Add option to sqfsdiff to extract regular files that are different

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

6 files changed, 95 insertions, 3 deletions

diff --git a/difftool/Makemodule.am b/difftool/Makemodule.am
index 1b0011b..83e5f83 100644
--- a/difftool/Makemodule.am
+++ b/difftool/Makemodule.am
@@ -6,6 +6,7 @@ fscompare_LDADD = libfstree.a libutil.a
 sqfsdiff_SOURCES = difftool/sqfsdiff.c difftool/difftool.h difftool/util.c
 sqfsdiff_SOURCES += difftool/compare_dir.c difftool/node_compare.c
 sqfsdiff_SOURCES += difftool/compare_files_sqfs.c difftool/super.c
+sqfsdiff_SOURCES += difftool/extract.c
 sqfsdiff_LDADD = libsquashfs.a libfstree.a libcompress.a libutil.a
 sqfsdiff_LDADD += $(XZ_LIBS) $(ZLIB_LIBS) $(LZO_LIBS) $(LZ4_LIBS) $(ZSTD_LIBS)
 
diff --git a/difftool/compare_files_sqfs.c b/difftool/compare_files_sqfs.c
index 76fae7a..32b97b1 100644
--- a/difftool/compare_files_sqfs.c
+++ b/difftool/compare_files_sqfs.c
@@ -15,7 +15,7 @@ int compare_files(file_info_t *a, file_info_t *b, const char *path)
 	ssize_t ret;
 
 	if (a->size != b->size)
-		return 1;
+		goto out_different;
 
 	if (compare_flags & COMPARE_NO_CONTENTS)
 		return 0;
@@ -41,8 +41,14 @@ int compare_files(file_info_t *a, file_info_t *b, const char *path)
 		}
 
 		if (memcmp(a_buf, b_buf, diff) != 0)
-			return 1;
+			goto out_different;
 	}
 
 	return 0;
+out_different:
+	if (compare_flags & COMPARE_EXTRACT_FILES) {
+		if (extract_files(a, b, path))
+			return -1;
+	}
+	return 1;
 }
diff --git a/difftool/difftool.h b/difftool/difftool.h
index 23a161c..bd04fd2 100644
--- a/difftool/difftool.h
+++ b/difftool/difftool.h
@@ -35,6 +35,7 @@ enum {
 	COMPARE_NO_CONTENTS = 0x04,
 	COMPARE_TIMESTAMP = 0x08,
 	COMPARE_INODE_NUM = 0x10,
+	COMPARE_EXTRACT_FILES = 0x20,
 };
 
 int compare_dir_entries(tree_node_t *a, tree_node_t *b);
@@ -47,4 +48,6 @@ int node_compare(tree_node_t *a, tree_node_t *b);
 
 int compare_super_blocks(const sqfs_super_t *a, const sqfs_super_t *b);
 
+int extract_files(file_info_t *a, file_info_t *b, const char *path);
+
 #endif /* DIFFTOOL_H */
diff --git a/difftool/extract.c b/difftool/extract.c
new file mode 100644
index 0000000..9d84265
--- /dev/null
+++ b/difftool/extract.c
@@ -0,0 +1,52 @@
+/* SPDX-License-Identifier: GPL-3.0-or-later */
+/*
+ * extract.c
+ *
+ * Copyright (C) 2019 David Oberhollenzer <goliath@infraroot.at>
+ */
+#include "difftool.h"
+
+static int extract(data_reader_t *data, file_info_t *fi,
+		   const char *path, char prefix)
+{
+	char *ptr, *temp = alloca(strlen(path) + 3);
+	int fd;
+
+	temp[0] = prefix;
+	temp[1] = '/';
+	strcpy(temp + 2, path);
+
+	ptr = strrchr(temp, '/');
+	*ptr = '\0';
+	if (mkdir_p(temp))
+		return -1;
+	*ptr = '/';
+
+	fd = open(temp, O_CREAT | O_EXCL | O_WRONLY, 0600);
+	if (fd < 0) {
+		perror(temp);
+		return -1;
+	}
+
+	if (data_reader_dump_file(data, fi, fd, true)) {
+		close(fd);
+		return -1;
+	}
+
+	close(fd);
+	return 0;
+}
+
+int extract_files(file_info_t *a, file_info_t *b, const char *path)
+{
+	if (a != NULL) {
+		if (extract(sqfs_a.data, a, path, 'a'))
+			return -1;
+	}
+
+	if (b != NULL) {
+		if (extract(sqfs_b.data, b, path, 'b'))
+			return -1;
+	}
+	return 0;
+}
diff --git a/difftool/sqfsdiff.c b/difftool/sqfsdiff.c
index 8818a37..4754557 100644
--- a/difftool/sqfsdiff.c
+++ b/difftool/sqfsdiff.c
@@ -13,11 +13,12 @@ static struct option long_opts[] = {
 	{ "timestamps", no_argument, NULL, 'T' },
 	{ "inode-num", no_argument, NULL, 'I' },
 	{ "super", no_argument, NULL, 'S' },
+	{ "extract", required_argument, NULL, 'e' },
 	{ "help", no_argument, NULL, 'h' },
 	{ "version", no_argument, NULL, 'V' },
 };
 
