A strange meta-package of a sort, rdigest is implemented as a jumbo

patch to the well-known digest package.  The output is the same for
files, but if rdigest is given a directory as an argument, it will
recursively checksum the directory in such a manner as to:

(1) give a repeatable checksum
(2) with no regard for file ownership
(3) or timestamps
(4) or modes
(5) while taking into account empty files and directories

such that the checksum of an extracted tar file will match the
checksum of the same tree as retrieved via cvs, or two allegedly
matching trees from different cvs servers, etc.

7 files changed, 424 insertions, 0 deletions

diff --git a/pkgtools/rdigest/DESCR b/pkgtools/rdigest/DESCR
new file mode 100644
index 00000000000..fec28fe43ca
--- /dev/null
+++ b/pkgtools/rdigest/DESCR
@@ -0,0 +1,50 @@
+This utility is a wrapper for the md5(3), sha1(3) and rmd160(3)
+routines.
+
+The utility is linked statically, so that it can be used without any
+shared libraries installed on the system.
+
+It is remarkably similar to the digest package, except that it
+can recursively checksum directory trees.
+
+While a simple checksum on a tar file is usually satisfactory, once
+said archive has been extracted, it is virtually impossible to
+repackage the extracted tree in a form that can yield a repeatable
+checksum.  To be specific, changes in timestamps and ownership, and
+changes in file ordering within directories can affect the checksum of
+the archive, while not really impacting the actual code at all.
+
+The algorithm used to checksum a directory in this implementation is
+as follows:
+
+	cd into directory
+	initialize MASTER_HASH
+	walk directory tree, sorting all entries
+	foreach entry
+		if it is a directory:
+			skip it if it is named "RCS", "CVS", or "SCCS"
+			initialize SLAVE_HASH
+			add the string "d " into SLAVE_HASH
+			add the pathname of the directory into SLAVE_HASH
+			finish SLAVE_HASH
+			fold the hexified SLAVE_HASH result into MASTER_HASH
+		if it is a file:
+			initialize SLAVE_HASH
+			add the string "f " into SLAVE_HASH
+			add the pathname of the file into SLAVE_HASH
+			finish SLAVE_HASH
+			fold the hexified SLAVE_HASH result into MASTER_HASH
+			initialize SLAVE_HASH
+			add the contents of the file into SLAVE_HASH
+			finish SLAVE_HASH
+			fold the hexified SLAVE_HASH result into MASTER_HASH
+		if it is a symbolic link
+			initialize SLAVE_HASH
+			add the string "l " into SLAVE_HASH
+			add the pathname of the link into SLAVE_HASH
+			finish SLAVE_HASH
+			fold the hexified SLAVE_HASH result into MASTER_HASH
+			initialize SLAVE_HASH
+			add the contents of the link into SLAVE_HASH
+			finish SLAVE_HASH
+			fold the hexified SLAVE_HASH result into MASTER_HASH
diff --git a/pkgtools/rdigest/Makefile b/pkgtools/rdigest/Makefile
new file mode 100644
