Some fairly major changes:

 o -N is implemented
 o -X is implemented
 o -x is implemented
 o diff.c has been rewritten and GNU long options are now supported
 o diffdir.c has been rewritten
    + no longer does fork + exec of /usr/bin/diff
    + can be called recursively (and will be for -r)
 o diff.h
    + don't include any .h files here any more, do it in the .c files
    + no Bell Labs code in this, gets a UCB copyright (the 32v sources
      only have a diff.c and there is nothing in common).
 o diffreg.c
    + most all remaining globals are now private to diffreg.c
    + files are only opened once
    + dynamically allocated objects are either freed or realloced
    + added missing UCB copyright (there were lots of UCB changes)
    + print correct thing when -s is specified
OK deraadt@

6 files changed, 798 insertions, 723 deletions

diff --git a/usr.bin/diff/diff.1 b/usr.bin/diff/diff.1
index 3518ca11573..c555da5ddd0 100644
--- a/usr.bin/diff/diff.1
+++ b/usr.bin/diff/diff.1
@@ -1,4 +1,4 @@
-.\" $OpenBSD: diff.1,v 1.10 2003/07/04 17:50:24 millert Exp $
+.\" $OpenBSD: diff.1,v 1.11 2003/07/06 20:48:59 millert Exp $
 .\"
 .\" Copyright (c) 1980, 1990, 1993
 .\"	The Regents of the University of California.  All rights reserved.
@@ -56,14 +56,18 @@
 .Op Fl U Ar number
 .Ar file1 file2
 .Nm diff
-.Op Fl abitw
+.Op Fl abiNtw
 .Oo
 .Fl c | Fl e | Fl f |
 .Fl n | Fl u
 .Oc
+.Bk -words
 .Op Fl r
 .Op Fl s
 .Op Fl S Ar name
+.Op Fl X Ar file
+.Op Fl x Ar pattern
+.Ek
 .Ar dir1 dir2
 .Sh DESCRIPTION
 The
@@ -78,8 +82,6 @@ No output is produced if the files are identical.
 .Pp
 Output options (mutually exclusive):
 .Bl -tag -width Ds
-.It Fl a
-Treat all files as ASCII.
 .It Fl c
 Produces a diff with 3 lines of context.
 With
@@ -168,6 +170,8 @@ lines of context.
 .Pp
 Comparison options:
 .Bl -tag -width Ds
+.It Fl a
+Treat all files as ASCII.
 .It Fl b
 Causes trailing blanks (spaces and tabs) to be ignored, and other
 strings of blanks to compare equal.
@@ -197,6 +201,9 @@ will compare equal to
 .Pp
 Directory comparison options:
 .Bl -tag -width Ds
+.It Fl N
+If a file is found in only one directory, act as if it was found in the
+other directory too but was of zero size.
 .It Fl r
 Causes application of
 .Nm
@@ -210,6 +217,21 @@ Re-starts a directory
 .Nm
 in the middle, beginning with file
 .Ar name .
+.It Fl X Ar file
+Exclude files and subdirectories from comparison whose basenames match
+lines in
+.Ar file .
+Multiple
+.Fl X
+options may be specified.
+.It Fl x Ar pattern
+Exclude files and subdirectories from comparison whose basenames match
+.Ar pattern .
+Patterns are matched using shell-style globbing via
+.Xr fnmatch 3 .
+Multiple
+.Fl x
+options may be specified.
 .El
 .Pp
 If both arguments are directories,
@@ -381,7 +403,8 @@ An error occurred.
 .Xr cmp 1 ,
 .Xr comm 1 ,
 .Xr diff3 1 ,
-.Xr ed 1
+.Xr ed 1 ,
+.Xr fnmatch 3
 .Sh HISTORY
 A
 .Nm
diff --git a/usr.bin/diff/diff.c b/usr.bin/diff/diff.c
index 9a6ad78dc5b..6c570cc98e4 100644
--- a/usr.bin/diff/diff.c
+++ b/usr.bin/diff/diff.c
@@ -1,156 +1,157 @@
-/*	$OpenBSD: diff.c,v 1.22 2003/07/04 17:52:35 millert Exp $	*/
+/*	$OpenBSD: diff.c,v 1.23 2003/07/06 20:48:59 millert Exp $	*/
 
 /*
- * Copyright (C) Caldera International Inc.  2001-2002.
- * All rights reserved.
+ * Copyright (c) 2003 Todd C. Miller <Todd.Miller@courtesan.com>
  *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code and documentation must retain the above
- *    copyright notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- * 3. All advertising materials mentioning features or use of this software
- *    must display the following acknowledgement:
- *	This product includes software developed or owned by Caldera
- *	International, Inc.
- * 4. Neither the name of Caldera International, Inc. nor the names of other
- *    contributors may be used to endorse or promote products derived from
- *    this software without specific prior written permission.
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
  *
- * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA
- * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT,
- * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
- * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
- * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
- * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
- * POSSIBILITY OF SUCH DAMAGE.
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ *
+ * Sponsored in part by the Defense Advanced Research Projects
+ * Agency (DARPA) and Air Force Research Laboratory, Air Force
+ * Materiel Command, USAF, under agreement number F39502-99-1-0512.
  */
 
+#ifndef lint
+static const char rcsid[] = "$OpenBSD: diff.c,v 1.23 2003/07/06 20:48:59 millert Exp $";
+#endif /* not lint */
+
+#include <sys/param.h>
+#include <sys/stat.h>
+
+#include <err.h>
 #include <errno.h>
+#include <getopt.h>
 #include <stdlib.h>
+#include <stdio.h>
 #include <stdarg.h>
 #include <string.h>
 #include <unistd.h>
 
 #include "diff.h"
-#include "pathnames.h"
 
-#if 0
-static char const sccsid[] = "@(#)diff.c 4.7 5/11/89";
-#endif
-
-/*
- * diff - driver and subroutines
- */
-int	opt;
-int	aflag;			/* treat all files as text */
-int	tflag;			/* expand tabs on output */
-/* Algorithm related options. */
-int	bflag;			/* ignore blanks in comparisons */
-int	wflag;			/* totally ignore blanks in comparisons */
-int	iflag;			/* ignore case in comparisons */
-/* Options on hierarchical diffs. */
-int	rflag;			/* recursively trace directories */
-int	sflag;			/* announce files which are same */
-char	*start;			/* do file only if name >= this */
-/* Variable for -D D_IFDEF option. */
-char	*ifdefname;		/* What we will print for #ifdef/#endif */
-/* Variables for -c and -u context option. */
-int	context;		/* lines of context to be printed */
-/* State for exit status. */
-int	status;
-int	anychange;
-/* Variables for diffdir. */
-char	**diffargv;		/* option list to pass to recursive diffs */
+int	 aflag, bflag, iflag, Nflag, rflag, sflag, tflag, wflag;
+int	 format, context, status;
+char	*start, *ifdefname, *diffargs;
+struct stat stb1, stb2;
+struct excludes *excludes_list;
 
-/*
- * Input file names.
- * With diffdir, file1 and file2 are allocated MAXPATHLEN space,
- * and padded with a '/', and then efile1 and efile2 point after
- * the '/'.
- */
-char	*file1, *file2, *efile1, *efile2;
-struct	stat stb1, stb2;
+#define	OPTIONS	"abC:cD:efhinNrS:stU:uwX:x:"
+static struct option longopts[] = {
+	{ "text",			no_argument,		0,	'a' },
+	{ "ignore-space-change",	no_argument,		0,	'b' },
+	{ "context",			optional_argument,	0,	'C' },
+	{ "ifdef",			required_argument,	0,	'D' },
+	{ "ed",				no_argument,		0,	'e' },
+	{ "forward-ed",			no_argument,		0,	'f' },
+	{ "ignore-case",		no_argument,		0,	'i' },
+	{ "new-file",			no_argument,		0,	'N' },
+	{ "rcs",			no_argument,		0,	'n' },
+	{ "recursive",			no_argument,		0,	'r' },
+	{ "report-identical-files",	no_argument,		0,	's' },
+	{ "starting-file",		required_argument,	0,	'S' },
+	{ "expand-tabs",		no_argument,		0,	't' },
+	{ "unified",			optional_argument,	0,	'U' },
+	{ "ignore-all-space",		no_argument,		0,	'w' },
+	{ "exclude",			required_argument,	0,	'x' },
+	{ "exclude-from",		required_argument,	0,	'X' },
+};
 
 __dead void usage(void);
+void push_excludes(char *);
+void read_excludes_file(char *file);
+void set_argstr(char **, char **);
 
 int
 main(int argc, char **argv)
 {
-	int ch;
+	char *ep, **oargv;
+	long  l;
+	int   ch, gotstdin;
 
-	status = 2;
-	diffargv = argv;
+	oargv = argv;
+	gotstdin = 0;
 
