v04i082: TENEX - pgm to convert DEC10 data to 8bit bytes

[Dan Ts'o]

Posting-number: Volume 4, Issue 82
Submitted-by: "Dan Ts'o" <dan at rna.UUCP>
Archive-name: tenex

	I had just spent a few hours uploading MSDOS stuff from SIMTEL20.
Back on my PC, I realized that while I had remembered to transfer everything
in binary mode (in FTP), that I had neglected to account for (is it called ?)
TENEX mode (since SIMTEL20 is a 36-bit machine). I decided it would be faster
to write a C program to convert the 36-bit data into good ol' 8-bit bytes,
than re-upload every. So here is the program. I used MSC 5.0. The program will
convert both binary (-b, default, from each 36-bit word, generate 4x8-bit bytes
and discard the remaining 4bits) and text (-t, from each 36-bit word, generate
5x7bit (with zero parity bit) ASCII bytes and discard the last bit).
The program has been tested on both MSDOS 3.1 and 4.3BSD UNIX. If anyone needs
a binary copy for MSDOS, I'll mail them one. I hope this wheel hasn't been
reinvented too many times...

					Cheers,
					Dan Ts'o
					tso at rockefeller.edu
					dan at rna.rockefeller.edu
					...cmcl2!rna!dan
					212-570-7671
					Dept Neurbiology
					The Rockefeller University
					1230 York Ave.
					NY, NY  10021
wc tenex.c:
266     752    4787
sum tenex.c: (VAX)
32982     5
tenex.c:
/*
 * TENEX convert - convert tenex 36-bit files into standard 8-bit
 *	byte stream by discarding every 9th nibble (hopefully its zero).
 *	Also convert TENEX 7-bit ASCII to standard 8-bit byte ASCII by
 *	discarding every 36th bit.
 *
 *	Written by Daniel Ts'o, dan at rna.rockefeller.edu, 9/7/88
 *
 *	Usage: tenex [-[t|b]] [-[v|q]] [files...]
 *
 *	-t	Convert 5x7bit ASCII (35+1bit packing) into 8-bit byte stream
 *	-b	Convert 4*8bit binary (32+4bit packing) into 8-bit byte stream
			(Default is -b)
 *	-v	Print progress on stderr
 *	-q	Quietly delete nonzero bits
 *
 *	If file arguments are given, they are converted "in place", using
 *		temporary file "tenex.tmp".
 *	If file arguments are missing, convert stdin to stdout (filter mode).
 *
 *	Compile with MSC 5.0 on MSDOS:
 *
 *		cl tenex.c \lib\setargv.obj /link /NOE
 *
 *	to permit wildcard expansion.
 */

#include <stdio.h>

#define	TEMPFILE	"tenex.tmp"

/*#define	RENAME			/* Some UNIX's don't have rename() */

#ifdef	MSDOS
#define	READMODE	"rb"
#define	WRITEMODE	"wb"
#include <fcntl.h>
#include <io.h>
#else
#define	READMODE	"r"
#define	WRITEMODE	"w"
#endif

char *nodename;
int vflag = 0;
int qflag = 0;
int tflag = 0;

main(c,v)
char **v;
{
	register int i,f;
	FILE *ifd, *ofd;
	char ibuf[BUFSIZ];
	char obuf[BUFSIZ];

	f = 0;
	nodename = *v;
	setbuf(stdin, ibuf);
	setbuf(stdout, obuf);
	while (c > 1 && v[1][0] == '-') {
		c--;
		v++;
		switch (v[0][1]) {
		case 'v':
			vflag++;
			break;
		case 'q':
			qflag++;
			break;
		case 't':
			tflag++;
			break;
		case 'b':
			tflag = 0;
			break;
		default:
			fprintf(stderr, "%s:%s: Bad option\n", nodename, *v);
			exit(-1);
		}
	}

