*: teach tar et. al. to understand .xz by heart

function                                             old     new   delta
unpack_xz_stream                                       -    4126   +4126
setup_unzip_on_fd                                     80     150     +70
open_zipped                                          113     131     +18
unpack_unxz                                            5      12      +7
send_tree                                            360     353      -7
unpack_xz_stream_stdin                              3953       -   -3953
------------------------------------------------------------------------------
(add/remove: 1/1 grow/shrink: 3/1 up/down: 4221/-3960)        Total: 261 bytes

Signed-off-by: Denys Vlasenko <vda.linux@googlemail.com>

archival/bbunzip.c

@@ -374,7 +374,7 @@ char* make_new_name_unxz(char *filename)
 static
 IF_DESKTOP(long long) int unpack_unxz(unpack_info_t *info UNUSED_PARAM)
 {
-	return unpack_xz_stream_stdin();
+	return unpack_xz_stream(STDIN_FILENO, STDOUT_FILENO);
 }
 int unxz_main(int argc, char **argv) MAIN_EXTERNALLY_VISIBLE;
 int unxz_main(int argc UNUSED_PARAM, char **argv)

archival/libunarchive/decompress_unxz.c

@@ -3,7 +3,7 @@
  * by Lasse Collin <lasse.collin@tukaani.org>
  * and Igor Pavlov <http://7-zip.org/>
  *
- * See README file in unxzbz/ directory for more information.
+ * See README file in unxz/ directory for more information.
  *
  * This file is:
  * Copyright (C) 2010 Denys Vlasenko <vda.linux@googlemail.com>
@@ -48,7 +48,7 @@ static uint32_t xz_crc32(uint32_t *crc32_table,
 #include "unxz/xz_stream.h"
 
 IF_DESKTOP(long long) int FAST_FUNC
-unpack_xz_stream_stdin(void)
+unpack_xz_stream(int src_fd, int dst_fd)
 {
 	struct xz_buf iobuf;
 	struct xz_dec *state;
@@ -79,7 +79,7 @@ unpack_xz_stream_stdin(void)
 		iobuf.in_pos = 0;
 		rd = IN_SIZE - insz;
 		if (rd) {
-			rd = safe_read(STDIN_FILENO, membuf + insz, rd);
+			rd = safe_read(src_fd, membuf + insz, rd);
 			if (rd < 0) {
 				bb_error_msg("read error");
 				total = -1;
@@ -94,10 +94,11 @@ unpack_xz_stream_stdin(void)
 //				iobuf.in_pos, iobuf.in_size, iobuf.out_pos, iobuf.out_size, r);
 		outpos = iobuf.out_pos;
 		if (outpos) {
-			xwrite(STDOUT_FILENO, iobuf.out, outpos);
+			xwrite(dst_fd, iobuf.out, outpos);
 			IF_DESKTOP(total += outpos;)
 		}
 		if (r == XZ_STREAM_END
+		/* this happens even with well-formed files: */
 		 || (r == XZ_BUF_ERROR && insz == 0 && outpos == 0)
 		) {
 			break;

include/libbb.h

@@ -656,7 +656,7 @@ extern void *xmalloc_open_read_close(const char *filename, size_t *maxsz_p) FAST
  || ENABLE_FEATURE_SEAMLESS_BZ2 \
  || ENABLE_FEATURE_SEAMLESS_GZ \
  /* || ENABLE_FEATURE_SEAMLESS_Z */
-extern int setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/) FAST_FUNC;
+extern void setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/) FAST_FUNC;
 #else
 # define setup_unzip_on_fd(...) ((void)0)
 #endif

include/unarchive.h

@@ -143,14 +143,15 @@ typedef struct inflate_unzip_result {
 } inflate_unzip_result;
 
 IF_DESKTOP(long long) int inflate_unzip(inflate_unzip_result *res, off_t compr_size, int src_fd, int dst_fd) FAST_FUNC;
-IF_DESKTOP(long long) int unpack_xz_stream_stdin(void) FAST_FUNC;
+/* xz unpacker takes .xz stream from offset 0 */
+IF_DESKTOP(long long) int unpack_xz_stream(int src_fd, int dst_fd) FAST_FUNC;
 /* lzma unpacker takes .lzma stream from offset 0 */
 IF_DESKTOP(long long) int unpack_lzma_stream(int src_fd, int dst_fd) FAST_FUNC;
 /* the rest wants 2 first bytes already skipped by the caller */
 IF_DESKTOP(long long) int unpack_bz2_stream(int src_fd, int dst_fd) FAST_FUNC;
 IF_DESKTOP(long long) int unpack_gz_stream(int src_fd, int dst_fd) FAST_FUNC;
 IF_DESKTOP(long long) int unpack_gz_stream_with_info(int src_fd, int dst_fd, unpack_info_t *info) FAST_FUNC;
-IF_DESKTOP(long long) int unpack_Z_stream(int fd_in, int fd_out) FAST_FUNC;
+IF_DESKTOP(long long) int unpack_Z_stream(int src_fd, int dst_fd) FAST_FUNC;
 /* wrapper which checks first two bytes to be "BZ" */
 IF_DESKTOP(long long) int unpack_bz2_stream_prime(int src_fd, int dst_fd) FAST_FUNC;
 

libbb/read.c

@@ -305,22 +305,26 @@ void* FAST_FUNC xmalloc_xopen_read_close(const char *filename, size_t *maxsz_p)
 	return buf;
 }
 
