unsquashfs: add file data extraction

Signed-off-by: David Oberhollenzer <david.oberhollenzer@sigma-star.at>

4 files changed, 183 insertions, 27 deletions

diff --git a/unpack/Makemodule.am b/unpack/Makemodule.am
index a199375..a8fd2e7 100644
--- a/unpack/Makemodule.am
+++ b/unpack/Makemodule.am
@@ -1,6 +1,6 @@
 unsquashfs_SOURCES = unpack/unsquashfs.c unpack/tree_node_from_inode.c
 unsquashfs_SOURCES += unpack/unsquashfs.h unpack/read_fstree.c
-unsquashfs_SOURCES += unpack/list_files.c
+unsquashfs_SOURCES += unpack/list_files.c unpack/extract_file.c
 unsquashfs_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
 
 if WITH_LZMA
diff --git a/unpack/extract_file.c b/unpack/extract_file.c
new file mode 100644
index 0000000..c18ace8
--- /dev/null
+++ b/unpack/extract_file.c
@@ -0,0 +1,97 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "unsquashfs.h"
+
+int extract_file(file_info_t *fi, compressor_t *cmp, size_t block_size,
+		 frag_reader_t *frag, int sqfsfd, int outfd)
+{
+	size_t i, count, fragsz;
+	bool compressed;
+	void *buffer;
+	uint32_t bs;
+	ssize_t ret;
+
+	buffer = malloc(block_size);
+	if (buffer == NULL) {
+		perror("allocating scratch buffer");
+		return -1;
+	}
+
+	count = fi->size / block_size;
+
+	if (count > 0) {
+		if (lseek(sqfsfd, fi->startblock, SEEK_SET) == (off_t)-1)
+			goto fail_seek;
+
+		for (i = 0; i < count; ++i) {
+			bs = fi->blocksizes[i];
+
+			compressed = (bs & (1 << 24)) == 0;
+			bs &= (1 << 24) - 1;
+
+			if (bs > block_size)
+				goto fail_bs;
+
+			ret = read_retry(sqfsfd, buffer, bs);
+			if (ret < 0)
+				goto fail_rd;
+
+			if ((size_t)ret < bs)
+				goto fail_trunc;
+
+			if (compressed) {
+				ret = cmp->do_block(cmp, buffer, bs);
+				if (ret <= 0)
+					goto fail;
+
+				bs = ret;
+			}
+
+			ret = write_retry(outfd, buffer, bs);
+			if (ret < 0)
+				goto fail_wr;
+
+			if ((size_t)ret < bs)
+				goto fail_wr_trunc;
+		}
+	}
+
+	fragsz = fi->size % block_size;
+
+	if (fragsz > 0) {
+		if (frag_reader_read(frag, fi->fragment, fi->fragment_offset,
+				     buffer, fragsz)) {
+			goto fail;
+		}
+
+		ret = write_retry(outfd, buffer, fragsz);
+		if (ret < 0)
+			goto fail_wr;
+
+		if ((size_t)ret < fragsz)
+			goto fail_wr_trunc;
+	}
+
+	free(buffer);
+	return 0;
+fail_seek:
+	perror("seek on squashfs");
+	goto fail;
+fail_wr:
+	perror("writing uncompressed block");
+	goto fail;
+fail_wr_trunc:
+	fputs("writing uncompressed block: truncated write\n", stderr);
+	goto fail;
+fail_rd:
+	perror("reading from squashfs");
+	goto fail;
+fail_trunc:
+	fputs("reading from squashfs: unexpected end of file\n", stderr);
+	goto fail;
+fail_bs:
+	fputs("found compressed block larger than block size\n", stderr);
+	goto fail;
+fail:
+	free(buffer);
+	return -1;
+}
diff --git a/unpack/unsquashfs.c b/unpack/unsquashfs.c
index b253bae..90a7583 100644
--- a/unpack/unsquashfs.c
+++ b/unpack/unsquashfs.c
@@ -1,8 +1,15 @@
 /* SPDX-License-Identifier: GPL-3.0-or-later */
 #include "unsquashfs.h"
 
+enum {
+	OP_NONE = 0,
+	OP_LS,
+	OP_CAT,
+};
+
 static struct option long_opts[] = {
 	{ "list", required_argument, NULL, 'l' },
+	{ "cat", required_argument, NULL, 'c' },
 	{ "help", no_argument, NULL, 'h' },
 	{ "version", no_argument, NULL, 'V' },
 };
@@ -17,6 +24,8 @@ static const char *help_string =
 "Possible options:\n"
 "  --list, -l <path>  Produce a directory listing for a given path in the\n"
 "                     squashfs image.\n"
+"  --cat, -c <path>   If the specified path is a regular file in the image,\n"
+"                     dump its contents to stdout.\n"
 "  --help, -h         Print help text and exit.\n"
 "  --version, -V      Print version information and exit.\n"
 "\n";
@@ -56,12 +65,35 @@ static tree_node_t *find_node(tree_node_t *n, const char *path)
 	return n;
 }
 
