[2]);
	  exit(1);
	  }
	for ( i = 0; i < lcount; i++ )
	  fputs(ltab[i], &iobuf);
	putc(CPMEOF, &iobuf);
	fflush(&iobuf);	
	fclose(&iobuf);
	puts("All done.\n");
}

/*
 * compare two strings via double deep pointers
 * ignore case and scan past blanks and punctuation in the comparison
 * i.e. the beginnings of a lexicographic compare
 */

TRIAD
cmpstr(s, t)
  char **s, **t;
{
/*	char *s1, *t1; ------ Made global for speed */
/*	int i;					    */

	s1 = *s;
	for ( i = 0; i < skipchars; ++i, ++s1 )
	  if ( *s1 == '\0' )
	    break;
	t1 = *t;
	for ( i = 0; i < skipchars; ++i, ++t1 )
	  if ( *t1 == '\0' )
	    break;
	while ( *s1 && *t1 ) { /* not at the end of either string */
	  if ( *s1 && ! ( isspace(*s1) || isalpha(*s1) || isdigit(*s1)) )
	    ++s1;
	  else if ( *t1 && ! ( isspace(*t1) || isalpha(*t1) || isdigit(*t1)) )
	    ++t1;
	  else if ( *s1 != *t1 )
	    break;
	  else { /* identical matching non-blanks */
	    ++s1;
	    ++t1;
	    }
	  }
	if ( i = (tolower(*s1) - tolower(*t1)) ) /* case independent compare */
	  return (i);
	else					 /* unless they match */
	  return (*s1 - *t1);
}



/*
 * abort with error message
 */

PROC
abort(s)
  char *s;
{
	printf("%s\n\n", s);
	printf("Usage:\n\n");
	printf("sortlex <infile> <outfile> [-l<maxlines>] [-s<skip chars>]\n");
	printf("DEFAULTS: <maxlines>=%d, <skip chars>=0\n", MAXLINES);
	printf("\nLast Revision Date: %s\n", REV_DATE);
	exit(1);
}