Add directory reader data structure

This moves a lot of the stuff that is done manually in the tree
deserializer to a generic helper in libsquashfs.

Due to how the fstree is implemented, as a work around, the inode needs
to be temporarily stored in the tree node, but some of the directory
details could be removed.

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

4 files changed, 331 insertions, 105 deletions

diff --git a/lib/sqfs/Makemodule.am b/lib/sqfs/Makemodule.am
index 7d228a4..807c104 100644
--- a/lib/sqfs/Makemodule.am
+++ b/lib/sqfs/Makemodule.am
@@ -14,6 +14,7 @@ libsquashfs_la_SOURCES += lib/sqfs/read_inode.c lib/sqfs/write_inode.c
 libsquashfs_la_SOURCES += lib/sqfs/dir_writer.c lib/sqfs/xattr_reader.c
 libsquashfs_la_SOURCES += lib/sqfs/read_table.c lib/sqfs/comp/compressor.c
 libsquashfs_la_SOURCES += lib/sqfs/io_stdin.c lib/sqfs/comp/internal.h
+libsquashfs_la_SOURCES += lib/sqfs/dir_reader.c
 libsquashfs_la_SOURCES += lib/sqfs/blk_proc/process_block.c lib/sqfs/io.c
 libsquashfs_la_SOURCES += lib/sqfs/blk_proc/internal.h
 libsquashfs_la_CPPFLAGS = $(AM_CPPFLAGS)
diff --git a/lib/sqfs/dir_reader.c b/lib/sqfs/dir_reader.c
new file mode 100644
index 0000000..d4bfd27
--- /dev/null
+++ b/lib/sqfs/dir_reader.c
@@ -0,0 +1,266 @@
+/* SPDX-License-Identifier: LGPL-3.0-or-later */
+/*
+ * fs_reader.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#define SQFS_BUILDING_DLL
+#include "config.h"
+
+#include "sqfs/meta_reader.h"
+#include "sqfs/compress.h"
+#include "sqfs/super.h"
+#include "sqfs/inode.h"
+#include "sqfs/error.h"
+#include "sqfs/dir.h"
+#include "util.h"
+
+#include <string.h>
+#include <stdlib.h>
+
+struct sqfs_dir_reader_t {
+	sqfs_meta_reader_t *meta_dir;
+	sqfs_meta_reader_t *meta_inode;
+	const sqfs_super_t *super;
+
+	sqfs_dir_header_t hdr;
+	uint64_t dir_block_start;
+	size_t entries;
+	size_t size;
+
+	size_t start_size;
+	uint16_t dir_offset;
+	uint16_t inode_offset;
+};
+
+sqfs_dir_reader_t *sqfs_dir_reader_create(const sqfs_super_t *super,
+					  sqfs_compressor_t *cmp,
+					  sqfs_file_t *file)
+{
+	sqfs_dir_reader_t *rd = calloc(1, sizeof(*rd));
+	uint64_t start, limit;
+
+	if (rd == NULL)
+		return NULL;
+
+	start = super->inode_table_start;
+	limit = super->directory_table_start;
+
+	rd->meta_inode = sqfs_meta_reader_create(file, cmp, start, limit);
+
+	if (rd->meta_inode == NULL) {
+		free(rd);
+		return NULL;
+	}
+
+	start = super->directory_table_start;
+	limit = super->id_table_start;
+
+	if (super->fragment_table_start < limit)
+		limit = super->fragment_table_start;
+
+	if (super->export_table_start < limit)
+		limit = super->export_table_start;
+
+	rd->meta_dir = sqfs_meta_reader_create(file, cmp, start, limit);
+
+	if (rd->meta_dir == NULL) {
+		sqfs_meta_reader_destroy(rd->meta_inode);
+		free(rd);
+		return NULL;
+	}
+
+	rd->super = super;
+	return rd;
+}
+
+void sqfs_dir_reader_destroy(sqfs_dir_reader_t *rd)
+{
+	sqfs_meta_reader_destroy(rd->meta_inode);
+	sqfs_meta_reader_destroy(rd->meta_dir);
+	free(rd);
+}
+
+int sqfs_dir_reader_open_dir(sqfs_dir_reader_t *rd,
+			     const sqfs_inode_generic_t *inode)
+{
+	uint64_t block_start;
+	size_t size, offset;
+
+	if (inode->base.type == SQFS_INODE_DIR) {
+		size = inode->data.dir.size;
+		offset = inode->data.dir.offset;
+		block_start = inode->data.dir.start_block;
+	} else if (inode->base.type == SQFS_INODE_EXT_DIR) {
+		size = inode->data.dir_ext.size;
+		offset = inode->data.dir_ext.offset;
+		block_start = inode->data.dir_ext.start_block;
+	} else {
+		return SQFS_ERROR_NOT_DIR;
+	}
+
+	memset(&rd->hdr, 0, sizeof(rd->hdr));
+	rd->size = size;