+static char *get_path(char *old, const char *arg)
+{
+	char *path;
+
+	free(old);
+
+	path = strdup(arg);
+	if (path == NULL) {
+		perror("processing arguments");
+		exit(EXIT_FAILURE);
+	}
+
+	if (canonicalize_name(path)) {
+		fprintf(stderr, "Invalid path: %s\n", arg);
+		free(path);
+		exit(EXIT_FAILURE);
+	}
+
+	return path;
+}
+
 int main(int argc, char **argv)
 {
-	int i, fd, status = EXIT_FAILURE;
-	char *lspath = NULL;
+	int i, fd, status = EXIT_FAILURE, op = OP_NONE;
+	frag_reader_t *frag = NULL;
+	char *cmdpath = NULL;
 	sqfs_super_t super;
 	compressor_t *cmp;
+	tree_node_t *n;
 	fstree_t fs;
 
 	for (;;) {
@@ -70,22 +102,13 @@ int main(int argc, char **argv)
 			break;
 
 		switch (i) {
+		case 'c':
+			op = OP_CAT;
+			cmdpath = get_path(cmdpath, optarg);
+			break;
 		case 'l':
-			if (optarg != NULL) {
-				free(lspath);
-
-				lspath = strdup(optarg ? optarg : "");
-				if (lspath == NULL) {
-					perror("processing arguments");
-					return EXIT_FAILURE;
-				}
-
-				if (canonicalize_name(lspath)) {
-					fprintf(stderr, "Invalid pth: %s\n",
-						optarg);
-					goto out_cmd;
-				}
-			}
+			op = OP_LS;
+			cmdpath = get_path(cmdpath, optarg);
 			break;
 		case 'h':
 			printf(help_string, __progname);
@@ -100,10 +123,14 @@ int main(int argc, char **argv)
 		}
 	}
 
+	if (op == OP_NONE) {
+		fputs("No opteration specified\n", stderr);
+		goto fail_arg;
+	}
+
 	if (optind >= argc) {
-		fprintf(stderr, "Usage: %s [OPTIONS] <filename>\n",
-			__progname);
-		goto out_cmd;
+		fputs("Missing image argument\n", stderr);
+		goto fail_arg;
 	}
 
 	fd = open(argv[optind], O_RDONLY);
@@ -145,29 +172,57 @@ int main(int argc, char **argv)
 	if (read_fstree(&fs, fd, &super, cmp))
 		goto out_cmp;
 
-	if (lspath != NULL) {
-		tree_node_t *n = find_node(fs.root, lspath);
-
+	switch (op) {
+	case OP_LS:
+		n = find_node(fs.root, cmdpath);
 		if (n == NULL) {
-			perror(lspath);
+			perror(cmdpath);
 			goto out_fs;
 		}
 
 		list_files(n);
+		break;
+	case OP_CAT:
+		n = find_node(fs.root, cmdpath);
+		if (n == NULL) {
+			perror(cmdpath);
+			goto out_fs;
+		}
+
+		if (!S_ISREG(n->mode)) {
+			fprintf(stderr, "/%s: not a regular file\n", cmdpath);
+			goto out_fs;
+		}
+
+		if (super.fragment_entry_count > 0 &&
+		    super.fragment_table_start < super.bytes_used &&
+		    !(super.flags & SQFS_FLAG_NO_FRAGMENTS)) {
+			frag = frag_reader_create(&super, fd, cmp);
+			if (frag == NULL)
+				goto out_fs;
+		}
+
+		if (extract_file(n->data.file, cmp, super.block_size,
+				 frag, fd, STDOUT_FILENO)) {
+			goto out_fs;
+		}
+		break;
 	}
 
 	status = EXIT_SUCCESS;
 out_fs:
+	if (frag != NULL)
+		frag_reader_destroy(frag);
 	fstree_cleanup(&fs);
 out_cmp:
 	cmp->destroy(cmp);
 out_fd:
 	close(fd);
 out_cmd:
-	free(lspath);
+	free(cmdpath);
 	return status;
 fail_arg:
 	fprintf(stderr, "Try `%s --help' for more information.\n", __progname);
-	free(lspath);
+	free(cmdpath);
 	return EXIT_FAILURE;
 }
diff --git a/unpack/unsquashfs.h b/unpack/unsquashfs.h
index 52b3889..0e8854b 100644
--- a/unpack/unsquashfs.h
+++ b/unpack/unsquashfs.h
@@ -3,6 +3,7 @@
 #define UNSQUASHFS_H
 
 #include "meta_reader.h"
+#include "frag_reader.h"
 #include "squashfs.h"
 #include "compress.h"
 #include "id_table.h"
@@ -27,4 +28,7 @@ int read_fstree(fstree_t *out, int fd, sqfs_super_t *super, compressor_t *cmp);
 
 void list_files(tree_node_t *node);
 
+int extract_file(file_info_t *fi, compressor_t *cmp, size_t block_size,
+		 frag_reader_t *frag, int sqfsfd, int outfd);
+
 #endif /* UNSQUASHFS_H */