index 00000000000..3286b2ac4ce
--- /dev/null
+++ b/pkgtools/rdigest/Makefile
@@ -0,0 +1,13 @@
+# $NetBSD: Makefile,v 1.1.1.1 2003/07/24 05:21:05 atatat Exp $
+
+DISTNAME=	rdigest-${VERSION}
+CATEGORIES=	# be quiet pkglint
+MASTER_SITES=	# be quiet pkglint
+
+MAINTAINER=	atatat@netbsd.org
+COMMENT=	Recursive message digest wrapper utility
+
+DIGESTDIR=	${.CURDIR}/../../pkgtools/digest
+FILESDIR=	${DIGESTDIR}/files
+
+.include "${DIGESTDIR}/Makefile"
diff --git a/pkgtools/rdigest/PLIST b/pkgtools/rdigest/PLIST
new file mode 100644
index 00000000000..5484aa3449d
--- /dev/null
+++ b/pkgtools/rdigest/PLIST
@@ -0,0 +1,3 @@
+@comment $NetBSD: PLIST,v 1.1.1.1 2003/07/24 05:21:04 atatat Exp $
+bin/rdigest
+man/man1/rdigest.1
diff --git a/pkgtools/rdigest/distinfo b/pkgtools/rdigest/distinfo
new file mode 100644
index 00000000000..dfb2312e8aa
--- /dev/null
+++ b/pkgtools/rdigest/distinfo
@@ -0,0 +1,5 @@
+$NetBSD: distinfo,v 1.1.1.1 2003/07/24 05:21:05 atatat Exp $
+
+SHA1 (patch-aa) = a1ed803a43925ef4c82e2c468bd2ee922b1998b6
+SHA1 (patch-ab) = b34e588443c030e0111edaff19ccf0882bf74a13
+SHA1 (patch-ac) = b67e1db9ec7bce6a1b1e3a398251584f9194a1c7
diff --git a/pkgtools/rdigest/patches/patch-aa b/pkgtools/rdigest/patches/patch-aa
new file mode 100644
index 00000000000..8320f06d514
--- /dev/null
+++ b/pkgtools/rdigest/patches/patch-aa
@@ -0,0 +1,60 @@
+$NetBSD: patch-aa,v 1.1.1.1 2003/07/24 05:21:05 atatat Exp $
+
+--- Makefile.in.orig	2002-12-20 23:06:12.000000000 -0500
++++ Makefile.in
+@@ -56,7 +56,7 @@ missing mkinstalldirs regress.sh
+ .c.o:
+ 	$(COMPILE) -c $< -o $@
+ 
+-all: digest
++all: rdigest
+ 
+ digest-types.h: bits
+ 	./bits digest-types.h
+@@ -64,10 +64,10 @@ digest-types.h: bits
+ bits: bits.o
+ 	$(LINK) $(bits_OBJS) $(LIBS)
+ 
+-digest: $(digest_OBJS)
++rdigest: $(digest_OBJS)
+ 	$(LINK) $(digest_OBJS) $(LIBS)
+ 
+-check: digest
++check: rdigest
+ 	@SHELL@ $(srcdir)/regress.sh
+ 
+ digest.o: digest-types.h
+@@ -81,7 +81,7 @@ sha1.o: digest-types.h
+ sha1hl.o: digest-types.h
+ 
+ clean:
+-	rm -f *.o digest bits digest-types.h
++	rm -f *.o rdigest bits digest-types.h
+ 
+ distclean: clean
+ 	rm -f Makefile config.h
+@@ -90,18 +90,18 @@ distclean: clean
+ maintainer-clean: distclean
+ 	rm -f configure config.h.in
+ 
+-install: digest
++install: rdigest
+ 	$(mkinstalldirs) $(DESTDIR)$(bindir)
+-	@f=`echo digest|sed '$(transform)'`; \
+-	echo "$(INSTALL_PROGRAM) digest $(DESTDIR)$(bindir)/$$f"; \
+-	$(INSTALL_PROGRAM) digest $(DESTDIR)$(bindir)/$$f
++	@f=`echo rdigest|sed '$(transform)'`; \
++	echo "$(INSTALL_PROGRAM) rdigest $(DESTDIR)$(bindir)/$$f"; \
++	$(INSTALL_PROGRAM) rdigest $(DESTDIR)$(bindir)/$$f
+ 	$(mkinstalldirs) $(DESTDIR)$(mandir)/man1
+-	@f=`echo digest.1|sed '$(transform)'`; \
++	@f=`echo rdigest.1|sed '$(transform)'`; \