-static const char *short_opts = "OPCTIShV";
+static const char *short_opts = "OPCTISe:hV";
 
 static const char *usagestr =
 "Usage: sqfsdiff [OPTIONS...] <first> <second>\n"
@@ -45,6 +46,11 @@ static const char *usagestr =
 "  --inode-num, -I             Compare inode numbers of all files.\n"
 "  --super, -S                 Also compare metadata in super blocks.\n"
 "\n"
+"  --extract, -e <path>        Extract files that differ to the specified\n"
+"                              directory. Contents of the first image end up\n"
+"                              in a subdirectory 'a' and of the second image\n"
+"                              in a subdirectory 'b'.\n"
+"\n"
 "  --help, -h                  Print help text and exit.\n"
 "  --version, -V               Print version information and exit.\n"
 "\n";
@@ -55,6 +61,7 @@ const char *second_path;
 sqfs_reader_t sqfs_a;
 sqfs_reader_t sqfs_b;
 static bool compare_super = false;
+static const char *extract_dir;
 
 static void process_options(int argc, char **argv)
 {
@@ -84,6 +91,10 @@ static void process_options(int argc, char **argv)
 		case 'S':
 			compare_super = true;
 			break;
+		case 'e':
+			compare_flags |= COMPARE_EXTRACT_FILES;
+			extract_dir = optarg;
+			break;
 		case 'h':
 			fputs(usagestr, stdout);
 			exit(EXIT_SUCCESS);
@@ -125,6 +136,11 @@ int main(int argc, char **argv)
 
 	process_options(argc, argv);
 
+	if (extract_dir != NULL) {
+		if (mkdir_p(extract_dir))
+			return EXIT_FAILURE;
+	}
+
 	if (sqfs_reader_open(&sqfs_a, first_path, 0))
 		return 2;
 
@@ -133,6 +149,14 @@ int main(int argc, char **argv)
 		goto out_sqfs_a;
 	}
 
+	if (extract_dir != NULL) {
+		if (chdir(extract_dir)) {
+			perror(extract_dir);
+			ret = -1;
+			goto out;
+		}
+	}
+
 	ret = node_compare(sqfs_a.fs.root, sqfs_b.fs.root);
 	if (ret != 0)
 		goto out;
diff --git a/doc/sqfsdiff.1 b/doc/sqfsdiff.1
index ca3818f..5222c19 100644
--- a/doc/sqfsdiff.1
+++ b/doc/sqfsdiff.1
@@ -38,6 +38,12 @@ Compare inode numbers of all files.
 \fB\-\-super\fR, \fB\-S\fR
 Also compare metadata in super blocks.
 .TP
+\fB\-\-extract\fR, \fB\-e\fR <path>
+Extract files that exist in both images but have different contents to the
+specified directory. Contents of the first image end up in a sub directory
+named \fBa\fR and the contents of the second image in a sub directory
+named \fBb\fR.
+.TP
 \fB\-\-help\fR, \fB\-h\fR
 Print help text and exit.
 .TP