Move file extraction code to generic data_reader_t in libsquashfs.a

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

8 files changed, 205 insertions, 157 deletions

diff --git a/include/data_reader.h b/include/data_reader.h
new file mode 100644
index 0000000..68788b4
--- /dev/null
+++ b/include/data_reader.h
@@ -0,0 +1,33 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#ifndef DATA_READER_H
+#define DATA_READER_H
+
+#include "squashfs.h"
+#include "compress.h"
+#include "fstree.h"
+
+typedef struct data_reader_t data_reader_t;
+
+/*
+  Create a data reader for accessing data blocks in a squashfs image.
+
+  Internally creates a fragment_reader_t (if applicable) to resolve
+  fragment indices.
+
+  Prints error messsages to stderr on failure.
+ */
+data_reader_t *data_reader_create(int fd, sqfs_super_t *super,
+				  compressor_t *cmp);
+
+void data_reader_destroy(data_reader_t *data);
+
+/*
+  Use a file_info_t to locate and extract all blocks of the coresponding
+  file and its fragment, if it has one. The entire data is dumped to the
+  given file descriptor.
+
+  Returns 0 on success, prints error messages to stderr on failure.
+ */
+int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd);
+
+#endif /* DATA_READER_H */
diff --git a/lib/Makemodule.am b/lib/Makemodule.am
index 3c1ef12..ddc78f6 100644
--- a/lib/Makemodule.am
+++ b/lib/Makemodule.am
@@ -23,8 +23,8 @@ libsquashfs_a_SOURCES += lib/sqfs/serialize_fstree.c
 libsquashfs_a_SOURCES += lib/sqfs/tree_node_from_inode.c
 libsquashfs_a_SOURCES += lib/sqfs/deserialize_fstree.c
 libsquashfs_a_SOURCES += lib/sqfs/data_writer.c lib/sqfs/write_xattr.c
-libsquashfs_a_SOURCES += include/data_writer.h
-libsquashfs_a_SOURCES += include/frag_reader.h
+libsquashfs_a_SOURCES += include/data_writer.h include/frag_reader.h
+libsquashfs_a_SOURCES += include/data_reader.h lib/sqfs/data_reader.c
 
 libutil_a_SOURCES = lib/util/canonicalize_name.c lib/util/write_retry.c
 libutil_a_SOURCES += lib/util/read_retry.c include/util.h
diff --git a/lib/sqfs/data_reader.c b/lib/sqfs/data_reader.c
new file mode 100644
index 0000000..22413f2
--- /dev/null
+++ b/lib/sqfs/data_reader.c
@@ -0,0 +1,157 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+#include "data_reader.h"
+#include "frag_reader.h"
+#include "util.h"
+
+#include <stdlib.h>
+#include <unistd.h>
+#include <stdio.h>
+
+struct data_reader_t {
+	compressor_t *cmp;
+	size_t block_size;
+	frag_reader_t *frag;
+	int sqfsfd;
+
+	void *buffer;
+	void *scratch;
+};
+
+data_reader_t *data_reader_create(int fd, sqfs_super_t *super,
+				  compressor_t *cmp)
+{
+	data_reader_t *data = calloc(1, sizeof(*data) + 2 * super->block_size);
+
+	if (data == NULL) {
+		perror("creating data reader");
+		return data;
+	}
+
+	if (super->fragment_entry_count > 0 &&
+	    !(super->flags & SQFS_FLAG_NO_FRAGMENTS)) {
+		if (super->fragment_table_start >= super->bytes_used) {
+			fputs("Fragment table start is past end of file\n",
+			      stderr);
+			free(data);
+			return NULL;
+		}
+
+		data->frag = frag_reader_create(super, fd, cmp);
+		if (data->frag == NULL) {
+			free(data);
+			return NULL;
+		}
+	}
+
+	data->buffer = (char *)data + sizeof(*data);
+	data->scratch = (char *)data->buffer + super->block_size;
+	data->sqfsfd = fd;
+	data->block_size = super->block_size;
+	data->cmp = cmp;
+	return data;
+}
+
+void data_reader_destroy(data_reader_t *data)
+{
+	if (data->frag != NULL)
+		frag_reader_destroy(data->frag);
+	free(data);
+}
+
+int data_reader_dump_file(data_reader_t *data, file_info_t *fi, int outfd)
+{
+	size_t i, count, fragsz;
+	bool compressed;
+	uint32_t bs;
+	ssize_t ret;
+	void *ptr;
+
+	count = fi->size / data->block_size;
+
+	if (count > 0) {
+		if (lseek(data->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 > data->block_size)
+				goto fail_bs;
+
+			ret = read_retry(data->sqfsfd, data->buffer, bs);
+			if (ret < 0)
+				goto fail_rd;
+
+			if ((size_t)ret < bs)
+				goto fail_trunc;
+
+			if (compressed) {
+				ret = data->cmp->do_block(data->cmp,
+							  data->buffer, bs,
+							  data->scratch,
+							  data->block_size);
+				if (ret <= 0)
+					return -1;
+
+				bs = ret;
+				ptr = data->scratch;
+			} else {
+				ptr = data->buffer;
+			}
+
+			ret = write_retry(outfd, ptr, bs);
+			if (ret < 0)
+				goto fail_wr;
+
+			if ((size_t)ret < bs)
+				goto fail_wr_trunc;
+		}
+	}
+
+	fragsz = fi->size % data->block_size;
+
+	if (fragsz > 0) {
+		if (data->frag == NULL)
+			goto fail_frag;
+
+		if (frag_reader_read(data->frag, fi->fragment,
+				     fi->fragment_offset, data->buffer,
+				     fragsz)) {
+			return -1;
+		}
+
+		ret = write_retry(outfd, data->buffer, fragsz);
+		if (ret < 0)
+			goto fail_wr;
+
+		if ((size_t)ret < fragsz)
+			goto fail_wr_trunc;
+	}
+
+	return 0;
+fail_seek:
+	perror("seek on squashfs");
+	return -1;
+fail_wr:
+	perror("writing uncompressed block");
+	return -1;
+fail_wr_trunc:
+	fputs("writing uncompressed block: truncated write\n", stderr);
+	return -1;
+fail_rd:
+	perror("reading from squashfs");
+	return -1;
+fail_trunc:
+	fputs("reading from squashfs: unexpected end of file\n", stderr);
+	return -1;
+fail_bs:
+	fputs("found compressed block larger than block size\n", stderr);
+	return -1;
+fail_frag:
+	fputs("file not a multiple of block size but no fragments available\n",
+	      stderr);
+	return -1;
+}
diff --git a/unpack/Makemodule.am b/unpack/Makemodule.am
index 5f4269e..c70026d 100644
--- a/unpack/Makemodule.am
+++ b/unpack/Makemodule.am
@@ -1,5 +1,5 @@
 rdsquashfs_SOURCES = unpack/rdsquashfs.c unpack/rdsquashfs.h
-rdsquashfs_SOURCES += unpack/list_files.c unpack/extract_file.c
+rdsquashfs_SOURCES += unpack/list_files.c
 rdsquashfs_SOURCES += unpack/restore_fstree.c unpack/describe.c
 rdsquashfs_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
 