-	while ((ch = getopt(argc, argv, "abC:cD:efhinrS:stU:uw")) != -1) {
+	while ((ch = getopt_long(argc, argv, OPTIONS, longopts, NULL)) != -1) {
 		switch (ch) {
 		case 'a':
-			aflag++;
+			aflag = 1;
 			break;
 		case 'b':
-			bflag++;
+			bflag = 1;
 			break;
 		case 'C':
-			opt = D_CONTEXT;
-			if (!isdigit(*optarg))
-				usage();
-			context = atoi(optarg);	/* XXX - use strtol */
-			break;
 		case 'c':
-			opt = D_CONTEXT;
-			context = 3;
+			format = D_CONTEXT;
+			if (optarg != NULL) {
+				l = strtol(optarg, &ep, 10);
+				if (*ep != '\0' || l < 0 || l >= INT_MAX)
+					usage();
+				context = (int)l;
+			} else
+				context = 3;
 			break;
 		case 'D':
-			opt = D_IFDEF;
+			format = D_IFDEF;
 			ifdefname = optarg;
 			break;
 		case 'e':
-			opt = D_EDIT;
+			format = D_EDIT;
 			break;
 		case 'f':
-			opt = D_REVERSE;
+			format = D_REVERSE;
 			break;
 		case 'h':
 			/* silently ignore for backwards compatibility */
 			break;
 		case 'i':
-			iflag++;
+			iflag = 1;
+			break;
+		case 'N':
+			Nflag = 1;
 			break;
 		case 'n':
-			opt = D_NREVERSE;
+			format = D_NREVERSE;
 			break;
 		case 'r':
-			rflag++;
+			rflag = 1;
 			break;
 		case 'S':
 			start = optarg;
 			break;
 		case 's':
-			sflag++;
+			sflag = 1;
 			break;
 		case 't':
-			tflag++;
+			tflag = 1;
 			break;
 		case 'U':
-			opt = D_UNIFIED;
-			if (!isdigit(*optarg))
-				usage();
-			context = atoi(optarg);	/* XXX - use strtol */
-			break;
 		case 'u':
-			opt = D_UNIFIED;
-			context = 3;
+			format = D_UNIFIED;
+			if (optarg != NULL) {
+				l = strtol(optarg, &ep, 10);
+				if (*ep != '\0' || l < 0 || l >= INT_MAX)
+					usage();
+				context = (int)l;
+			} else
+				context = 3;
 			break;
 		case 'w':
-			wflag++;
+			wflag = 1;
+			break;
+		case 'X':
+			read_excludes_file(optarg);
+			break;
+		case 'x':
+			push_excludes(optarg);
 			break;
 		default:
 			usage();
@@ -160,49 +161,42 @@ main(int argc, char **argv)
 	argc -= optind;
 	argv += optind;
 
+	/*
+	 * Do sanity checks, fill in stb1 and stb2 and call the appropriate
+	 * driver routine.  Both drivers use the contents of stb1 and stb2.
+	 */
 	if (argc != 2)
-		errorx("two filename arguments required");
-	file1 = argv[0];
-	file2 = argv[1];
-	if (!strcmp(file1, "-"))
+		usage();
+	if (strcmp(argv[0], "-") == 0) {
 		stb1.st_mode = S_IFREG;
-	else if (stat(file1, &stb1) < 0)
-		error("%s", file1);
-	if (!strcmp(file2, "-"))
+		gotstdin = 1;
+	} else if (stat(argv[0], &stb1) != 0)
+		error("%s", argv[0]);
+	if (strcmp(argv[1], "-") == 0) {
 		stb2.st_mode = S_IFREG;
-	else if (stat(file2, &stb2) < 0)
-		error("%s", file2);
-	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode))
-		diffdir(argv);
-	else
-		diffreg();
-	done(0);
-}
-
-int
-min(int a, int b)
-{
-
-	return (a < b ? a : b);
-}
-
-int
-max(int a, int b)
-{
-
-	return (a > b ? a : b);
+		gotstdin = 1;
+	} else if (stat(argv[1], &stb2) != 0)
+		error("%s", argv[1]);
+	if (gotstdin && (S_ISDIR(stb1.st_mode) || S_ISDIR(stb2.st_mode)))
+		errorx("can't compare - to a directory");
+	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
+		if (format == D_IFDEF)
+			errorx("-D option not supported with directories");
+		set_argstr(oargv, argv);
+		diffdir(argv[0], argv[1]);
+	} else
+		diffreg(argv[0], argv[1], 0);
+	exit(status);
 }
 
-__dead void
-done(int sig)
+void
+quit(int signo)
 {
 	if (tempfiles[0] != NULL)
 		unlink(tempfiles[0]);
 	if (tempfiles[1] != NULL)
 		unlink(tempfiles[1]);
-	if (sig)
-		_exit(status);
-	exit(status);
+	_exit(status);
 }
 
 void *
@@ -237,7 +231,7 @@ error(const char *fmt, ...)
 		unlink(tempfiles[1]);
 	errno = sverrno;
 	va_start(ap, fmt);
-	verr(status, fmt, ap);
+	verr(2, fmt, ap);
 	va_end(ap);
 }
 
@@ -251,10 +245,67 @@ errorx(const char *fmt, ...)
 	if (tempfiles[1] != NULL)
 		unlink(tempfiles[1]);
 	va_start(ap, fmt);
-	verrx(status, fmt, ap);
+	verrx(2, fmt, ap);
 	va_end(ap);
 }
 
+void
+set_argstr(char **av, char **ave)
+{
+	size_t argsize;
+	char **ap;
+
+	argsize = 4 + (char *)ave - (char *)av + 1;
+	diffargs = emalloc(argsize);
+	strlcpy(diffargs, "diff", argsize);
+	for (ap = av + 1; ap < ave; ap++) {
+		if (strcmp(*ap, "--") != 0) {
+			strlcat(diffargs, " ", argsize);
+			strlcat(diffargs, *ap, argsize);
+		}
+	}
+}
+
+/*
+ * Read in an excludes file and push each line.
+ */
+void
+read_excludes_file(char *file)
+{
+	FILE *fp;
+	char *buf, *pattern;
+	size_t len;
+
+	if (strcmp(file, "-") == 0)
+		fp = stdin;
+	else if ((fp = fopen(file, "r")) == NULL)
+		error("%s", file);
+	while ((buf = fgetln(fp, &len)) != NULL) {
+		if (buf[len - 1] == '\n')
+			len--;
+		pattern = emalloc(len + 1);
+		memcpy(pattern, buf, len);
+		pattern[len] = '\0';
+		push_excludes(pattern);
+	}
+	if (strcmp(file, "-") != 0)
+		fclose(fp);
+}
+
+/*
+ * Push a pattern onto the excludes list.
+ */
+void
+push_excludes(char *pattern)
+{
+	struct excludes *entry;
+
+	entry = emalloc(sizeof(*entry));
+	entry->pattern = pattern;
+	entry->next = excludes_list;
+	excludes_list = entry;
+}
+
 __dead void
 usage(void)
 {
@@ -263,8 +314,8 @@ usage(void)
 	    "       diff [-bitw] -C number file1 file2\n"
 	    "       diff [-bitw] -D string file1 file2\n"
 	    "       diff [-bitw] -U number file1 file2\n"
-	    "       diff [-biwt] [-c | -e | -f | -n | -u ] "
-	    "[-r] [-s] [-S name]\n            dir1 dir2\n");
+	    "       diff [-biNwt] [-c | -e | -f | -n | -u ] [-r] [-s] [-S name]"
+	    " [-X file]\n            [-x pattern] dir1 dir2\n");
 
 	exit(2);
 }
diff --git a/usr.bin/diff/diff.h b/usr.bin/diff/diff.h
index 971c9b3269a..276660f5954 100644
--- a/usr.bin/diff/diff.h
+++ b/usr.bin/diff/diff.h
@@ -1,53 +1,36 @@
-/*	$OpenBSD: diff.h,v 1.14 2003/07/04 17:50:24 millert Exp $	*/
+/*	$OpenBSD: diff.h,v 1.15 2003/07/06 20:48:59 millert Exp $	*/
 
-/*
- * Copyright (C) Caldera International Inc.  2001-2002.
- * All rights reserved.
+/*-
+ * Copyright (c) 1991, 1993
+ *	The Regents of the University of California.  All rights reserved.
  *
  * Redistribution and use in source and binary forms, with or without
  * modification, are permitted provided that the following conditions
  * are met:
- * 1. Redistributions of source code and documentation must retain the above
- *    copyright notice, this list of conditions and the following disclaimer.
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
  * 2. Redistributions in binary form must reproduce the above copyright
  *    notice, this list of conditions and the following disclaimer in the
  *    documentation and/or other materials provided with the distribution.
- * 3. All advertising materials mentioning features or use of this software
- *    must display the following acknowledgement:
- *	This product includes software developed or owned by Caldera
- *	International, Inc.
- * 4. Neither the name of Caldera International, Inc. nor the names of other
- *    contributors may be used to endorse or promote products derived from
- *    this software without specific prior written permission.
+ * 3. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
  *
- * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA
- * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT,
- * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
- * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
- * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
- * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
- * POSSIBILITY OF SUCH DAMAGE.
- */
-
-/*	diff.h	4.7	85/08/16	*/
-
-/*
- * diff - common declarations
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *	@(#)diff.h	8.1 (Berkeley) 6/6/93
  */
 
-#include <sys/param.h>
-#include <sys/stat.h>
-
-#include <ctype.h>
-#include <err.h>
-#include <signal.h>
-#include <stdio.h>
-
 /*
  * Output format options
  */
@@ -60,21 +43,31 @@
 #define	D_NREVERSE	5	/* Reverse ed script with numbered
 				   lines and no trailing . */
 
-extern int	aflag, bflag, iflag, rflag, sflag, tflag, wflag;
+/*
+ * Output flags
+ */
+#define	D_HEADER	1	/* Print a header/footer between files */
+#define	D_EMPTY1	2	/* Treat first file as empty (/dev/null) */
+#define	D_EMPTY2	4	/* Treat second file as empty (/dev/null) */
+
+struct excludes {
+	char *pattern;
+	struct excludes *next;
+};
+
+extern int	aflag, bflag, iflag, Nflag, rflag, sflag, tflag, wflag;
 extern char	*start, *ifdefname;
-extern int	opt, wantelses, context, status, anychange;
-extern char	*tempfiles[], **diffargv;
-extern char	*file1, *file2, *efile1, *efile2;
+extern int	format, context, status, anychange;
+extern char	*tempfiles[], *diffargs;
 extern struct	stat stb1, stb2;
+extern struct	excludes *excludes_list;
 
