dpkg_deb: slight code shrink

ar: reuse existing ar unpacking code
get_header_tar: handle autodetection for tiny .tar.gz files too
unarchive.h: do not include CONFIGed out things

function                                             old     new   delta
get_header_tar                                      1521    1534     +13
dpkg_deb_main                                        400     380     -20
ar_main                                              260     196     -64
------------------------------------------------------------------------------
(add/remove: 0/0 grow/shrink: 1/2 up/down: 13/-84)            Total: -71 bytes

archival/Config.in

@@ -90,10 +90,11 @@ config DPKG
 	bool "dpkg"
 	default n
 	help
-	  dpkg is a medium-level tool to install, build, remove and manage Debian packages.
+	  dpkg is a medium-level tool to install, build, remove and manage
+	  Debian packages.
 
-	  This implementation of dpkg has a number of limitations, you should use the
-	  official dpkg if possible.
+	  This implementation of dpkg has a number of limitations,
+	  you should use the official dpkg if possible.
 
 config DPKG_DEB
 	bool "dpkg_deb"
@@ -103,8 +104,8 @@ config DPKG_DEB
 
 	  This implementation of dpkg-deb cannot pack archives.
 
-	  Unless you have a specific application which requires dpkg-deb, you should
-	  probably say N here.
+	  Unless you have a specific application which requires dpkg-deb,
+	  say N here.
 
 config FEATURE_DPKG_DEB_EXTRACT_ONLY
 	bool "Extract only (-x)"

archival/ar.c

@@ -45,7 +45,6 @@ int ar_main(int argc, char **argv)
 
 	archive_handle_t *archive_handle;
 	unsigned opt;
-	char magic[8];
 
 	archive_handle = init_handle();
 
@@ -82,14 +81,7 @@ int ar_main(int argc, char **argv)
 		llist_add_to(&(archive_handle->accept), argv[optind++]);
 	}
 
-	xread(archive_handle->src_fd, magic, 7);
-	if (strncmp(magic, "!<arch>", 7) != 0) {
-		bb_error_msg_and_die("invalid ar magic");
-	}
-	archive_handle->offset += 7;
-
-	while (get_header_ar(archive_handle) == EXIT_SUCCESS)
-		continue;
+	unpack_ar_archive(archive_handle);
 
 	return EXIT_SUCCESS;
 }

archival/dpkg_deb.c

@@ -20,8 +20,8 @@ int dpkg_deb_main(int argc, char **argv)
 	archive_handle_t *tar_archive;
 	llist_t *control_tar_llist = NULL;
 	unsigned opt;
-	const char *extract_dir = NULL;
-	short argcount = 1;
+	const char *extract_dir;
+	int need_args;
 
 	/* Setup the tar archive handle */
 	tar_archive = init_handle();
@@ -43,17 +43,21 @@ int dpkg_deb_main(int argc, char **argv)
 
 	opt_complementary = "c--efXx:e--cfXx:f--ceXx:X--cefx:x--cefX";
 	opt = getopt32(argv, "cefXx");
+	argv += optind;
+	argc -= optind;
 
 	if (opt & DPKG_DEB_OPT_CONTENTS) {
 		tar_archive->action_header = header_verbose_list;
 	}
+	extract_dir = NULL;
+	need_args = 1;
 	if (opt & DPKG_DEB_OPT_CONTROL) {
 		ar_archive->accept = control_tar_llist;
 		tar_archive->action_data = data_extract_all;
-		if (optind + 1 == argc) {
+		if (1 == argc) {
 			extract_dir = "./DEBIAN";
 		} else {
-			argcount++;
+			need_args++;
 		}
 	}
 	if (opt & DPKG_DEB_OPT_FIELD) {
@@ -70,28 +74,31 @@ int dpkg_deb_main(int argc, char **argv)
 	}
 	if (opt & (DPKG_DEB_OPT_EXTRACT_VERBOSE | DPKG_DEB_OPT_EXTRACT)) {
 		tar_archive->action_data = data_extract_all;
-		argcount = 2;
+		need_args = 2;
 	}
 
-	if ((optind + argcount) != argc) {
+	if (need_args != argc) {
 		bb_show_usage();
 	}
 
-	tar_archive->src_fd = ar_archive->src_fd = xopen(argv[optind++], O_RDONLY);
+	tar_archive->src_fd = ar_archive->src_fd = xopen(argv[0], O_RDONLY);
 