+	rd->entries = 0;
+
+	if (rd->size <= sizeof(rd->hdr))
+		return 0;
+
+	block_start += rd->super->directory_table_start;
+
+	rd->dir_block_start = block_start;
+	rd->dir_offset = offset;
+	rd->start_size = size;
+
+	return sqfs_meta_reader_seek(rd->meta_dir, block_start, offset);
+}
+
+int sqfs_dir_reader_read(sqfs_dir_reader_t *rd, sqfs_dir_entry_t **out)
+{
+	sqfs_dir_entry_t *ent;
+	size_t count;
+	int err;
+
+	if (!rd->entries) {
+		if (rd->size < sizeof(rd->hdr))
+			return 1;
+
+		err = sqfs_meta_reader_read_dir_header(rd->meta_dir, &rd->hdr);
+		if (err)
+			return err;
+
+		rd->size -= sizeof(rd->hdr);
+		rd->entries = rd->hdr.count + 1;
+	}
+
+	err = sqfs_meta_reader_read_dir_ent(rd->meta_dir, &ent);
+	if (err)
+		return err;
+
+	count = sizeof(*ent) + strlen((const char *)ent->name);
+
+	if (count > rd->size) {
+		rd->size = 0;
+		rd->entries = 0;
+	} else {
+		rd->size -= count;
+		rd->entries -= 1;
+	}
+
+	rd->inode_offset = ent->offset;
+	*out = ent;
+	return 0;
+}
+
+int sqfs_dir_reader_rewind(sqfs_dir_reader_t *rd)
+{
+	memset(&rd->hdr, 0, sizeof(rd->hdr));
+	rd->size = rd->start_size;
+	rd->entries = 0;
+
+	return sqfs_meta_reader_seek(rd->meta_dir, rd->dir_block_start,
+				     rd->dir_offset);
+}
+
+int sqfs_dir_reader_find(sqfs_dir_reader_t *rd, const char *name)
+{
+	sqfs_dir_entry_t *ent;
+	int ret;
+
+	if (rd->size != rd->start_size) {
+		ret = sqfs_dir_reader_rewind(rd);
+		if (ret)
+			return ret;
+	}
+
+	do {
+		ret = sqfs_dir_reader_read(rd, &ent);
+		if (ret < 0)
+			return ret;
+		if (ret > 0)
+			return SQFS_ERROR_NO_ENTRY;
+
+		ret = strcmp((const char *)ent->name, name);
+		free(ent);
+	} while (ret < 0);
+
+	return ret == 0 ? 0 : SQFS_ERROR_NO_ENTRY;
+}
+
+int sqfs_dir_reader_get_inode(sqfs_dir_reader_t *rd,
+			      sqfs_inode_generic_t **inode)
+{
+	uint64_t block_start;
+
+	block_start = rd->hdr.start_block;
+
+	return sqfs_meta_reader_read_inode(rd->meta_inode, rd->super,
+					   block_start, rd->inode_offset,
+					   inode);
+}
+
+int sqfs_dir_reader_get_root_inode(sqfs_dir_reader_t *rd,
+				   sqfs_inode_generic_t **inode)
+{
+	uint64_t block_start = rd->super->root_inode_ref >> 16;
+	uint16_t offset = rd->super->root_inode_ref & 0xFFFF;
+
+	return sqfs_meta_reader_read_inode(rd->meta_inode, rd->super,
+					   block_start, offset, inode);
+}
+
+int sqfs_dir_reader_find_by_path(sqfs_dir_reader_t *rd, const char *path,
+				 sqfs_inode_generic_t **out)
+{
+	sqfs_inode_generic_t *inode;
+	sqfs_dir_entry_t *ent;
+	const char *ptr;
+	int ret;
+
+	ret = sqfs_dir_reader_get_root_inode(rd, &inode);
+	if (ret)
+		return ret;
+
+	while (*path != '\0') {
+		if (*path == '/' || *path == '\\') {
+			while (*path == '/' || *path == '\\')
+				++path;
+			continue;
+		}
+
+		ret = sqfs_dir_reader_open_dir(rd, inode);
+		free(inode);
+		if (ret)
+			return ret;
+
+		ptr = strchr(path, '/');
+		if (ptr == NULL)
+			ptr = strchrnul(path, '\\');
+
+		do {
+			ret = sqfs_dir_reader_read(rd, &ent);
+			if (ret < 0)
+				return ret;
+
+			if (ret == 0) {
+				ret = strncmp((const char *)ent->name,
+					      path, ptr - path);
+				if (ret == 0)
+					ret = ent->name[ptr - path];
+				free(ent);
+			}
+		} while (ret < 0);
+
+		if (ret > 0)
+			return SQFS_ERROR_NO_ENTRY;
+
+		ret = sqfs_dir_reader_get_inode(rd, &inode);
+		if (ret)
+			return ret;
+
+		path = ptr;
+	}
+
+	*out = inode;
+	return 0;
+}
diff --git a/lib/sqfshelper/deserialize_fstree.c b/lib/sqfshelper/deserialize_fstree.c
index ef000d4..e0f53e7 100644
--- a/lib/sqfshelper/deserialize_fstree.c
+++ b/lib/sqfshelper/deserialize_fstree.c
@@ -93,102 +93,70 @@ fail:
 	return false;
 }
 