+char	*copytemp(const char *, int);
+char	*splice(char *, char *);
 void	*emalloc(size_t);
 void	*erealloc(void *, size_t);
-char	*splice(char *, char *);
-char	*copytemp(const char *, int);
-void	diffdir(char **);
-void	diffreg(void);
-int	max(int, int);
-int	min(int, int);
+void	diffdir(char *, char *);
+void	diffreg(char *, char *, int);
+void	quit(int);
 __dead void error(const char *, ...);
 __dead void errorx(const char *, ...);
-__dead void done(int);
diff --git a/usr.bin/diff/diffdir.c b/usr.bin/diff/diffdir.c
index a197b3e357d..6b3ebd01864 100644
--- a/usr.bin/diff/diffdir.c
+++ b/usr.bin/diff/diffdir.c
@@ -1,365 +1,292 @@
-/*	$OpenBSD: diffdir.c,v 1.18 2003/07/06 02:11:12 millert Exp $	*/
+/*	$OpenBSD: diffdir.c,v 1.19 2003/07/06 20:48:59 millert Exp $	*/
 
 /*
- * Copyright (C) Caldera International Inc.  2001-2002.
- * All rights reserved.
+ * Copyright (c) 2003 Todd C. Miller <Todd.Miller@courtesan.com>
  *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code and documentation must retain the above
- *    copyright notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- * 3. All advertising materials mentioning features or use of this software
- *    must display the following acknowledgement:
- *	This product includes software developed or owned by Caldera
- *	International, Inc.
- * 4. Neither the name of Caldera International, Inc. nor the names of other
- *    contributors may be used to endorse or promote products derived from
- *    this software without specific prior written permission.
+ * Permission to use, copy, modify, and distribute this software for any
+ * purpose with or without fee is hereby granted, provided that the above
+ * copyright notice and this permission notice appear in all copies.
  *
- * USE OF THE SOFTWARE PROVIDED FOR UNDER THIS LICENSE BY CALDERA
- * INTERNATIONAL, INC. AND CONTRIBUTORS ``AS IS'' AND ANY EXPRESS OR
- * IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES
- * OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED.
- * IN NO EVENT SHALL CALDERA INTERNATIONAL, INC. BE LIABLE FOR ANY DIRECT,
- * INDIRECT INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES
- * (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR
- * SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT,
- * STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING
- * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
- * POSSIBILITY OF SUCH DAMAGE.
+ * THE SOFTWARE IS PROVIDED "AS IS" AND THE AUTHOR DISCLAIMS ALL WARRANTIES
+ * WITH REGARD TO THIS SOFTWARE INCLUDING ALL IMPLIED WARRANTIES OF
+ * MERCHANTABILITY AND FITNESS. IN NO EVENT SHALL THE AUTHOR BE LIABLE FOR
+ * ANY SPECIAL, DIRECT, INDIRECT, OR CONSEQUENTIAL DAMAGES OR ANY DAMAGES
+ * WHATSOEVER RESULTING FROM LOSS OF USE, DATA OR PROFITS, WHETHER IN AN
+ * ACTION OF CONTRACT, NEGLIGENCE OR OTHER TORTIOUS ACTION, ARISING OUT OF
+ * OR IN CONNECTION WITH THE USE OR PERFORMANCE OF THIS SOFTWARE.
+ *
+ * Sponsored in part by the Defense Advanced Research Projects
+ * Agency (DARPA) and Air Force Research Laboratory, Air Force
+ * Materiel Command, USAF, under agreement number F39502-99-1-0512.
  */
 
-#include <sys/types.h>
-#include <sys/wait.h>
+#ifndef lint
+static const char rcsid[] = "$OpenBSD: diffdir.c,v 1.19 2003/07/06 20:48:59 millert Exp $";
+#endif /* not lint */
+
+#include <sys/param.h>
+#include <sys/stat.h>
 
 #include <dirent.h>
+#include <err.h>
 #include <errno.h>
 #include <fcntl.h>
+#include <fnmatch.h>
+#include <paths.h>
 #include <stdio.h>
 #include <stdlib.h>
 #include <string.h>
 #include <unistd.h>
 
 #include "diff.h"
-#include "pathnames.h"
 
-#if 0
-static const char sccsid[] = "@(#)diffdir.c	4.12 (Berkeley) 4/30/89";
-#endif
+static int dircompare(const void *, const void *);
+static int excluded(const char *);
+static struct dirent **slurpdir(char *, char **);
+static void diffit(struct dirent *, char *, size_t, char *, size_t);
 
 /*
- * diff - directory comparison
+ * Diff directory traveral. Will be called recursively if -r was specified.
  */
-#define	d_flags	d_ino
-
-#define	DIRECT	1		/* Directory */
-
-struct dir {
-	u_long d_ino;
-	short d_reclen;
-	short d_namlen;
-	char *d_entry;
-};
+void
+diffdir(char *p1, char *p2)
+{
+	struct dirent **dirp1, **dirp2, **dp1, **dp2;
+	struct dirent *dent1, *dent2;
+	size_t dirlen1, dirlen2;
+	char path1[MAXPATHLEN], path2[MAXPATHLEN];
+	char *dirbuf1, *dirbuf2;
+	int pos;
 
-static int dirstatus;		/* exit status from diffdir */
-static char title[2 * BUFSIZ];
+	dirlen1 = strlcpy(path1, *p1 ? p1 : ".", sizeof(path1));
+	if (dirlen1 >= sizeof(path1) - 1) {
+		warnx("%s: %s", p1, strerror(ENAMETOOLONG));
+		status = 2;
+		return;
+	}
+	if (path1[dirlen1 - 1] != '/') {
+		path1[dirlen1++] = '/';
+		path1[dirlen1] = '\0';
+	}
+	dirlen2 = strlcpy(path2, *p2 ? p2 : ".", sizeof(path2));
+	if (dirlen2 >= sizeof(path2) - 1) {
+		warnx("%s: %s", p2, strerror(ENAMETOOLONG));
+		status = 2;
+		return;
+	}
+	if (path2[dirlen2 - 1] != '/') {
+		path2[dirlen2++] = '/';
+		path2[dirlen2] = '\0';
+	}
 
+	/* get a list of the entries in each directory */
+	dp1 = dirp1 = slurpdir(path1, &dirbuf1);
+	dp2 = dirp2 = slurpdir(path2, &dirbuf2);
+	if (dirp1 == NULL || dirp2 == NULL)
+		return;
 
-static struct dir *setupdir(char *);
-static int ascii(int);
-static void compare(struct dir *);
-static void calldiff(void);
-static void setfile(char **fpp, char **epp, char *file);
-static int useless(char *);
-static void only(struct dir *dp, int which);
-static int entcmp(const void *, const void *);
+	/*
+	 * If we were given a starting point, find it.
+	 */
+	if (start != NULL) {
+		while (*dp1 != NULL && strcmp((*dp1)->d_name, start) < 0)
+			dp1++;
+		while (*dp2 != NULL && strcmp((*dp2)->d_name, start) < 0)
+			dp2++;
+	}
 
-void
-diffdir(char **argv)
-{
-	struct dir *dir1, *dir2;
-	struct dir *d1, *d2;
-	int i, cmp;
+	/*
+	 * Iterate through the two directory lists, diffing as we go.
+	 */
+	while (*dp1 != NULL || *dp2 != NULL) {
+		dent1 = *dp1;
+		dent2 = *dp2;
 