-	/* Workout where to extract the files */
+	/* Work out where to extract the files */
 	/* 2nd argument is a dir name */
-	if (argv[optind]) {
-		extract_dir = argv[optind];
+	if (argv[1]) {
+		extract_dir = argv[1];
 	}
 	if (extract_dir) {
 		mkdir(extract_dir, 0777); /* bb_make_directory(extract_dir, 0777, 0) */
 		xchdir(extract_dir);
 	}
+
+	/* Do it */
 	unpack_ar_archive(ar_archive);
 
 	/* Cleanup */
-	close(ar_archive->src_fd);
+	if (ENABLE_FEATURE_CLEAN_UP)
+		close(ar_archive->src_fd);
 
 	return EXIT_SUCCESS;
 }

archival/libunarchive/filter_accept_list_reassign.c

@@ -8,6 +8,8 @@
 #include "libbb.h"
 #include "unarchive.h"
 
+/* Built and used only if ENABLE_DPKG || ENABLE_DPKG_DEB */
+
 /*
  * Reassign the subarchive metadata parser based on the filename extension
  * e.g. if its a .tar.gz modify archive_handle->sub_archive to process a .tar.gz
@@ -19,23 +21,25 @@ char FAST_FUNC filter_accept_list_reassign(archive_handle_t *archive_handle)
 	if (find_list_entry(archive_handle->accept, archive_handle->file_header->name)) {
 		const char *name_ptr;
 
-		/* Extract the last 2 extensions */
+		/* Find extension */
 		name_ptr = strrchr(archive_handle->file_header->name, '.');
 
 		/* Modify the subarchive handler based on the extension */
-#if ENABLE_FEATURE_DEB_TAR_GZ
-		if (strcmp(name_ptr, ".gz") == 0) {
+		if (ENABLE_FEATURE_DEB_TAR_GZ
+		 && strcmp(name_ptr, ".gz") == 0
+		) {
 			archive_handle->action_data_subarchive = get_header_tar_gz;
 			return EXIT_SUCCESS;
 		}
-#endif
-#if ENABLE_FEATURE_DEB_TAR_BZ2
-		if (strcmp(name_ptr, ".bz2") == 0) {
+		if (ENABLE_FEATURE_DEB_TAR_BZ2
+		 && strcmp(name_ptr, ".bz2") == 0
+		) {
 			archive_handle->action_data_subarchive = get_header_tar_bz2;
 			return EXIT_SUCCESS;
 		}
-#endif
-		if (ENABLE_FEATURE_DEB_TAR_LZMA && !strcmp(name_ptr, ".lzma")) {
+		if (ENABLE_FEATURE_DEB_TAR_LZMA
+		 && strcmp(name_ptr, ".lzma") == 0
+		) {
 			archive_handle->action_data_subarchive = get_header_tar_lzma;
 			return EXIT_SUCCESS;
 		}

archival/libunarchive/get_header_ar.c

@@ -108,12 +108,13 @@ char FAST_FUNC get_header_ar(archive_handle_t *archive_handle)
 
 	if (archive_handle->filter(archive_handle) == EXIT_SUCCESS) {
 		archive_handle->action_header(typed);
+#if ENABLE_DPKG || ENABLE_DPKG_DEB
 		if (archive_handle->sub_archive) {
 			while (archive_handle->action_data_subarchive(archive_handle->sub_archive) == EXIT_SUCCESS)
 				continue;
-		} else {
+		} else
+#endif
 			archive_handle->action_data(archive_handle);
-		}
 	} else {
 		data_skip(archive_handle);
 	}

archival/libunarchive/get_header_tar.c

@@ -92,22 +92,32 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
  again_after_align:
 