diff --git a/unpack/extract_file.c b/unpack/extract_file.c
deleted file mode 100644
index 62d1086..0000000
--- a/unpack/extract_file.c
+++ /dev/null
@@ -1,93 +0,0 @@
-/* SPDX-License-Identifier: GPL-3.0-or-later */
-#include "rdsquashfs.h"
-
-int extract_file(file_info_t *fi, unsqfs_info_t *info, int outfd)
-{
-	size_t i, count, fragsz;
-	bool compressed;
-	uint32_t bs;
-	ssize_t ret;
-	void *ptr;
-
-	count = fi->size / info->block_size;
-
-	if (count > 0) {
-		if (lseek(info->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 > info->block_size)
-				goto fail_bs;
-
-			ret = read_retry(info->sqfsfd, info->buffer, bs);
-			if (ret < 0)
-				goto fail_rd;
-
-			if ((size_t)ret < bs)
-				goto fail_trunc;
-
-			if (compressed) {
-				ret = info->cmp->do_block(info->cmp,
-							  info->buffer, bs,
-							  info->scratch,
-							  info->block_size);
-				if (ret <= 0)
-					return -1;
-
-				bs = ret;
-				ptr = info->scratch;
-			} else {
-				ptr = info->buffer;
-			}
-
-			ret = write_retry(outfd, ptr, bs);
-			if (ret < 0)
-				goto fail_wr;
-
-			if ((size_t)ret < bs)
-				goto fail_wr_trunc;
-		}
-	}
-
-	fragsz = fi->size % info->block_size;
-
-	if (fragsz > 0) {
-		if (frag_reader_read(info->frag, fi->fragment,
-				     fi->fragment_offset, info->buffer,
-				     fragsz)) {
-			return -1;
-		}
-
-		ret = write_retry(outfd, info->buffer, fragsz);
-		if (ret < 0)
-			goto fail_wr;
-
-		if ((size_t)ret < fragsz)
-			goto fail_wr_trunc;
-	}
-
-	return 0;
-fail_seek:
-	perror("seek on squashfs");
-	return -1;
-fail_wr:
-	perror("writing uncompressed block");
-	return -1;
-fail_wr_trunc:
-	fputs("writing uncompressed block: truncated write\n", stderr);
-	return -1;
-fail_rd:
-	perror("reading from squashfs");
-	return -1;
-fail_trunc:
-	fputs("reading from squashfs: unexpected end of file\n", stderr);
-	return -1;
-fail_bs:
-	fputs("found compressed block larger than block size\n", stderr);
-	return -1;
-}
diff --git a/unpack/rdsquashfs.c b/unpack/rdsquashfs.c
index 81da226..7fd2947 100644
--- a/unpack/rdsquashfs.c
+++ b/unpack/rdsquashfs.c
@@ -121,43 +121,6 @@ static char *get_path(char *old, const char *arg)
 	return path;
 }
 