-	if (opt == D_IFDEF)
-		warnx("can't specify -I with directories");
-	if (opt == D_EDIT && sflag)
-		warnx("warning: shouldn't give -s with -e");
-	strlcpy(title, "diff ", sizeof title);
-	for (i = 1; diffargv[i + 2]; i++) {
-		if (!strcmp(diffargv[i], "-"))
-			continue;	/* was -S, dont look silly */
-		strlcat(title, diffargv[i], sizeof title);
-		strlcat(title, " ", sizeof title);
-	}
-	setfile(&file1, &efile1, file1);
-	setfile(&file2, &efile2, file2);
-	argv[0] = file1;
-	argv[1] = file2;
-	dir1 = setupdir(file1);
-	dir2 = setupdir(file2);
-	d1 = dir1;
-	d2 = dir2;
-	while (d1->d_entry != 0 || d2->d_entry != 0) {
-		if (d1->d_entry && useless(d1->d_entry)) {
-			d1++;
-			continue;
-		}
-		if (d2->d_entry && useless(d2->d_entry)) {
-			d2++;
-			continue;
-		}
-		if (d1->d_entry == 0)
-			cmp = 1;
-		else if (d2->d_entry == 0)
-			cmp = -1;
-		else
-			cmp = strcmp(d1->d_entry, d2->d_entry);
-		if (cmp < 0) {
-			if (opt == D_NORMAL || opt == D_CONTEXT ||
-			    opt == D_UNIFIED)
-				only(d1, 1);
-			d1++;
-			dirstatus |= 1;
-		} else if (cmp == 0) {
-			compare(d1);
-			d1++;
-			d2++;
+		pos = dent1 == NULL ? 1 : dent2 == NULL ? -1 :
+		    strcmp(dent1->d_name, dent2->d_name);
+		if (pos == 0) {
+			/* file exists in both dirs, diff it */
+			diffit(dent1, path1, dirlen1, path2, dirlen2);
+			dp1++;
+			dp2++;
+		} else if (pos < 0) {
+			/* file only in first dir, only diff if -N */
+			if (Nflag)
+				diffit(dent1, path1, dirlen1, path2, dirlen2);
+			else if (format == D_NORMAL || format == D_CONTEXT ||
+			    format == D_UNIFIED)
+				/* XXX GNU diff always prints this XXX */
+				printf("Only in %.*s: %s\n", (int)(dirlen1 - 1),
+				    path1, dent1->d_name);
+			dp1++;
 		} else {
-			if (opt == D_NORMAL || opt == D_CONTEXT ||
-			    opt == D_UNIFIED)
-				only(d2, 2);
-			d2++;
-			dirstatus |= 1;
+			/* file only in second dir, only diff if -N */
+			if (Nflag)
+				diffit(dent2, path1, dirlen1, path2, dirlen2);
+			else if (format == D_NORMAL || format == D_CONTEXT ||
+			    format == D_UNIFIED)
+				/* XXX GNU diff always prints this XXX */
+				printf("Only in %.*s: %s\n", (int)(dirlen2 - 1),
+				    path2, dent2->d_name);
+			dp2++;
 		}
 	}
-	if (rflag) {
-		for (d1 = dir1; d1->d_entry; d1++) {
-			if ((d1->d_flags & DIRECT) == 0)
-				continue;
-			strlcpy(efile1, d1->d_entry,
-			    file1 + MAXPATHLEN - efile1);
-			strlcpy(efile2, d1->d_entry,
-			    file2 + MAXPATHLEN - efile2);
-			calldiff();
-		}
+
+	if (dirbuf1 != NULL) {
+		free(dirp1);
+		free(dirbuf1);
+	}
+	if (dirbuf2 != NULL) {
+		free(dirp2);
+		free(dirbuf2);
 	}
-	status = dirstatus;
 }
 
-void
-setfile(char **fpp, char **epp, char *file)
+/*
+ * Read in a whole directory's worth of struct dirents, culling
+ * out the "excluded" ones.
+ * Returns an array of struct dirent *'s that point into the buffer
+ * returned via bufp.  Caller is responsible for free()ing both of these.
+ */
+static struct dirent **
+slurpdir(char *path, char **bufp)
 {
-	char *cp;
-	size_t len;
+	char *buf, *ebuf, *cp;
+	size_t bufsize;
+	long base;
+	int fd, nbytes, entries;
+	struct stat sb;
+	struct dirent **dirlist, *dp;
 
-	if (*file == '\0')
-		file = ".";
-	*fpp = emalloc(MAXPATHLEN);
-	len = strlcpy(*fpp, file, MAXPATHLEN);
-	if (len >= MAXPATHLEN - 1)
-		errorx("%s: %s", file, strerror(ENAMETOOLONG));
-	cp = *fpp + len - 1;
-	if (*cp == '/')
-		++cp;
-	else {
-		*++cp = '/';
-		*++cp = '\0';
-	}
-	*epp = cp;
-}
+	*bufp = NULL;
+	if ((fd = open(path, O_RDONLY, 0644)) == -1) {
+		static struct dirent *dummy;
 
-void
-only(struct dir *dp, int which)
-{
-	char *file = which == 1 ? file1 : file2;
-	char *efile = which == 1 ? efile1 : efile2;
+		if (!Nflag) {
+			warn("%s", path);
+			return (NULL);
+		}
+		return (&dummy);
+	}
+	fstat(fd, &sb);
 
-	printf("Only in %.*s: %s\n", (int)(efile - file - 1), file, dp->d_entry);
-}
+	bufsize = sb.st_size;
+	if (bufsize < sb.st_blksize)
+		bufsize = sb.st_blksize;
+	buf = emalloc(bufsize);
 
-struct dir *
-setupdir(char *cp)
-{
-	struct dir *dp, *ep;
-	struct dirent *rp;
-	int nitems;
-	DIR *dirp;
+	nbytes = getdirentries(fd, buf, bufsize, &base);
+	if (nbytes <= 0) {
+		free(buf);
+		warn("%s", path);
+		return (NULL);
+	}
+	ebuf = buf + nbytes;
+	close(fd);
 
-	dirp = opendir(cp);
-	if (dirp == NULL)
-		error("%s", cp);
-	nitems = 0;
-	dp = emalloc(sizeof(struct dir));
-	while ((rp = readdir(dirp))) {
-		ep = &dp[nitems++];
-		ep->d_reclen = rp->d_reclen;
-		ep->d_namlen = rp->d_namlen;
-		ep->d_entry = 0;
-		ep->d_flags = 0;
-		if (ep->d_namlen > 0) {
-			ep->d_entry = emalloc(ep->d_namlen + 1);
-			strlcpy(ep->d_entry, rp->d_name, ep->d_namlen + 1);
-		}
-		dp = erealloc(dp, (nitems + 1) * sizeof(struct dir));
+	/*
+	 * We now have all the directory entries in our buffer.
+	 * However, in order to easily sort them we need to convert
+	 * the buffer into an array.
+	 */
+	for (entries = 0, cp = buf; cp < ebuf; ) {
+		dp = (struct dirent *)cp;
+		if (dp->d_fileno != 0 && dp->d_type != DT_WHT)
+			entries++;
+		if (dp->d_reclen <= 0)
+			break;
+		cp += dp->d_reclen;
+	}
+	dirlist = emalloc(sizeof(struct dirent *) * (entries + 1));
+	for (entries = 0, cp = buf; cp < ebuf; ) {
+		dp = (struct dirent *)cp;
+		if (dp->d_fileno != 0 && dp->d_type != DT_WHT &&
+		    !excluded(dp->d_name))
+			dirlist[entries++] = dp;
+		if (dp->d_reclen <= 0)
+			break;
+		cp += dp->d_reclen;
 	}
-	dp[nitems].d_entry = 0;	/* delimiter */
-	closedir(dirp);
-	qsort(dp, nitems, sizeof(struct dir), entcmp);
-	return (dp);
+	dirlist[entries] = NULL;
+
+	qsort(dirlist, entries, sizeof(struct dir *), dircompare);
+
+	*bufp = buf;
+	return (dirlist);
 }
 
+/*
+ * Compare d_name in two dirent structures; for qsort(3).
+ */
 static int
-entcmp(const void *v1, const void *v2)
+dircompare(const void *vp1, const void *vp2)
 {
-	const struct dir *d1, *d2;
+	struct dirent *dp1 = *((struct dirent **) vp1);
+	struct dirent *dp2 = *((struct dirent **) vp2);
 
-	d1 = v1;
-	d2 = v2;
-	return (strcmp(d1->d_entry, d2->d_entry));
+	return (strcmp(dp1->d_name, dp2->d_name));
 }
 
+/*
+ * Do the actual diff by calling either diffreg() or diffdir().
+ */
 static void
-compare(struct dir *dp)
+diffit(struct dirent *dp, char *path1, size_t plen1, char *path2, size_t plen2)
 {
-	char buf1[BUFSIZ], buf2[BUFSIZ];
-	int i, j, f1, f2, fmt1, fmt2;
-	struct stat stb1, stb2;
+	int flags = D_HEADER;
 
-	strlcpy(efile1, dp->d_entry, file1 + MAXPATHLEN - efile1);
-	strlcpy(efile2, dp->d_entry, file2 + MAXPATHLEN - efile2);
-	f1 = open(file1, 0);
-	if (f1 < 0) {
-		warn("%s", file1);
-		return;
-	}
-	f2 = open(file2, 0);
-	if (f2 < 0) {
-		warn("%s", file2);
-		close(f1);
-		return;
-	}
-	fstat(f1, &stb1);
-	fstat(f2, &stb2);
-	fmt1 = stb1.st_mode & S_IFMT;
-	fmt2 = stb2.st_mode & S_IFMT;
-	if (fmt1 != S_IFREG || fmt2 != S_IFREG) {
-		if (fmt1 == fmt2) {
-			if (fmt1 != S_IFDIR && stb1.st_rdev == stb2.st_rdev)
-				goto same;
-			if (fmt1 == S_IFDIR) {
-				dp->d_flags = DIRECT;
-				if (opt == D_EDIT)
-					goto closem;
-				printf("Common subdirectories: %s and %s\n",
-				    file1, file2);
-				goto closem;
-			}
+	strlcpy(path1 + plen1, dp->d_name, MAXPATHLEN - plen1);
+	if (stat(path1, &stb1) != 0) {
+		if (!Nflag || errno != ENOENT) {
+			warn("%s", path1);
+			return;
 		}
-		goto notsame;
-	}
-	if (stb1.st_size != stb2.st_size)
-		goto notsame;
-	for (;;) {
-		i = read(f1, buf1, BUFSIZ);
-		j = read(f2, buf2, BUFSIZ);
-		if (i < 0 || j < 0 || i != j)
-			goto notsame;
-		if (i == 0 && j == 0)
-			goto same;
-		for (j = 0; j < i; j++)
-			if (buf1[j] != buf2[j])
-				goto notsame;
-	}
-same:
-	if (sflag != 0)
-		printf("Files %s and %s are identical\n", file1, file2);
-	goto closem;
-notsame:
-	dirstatus |= 1;
-	if (!ascii(f1) || !ascii(f2)) {
-		if (opt == D_NORMAL || opt == D_CONTEXT || opt == D_UNIFIED)
-			printf("Binary files %s and %s differ\n",
-			    file1, file2);
-		goto closem;
+		flags |= D_EMPTY1;
+		memset(&stb1, 0, sizeof(stb1));
 	}
-	close(f1);
-	close(f2);
-	anychange = 1;
-	if (opt == D_EDIT) {
-		printf("ed - %s << '-*-END-*-'\n", dp->d_entry);
-		calldiff();
-	} else {
-		printf("%s%s %s\n", title, file1, file2);
-		calldiff();
-	}
-	if (opt == D_EDIT)
-		printf("w\nq\n-*-END-*-\n");
-	return;
-closem:
-	close(f1);
-	close(f2);
-}
 