 #if ENABLE_DESKTOP
+	/* to prevent misdetection of bz2 sig */
+	*(uint32_t*)(&tar) = 0;
 	i = full_read(archive_handle->src_fd, &tar, 512);
 	/* If GNU tar sees EOF in above read, it says:
 	 * "tar: A lone zero block at N", where N = kilobyte
 	 * where EOF was met (not EOF block, actual EOF!),
-	 * and tar will exit with error code 0.
+	 * and exits with EXIT_SUCCESS.
 	 * We will mimic exit(EXIT_SUCCESS), although we will not mimic
 	 * the message and we don't check whether we indeed
 	 * saw zero block directly before this. */
-	if (i == 0)
+	if (i == 0) {
 		xfunc_error_retval = 0;
-	if (i != 512)
+ short_read:
 		bb_error_msg_and_die("short read");
+	}
+	if (i != 512) {
+		if (ENABLE_FEATURE_TAR_AUTODETECT)
+			goto autodetect;
+		goto short_read;
+	}
+
 #else
-	xread(archive_handle->src_fd, &tar, 512);
+	i = 512;
+	xread(archive_handle->src_fd, &tar, i);
 #endif
-	archive_handle->offset += 512;
+	archive_handle->offset += i;
 
 	/* If there is no filename its an empty header */
 	if (tar.name[0] == 0 && tar.prefix[0] == 0) {
@@ -133,6 +143,7 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 #if ENABLE_FEATURE_TAR_AUTODETECT
 		char FAST_FUNC (*get_header_ptr)(archive_handle_t *);
 
+ USE_DESKTOP(autodetect:)
 		/* tar gz/bz autodetect: check for gz/bz2 magic.
 		 * If we see the magic, and it is the very first block,
 		 * we can switch to get_header_tar_gz/bz2/lzma().
@@ -154,7 +165,7 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 		/* Two different causes for lseek() != 0:
 		 * unseekable fd (would like to support that too, but...),
 		 * or not first block (false positive, it's not .gz/.bz2!) */
-		if (lseek(archive_handle->src_fd, -512, SEEK_CUR) != 0)
+		if (lseek(archive_handle->src_fd, -i, SEEK_CUR) != 0)
 			goto err;
 		while (get_header_ptr(archive_handle) == EXIT_SUCCESS)
 			continue;
@@ -328,7 +339,7 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 		p_linkname = NULL;
 	}
 #endif
-	if (!strncmp(file_header->name, "/../"+1, 3)
+	if (strncmp(file_header->name, "/../"+1, 3) == 0
 	 || strstr(file_header->name, "/../")
 	) {
 		bb_error_msg_and_die("name with '..' encountered: '%s'",
@@ -340,7 +351,7 @@ char FAST_FUNC get_header_tar(archive_handle_t *archive_handle)
 	cp = last_char_is(file_header->name, '/');
 
 	if (archive_handle->filter(archive_handle) == EXIT_SUCCESS) {
-		archive_handle->action_header(archive_handle->file_header);
+		archive_handle->action_header(/*archive_handle->*/ file_header);
 		/* Note that we kill the '/' only after action_header() */
 		/* (like GNU tar 1.15.1: verbose mode outputs "dir/dir/") */
 		if (cp) *cp = '\0';

archival/libunarchive/get_header_tar_gz.c

@@ -20,6 +20,7 @@ char FAST_FUNC get_header_tar_gz(archive_handle_t *archive_handle)
 	 * need the header. */
 #if BB_MMU
 	xread(archive_handle->src_fd, &magic, 2);
+	/* Can skip this check, but error message will be less clear */
 	if ((magic[0] != 0x1f) || (magic[1] != 0x8b)) {
 		bb_error_msg_and_die("invalid gzip magic");
 	}

archival/libunarchive/unpack_ar_archive.c

@@ -16,5 +16,6 @@ void FAST_FUNC unpack_ar_archive(archive_handle_t *ar_archive)
 	}
 	ar_archive->offset += 7;
 
-	while (get_header_ar(ar_archive) == EXIT_SUCCESS);
+	while (get_header_ar(ar_archive) == EXIT_SUCCESS)
+		continue;
 }

include/unarchive.h

@@ -47,11 +47,13 @@ typedef struct archive_handle_t {
 	/* Process the data component, e.g. extract to filesystem */
 	void FAST_FUNC (*action_data)(struct archive_handle_t *);
 
+#if ENABLE_DPKG || ENABLE_DPKG_DEB
+	/* "subarchive" is used only by dpkg[-deb] applets */
 	/* How to process any sub archive, e.g. get_header_tar_gz */
 	char FAST_FUNC (*action_data_subarchive)(struct archive_handle_t *);
-
 	/* Contains the handle to a sub archive */
 	struct archive_handle_t *sub_archive;
+#endif
 
 	/* The raw stream as read from disk or stdin */
 	int src_fd;