-static int fill_dir(sqfs_meta_reader_t *ir, sqfs_meta_reader_t *dr,
-		    tree_node_t *root, sqfs_super_t *super,
-		    sqfs_id_table_t *idtbl,
+static int fill_dir(sqfs_dir_reader_t *dr,
+		    tree_node_t *root, sqfs_id_table_t *idtbl,
 		    fstree_t *fs, sqfs_xattr_reader_t *xr, int flags)
 {
 	sqfs_inode_generic_t *inode;
-	sqfs_dir_header_t hdr;
 	sqfs_dir_entry_t *ent;
 	tree_node_t *n, *prev;
-	uint64_t block_start;
-	size_t size, diff;
-	uint32_t i;
 	int err;
 
-	size = root->data.dir->size;
-	if (size <= sizeof(hdr))
-		return 0;
-
-	block_start = root->data.dir->start_block;
-	block_start += super->directory_table_start;
-
-	if (sqfs_meta_reader_seek(dr, block_start,
-				  root->data.dir->block_offset)) {
-		return -1;
-	}
-
-	while (size > sizeof(hdr)) {
-		if (sqfs_meta_reader_read_dir_header(dr, &hdr))
+	for (;;) {
+		err = sqfs_dir_reader_read(dr, &ent);
+		if (err > 0)
+			break;
+		if (err < 0)
 			return -1;
 