-static void
-calldiff(void)
-{
-	int lstatus;
-	pid_t pid;
+	strlcpy(path2 + plen2, dp->d_name, MAXPATHLEN - plen2);
+	if (stat(path2, &stb2) != 0) {
+		if (!Nflag || errno != ENOENT) {
+			warn("%s", path2);
+			return;
+		}
+		flags |= D_EMPTY2;
+		memset(&stb2, 0, sizeof(stb2));
+		stb2.st_mode = stb1.st_mode;
+	}
+	if (stb1.st_mode == 0)
+		stb1.st_mode = stb2.st_mode;
 
-	fflush(stdout);
-	pid = fork();
-	if (pid == -1)
-		errorx("No more processes");
-	if (pid == 0) {
-		execv(_PATH_DIFF, diffargv);
-		error("%s", _PATH_DIFF);
+	if (S_ISDIR(stb1.st_mode) && S_ISDIR(stb2.st_mode)) {
+		/* XXX GNU diff always prints this for dirs XXX */
+		if (format != D_EDIT)
+			printf("Common subdirectories: %s and %s\n",
+			    path1, path2);
+		if (rflag)
+			diffdir(path1, path2);
+		return;
 	}
-	while (wait(&lstatus) != pid)
-		continue;
-	/*
-		if ((lstatus >> 8) >= 2)
-			done(0);
-	*/
-	dirstatus |= lstatus >> 8;
+	diffreg(path1, path2, flags);
 }
 
-int
-ascii(int f)
+/*
+ * Exclude the given directory entry?
+ */
+static int
+excluded(const char *entry)
 {
-	char buf[BUFSIZ], *cp;
-	int cnt;
+	struct excludes *excl;
 
-	if (aflag)
+	/* always skip "." and ".." */
+	if (entry[0] == '.' &&
+	    (entry[1] == '\0' || (entry[1] == '.' && entry[2] == '\0')))
 		return (1);
 
-	lseek(f, (off_t)0, SEEK_SET);
-	cnt = read(f, buf, BUFSIZ);
-	cp = buf;
-	while (--cnt >= 0)
-		if (*cp++ & 0200)
-			return (0);
-	return (1);
-}
+	/* check excludes list */
+	for (excl = excludes_list; excl != NULL; excl = excl->next)
+		if (fnmatch(excl->pattern, entry, FNM_PATHNAME) == 0)
+			return (1);
 
-/*
- * THIS IS CRUDE.
- */
-int
-useless(char *cp)
-{
-	if (cp[0] == '.') {
-		if (cp[1] == '\0')
-			return (1);	/* directory "." */
-		if (cp[1] == '.' && cp[2] == '\0')
-			return (1);	/* directory ".." */
-	}
-	if (start && strcmp(start, cp) > 0)
-		return (1);
 	return (0);
 }
diff --git a/usr.bin/diff/diffreg.c b/usr.bin/diff/diffreg.c
index c8dbc50c5d5..3a5852125e5 100644
--- a/usr.bin/diff/diffreg.c
+++ b/usr.bin/diff/diffreg.c
@@ -1,4 +1,4 @@
-/*	$OpenBSD: diffreg.c,v 1.26 2003/07/04 17:37:07 millert Exp $	*/
+/*	$OpenBSD: diffreg.c,v 1.27 2003/07/06 20:48:59 millert Exp $	*/
 
 /*
  * Copyright (C) Caldera International Inc.  2001-2002.
@@ -33,20 +33,56 @@
  * IN ANY WAY OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE
  * POSSIBILITY OF SUCH DAMAGE.
  */
+/*-
+ * Copyright (c) 1991, 1993
+ *	The Regents of the University of California.  All rights reserved.
+ *
+ * Redistribution and use in source and binary forms, with or without
+ * modification, are permitted provided that the following conditions
+ * are met:
+ * 1. Redistributions of source code must retain the above copyright
+ *    notice, this list of conditions and the following disclaimer.
+ * 2. Redistributions in binary form must reproduce the above copyright
+ *    notice, this list of conditions and the following disclaimer in the
+ *    documentation and/or other materials provided with the distribution.
+ * 3. Neither the name of the University nor the names of its contributors
+ *    may be used to endorse or promote products derived from this software
+ *    without specific prior written permission.
+ *
+ * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
+ * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
+ * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
+ * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
+ * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
+ * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
+ * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
+ * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
+ * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
+ * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
+ * SUCH DAMAGE.
+ *
+ *	@(#)diffreg.c   8.1 (Berkeley) 6/6/93
+ */
+
+#ifndef lint
+static const char rcsid[] = "$OpenBSD: diffreg.c,v 1.27 2003/07/06 20:48:59 millert Exp $";
+#endif /* not lint */
 
 #include <sys/types.h>
+#include <sys/stat.h>
 
-#include <stdlib.h>
-#include <unistd.h>
+#include <ctype.h>
+#include <err.h>
 #include <fcntl.h>
+#include <libgen.h>
+#include <paths.h>
+#include <signal.h>
+#include <stdio.h>
+#include <stdlib.h>
 #include <string.h>
+#include <unistd.h>
 
 #include "diff.h"
-#include "pathnames.h"
-
-#if 0
-static char const sccsid[] = "@(#)diffreg.c 4.21 4/6/90";
-#endif
 
 /*
  * diff - compare two files.
@@ -115,10 +151,6 @@ static char const sccsid[] = "@(#)diffreg.c 4.21 4/6/90";
  *	6n words for files of length n.
  */
 
-#define	prints(s)	fputs(s,stdout)
-
-FILE *input[2];
-
 struct cand {
 	int x;
 	int y;
@@ -130,40 +162,42 @@ struct line {
 	int value;
 } *file[2];
 
-int len[2];
-struct line *sfile[2];	/* shortened by pruning common prefix and suffix */
-int slen[2];
-int pref, suff;			/* length of prefix and suffix */
-int inifdef;			/* whether or not we are in a #ifdef block */
-int *class;			/* will be overlaid on file[0] */
-int *member;			/* will be overlaid on file[1] */
-int *klist;			/* will be overlaid on file[0] after class */
-struct cand *clist;		/* merely a free storage pot for candidates */
-int clen = 0;
-int *J;				/* will be overlaid on class */
-long *ixold;			/* will be overlaid on klist */
-long *ixnew;			/* will be overlaid on file[1] */
-u_char *chrtran;		/* translation table for case-folding */
+static int  *J;			/* will be overlaid on class */
+static int  *class;		/* will be overlaid on file[0] */
+static int  *klist;		/* will be overlaid on file[0] after class */
+static int  *member;		/* will be overlaid on file[1] */
+static int   clen;
+static int   inifdef;		/* whether or not we are in a #ifdef block */
+static int   len[2];
+static int   pref, suff;	/* length of prefix and suffix */
+static int   slen[2];
+static int   anychange;
+static long *ixnew;		/* will be overlaid on file[1] */
+static long *ixold;		/* will be overlaid on klist */
+static struct cand *clist;	/* merely a free storage pot for candidates */
+static struct line *sfile[2];	/* shortened by pruning common prefix/suffix */
+static u_char *chrtran;		/* translation table for case-folding */
 
 static void fetch(long *, int, int, FILE *, char *, int);
-static void output(void);
-static void check(void);
+static void output(char *, FILE *, char *, FILE *);
+static void check(char *, FILE *, char *, FILE *);
 static void range(int, int, char *);
-static void dump_context_vec(void);
-static void dump_unified_vec(void);
+static void dump_context_vec(FILE *, FILE *);
+static void dump_unified_vec(FILE *, FILE *);
 static void prepare(int, FILE *);
 static void prune(void);
 static void equiv(struct line *, int, struct line *, int, int *);
 static void unravel(int);
 static void unsort(struct line *, int, int *);
-static void change(int, int, int, int);
+static void change(char *, FILE *, char *, FILE *, int, int, int, int);
 static void sort(struct line *, int);
-static int newcand(int, int, int);
-static int search(int *, int, int);
-static int skipline(int);
-static int asciifile(FILE *);
-static int stone(int *, int, int *, int *);
-static int readhash(FILE *);
+static int  asciifile(FILE *);
+static int  newcand(int, int, int);
+static int  search(int *, int, int);
+static int  skipline(FILE *);
+static int  stone(int *, int, int *, int *);
+static int  readhash(FILE *);
+static int  files_differ(FILE *, FILE *, int);
 
 /*
  * chrtran points to one of 2 translation tables: cup2low if folding upper to
@@ -224,70 +258,100 @@ u_char cup2low[256] = {
 };
 
 void
-diffreg(void)
+diffreg(char *ofile1, char *ofile2, int flags)
 {
-	char buf1[BUFSIZ], buf2[BUFSIZ];
-	FILE *f1, *f2;
-	int i, j;
+	char *file1 = ofile1;
+	char *file2 = ofile2;
+	FILE *f1 = NULL;
+	FILE *f2 = NULL;
+	int i;
 