-static int alloc_buffers(unsqfs_info_t *info, sqfs_super_t *super)
-{
-	info->buffer = malloc(super->block_size);
-	if (info->buffer == NULL) {
-		perror("allocating block buffer");
-		return -1;
-	}
-
-	info->scratch = malloc(super->block_size);
-	if (info->scratch == NULL) {
-		perror("allocating scrtach buffer for extracting blocks");
-		return -1;
-	}
-
-	return 0;
-}
-
-static int load_fragment_table(unsqfs_info_t *info, sqfs_super_t *super)
-{
-	if (super->fragment_entry_count == 0)
-		return 0;
-
-	if (super->flags & SQFS_FLAG_NO_FRAGMENTS)
-		return 0;
-
-	if (super->fragment_table_start >= super->bytes_used) {
-		fputs("Fragment table start is past end of file\n", stderr);
-		return -1;
-	}
-
-	info->frag = frag_reader_create(super, info->sqfsfd, info->cmp);
-	if (info->frag == NULL)
-		return -1;
-
-	return 0;
-}
-
 int main(int argc, char **argv)
 {
 	int i, status = EXIT_FAILURE, op = OP_NONE;
@@ -281,8 +244,6 @@ int main(int argc, char **argv)
 		goto out_cmp;
 	}
 
-	info.block_size = super.block_size;
-
 	switch (op) {
 	case OP_LS:
 		n = find_node(fs.root, cmdpath);
@@ -305,14 +266,14 @@ int main(int argc, char **argv)
 			goto out_fs;
 		}
 
-		if (load_fragment_table(&info, &super))
+		info.data = data_reader_create(info.sqfsfd, &super, info.cmp);
+		if (info.data == NULL)
 			goto out_fs;
 
-		if (alloc_buffers(&info, &super))
-			goto out_fs;
-
-		if (extract_file(n->data.file, &info, STDOUT_FILENO))
+		if (data_reader_dump_file(info.data, n->data.file,
+					  STDOUT_FILENO)) {
 			goto out_fs;
+		}
 		break;
 	case OP_UNPACK:
 		n = find_node(fs.root, cmdpath);
@@ -321,10 +282,8 @@ int main(int argc, char **argv)
 			goto out_fs;
 		}
 
-		if (load_fragment_table(&info, &super))
-			goto out_fs;
-
-		if (alloc_buffers(&info, &super))
+		info.data = data_reader_create(info.sqfsfd, &super, info.cmp);
+		if (info.data == NULL)
 			goto out_fs;
 
 		if (restore_fstree(unpack_root, n, &info))
@@ -337,8 +296,8 @@ int main(int argc, char **argv)
 
 	status = EXIT_SUCCESS;
 out_fs:
-	if (info.frag != NULL)
-		frag_reader_destroy(info.frag);
+	if (info.data != NULL)
+		data_reader_destroy(info.data);
 	fstree_cleanup(&fs);
 out_cmp:
 	info.cmp->destroy(info.cmp);
@@ -346,8 +305,6 @@ out_fd:
 	close(info.sqfsfd);
 out_cmd:
 	free(cmdpath);
-	free(info.buffer);
-	free(info.scratch);
 	return status;
 fail_arg:
 	fprintf(stderr, "Try `%s --help' for more information.\n", __progname);
diff --git a/unpack/rdsquashfs.h b/unpack/rdsquashfs.h
index a615bb5..b927bab 100644
--- a/unpack/rdsquashfs.h
+++ b/unpack/rdsquashfs.h
@@ -3,7 +3,7 @@
 #define RDSQUASHFS_H
 
 #include "meta_reader.h"
-#include "frag_reader.h"
+#include "data_reader.h"
 #include "highlevel.h"
 #include "squashfs.h"
 #include "compress.h"
@@ -27,21 +27,15 @@ enum UNPACK_FLAGS {
 };
 
 typedef struct {
+	data_reader_t *data;
 	compressor_t *cmp;
-	size_t block_size;
-	frag_reader_t *frag;
 	int rdtree_flags;
 	int sqfsfd;
 	int flags;
-
-	void *buffer;
-	void *scratch;
 } unsqfs_info_t;
 
 void list_files(tree_node_t *node);
 
-int extract_file(file_info_t *fi, unsqfs_info_t *info, int outfd);
-
 int restore_fstree(const char *rootdir, tree_node_t *root,
 		   unsqfs_info_t *info);
 
diff --git a/unpack/restore_fstree.c b/unpack/restore_fstree.c
index 355bbb5..a39ec97 100644
--- a/unpack/restore_fstree.c
+++ b/unpack/restore_fstree.c
@@ -63,7 +63,7 @@ static int create_node(tree_node_t *n, unsqfs_info_t *info)
 			return -1;
 		}
 
-		if (extract_file(n->data.file, info, fd)) {
+		if (data_reader_dump_file(info->data, n->data.file, fd)) {
 			close(fd);
 			return -1;
 		}