-		size -= sizeof(hdr);
-
-		for (i = 0; i <= hdr.count && size > sizeof(*ent); ++i) {
-			if (sqfs_meta_reader_read_dir_ent(dr, &ent))
-				return -1;
-
-			diff = sizeof(*ent) + strlen((char *)ent->name);
-			if (diff > size) {
-				free(ent);
-				break;
-			}
-			size -= diff;
-
-			if (should_skip(ent->type, flags)) {
-				free(ent);
-				continue;
-			}
+		if (should_skip(ent->type, flags)) {
+			free(ent);
+			continue;
+		}
 
-			if (!is_name_sane((const char *)ent->name)) {
-				free(ent);
-				continue;
-			}
+		if (!is_name_sane((const char *)ent->name)) {
+			free(ent);
+			continue;
+		}
 
-			err = sqfs_meta_reader_read_inode(ir, super,
-							  hdr.start_block,
-							  ent->offset, &inode);
-			if (err) {
-				free(ent);
-				return err;
-			}
+		err = sqfs_dir_reader_get_inode(dr, &inode);
+		if (err) {
+			free(ent);
+			return err;
+		}
 
-			n = tree_node_from_inode(inode, idtbl,
-						 (char *)ent->name);
+		n = tree_node_from_inode(inode, idtbl, (char *)ent->name);
+		if (n == NULL) {
+			free(ent);
+			free(inode);
+			return -1;
+		}
 
-			if (n == NULL) {
-				free(ent);
-				free(inode);
-				return -1;
-			}
+		if (node_would_be_own_parent(root, n)) {
+			fputs("WARNING: Found a directory that "
+			      "contains itself, skipping loop back "
+			      "reference!\n", stderr);
+			free(n);
+			free(ent);
+			free(inode);
+			continue;
+		}
 
-			if (node_would_be_own_parent(root, n)) {
-				fputs("WARNING: Found a directory that "
-				      "contains itself, skipping loop back "
-				      "reference!\n", stderr);
+		if (flags & RDTREE_READ_XATTR) {
+			if (restore_xattr(xr, fs, n, inode)) {
 				free(n);
 				free(ent);
 				free(inode);
-				continue;
+				return -1;
 			}
+		}
 
-			if (flags & RDTREE_READ_XATTR) {
-				if (restore_xattr(xr, fs, n, inode)) {
-					free(n);
-					free(ent);
-					free(inode);
-					return -1;
-				}
-			}
+		free(ent);
 
-			free(ent);
-			free(inode);
-
-			n->parent = root;
-			n->next = root->data.dir->children;
-			root->data.dir->children = n;
-		}
+		n->inode = inode;
+		n->parent = root;
+		n->next = root->data.dir->children;
+		root->data.dir->children = n;
 	}
 
 	n = root->data.dir->children;
@@ -196,11 +164,16 @@ static int fill_dir(sqfs_meta_reader_t *ir, sqfs_meta_reader_t *dr,
 
 	while (n != NULL) {
 		if (S_ISDIR(n->mode)) {
-			if (fill_dir(ir, dr, n, super, idtbl, fs, xr, flags))
+			err = sqfs_dir_reader_open_dir(dr, n->inode);
+			if (err)
+				return -1;
+
+			if (fill_dir(dr, n, idtbl, fs, xr, flags))
 				return -1;
 
 			if (n->data.dir->children == NULL &&
 			    (flags & RDTREE_NO_EMPTY)) {
+				free(n->inode);
 				if (prev == NULL) {
 					root->data.dir->children = n->next;
 					free(n);
@@ -214,6 +187,9 @@ static int fill_dir(sqfs_meta_reader_t *ir, sqfs_meta_reader_t *dr,
 			}
 		}
 
