/*
 * Merge two files by lines.
 * Ignore punctuation and case in the comparison.
 * Optionally skip leading chars in comparison.
 *
 *
 * Usage:
 *
 *   sortlex <infile1> <infile2> <outfile> [-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 BOOL int
#define TRIAD int	/* -1, 0, or +1 */
#define NATURAL unsigned	/* Natural or counting numbers */

#define YES 1
#define NO  0


/*
 * forward declared functions
 */

TRIAD cmpstr();

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

char *s1, *t1;
int i;

/*
 * Globals
 */

FILE ibuf1, ibuf2, obuf;
int lcount;
NATURAL maxlines;
NATURAL skipchars;
char linbuf1[300], linbuf2[300];



PROC
main(argc, argv)
  char **argv;
{
	int maxwidth, width, wideline;
	BOOL mor_in_1, mor_in_2;
	int i;

	skipchars = maxwidth = 0;
	if ( argc < 3 || argc >5 )
	  abort("incorrect number of arguments");
	if ( fopen(argv[1], &ibuf1) == ERROR ) {
	  printf("Can't open %s\n", argv[1]);
	  exit(1);
	  }
	if ( fopen(argv[2], &ibuf2) == ERROR ) {
	  printf("Can't open %s\n", argv[2]);
	  exit(1);
	  }
	if ( fcreat(argv[3], &obuf) == ERROR ) {
	  printf("Can't create %s\n", argv[3]);
	  exit(1);
	  }
	for ( i = 4; 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");
	    }
	  }
	lcount = 0;
	if ( ! fgets(linbuf1, &ibuf1) )
	  abort("First File is Empty");
	if ( ! fgets(linbuf2, &ibuf2) )
	  abort("Second File is empty");
	mor_in_1 = mor_in_2 = YES;
	do {
	  if ( cmpstr(linbuf1, linbuf2) < 0 ) {
	    fputs(linbuf1, &obuf);
	    ++lcount;
	    if ( mor_in_1 )
	      if ( ! fgets(linbuf1, &ibuf1) ) {
	        mor_in_1 = NO;
	        linbuf1[0] = 0xff;
	        linbuf1[1] = '\0';	/* Greatest Possible String */
	        }
	    }
	  else {
	    fputs(linbuf2, &obuf);
	    ++lcount;
	    if ( mor_in_2 )
	      if ( ! fgets(linbuf2, &ibuf2) ) {
	        mor_in_2 = NO;
	        linbuf2[0] = 0xff;
	        linbuf2[1] = '\0';	/* Greatest Possible String */
	        }
	    }
	  } while ( mor_in_1 || mor_in_2 );
	fclose(&ibuf1);
	fclose(&ibuf2);
	putc(CPMEOF, &obuf);
	fflush(&obuf);	
	fclose(&obuf);
	printf("%d lines were merged\n", lcount);
}

/*
 * compare two strings
 * 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;
	if ( *s1 != 0xff ) /* max string mark */
	  for ( i = 0; i < skipchars; ++i, ++s1 )
	    if ( *s1 == '\0' )
	      break;
	t1 = t;
	if ( *t1 != 0xff ) /* max string mark */
	  for ( i = 0; i < skipchars; ++i, ++t1 )
	    if ( *t1 == '\0' )
	      break;
	if ( *s1 == 0xff || *t1 == 0xff )
	  return (tolower(*s1) - tolower(*t1));
	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);
	puts("Usage:\n\n");
	puts("mergelex <infile1> <infile2> <outfile> [-s<skip chars>]\n");
	printf("\nLast Revision Date: %s\n", REV_DATE);
	exit(1);
}