+	anychange = 0;
 	chrtran = (iflag ? cup2low : clow2low);
 	if (strcmp(file1, "-") == 0 && strcmp(file2, "-") == 0)
-		errorx("can't specify - -");
-	if (S_ISDIR(stb1.st_mode)) {
-		file1 = splice(file1, file2);
-		if (stat(file1, &stb1) < 0)
-			error("%s", file1);
-	} else if (strcmp(file1, "-") == 0 ||
-	    (!S_ISREG(stb1.st_mode) && strcmp(file1, _PATH_DEVNULL) != 0)) {
-		file1 = copytemp(file1, 1);
-		if (stat(file1, &stb1) < 0)
-			error("%s", file1);
+		goto notsame;
+
+	/* XXX - only make temp file for stdin if not seekable? (millert) */
+	if (flags & D_EMPTY1)
+		f1 = fopen(_PATH_DEVNULL, "r");
+	else {
+		if (S_ISDIR(stb1.st_mode)) {
+			file1 = splice(file1, file2);
+			if (stat(file1, &stb1) < 0) {
+				warn("%s", file1);
+				status |= 2;
+				goto closem;
+			}
+		} else if (strcmp(file1, "-") == 0 || !S_ISREG(stb1.st_mode)) {
+			file1 = copytemp(file1, 1);
+			if (file1 == NULL || stat(file1, &stb1) < 0) {
+				warn("%s", file1);
+				status |= 2;
+				goto closem;
+			}
+		}
+		f1 = fopen(file1, "r");
 	}
-	if (S_ISDIR(stb2.st_mode)) {
-		file2 = splice(file2, file1);
-		if (stat(file2, &stb2) < 0)
-			error("%s", file2);
-	} else if (strcmp(file2, "-") == 0 ||
-	    (!S_ISREG(stb2.st_mode) && strcmp(file2, _PATH_DEVNULL) != 0)) {
-		file2 = copytemp(file2, 2);
-		if (stat(file2, &stb2) < 0)
-			error("%s", file2);
+	if (f1 == NULL) {
+		warn("%s", file1);
+		status |= 2;
+		goto closem;
 	}
-	if ((f1 = fopen(file1, "r")) == NULL)
-		error("%s", file1);
-	if ((f2 = fopen(file2, "r")) == NULL)
-		error("%s", file2);
-	if ((stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT) ||
-	    stb1.st_size != stb2.st_size)
-		goto notsame;
-	for (;;) {
-		i = fread(buf1, 1, BUFSIZ, f1);
-		j = fread(buf2, 1, BUFSIZ, f2);
-		if (i < 0 || j < 0 || i != j)
-			goto notsame;
-		if (i == 0 && j == 0) {
-			fclose(f1);
-			fclose(f2);
-			status = 0;	/* files don't differ */
-			goto same;
+
+	if (flags & D_EMPTY2)
+		f2 = fopen(_PATH_DEVNULL, "r");
+	else {
+		if (S_ISDIR(stb2.st_mode)) {
+			file2 = splice(file2, file1);
+			if (stat(file2, &stb2) < 0) {
+				warn("%s", file2);
+				status |= 2;
+				goto closem;
+			}
+		} else if (strcmp(file2, "-") == 0 || !S_ISREG(stb2.st_mode)) {
+			file2 = copytemp(file2, 2);
+			if (file2 == NULL || stat(file2, &stb2) < 0) {
+				warn("%s", file2);
+				status |= 2;
+				goto closem;
+			}
 		}
-		for (j = 0; j < i; j++)
-			if (buf1[j] != buf2[j])
-				goto notsame;
+		f2 = fopen(file2, "r");
+	}
+	if (f2 == NULL) {
+		warn("%s", file2);
+		status |= 2;
+		goto closem;
 	}
+
+	switch (files_differ(f1, f2, flags)) {
+	case 0:
+		goto same;
+	case 1:
+		break;
+	default:
+		/* error */
+		status |= 2;
+		goto closem;
+	}
+
 notsame:
 	/*
 	 * Files certainly differ at this point; set status accordingly
 	 */
-	status = 1;
+	status |= 1;
+	if (flags & D_HEADER) {
+		if (format == D_EDIT)
+			printf("ed - %s << '-*-END-*-'\n", basename(file1));
+		else
+			printf("%s %s %s\n", diffargs, file1, file2);
+	}
 	if (!asciifile(f1) || !asciifile(f2)) {
 		printf("Binary files %s and %s differ\n", file1, file2);
-		exit(status);
+		goto closem;
 	}
 	prepare(0, f1);
 	prepare(1, f2);
-	fclose(f1);
-	fclose(f2);
 	prune();
 	sort(sfile[0], slen[0]);
 	sort(sfile[1], slen[1]);
@@ -306,23 +370,74 @@ notsame:
 	free(member);
 	free(class);
 
-	J = emalloc((len[0] + 2) * sizeof(int));
+	J = erealloc(J, (len[0] + 2) * sizeof(int));
 	unravel(klist[i]);
 	free(clist);
 	free(klist);
 
-	ixold = emalloc((len[0] + 2) * sizeof(long));
-	ixnew = emalloc((len[1] + 2) * sizeof(long));
-	check();
-	output();
-	status = anychange;
+	ixold = erealloc(ixold, (len[0] + 2) * sizeof(long));
+	ixnew = erealloc(ixnew, (len[1] + 2) * sizeof(long));
+	check(file1, f1, file2, f2);
+	output(file1, f1, file2, f2);
+	if ((flags & D_HEADER) && format == D_EDIT)
+		printf("w\nq\n-*-END-*-\n");
 same:
-	if (anychange == 0 && (opt == D_CONTEXT || opt == D_UNIFIED))
-		printf("No differences encountered\n");
+	if (anychange == 0 && sflag != 0)
+		printf("Files %s and %s are identical\n", file1, file2);
+
+closem:
+	if (f1 != NULL)
+		fclose(f1);
+	if (f2 != NULL)
+		fclose(f2);
+	if (tempfiles[0] != NULL) {
+		unlink(tempfiles[0]);
+		free(tempfiles[0]);
+		tempfiles[0] = NULL;
+	}
+	if (tempfiles[1] != NULL) {
+		unlink(tempfiles[1]);
+		free(tempfiles[1]);
+		tempfiles[1] = NULL;
+	}
+	if (file1 != ofile1)
+		free(file1);
+	if (file2 != ofile2)
+		free(file2);
+}
+
+/*
+ * Check to see if the given files differ.
+ * Returns 0 if they are the same, 1 if different, and -1 on error.
+ * XXX - could use code from cmp(1) [faster]
+ */
+static int
+files_differ(FILE *f1, FILE *f2, int flags)
+{
+	char buf1[BUFSIZ], buf2[BUFSIZ];
+	size_t i, j;
+
+	if ((flags & (D_EMPTY1|D_EMPTY2)) || stb1.st_size != stb2.st_size ||
+	    (stb1.st_mode & S_IFMT) != (stb2.st_mode & S_IFMT))
+		return (1);
+	for (;;) {
+		i = fread(buf1, 1, sizeof(buf1), f1);
+		j = fread(buf2, 1, sizeof(buf2), f2);
+		if (i != j)
+			return (1);
+		if (i == 0 && j == 0) {
+			if (ferror(f1) || ferror(f2))
+				return (1);
+			return (0);
+		}
+		if (memcmp(buf1, buf2, i) != 0)
+			return (1);
+	}
 }
 
 char *tempfiles[2];
 
+/* XXX - pass back a FILE * too (millert) */
 char *
 copytemp(const char *file, int n)
 {
@@ -335,24 +450,24 @@ copytemp(const char *file, int n)
 	if (strcmp(file, "-") == 0)
 		ifd = STDIN_FILENO;
 	else if ((ifd = open(file, O_RDONLY, 0644)) < 0)
-		error("%s", file);
+		return (NULL);
 
 	if ((tempdir = getenv("TMPDIR")) == NULL)
 		tempdir = _PATH_TMP;
 	if (asprintf(&tempfile, "%s/diff%d.XXXXXXXX", tempdir, n) == -1)
-		error(NULL);
+		return (NULL);
 	tempfiles[n - 1] = tempfile;
 
-	signal(SIGHUP, done);
-	signal(SIGINT, done);
-	signal(SIGPIPE, done);
-	signal(SIGTERM, done);
+	signal(SIGHUP, quit);
+	signal(SIGINT, quit);
+	signal(SIGPIPE, quit);
+	signal(SIGTERM, quit);
 	ofd = mkstemp(tempfile);
 	if (ofd < 0)
-		error("%s", tempfile);
+		return (NULL);
 	while ((i = read(ifd, buf, BUFSIZ)) > 0) {
 		if (write(ofd, buf, i) != i)
-			error("%s", tempfile);
+			return (NULL);
 	}
 	close(ifd);
 	close(ofd);
@@ -365,8 +480,6 @@ splice(char *dir, char *file)
 	char *tail, *buf;
 	size_t len;
 
-	if (!strcmp(file, "-"))
-		errorx("can't specify - with other arg directory");
 	tail = strrchr(file, '/');
 	if (tail == NULL)
 		tail = file;
@@ -384,7 +497,7 @@ prepare(int i, FILE *fd)
 	struct line *p;
 	int j, h;
 