+/* Used by e.g. rpm which gives us a fd without filename,
+ * thus we can't guess the format from filename's extension.
+ */
 #if ZIPPED
-int FAST_FUNC setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/)
+void FAST_FUNC setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/)
 {
 	const int fail_if_not_detected = 1;
-	unsigned char magic[2];
+	unsigned char magic[8];
-#if BB_MMU
+	int offset = -2;
+# if BB_MMU
 	IF_DESKTOP(long long) int FAST_FUNC (*xformer)(int src_fd, int dst_fd);
 	enum { xformer_prog = 0 };
-#else
+# else
 	enum { xformer = 0 };
 	const char *xformer_prog;
-#endif
+# endif
 
 	/* .gz and .bz2 both have 2-byte signature, and their
 	 * unpack_XXX_stream wants this header skipped. */
-	xread(fd, &magic, 2);
+	xread(fd, magic, 2);
 	if (ENABLE_FEATURE_SEAMLESS_GZ
 	 && magic[0] == 0x1f && magic[1] == 0x8b
 	) {
@@ -341,28 +345,41 @@ int FAST_FUNC setup_unzip_on_fd(int fd /*, int fail_if_not_detected*/)
 # endif
 		goto found_magic;
 	}
-// TODO: xz format support. rpm adopted it, "rpm -i FILE.rpm" badly needs this.
+	if (ENABLE_FEATURE_SEAMLESS_XZ
-// Signature: 0xFD, '7', 'z', 'X', 'Z', 0x00
+	 && magic[0] == 0xfd && magic[1] == '7'
-// More info at: http://tukaani.org/xz/xz-file-format.txt
+	) {
+		/* .xz signature: 0xfd, '7', 'z', 'X', 'Z', 0x00 */
+		/* More info at: http://tukaani.org/xz/xz-file-format.txt */
+		offset = -6;
+		xread(fd, magic + 2, 4);
+		if (strcmp((char*)magic + 2, "zXZ") == 0) {
+# if BB_MMU
+			xformer = unpack_xz_stream;
+# else
+			xformer_prog = "unxz";
+# endif
+			xlseek(fd, offset, SEEK_CUR);
+			goto found_magic;
+		}
+	}
 
 	/* No known magic seen */
 	if (fail_if_not_detected)
 		bb_error_msg_and_die("no gzip"
 			IF_FEATURE_SEAMLESS_BZ2("/bzip2")
+			IF_FEATURE_SEAMLESS_XZ("/xz")
 			" magic");
-	xlseek(fd, -2, SEEK_CUR);
+	xlseek(fd, offset, SEEK_CUR);
-	return fd;
+	return;
 
  found_magic:
 # if !BB_MMU
 	/* NOMMU version of open_transformer execs
 	 * an external unzipper that wants
 	 * file position at the start of the file */
-	xlseek(fd, -2, SEEK_CUR);
+	xlseek(fd, offset, SEEK_CUR);
 # endif
 	open_transformer(fd, xformer, xformer_prog);
-
-	return fd;
 }
 #endif /* ZIPPED */
 
@@ -380,12 +397,14 @@ int FAST_FUNC open_zipped(const char *fname)
 
 	sfx = strrchr(fname, '.');
 	if (sfx) {
-		if (ENABLE_FEATURE_SEAMLESS_LZMA && strcmp(sfx, ".lzma") == 0)
+		sfx++;
+		if (ENABLE_FEATURE_SEAMLESS_LZMA && strcmp(sfx, "lzma") == 0)
 			/* .lzma has no header/signature, just trust it */
 			open_transformer(fd, unpack_lzma_stream, "unlzma");
 		else
-		if ((ENABLE_FEATURE_SEAMLESS_GZ && strcmp(sfx, ".gz") == 0)
+		if ((ENABLE_FEATURE_SEAMLESS_GZ && strcmp(sfx, "gz") == 0)
-		 || (ENABLE_FEATURE_SEAMLESS_BZ2 && strcmp(sfx, ".bz2") == 0)
+		 || (ENABLE_FEATURE_SEAMLESS_BZ2 && strcmp(sfx, "bz2") == 0)
+		 || (ENABLE_FEATURE_SEAMLESS_XZ && strcmp(sfx, "xz") == 0)
 		) {
 			setup_unzip_on_fd(fd /*, fail_if_not_detected: 1*/);
 		}