+ 	echo "$(INSTALL_DATA) digest.1 $(DESTDIR)$(mandir)/man1/$$f"; \
+ 	$(INSTALL_DATA) digest.1 $(DESTDIR)$(mandir)/man1/$$f
+ 
+ uninstall:
+-	@f=`echo digest|sed '$(transform)'`; \
++	@f=`echo rdigest|sed '$(transform)'`; \
+ 	echo " rm -f $(DESTDIR)$(bindir)/$$f"; \
+ 	rm -f $(DESTDIR)$(bindir)/$$f
+ 
diff --git a/pkgtools/rdigest/patches/patch-ab b/pkgtools/rdigest/patches/patch-ab
new file mode 100644
index 00000000000..c25703f6988
--- /dev/null
+++ b/pkgtools/rdigest/patches/patch-ab
@@ -0,0 +1,89 @@
+$NetBSD: patch-ab,v 1.1.1.1 2003/07/24 05:21:07 atatat Exp $
+
+--- digest.1.orig	2002-02-16 15:24:32.000000000 -0500
++++ digest.1
+@@ -31,21 +31,29 @@
+ .\" SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
+ .\"
+ .\"
+-.Dd February 28, 2001
+-.Dt DIGEST 1
++.Dd July 24, 2003
++.Dt RDIGEST 1
+ .Os
+ .Sh NAME
+-.Nm digest
+-.Nd calculate message digests
++.Nm rdigest
++.Nd calculate message digests of files and directories
+ .Sh SYNOPSIS
+ .Nm
++.Op Fl x Ar directory
+ .Ar algorithm
+ .Op file ...
+ .Sh DESCRIPTION
+ The
+ .Nm
+-utility calculates message digests of files or,
++utility calculates message digests of files, directories, or,
+ if no file is specified, standard input.
++The
++.Fl x
++option can be used as many times as required to indicate directories
++not to be included in a directory's hash.
++This allows you to skip, eg, CVS subdirectories in a given tree, which
++can have relatively volatile contents not pertinent to the hash.
++.Pp
+ The list of possible algorithms is:
+ .Bl -tag -width Ds
+ .It md5
+@@ -81,11 +89,32 @@ algorithm implementations, which are loc
+ C library, and was designed to be scalable as new message digest
+ algorithms are developed.
+ .Pp
++The checksum for a directory is implemented as the hash over a list of
++one or two hashes for each of the items in the tree.
++.Bl -bullet
++.It
++For a file, the string
++.Li Dq f \& 
++along with the path make up the data for one hash, and the contents of
++the file is used for the second hash.
++.It
++For a symbolic link, the string
++.Li Dq l \&
++along with the path to the symbolic link make up the data for one
++hash, and the target of the symbolic link is used for the second hash.
++.It
++For a directory, the string
++.Li Dq d \&
++along with the path make up the data for one hash.
++There is no second hash for a directory.
++.El
++.Pp
+ The
+ .Nm
+ utility exits 0 on success, and >0 if an error occurs.
+ .Sh SEE ALSO
+ .Xr cksum 1 ,
++.Xr digest 1 ,
+ .Xr md5 3 ,
+ .Xr rmd160 3 ,
+ .Xr sha1 3
+@@ -93,9 +122,13 @@ utility exits 0 on success, and >0 if an
+ The
+ .Nm
+ utility first appeared in
+-.Nx 1.6 .
++.Nx 2.0 .
+ .Sh AUTHORS
+ The
++.Xr digest 1
++utility was written by Alistair G. Crooks \*[Lt]agc@netbsd.org\*[Gt].
+ .Nm
+-utility was written by
+-.An Alistair G. Crooks Aq agc@netbsd.org .