-	fseek(fd, 0L, SEEK_SET);
+	rewind(fd);
 	p = emalloc(3 * sizeof(struct line));
 	for (j = 0; (h = readhash(fd));) {
 		p = erealloc(p, (++j + 3) * sizeof(struct line));
@@ -538,32 +651,30 @@ unravel(int p)
  *  2.  collect random access indexes to the two files
  */
 static void
-check(void)
+check(char *file1, FILE *f1, char *file2, FILE *f2)
 {
 	int i, j, jackpot, c, d;
 	long ctold, ctnew;
 
-	if ((input[0] = fopen(file1, "r")) == NULL)
-		error("%s", file1);
-	if ((input[1] = fopen(file2, "r")) == NULL)
-		error("%s", file2);
+	rewind(f1);
+	rewind(f2);
 	j = 1;
 	ixold[0] = ixnew[0] = 0;
 	jackpot = 0;
 	ctold = ctnew = 0;
 	for (i = 1; i <= len[0]; i++) {
 		if (J[i] == 0) {
-			ixold[i] = ctold += skipline(0);
+			ixold[i] = ctold += skipline(f1);
 			continue;
 		}
 		while (j < J[i]) {
-			ixnew[j] = ctnew += skipline(1);
+			ixnew[j] = ctnew += skipline(f2);
 			j++;
 		}
 		if (bflag || wflag || iflag) {
 			for (;;) {
-				c = getc(input[0]);
-				d = getc(input[1]);
+				c = getc(f1);
+				d = getc(f2);
 				ctold++;
 				ctnew++;
 				if (bflag && isspace(c) && isspace(d)) {
@@ -571,19 +682,19 @@ check(void)
 						if (c == '\n')
 							break;
 						ctold++;
-					} while (isspace(c = getc(input[0])));
+					} while (isspace(c = getc(f1)));
 					do {
 						if (d == '\n')
 							break;
 						ctnew++;
-					} while (isspace(d = getc(input[1])));
+					} while (isspace(d = getc(f2)));
 				} else if (wflag) {
 					while (isspace(c) && c != '\n') {
-						c = getc(input[0]);
+						c = getc(f1);
 						ctold++;
 					}
 					while (isspace(d) && d != '\n') {
-						d = getc(input[1]);
+						d = getc(f2);
 						ctnew++;
 					}
 				}
@@ -591,9 +702,9 @@ check(void)
 					jackpot++;
 					J[i] = 0;
 					if (c != '\n')
-						ctold += skipline(0);
+						ctold += skipline(f1);
 					if (d != '\n')
-						ctnew += skipline(1);
+						ctnew += skipline(f2);
 					break;
 				}
 				if (c == '\n')
@@ -603,13 +714,13 @@ check(void)
 			for (;;) {
 				ctold++;
 				ctnew++;
-				if ((c = getc(input[0])) != (d = getc(input[1]))) {
+				if ((c = getc(f1)) != (d = getc(f2))) {
 					/* jackpot++; */
 					J[i] = 0;
 					if (c != '\n')
-						ctold += skipline(0);
+						ctold += skipline(f1);
 					if (d != '\n')
-						ctnew += skipline(1);
+						ctnew += skipline(f2);
 					break;
 				}
 				if (c == '\n')
@@ -620,11 +731,8 @@ check(void)
 		ixnew[j] = ctnew;
 		j++;
 	}
-	for (; j <= len[1]; j++) {
-		ixnew[j] = ctnew += skipline(1);
-	}
-	fclose(input[0]);
-	fclose(input[1]);
+	for (; j <= len[1]; j++)
+		ixnew[j] = ctnew += skipline(f2);
 	/*
 	 * if (jackpot)
 	 *	fprintf(stderr, "jackpot\n");
@@ -678,27 +786,27 @@ unsort(struct line *f, int l, int *b)
 }
 
 static int
-skipline(int f)
+skipline(FILE *f)
 {
 	int i, c;
 
-	for (i = 1; (c = getc(input[f])) != '\n'; i++)
+	for (i = 1; (c = getc(f)) != '\n'; i++)
 		if (c < 0)
 			return (i);
 	return (i);
 }
 
 static void
-output(void)
+output(char *file1, FILE *f1, char *file2, FILE *f2)
 {
 	int m, i0, i1, j0, j1;
 
-	input[0] = fopen(file1, "r");
-	input[1] = fopen(file2, "r");
+	rewind(f1);
+	rewind(f2);
 	m = len[0];
 	J[0] = 0;
 	J[m + 1] = len[1] + 1;
-	if (opt != D_EDIT) {
+	if (format != D_EDIT) {
 		for (i0 = 1; i0 <= m; i0 = i1 + 1) {
 			while (i0 <= m && J[i0] == J[i0 - 1] + 1)
 				i0++;
@@ -708,7 +816,7 @@ output(void)
 				i1++;
 			j1 = J[i1 + 1] - 1;
 			J[i1] = j1;
-			change(i0, i1, j0, j1);
+			change(file1, f1, file2, f2, i0, i1, j0, j1);
 		}
 	} else {
 		for (i0 = m; i0 >= 1; i0 = i1 - 1) {
@@ -720,15 +828,15 @@ output(void)
 				i1--;
 			j1 = J[i1 - 1] + 1;
 			J[i1] = j1;
-			change(i1, i0, j1, j0);
+			change(file1, f1, file2, f2, i1, i0, j1, j0);
 		}
 	}
 	if (m == 0)
-		change(1, 0, 1, len[1]);
-	if (opt == D_IFDEF) {
+		change(file1, f1, file2, f2, 1, 0, 1, len[1]);
+	if (format == D_IFDEF) {
 		for (;;) {
 #define	c i0
-			c = getc(input[0]);
+			c = getc(f1);
 			if (c < 0)
 				return;
 			putchar(c);
@@ -736,17 +844,17 @@ output(void)
 #undef c
 	}
 	if (anychange != 0) {
-		if (opt == D_CONTEXT)
-			dump_context_vec();
-		else if (opt == D_UNIFIED)
-			dump_unified_vec();
+		if (format == D_CONTEXT)
+			dump_context_vec(f1, f2);
+		else if (format == D_UNIFIED)
+			dump_unified_vec(f1, f2);
 	}
 }
 
 /*
  * The following struct is used to record change information when
- * doing a "context" diff.  (see routine "change" to understand the
- * highly mneumonic field names)
+ * doing a "context" or "unified" diff.  (see routine "change" to
+ * understand the highly mnemonic field names)
  */
 struct context_vec {
 	int a;			/* start line in old file */
@@ -760,35 +868,32 @@ struct context_vec *context_vec_start, *context_vec_end, *context_vec_ptr;
 #define	MAX_CONTEXT	128
 
 /*
- * indicate that there is a difference between lines a and b of the from file
- * to get to lines c to d of the to file. If a is greater then b then there
+ * Indicate that there is a difference between lines a and b of the from file
+ * to get to lines c to d of the to file.  If a is greater then b then there
  * are no lines in the from file involved and this means that there were
- * lines appended (beginning at b). If c is greater than d then there are
+ * lines appended (beginning at b).  If c is greater than d then there are
  * lines missing from the to file.
  */
 static void
-change(int a, int b, int c, int d)
+change(char *file1, FILE *f1, char *file2, FILE *f2, int a, int b, int c, int d)
 {
-	struct stat stbuf;
-
-	if (opt != D_IFDEF && a > b && c > d)
+	if (format != D_IFDEF && a > b && c > d)
 		return;
 	if (anychange == 0) {
 		anychange = 1;
-		if (opt == D_CONTEXT || opt == D_UNIFIED) {
-			stat(file1, &stbuf);
-			printf("%s %s	%s", opt == D_CONTEXT ? "***" : "---",
-			   file1, ctime(&stbuf.st_mtime));
-			stat(file2, &stbuf);
-			printf("%s %s	%s", opt == D_CONTEXT ? "---" : "+++",
-			    file2, ctime(&stbuf.st_mtime));
-			context_vec_start = emalloc(MAX_CONTEXT *
-			    sizeof(struct context_vec));
+		if (format == D_CONTEXT || format == D_UNIFIED) {
+			printf("%s %s	%s", format == D_CONTEXT ? "***" : "---",
+			   file1, ctime(&stb1.st_mtime));
+			printf("%s %s	%s", format == D_CONTEXT ? "---" : "+++",
+			    file2, ctime(&stb2.st_mtime));
+			if (context_vec_start == NULL)
+				context_vec_start = emalloc(MAX_CONTEXT *
+				    sizeof(struct context_vec));
 			context_vec_end = context_vec_start + MAX_CONTEXT;
 			context_vec_ptr = context_vec_start - 1;
 		}
 	}
-	if (opt == D_CONTEXT || opt == D_UNIFIED) {
+	if (format == D_CONTEXT || format == D_UNIFIED) {
 		/*
 		 * If this new change is within 'context' lines of
 		 * the previous change, just add it to the change
@@ -800,10 +905,10 @@ change(int a, int b, int c, int d)
 		    (context_vec_ptr >= context_vec_start &&
 		    a > (context_vec_ptr->b + 2 * context) &&
 		    c > (context_vec_ptr->d + 2 * context))) {
-			if (opt == D_CONTEXT)
-				dump_context_vec();
+			if (format == D_CONTEXT)
+				dump_context_vec(f1, f2);
 			else
-				dump_unified_vec();
+				dump_unified_vec(f1, f2);
 		}
 		context_vec_ptr++;
 		context_vec_ptr->a = a;
@@ -812,13 +917,13 @@ change(int a, int b, int c, int d)
 		context_vec_ptr->d = d;
 		return;
 	}
-	switch (opt) {
+	switch (format) {
 
 	case D_NORMAL:
 	case D_EDIT:
 		range(a, b, ",");
 		putchar(a > b ? 'a' : c > d ? 'd' : 'c');
-		if (opt == D_NORMAL)
+		if (format == D_NORMAL)
 			range(c, d, ",");
 		putchar('\n');
 		break;
@@ -838,14 +943,14 @@ change(int a, int b, int c, int d)
 		}
 		break;
 	}
-	if (opt == D_NORMAL || opt == D_IFDEF) {
-		fetch(ixold, a, b, input[0], "< ", 1);
-		if (a <= b && c <= d && opt == D_NORMAL)
-			prints("---\n");
+	if (format == D_NORMAL || format == D_IFDEF) {
+		fetch(ixold, a, b, f1, "< ", 1);
+		if (a <= b && c <= d && format == D_NORMAL)
+			puts("---");
 	}
-	fetch(ixnew, c, d, input[1], opt == D_NORMAL ? "> " : "", 0);
-	if ((opt == D_EDIT || opt == D_REVERSE) && c <= d)
-		prints(".\n");
+	fetch(ixnew, c, d, f2, format == D_NORMAL ? "> " : "", 0);
+	if ((format == D_EDIT || format == D_REVERSE) && c <= d)
+		puts(".");
 	if (inifdef) {
 		fprintf(stdout, "#endif /* %s */\n", ifdefname);
 		inifdef = 0;
@@ -869,7 +974,7 @@ fetch(long *f, int a, int b, FILE *lb, char *s, int oldfile)
 	 * When doing #ifdef's, copy down to current line
 	 * if this is the first file, so that stuff makes it to output.
 	 */
-	if (opt == D_IFDEF && oldfile) {
+	if (format == D_IFDEF && oldfile) {
 		long curpos = ftell(lb);
 		/* print through if append (a>b), else to (nb: 0 vs 1 orig) */
 		nc = f[a > b ? b : a - 1] - curpos;
@@ -878,7 +983,7 @@ fetch(long *f, int a, int b, FILE *lb, char *s, int oldfile)
 	}
 	if (a > b)
 		return;
-	if (opt == D_IFDEF) {
+	if (format == D_IFDEF) {
 		if (inifdef) {
 			fprintf(stdout, "#else /* %s%s */\n",
 			    oldfile == 1 ? "!" : "", ifdefname);
@@ -893,8 +998,8 @@ fetch(long *f, int a, int b, FILE *lb, char *s, int oldfile)
 	for (i = a; i <= b; i++) {
 		fseek(lb, f[i - 1], SEEK_SET);
 		nc = f[i] - f[i - 1];
-		if (opt != D_IFDEF)
-			prints(s);
+		if (format != D_IFDEF)
+			fputs(s, stdout);
 		col = 0;
 		for (j = 0; j < nc; j++) {
 			c = getc(lb);
@@ -984,17 +1089,17 @@ readhash(FILE *f)
 	return ((short) low(sum) + (short) high(sum));
 }
 
-static int
+int
 asciifile(FILE *f)
 {
 	char buf[BUFSIZ], *cp;
 	int cnt;
 
-	if (aflag)
+	if (aflag || f == NULL)
 		return (1);
 
-	fseek(f, 0L, SEEK_SET);
-	cnt = fread(buf, 1, BUFSIZ, f);
+	rewind(f);
+	cnt = fread(buf, 1, sizeof(buf), f);
 	cp = buf;
 	while (--cnt >= 0)
 		if (*cp++ & 0200)
@@ -1002,9 +1107,19 @@ asciifile(FILE *f)
 	return (1);
 }
 
+static __inline int min(int a, int b)
+{
+	return (a < b ? a : b);
+}
+
+static __inline int max(int a, int b)
+{
+	return (a > b ? a : b);
+}
+
 /* dump accumulated "context" diff changes */
 static void
-dump_context_vec(void)
+dump_context_vec(FILE *f1, FILE *f2)
 {
 	struct context_vec *cvp = context_vec_start;
 	int lowa, upb, lowc, upd, do_output;
@@ -1025,7 +1140,7 @@ dump_context_vec(void)
 	printf(" ****\n");
 
 	/*
-	 * output changes to the "old" file.  The first loop suppresses
+	 * Output changes to the "old" file.  The first loop suppresses
 	 * output if there were no changes to the "old" file (we'll see
 	 * the "old" lines as context in the "new" list).
 	 */
@@ -1049,16 +1164,16 @@ dump_context_vec(void)
 				ch = (a <= b) ? 'd' : 'a';
 
 			if (ch == 'a')
-				fetch(ixold, lowa, b, input[0], "  ", 0);
+				fetch(ixold, lowa, b, f1, "  ", 0);
 			else {
-				fetch(ixold, lowa, a - 1, input[0], "  ", 0);
-				fetch(ixold, a, b, input[0],
+				fetch(ixold, lowa, a - 1, f1, "  ", 0);
+				fetch(ixold, a, b, f1,
 				    ch == 'c' ? "! " : "- ", 0);
 			}
 			lowa = b + 1;
 			cvp++;
 		}
-		fetch(ixold, b + 1, upb, input[0], "  ", 0);
+		fetch(ixold, b + 1, upb, f1, "  ", 0);
 	}
 	/* output changes to the "new" file */
 	printf("--- ");
@@ -1085,23 +1200,23 @@ dump_context_vec(void)
 				ch = (a <= b) ? 'd' : 'a';
 
 			if (ch == 'd')
-				fetch(ixnew, lowc, d, input[1], "  ", 0);
+				fetch(ixnew, lowc, d, f2, "  ", 0);
 			else {
-				fetch(ixnew, lowc, c - 1, input[1], "  ", 0);
-				fetch(ixnew, c, d, input[1],
+				fetch(ixnew, lowc, c - 1, f2, "  ", 0);
+				fetch(ixnew, c, d, f2,
 				    ch == 'c' ? "! " : "+ ", 0);
 			}
 			lowc = d + 1;
 			cvp++;
 		}
-		fetch(ixnew, d + 1, upd, input[1], "  ", 0);
+		fetch(ixnew, d + 1, upd, f2, "  ", 0);
 	}
 	context_vec_ptr = context_vec_start - 1;
 }
 
 /* dump accumulated "unified" diff changes */
 static void
-dump_unified_vec(void)
+dump_unified_vec(FILE *f1, FILE *f2)
 {
 	struct context_vec *cvp = context_vec_start;
 	int lowa, upb, lowc, upd;
@@ -1142,23 +1257,23 @@ dump_unified_vec(void)
 
 		switch (ch) {
 		case 'c':
-			fetch(ixold, lowa, a - 1, input[0], " ", 0);
-			fetch(ixold, a, b, input[0], "-", 0);
-			fetch(ixnew, c, d, input[1], "+", 0);
+			fetch(ixold, lowa, a - 1, f1, " ", 0);
+			fetch(ixold, a, b, f1, "-", 0);
+			fetch(ixnew, c, d, f2, "+", 0);
 			break;
 		case 'd':
-			fetch(ixold, lowa, a - 1, input[0], " ", 0);
-			fetch(ixold, a, b, input[0], "-", 0);
+			fetch(ixold, lowa, a - 1, f1, " ", 0);
+			fetch(ixold, a, b, f1, "-", 0);
 			break;
 		case 'a':
-			fetch(ixnew, lowc, c - 1, input[1], " ", 0);
-			fetch(ixnew, c, d, input[1], "+", 0);
+			fetch(ixnew, lowc, c - 1, f2, " ", 0);
+			fetch(ixnew, c, d, f2, "+", 0);
 			break;
 		}
 		lowa = b + 1;
 		lowc = d + 1;
 	}
-	fetch(ixnew, d + 1, upd, input[1], " ", 0);
+	fetch(ixnew, d + 1, upd, f2, " ", 0);
 
 	context_vec_ptr = context_vec_start - 1;
 }
diff --git a/usr.bin/diff/pathnames.h b/usr.bin/diff/pathnames.h
deleted file mode 100644
index 83b4235e923..00000000000
--- a/usr.bin/diff/pathnames.h
+++ /dev/null
@@ -1,34 +0,0 @@
-/*-
- * Copyright (c) 1989, 1993
- *	The Regents of the University of California.  All rights reserved.
- *
- * Redistribution and use in source and binary forms, with or without
- * modification, are permitted provided that the following conditions
- * are met:
- * 1. Redistributions of source code must retain the above copyright
- *    notice, this list of conditions and the following disclaimer.
- * 2. Redistributions in binary form must reproduce the above copyright
- *    notice, this list of conditions and the following disclaimer in the
- *    documentation and/or other materials provided with the distribution.
- * 3. Neither the name of the University nor the names of its contributors
- *    may be used to endorse or promote products derived from this software
- *    without specific prior written permission.
- *
- * THIS SOFTWARE IS PROVIDED BY THE REGENTS AND CONTRIBUTORS ``AS IS'' AND
- * ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO, THE
- * IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE
- * ARE DISCLAIMED.  IN NO EVENT SHALL THE REGENTS OR CONTRIBUTORS BE LIABLE
- * FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL
- * DAMAGES (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS
- * OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR BUSINESS INTERRUPTION)
- * HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT
- * LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY
- * OUT OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF
- * SUCH DAMAGE.
- *
- *	@(#)pathnames.h	8.1 (Berkeley) 6/6/93
- */
-
-#include <paths.h>
-
-#define	_PATH_DIFF	"/usr/bin/diff"