#ifdef	MSDOS
	if (setmode(fileno(stdin), O_BINARY) == -1
		|| setmode(fileno(stdout), O_BINARY) == -1) {
		fprintf(stderr, "%s: Cannot setmode on stdio\n", nodename);
		exit(1);
	}
#endif

	if (--c <= 0)
		exit(xfer(stdin, stdout, "(stdin)"));
	else {
		while (c--) {
			v++;
			if (**v == '-' && v[0][1] == 0) {
				xfer(stdin, stdout, "(stdin)");
				clearerr(stdin);
			}
			else {
				unlink(TEMPFILE);
				if ((ifd = fopen(*v, READMODE)) == NULL) {
					fprintf(stderr, "%s: %s: Can't open\n",
						nodename, *v);
					f++;
					break;
				}
				else if ((ofd = fopen(TEMPFILE, WRITEMODE))
					== NULL) {
					fprintf(stderr, "%s: %s: Can't open\n",
						nodename, TEMPFILE);
					f++;
					break;
				}
				else if (xfer(ifd, ofd, *v) == 0) {
					fclose(ifd);
					fclose(ofd);
					if (unlink(*v)) {
						fprintf(stderr,
						"%s: %s: Can't unlink\n",
						nodename, *v);
						f++;
						break;
					}
					if (rename(TEMPFILE, *v)) {
						fprintf(stderr,
						"%s: %s: Can't rename\n",
						nodename, *v);
						f++;
						break;
					}
				}
			}
		}
	}
	exit(f ? -1 : 0);
}

xfer(ifd, ofd, name)
FILE *ifd;
FILE *ofd;
char *name;
{
	if (vflag) {
		fprintf(stderr, "%s: %s", nodename, name);
		if (tflag)
			fprintf(stderr, " (text mode)\n");
		else
			fprintf(stderr, " (binary mode)\n");
	}
	if (tflag)
		return txfer(ifd, ofd, name);
	else
		return bxfer(ifd, ofd, name);
}

bxfer(ifd, ofd, name)
FILE *ifd;
FILE *ofd;
char *name;
{
	register int icnt, ocnt, ncnt, ci, co, c;

	icnt = 0;
	ocnt = 0;
	ncnt = 0;
	co = ci = c = 0;
	for (;;) {
		if (icnt <= 0) {
			if ((ci = getc(ifd)) == EOF) {
				if (ocnt > 0) {
					fprintf(stderr, "%s: 0x%x: Warning, unpaired ending nibble in file %s\n", nodename, co, name);
					putc(co, ofd);
				}
				break;
			}
			else {
				icnt = 1;
				c = ci>>4;
			}
		}
		else {
			icnt = 0;
			c = ci&017;
		}
		if (++ncnt >= 9) {
			ncnt = 0;
			if (c != 0 && !qflag)
				fprintf(stderr, "%s: 0x%x: Warning, nonzero nibble discarded from file %s\n", nodename, c, name);
			continue;
		}
		if (++ocnt >= 2) {
			co |= c;
			putc(co, ofd);
			if (ferror(ofd))
				break;
			co = 0;
			ocnt = 0;
		}
		else
			co |= c<<4;
	}
	fflush(ofd);
}

txfer(ifd, ofd, name)
FILE *ifd;
FILE *ofd;
char *name;
{
	register int icnt, ocnt, ncnt, ci, co, c;

	icnt = 0;
	ocnt = 0;
	ncnt = 0;
	co = ci = c = 0;
	for (;;) {
		if (icnt <= 0) {
			if ((ci = getc(ifd)) == EOF) {
				if (ocnt > 0) {
					fprintf(stderr, "%s: 0x%x: Warning, incomplete ending byte in file %s\n", nodename, co, name);
					putc(co, ofd);
				}
				break;
			}
			else {
				ci &= 0377;
				icnt = 7;
				c = ci>>7;
			}
		}
		else {
			icnt--;
			c = (ci>>icnt)&1;
		}
		if (++ncnt >= 36) {
			ncnt = 0;
			if (c != 0 && !qflag)
				fprintf(stderr, "%s: Warning, nonzero bit discarded from file %s\n", nodename, name);
			continue;
		}
		if (++ocnt >= 7) {
			co |= c;
			putc(co, ofd);
			if (ferror(ofd))
				break;
			co = 0;
			ocnt = 0;
		}
		else
			co |= c<<(7-ocnt);
	}
	fflush(ofd);
}

#ifdef	RENAME
rename(a, b)
char *a, *b;
{
	register int f;

	if (f = link(a, b))
		return f;
	return unlink(a);
}
#endif