+		free(n->inode);
+		n->inode = NULL;
+
 		prev = n;
 		n = n->next;
 	}
@@ -224,29 +200,15 @@ static int fill_dir(sqfs_meta_reader_t *ir, sqfs_meta_reader_t *dr,
 int deserialize_fstree(fstree_t *out, sqfs_super_t *super,
 		       sqfs_compressor_t *cmp, sqfs_file_t *file, int flags)
 {
-	uint64_t block_start, limit;
-	sqfs_meta_reader_t *ir, *dr;
 	sqfs_inode_generic_t *root;
 	sqfs_xattr_reader_t *xr;
 	sqfs_id_table_t *idtbl;
+	sqfs_dir_reader_t *dr;
 	int status = -1;
-	size_t offset;
-
-	ir = sqfs_meta_reader_create(file, cmp, super->inode_table_start,
-				     super->directory_table_start);
-	if (ir == NULL)
-		return -1;
-
-	limit = super->id_table_start;
-	if (super->export_table_start < limit)
-		limit = super->export_table_start;
-	if (super->fragment_table_start < limit)
-		limit = super->fragment_table_start;
 
-	dr = sqfs_meta_reader_create(file, cmp, super->directory_table_start,
-				     limit);
+	dr = sqfs_dir_reader_create(super, cmp, file);
 	if (dr == NULL)
-		goto out_ir;
+		return -1;
 
 	idtbl = sqfs_id_table_create();
 	if (idtbl == NULL)
@@ -262,9 +224,7 @@ int deserialize_fstree(fstree_t *out, sqfs_super_t *super,
 	if (sqfs_xattr_reader_load_locations(xr))
 		goto out_xr;
 
-	block_start = super->root_inode_ref >> 16;
-	offset = super->root_inode_ref & 0xFFFF;
-	if (sqfs_meta_reader_read_inode(ir, super, block_start, offset, &root))
+	if (sqfs_dir_reader_get_root_inode(dr, &root))
 		goto out_xr;
 
 	if (root->base.type != SQFS_INODE_DIR &&
@@ -308,9 +268,14 @@ int deserialize_fstree(fstree_t *out, sqfs_super_t *super,
 		}
 	}
 
+	if (sqfs_dir_reader_open_dir(dr, root)) {
+		free(root);
+		goto fail_fs;
+	}
+
 	free(root);
 
-	if (fill_dir(ir, dr, out->root, super, idtbl, out, xr, flags))
+	if (fill_dir(dr, out->root, idtbl, out, xr, flags))
 		goto fail_fs;
 
 	tree_node_sort_recursive(out->root);
@@ -321,9 +286,7 @@ out_xr:
 out_id:
 	sqfs_id_table_destroy(idtbl);
 out_dr:
-	sqfs_meta_reader_destroy(dr);
-out_ir:
-	sqfs_meta_reader_destroy(ir);
+	sqfs_dir_reader_destroy(dr);
 	return status;
 fail_fs:
 	fstree_cleanup(out);
diff --git a/lib/sqfshelper/tree_node_from_inode.c b/lib/sqfshelper/tree_node_from_inode.c
index f5b643c..3ada722 100644
--- a/lib/sqfshelper/tree_node_from_inode.c
+++ b/lib/sqfshelper/tree_node_from_inode.c
@@ -82,16 +82,12 @@ tree_node_t *tree_node_from_inode(sqfs_inode_generic_t *inode,
 		out->name += sizeof(dir_info_t);
 
 		out->data.dir->size = inode->data.dir.size;
-		out->data.dir->start_block = inode->data.dir.start_block;
-		out->data.dir->block_offset = inode->data.dir.offset;
 		break;
 	case SQFS_INODE_EXT_DIR:
 		out->data.dir = (dir_info_t *)out->payload;
 		out->name += sizeof(dir_info_t);
 
 		out->data.dir->size = inode->data.dir_ext.size;
-		out->data.dir->start_block = inode->data.dir_ext.start_block;
-		out->data.dir->block_offset = inode->data.dir_ext.offset;
 		break;
 	case SQFS_INODE_FILE:
 		out->data.file = (file_info_t *)out->payload;