add support for xz compressed rpmsv9.0.0.4g

Newer RPM distros are moving to XZ compression, so supporting that would
probably be a good thing.

URL: http://bugs.gentoo.org/292057
Reported-by: Rafał Mużyło <galtgendo@o2.pl>
Signed-off-by: Mike Frysinger <vapier@gentoo.org>

3 files changed, 54 insertions, 26 deletions

diff --git a/Makefile b/Makefile
index f051455..5b720f7 100644
--- a/Makefile
+++ b/Makefile
@@ -30,7 +30,7 @@ install: rpmoffset
 	$(dodir) $(DESTDIR)$(bindir)
 	$(dobin) rpm2targz rpmoffset $(DESTDIR)$(bindir)
 	set -e ; \
-	for t in tar tarbz2 tbz2 tarlzma tgz ; do \
+	for t in tar tarbz2 tbz2 tarlzma tgz tarxz txz ; do \
 		$(dosym) rpm2targz $(DESTDIR)$(bindir)/rpm2$$t ; \
 	done
 	$(dosym) rpm2targz $(DESTDIR)$(bindir)/rpmunpack
diff --git a/rpm2targz b/rpm2targz
index e439b08..1fe7a3a 100755
--- a/rpm2targz
+++ b/rpm2targz
@@ -71,14 +71,16 @@ else
 fi
 
 compress="cat"
-case ${argv0} in
-	rpmunpack)                    suffix="";;
-	rpm2tar)                      suffix=".tar";;
-	rpm2tarbz2)  compress="bzip2" suffix=".tar.bz2";;
-	rpm2tbz2)    compress="bzip2" suffix=".tbz2";;
-	rpm2tarlzma) compress="lzma"  suffix=".tar.lzma";;
-	rpm2tgz)     compress="gzip"  suffix=".tgz";;
-	rpm2targz|*) compress="gzip"  suffix=".tar.gz";;
+case ${argv0#rpm} in
+	unpack)                    suffix="";;
+	2tar)                      suffix=".tar";;
+	2tarbz2)  compress="bzip2" suffix=".tar.bz2";;
+	2tbz2)    compress="bzip2" suffix=".tbz2";;
+	2tarlzma) compress="lzma"  suffix=".tar.lzma";;
+	2tgz)     compress="gzip"  suffix=".tgz";;
+	2tarxz)   compress="xz"    suffix=".tar.xz";;
+	2txz)     compress="xz"    suffix=".txz";;
+	2targz|*) compress="gzip"  suffix=".tar.gz";;
 esac
 case ${argv0} in
 	rpm2tar*) strip=true;;
@@ -143,17 +145,18 @@ for file; do
 	# extract the CPIO from the RPM and unpack it
 	(
 		decompressor=""
-		if command -v rpm2cpio 2>/dev/null 1>&2 ; then
+		# Disable rpm2cpio as it might be broken #249769,
+		# or it might be too old #292057
+		#if command -v rpm2cpio >/dev/null 2>&1 ; then
+		if false ; then
 			decompressor="rpm2cpio"
 			rpm2cpio "${file}"
 		else
 			# do it by hand :/
-			offset=$(rpmoffset < "${file}")
-			magic=`(dd ibs=${offset} skip=1 if="${file}" count=1 | dd bs=10 count=1 | od -b) 2>/dev/null`
-			case ${magic} in
-				"0000000 102 132 150 "*)   decompressor="bzip2";;
-				"0000000 037 213 010 "*|*) decompressor="gzip";;
-			esac
+			set -- $(${rpmoffset} -v < "${file}")
+			decompressor=$1
+			offset=$2
+			[ -z "${offset}" ] && err "unable to locate cpio offset (broken/unknown compression?)"
 			dd ibs=${offset} skip=1 if="${file}" 2>/dev/null | ${decompressor} -dc
 		fi
 		[ $? -ne 0 ] && echo "${argv0}: ${file}: failed to extract cpio via ${decompressor} (not actually an RPM?)" 1>&2
diff --git a/rpmoffset.c b/rpmoffset.c
index b8f5a04..0a3c488 100644
--- a/rpmoffset.c
+++ b/rpmoffset.c
@@ -19,18 +19,36 @@
 # define BUFSIZ 8192
 #endif
 
-#define MAGIC_SIZE 3
+typedef struct {
+	const char *type;
+	const unsigned char *magic;
+	const size_t len;
+} magic_t;
+
+static const unsigned char magic_gzip[]  = { '\037', '\213', '\010' };
+static const unsigned char magic_bzip2[] = { 'B', 'Z', 'h' };
+static const unsigned char magic_xz[]    = { 0xFD, '7', 'z', 'X', 'Z', 0x00 };
+static const magic_t magics[] = {
+#define DECLARE_MAGIC_T(t) { .type = #t, .magic = magic_##t, .len = sizeof(magic_##t), },
+	DECLARE_MAGIC_T(gzip)
+	DECLARE_MAGIC_T(bzip2)
+	DECLARE_MAGIC_T(xz)
+#undef DECLARE_MAGIC_T
+};
+#define MAGIC_SIZE_MIN 3
+#define MAGIC_SIZE_MAX 6
 
 int main(int argc, char *argv[])
 {
+	int show_magic = 0;
 	size_t i, read_cnt, offset, left;
 	FILE *fp = stdin;
 	char p[BUFSIZ];
 
-	const char magics[][MAGIC_SIZE] = {
-		{ '\037', '\213', '\010' },   /* gzip */
-		{ 'B', 'Z', 'h' },            /* bzip */
-	};
+	if (argc == 2 && !strcmp(argv[1], "-v")) {
+		show_magic = 1;
+		--argc;
+	}
 
 	if (argc != 1) {
 		puts("Usage: rpmoffset < rpmfile");
@@ -41,23 +59,30 @@ int main(int argc, char *argv[])
 	offset = left = 0;
 	while (1) {
 		read_cnt = fread(p + left, 1, sizeof(p) - left, fp);
-		if (read_cnt + left < MAGIC_SIZE)
+		if (read_cnt + left < MAGIC_SIZE_MIN)
 			break;
 
 		for (i = 0; i < ARRAY_SIZE(magics); ++i) {
-			char *needle = memmem(p, sizeof(p), magics[i], MAGIC_SIZE);
+			const char *needle;
+
+			if (read_cnt + left < magics[i].len)
+				continue;
+
+			needle = memmem(p, sizeof(p), magics[i].magic, magics[i].len);
 			if (needle) {
+				if (show_magic)
+					printf("%s ", magics[i].type);
 				printf("%zu\n", offset + (needle - p));
 				return 0;
 			}
 		}
 
-		memmove(p, p + left + read_cnt - MAGIC_SIZE + 1, MAGIC_SIZE - 1);
+		memmove(p, p + left + read_cnt - MAGIC_SIZE_MIN + 1, MAGIC_SIZE_MIN - 1);
 
 		offset += read_cnt;
 		if (left == 0) {
-			offset -= MAGIC_SIZE - 1;
-			left = MAGIC_SIZE - 1;
+			offset -= MAGIC_SIZE_MIN - 1;
+			left = MAGIC_SIZE_MIN - 1;
 		}
 	}