adding support for adding extra data to deltatar indexes

diff --git a/deltatar/deltatar.py b/deltatar/deltatar.py
index 9bc7f62..6921669 100644 (file)
--- a/deltatar/deltatar.py
+++ b/deltatar/deltatar.py
@@ -83,6 +83,10 @@ class DeltaTar(object):
     # current time for this backup. Used for file names and file creation checks
     current_time = None
 
+    # extra data to included in the header of the index file when creating a
+    # backup
+    extra_data = dict()
+
     # valid tarfile modes and their corresponding default file extension
     __file_extensions_dict = {
         '': '',
@@ -494,7 +498,7 @@ class DeltaTar(object):
                        key_length=key_length)
 
     def create_full_backup(self, source_path, backup_path,
-                           max_volume_size=None):
+                           max_volume_size=None, extra_data=dict()):
         '''
         Creates a full backup.
 
@@ -504,6 +508,8 @@ class DeltaTar(object):
           be created if not existent.
         - max_volume_size: maximum volume size in megabytes. Used to split the
           backup in volumes. Optional (won't split in volumes by default).
+        - extra_data: a json-serializable dictionary with information that you
+          want to be included in the header of the index file
         '''
         # check input
         if not isinstance(source_path, str):
@@ -522,6 +528,14 @@ class DeltaTar(object):
         if max_volume_size != None:
             max_volume_size = max_volume_size*1024*1024
 
+        if not isinstance(extra_data, dict):
+            raise Exception('extra_data must be a dictionary')
+
+        try:
+            extra_data_str = json.dumps(extra_data)
+        except:
+            raise Exception('extra_data is not json-serializable')
+
         if not os.access(source_path, os.R_OK):
             raise Exception('Source path "%s" is not readable' % source_path)
 
@@ -578,7 +592,7 @@ class DeltaTar(object):
         # wraps some args from context into the handler
         new_volume_handler = partial(new_volume_handler, self, cwd, backup_path)
 
-        index_fd.write(bytes('{"type": "python-delta-tar-index", "version": 1, "backup-type": "full" }\n', 'UTF-8'))
+        index_fd.write(bytes('{"type": "python-delta-tar-index", "version": 1, "backup-type": "full", "extra_data": %s}\n' % extra_data_str, 'UTF-8'))
 
         s = bytes('{"type": "BEGIN-FILE-LIST"}\n', 'UTF-8')
         # calculate checksum and write into the stream
@@ -626,7 +640,7 @@ class DeltaTar(object):
         tarobj.close()
 
     def create_diff_backup(self, source_path, backup_path, previous_index_path,
-                           max_volume_size=None):
+                           max_volume_size=None, extra_data=dict()):
         '''
         Creates a backup.
 
@@ -653,6 +667,14 @@ class DeltaTar(object):
             raise Exception('Source path "%s" does not exist or is not a '\
                             'directory' % source_path)
 
+        if not isinstance(extra_data, dict):
+            raise Exception('extra_data must be a dictionary')
+
+        try:
+            extra_data_str = json.dumps(extra_data)
+        except:
+            raise Exception('extra_data is not json-serializable')
+
         if not os.access(source_path, os.R_OK):
             raise Exception('Source path "%s" is not readable' % source_path)
 
@@ -724,7 +746,7 @@ class DeltaTar(object):
         # wraps some args from context into the handler
         new_volume_handler = partial(new_volume_handler, self, cwd, backup_path)
 
-        index_fd.write(bytes('{"type": "python-delta-tar-index", "version": 1, "backup-type": "diff" }\n', 'UTF-8'))
+        index_fd.write(bytes('{"type": "python-delta-tar-index", "version": 1, "backup-type": "diff", "extra_data": %s}\n' % extra_data_str, 'UTF-8'))
 
         s = bytes('{"type": "BEGIN-FILE-LIST"}\n', 'UTF-8')
         # calculate checksum and write into the stream
@@ -846,6 +868,7 @@ class DeltaTar(object):
                 self.delta_tar = delta_tar
                 self.index_path = index_path
                 self.f = None
+                self.extra_data = dict()
                 self.__enter__()
 
             def __iter__(self):
@@ -867,6 +890,8 @@ class DeltaTar(object):
                             j.get('version', -1) != 1:
                         raise Exception("invalid index file format: %s" % json.dumps(j))
 
+                    self.extra_data = j.get('extra_data', dict())
+
                     # find BEGIN-FILE-LIST, ignore other headers
                     while True:
                         j, l_no = self.delta_tar._parse_json_line(self.f, l_no)

diff --git a/testing/test_deltatar.py b/testing/test_deltatar.py
index 90583f3..1f97eee 100644 (file)
--- a/testing/test_deltatar.py
+++ b/testing/test_deltatar.py
@@ -173,6 +173,66 @@ class DeltaTarTest(BaseTest):
             if value:
                 assert value == self.md5sum(key)
 
+    def test_full_backup_index_extra_data(self):
+        '''
+        Tests that the index file for a full backup can store extra_data and
+        that this data can be retrieved.
+        '''
+        deltatar = DeltaTar(mode=self.MODE, password=self.PASSWORD,
+                            logger=self.consoleLogger)
+
+        extra_data = dict(
+            hola="caracola",
+            otra_cosa=[1, "lista"],
+            y_otra=dict(bola=1.1)
+        )
+
+        deltatar.create_full_backup(
+            source_path="source_dir",
+            backup_path="backup_dir",
+            extra_data=extra_data)
+
+        index_filename = deltatar.index_name_func(is_full=True)
+        index_path = os.path.join("backup_dir", index_filename)
+
+        # iterate_index_path retrieves extra_data, and thus we can then compare
+        index_it = deltatar.iterate_index_path(index_path)
+        self.assertEqual(index_it.extra_data, extra_data)
+
+
+    def test_diff_backup_index_extra_data(self):
+        '''
+        Tests that the index file for a diff backup can store extra_data and
+        that this data can be retrieved.
+        '''
+        deltatar = DeltaTar(mode=self.MODE, password=self.PASSWORD,
+                            logger=self.consoleLogger)
+
+        extra_data = dict(
+            hola="caracola",
+            otra_cosa=[1, "lista"],
+            y_otra=dict(bola=1.1)
+        )
+        # do first backup
+        deltatar.create_full_backup(
+            source_path="source_dir",
+            backup_path="backup_dir")
+
+
+        prev_index_filename = deltatar.index_name_func(is_full=True)
+        prev_index_path = os.path.join("backup_dir", prev_index_filename)
+
+        # create empty diff backup
+        deltatar.create_diff_backup("source_dir", "backup_dir2",
+                                    prev_index_path, extra_data=extra_data)
+
+        index_filename = deltatar.index_name_func(is_full=False)
+        index_path = os.path.join("backup_dir2", index_filename)
+
+        # iterate_index_path retrieves extra_data, and thus we can then compare
+        index_it = deltatar.iterate_index_path(index_path)
+        self.assertEqual(index_it.extra_data, extra_data)
+
     def test_restore_multivol2(self):
         '''
         Creates a full backup without any filtering with multiple volumes and