++was implemented by Andrew Brown \*[Lt]atatat@netbsd.org\*[Gt] as a
++patch to
++.Xr digest 1
++to add recursive capabilities.
diff --git a/pkgtools/rdigest/patches/patch-ac b/pkgtools/rdigest/patches/patch-ac
new file mode 100644
index 00000000000..a3bf6e32e2f
--- /dev/null
+++ b/pkgtools/rdigest/patches/patch-ac
@@ -0,0 +1,204 @@
+$NetBSD: patch-ac,v 1.1.1.1 2003/07/24 05:21:07 atatat Exp $
+
+--- digest.c.orig	2003-07-23 20:27:09.000000000 -0400
++++ digest.c
+@@ -43,9 +43,14 @@ __RCSID("$NetBSD: digest.c,v 1.8 2003/07
+ #endif
+ 
+ 
++#include <sys/queue.h>
++#include <sys/stat.h>
++
+ #ifdef HAVE_ERRNO_H
+ #include <errno.h>
+ #endif
++#include <fcntl.h>
++#include <fts.h>
+ #ifdef HAVE_LOCALE_H
+ #include <locale.h>
+ #endif
+@@ -147,21 +152,172 @@ digest_file(char *fn, alg_t *alg)
+ 	return (rc);
+ }
+ 
++struct excl {
++	LIST_ENTRY(excl) n;
++	const char *p;
++};
++
++LIST_HEAD(, excl) excl = LIST_HEAD_INITIALIZER(excl);
++
++static void
++exclude(const char *p)
++{
++	struct excl *e;
++
++	e = malloc(sizeof(struct excl));
++	e->p = p;
++	LIST_INSERT_HEAD(&excl, e, n);
++}
++
++static int
++skip(const char *p)
++{
++	struct excl *e;
++
++	LIST_FOREACH(e, &excl, n)
++		if (strcmp(e->p, p) == 0)
++			return (1);
++
++	return (0);
++}
++
++static int
++compar(const FTSENT **fa, const FTSENT **fb)
++{
++	return (strcmp((*fa)->fts_name, (*fb)->fts_name));
++}
++
++static int
++digest_directory(char *dn, alg_t *alg)
++{
++	char	in[BUFSIZ * 20], dot[2];
++	char	*digest;
++	int	cc, rc, l, fd, cwd;
++	char	*pathlist[2];
++	FTS	*ftsp;
++	FTSENT	*f;
++
++	rc = 1;
++	l = alg->hash_len * 2;
++	digest = malloc(l + 1);
++	sprintf(dot, ".");
++	pathlist[0] = dot;
++	pathlist[1] = NULL;
++
++	if ((cwd = open(".", O_RDONLY)) == -1 ||
++	    chdir(dn) == -1 ||
++	    (ftsp = fts_open(pathlist,
++			     FTS_COMFOLLOW | FTS_NOCHDIR | FTS_PHYSICAL,
++			     compar)) == NULL) {
++		(void) fprintf(stderr, "%s\n", dn);
++		free(digest);
++		return (0);
++	}
++
++	(*alg->hash_init)(&alg->hash_ctx);
++
++	while ((f = fts_read(ftsp)) != NULL) {
++		/* skip the second pass on a directory */
++		if (f->fts_info == FTS_DP)
++			continue;
++
++		/* skip directories named CVS, RCS, or SCCS */
++		if ((f->fts_info == FTS_NS ||
++		     S_ISDIR(f->fts_statp->st_mode)) &&
++		    skip(f->fts_name)) {
++			fts_set(ftsp, f, FTS_SKIP);
++			continue;
++		}
++
++		/* try to handle things based on stat info */
++		if (f->fts_info != FTS_NS) {
++				/* only mention directories */
++			if (S_ISDIR(f->fts_statp->st_mode)) {
++				(*alg->hash_init)(&alg->hash_ctx2);
++				(*alg->hash_update)(&alg->hash_ctx2, "d ", 2);
++				(*alg->hash_update)(&alg->hash_ctx2, f->fts_path, f->fts_pathlen);
++				(*alg->hash_end)(&alg->hash_ctx2, digest);
++				digest[l] = '\n';
