def test_process_users(self) -> None: fixture_dir_name = self.fixture_file_name("", "rocketchat_fixtures") rocketchat_data = rocketchat_data_to_dict(fixture_dir_name) user_id_to_user_map = map_user_id_to_user(rocketchat_data["user"]) realm_id = 3 domain_name = "zulip.com" user_handler = UserHandler() user_id_mapper = IdMapper() process_users( user_id_to_user_map=user_id_to_user_map, realm_id=realm_id, domain_name=domain_name, user_handler=user_handler, user_id_mapper=user_id_mapper, ) self.assert_length(user_handler.get_all_users(), 6) self.assertTrue(user_id_mapper.has(rocketchat_data["user"][0]["_id"])) self.assertTrue(user_id_mapper.has(rocketchat_data["user"][4]["_id"])) user_id = user_id_mapper.get(rocketchat_data["user"][0]["_id"]) user = user_handler.get_user(user_id) self.assertEqual(user["full_name"], rocketchat_data["user"][0]["name"]) self.assertEqual(user["avatar_source"], "G") self.assertEqual(user["delivery_email"], "*****@*****.**") self.assertEqual(user["email"], "*****@*****.**") self.assertEqual(user["full_name"], "Rocket.Cat") self.assertEqual(user["id"], 1) self.assertEqual(user["is_active"], False) self.assertEqual(user["is_mirror_dummy"], False) self.assertEqual(user["is_bot"], True) self.assertEqual(user["bot_type"], 1) self.assertEqual(user["bot_owner"], 2) self.assertEqual(user["role"], UserProfile.ROLE_MEMBER) self.assertEqual(user["realm"], realm_id) self.assertEqual(user["short_name"], "rocket.cat") self.assertEqual(user["timezone"], "UTC") user_id = user_id_mapper.get(rocketchat_data["user"][2]["_id"]) user = user_handler.get_user(user_id) self.assertEqual(user["full_name"], rocketchat_data["user"][2]["name"]) self.assertEqual(user["avatar_source"], "G") self.assertEqual(user["delivery_email"], "*****@*****.**") self.assertEqual(user["email"], "*****@*****.**") self.assertEqual(user["full_name"], "Harry Potter") self.assertEqual(user["id"], 3) self.assertEqual(user["is_active"], True) self.assertEqual(user["is_mirror_dummy"], False) self.assertEqual(user["is_bot"], False) self.assertEqual(user["bot_type"], None) self.assertEqual(user["bot_owner"], None) self.assertEqual(user["role"], UserProfile.ROLE_REALM_OWNER) self.assertEqual(user["realm"], realm_id) self.assertEqual(user["short_name"], "harry.potter") self.assertEqual(user["timezone"], "UTC") # Test `is_mirror_dummy` set for users of type `unknown` rocketchat_data["user"].append({ "_id": "s0m34ndmID", "createdAt": datetime.datetime(2019, 11, 6, 0, 38, 42, 796000), "type": "unknown", "roles": ["unknown"], "name": "Unknown user", "username": "******", }) user_id_to_user_map = map_user_id_to_user(rocketchat_data["user"]) process_users( user_id_to_user_map=user_id_to_user_map, realm_id=realm_id, domain_name=domain_name, user_handler=user_handler, user_id_mapper=user_id_mapper, ) self.assert_length(user_handler.get_all_users(), 7) self.assertTrue(user_id_mapper.has(rocketchat_data["user"][6]["_id"])) user_id = user_id_mapper.get(rocketchat_data["user"][6]["_id"]) user = user_handler.get_user(user_id) self.assertEqual(user["id"], 7) self.assertEqual(user["is_active"], False) self.assertEqual(user["is_mirror_dummy"], True) self.assertEqual(user["is_bot"], False)
def test_convert_stream_subscription_data(self) -> None: fixture_dir_name = self.fixture_file_name("", "rocketchat_fixtures") rocketchat_data = rocketchat_data_to_dict(fixture_dir_name) realm_id = 3 domain_name = "zulip.com" user_handler = UserHandler() subscriber_handler = SubscriberHandler() user_id_mapper = IdMapper() stream_id_mapper = IdMapper() user_id_to_user_map = map_user_id_to_user(rocketchat_data["user"]) process_users( user_id_to_user_map=user_id_to_user_map, realm_id=realm_id, domain_name=domain_name, user_handler=user_handler, user_id_mapper=user_id_mapper, ) room_id_to_room_map: Dict[str, Dict[str, Any]] = {} team_id_to_team_map: Dict[str, Dict[str, Any]] = {} dsc_id_to_dsc_map: Dict[str, Dict[str, Any]] = {} direct_id_to_direct_map: Dict[str, Dict[str, Any]] = {} huddle_id_to_huddle_map: Dict[str, Dict[str, Any]] = {} categorize_channels_and_map_with_id( channel_data=rocketchat_data["room"], room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, ) zerver_stream = convert_channel_data( room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, stream_id_mapper=stream_id_mapper, realm_id=realm_id, ) convert_stream_subscription_data( user_id_to_user_map=user_id_to_user_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, zerver_stream=zerver_stream, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, subscriber_handler=subscriber_handler, ) priyansh_id = user_id_mapper.get(rocketchat_data["user"][1]["_id"]) harry_id = user_id_mapper.get(rocketchat_data["user"][2]["_id"]) hermione_id = user_id_mapper.get(rocketchat_data["user"][3]["_id"]) ron_id = user_id_mapper.get(rocketchat_data["user"][4]["_id"]) voldemort_id = user_id_mapper.get(rocketchat_data["user"][5]["_id"]) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[0]["id"]), {priyansh_id, harry_id, ron_id, hermione_id, voldemort_id}, ) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[1]["id"]), {priyansh_id, harry_id}) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[2]["id"]), {harry_id, hermione_id}) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[3]["id"]), {harry_id, ron_id, hermione_id}, ) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[4]["id"]), {harry_id}) self.assertEqual( subscriber_handler.get_users(stream_id=zerver_stream[5]["id"]), {harry_id}) # