/***********************************************************************
* uniq.c
*	Removes adjacent duplicate lines from a file.
* notes
*	#define DEBUG to produce debugging code.
*
* Copyright (c) David R Tribble, May85.
* 1.0 05-06-85 drt.
* 2.0 08-13-87 drt. Reworked program. Added hash comparison for
*	speed.
*/

static char	prog[] =	"uniq";

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

#ifdef DEBUG
 #define D
#else
 #define D	if (0)
#endif

#define LINESZ		4096		/* max line buffer size		*/

#define strequal(a,b)	(! strcmp(a, b))

/***********************************************************************
* getlin
* returns
*	Hash value or EOF.
*
* 1.0 05-06-85 drt.
* 2.0 08-13-87 drt. Reworked program. Added hash comparison for
*	speed.
*/

static int getlin (in, buf)
    FILE *	in;
    char *	buf;
{
    int		cnt;
    int		c;
    int		hash;
    char *	b0;

    b0 = buf;
    hash = 0;
    cnt = 1;
    for (;;)
    {
        c = getc(in);
        if (c == EOF)
            break;
        *buf = c;
        ++buf;
        ++cnt;
        hash += c;

        if (cnt >= LINESZ)
            break;
        if (c == '\n'  /** ||  c == '\f' **/)
            break;
        /* note: this won't handle NULs very well */
    }
    *buf = '\0';
    if (hash == EOF)
        hash = EOF + 1;

    D   fprintf(stderr, "getlin(): %04X \"%s\"\n", hash, b0);

    if (cnt > 1)
        return hash;
    return EOF;
}

/***********************************************************************
* uniq
* returns
*	void.
*
* 1.0 05-06-85 drt.
* 2.0 08-13-87 drt. Reworked program. Added hash comparison for
*	speed.
*/

static char	buf1 [LINESZ+1];
static char	buf2 [LINESZ+1];

void uniq (in, out)
    FILE *	in;
    FILE *	out;
{
    int		hash1;
    int		hash2;
    long	linec, dupc, outc;

    linec = dupc = outc = 0;
    hash1 = getlin(in, buf1);
    while ((hash2 = getlin(in, buf2)) != EOF)
    {
        ++linec;
        if (hash1 != hash2  ||  !strequal(buf1, buf2))
        {
            D   fprintf(stderr, "uniq(): lines differ, %04X %04X\n",
                    hash1, hash2);
            fputs(buf1, out);
            ++outc;
            strcpy(buf1, buf2);
            hash1 = hash2;
        }
        else
        {
            ++dupc;
            D   fprintf(stderr, "uniq(): lines match, %04X\n", hash1);
        }
    }
    fputs(buf1, out);

    fprintf(stderr, "%lu in, %lu out (%lu duplicates)\n",
        linec, outc, dupc);
}

/***********************************************************************
* main
*
* 1.0 05-06-85 drt.
* 2.0 08-13-87 drt. Reworked program. Added hash comparison for
*	speed.
*/

int main (argc, argv)
    int		argc;
    char	** argv;
{
    if (argc > 1  ||  isatty(fileno(stdin)))
        goto usage;

    uniq(stdin, stdout);
    return 0;

usage:
    fprintf(stderr, "Remove adjacent duplicate lines.\n\n");
    fprintf(stderr, "usage:\t%s <input\n", prog);
    return -1;
}

/* end uniq.c */