++				(*alg->hash_update)(&alg->hash_ctx, digest, l + 1);
++
++				/* hash the filename and then the contents separately */
++			} else if (S_ISREG(f->fts_statp->st_mode)) {
++				if ((fd = open(f->fts_path, O_RDONLY)) != -1) {
++					(*alg->hash_init)(&alg->hash_ctx2);
++					(*alg->hash_update)(&alg->hash_ctx2, "f ", 2);
++					(*alg->hash_update)(&alg->hash_ctx2, f->fts_path, f->fts_pathlen);
++					(*alg->hash_end)(&alg->hash_ctx2, &digest[0]);
++					digest[l] = '\n';
++					(*alg->hash_update)(&alg->hash_ctx, digest, 33);
++
++					(*alg->hash_init)(&alg->hash_ctx2);
++					while ((cc = read(fd, in, sizeof(in))) > 0) {
++						(*alg->hash_update)(&alg->hash_ctx2, in, cc);
++					}
++					close(fd);
++					(*alg->hash_end)(&alg->hash_ctx2, digest);
++					digest[l] = '\n';
++					(*alg->hash_update)(&alg->hash_ctx, digest, l + 1);
++				} else {
++					(void) fprintf(stderr, "%s\n", f->fts_path);
++					rc = 0;
++				}
++
++			/* hash in symlinks as well, along with the link contents */
++			} else if (S_ISLNK(f->fts_statp->st_mode)) {
++				if ((cc = readlink(f->fts_path, in, sizeof(in))) > 0) {
++					(*alg->hash_init)(&alg->hash_ctx2);
++					(*alg->hash_update)(&alg->hash_ctx2, "l ", 2);
++					(*alg->hash_update)(&alg->hash_ctx2, f->fts_path, f->fts_pathlen);
++					(*alg->hash_end)(&alg->hash_ctx2, digest);
++					digest[l] = '\n';
++					(*alg->hash_update)(&alg->hash_ctx, digest, l + 1);
++
++					(*alg->hash_init)(&alg->hash_ctx2);
++					(*alg->hash_update)(&alg->hash_ctx2, in, cc);
++					(*alg->hash_end)(&alg->hash_ctx2, digest);
++					digest[l] = '\n';
++					(*alg->hash_update)(&alg->hash_ctx, digest, l + 1);
++				} else {
++					(void) fprintf(stderr, "%s\n", f->fts_path);
++					rc = 0;
++				}
++			}
++		}
++	}
++
++	fts_close(ftsp);
++	fchdir(cwd);
++	close(cwd);
++
++	if (rc == 1) {
++		(*alg->hash_end)(&alg->hash_ctx, digest);
++		(void) printf("%s (%s) = %s\n", alg->name, dn, digest);
++	}
++
++	free(digest);
++	return (rc);
++}
++
+ int
+ main(int argc, char **argv)
+ {
+ 	alg_t  *alg;
+ 	int	rval;
+ 	int	i;
++	struct stat st;
+ 
+ #ifdef HAVE_SETLOCALE
+ 	(void) setlocale(LC_ALL, "");
+ #endif
+-	while ((i = getopt(argc, argv, "V")) != -1) {
++	while ((i = getopt(argc, argv, "Vx:")) != -1) {
+ 		switch(i) {
+ 		case 'V':
+ 			printf("%s\n", VERSION);
+ 			return EXIT_SUCCESS;
++		case 'x':
++			exclude(optarg);
++			break;
+ 		}
+ 	}
+ 	argc -= optind;
+@@ -186,7 +342,9 @@ main(int argc, char **argv)
+ 		}
+ 	} else {
+ 		for (i = 0 ; i < argc ; i++) {
+-			if (!digest_file(argv[i], alg)) {
++			if (stat(argv[i], &st) == -1 ||
++			    (S_ISREG(st.st_mode) && !digest_file(argv[i], alg)) ||
++			    (S_ISDIR(st.st_mode) && !digest_directory(argv[i], alg))) {
+ 				(void) fprintf(stderr, "%s\n", argv[i]);
+ 				rval = EXIT_FAILURE;
+ 			}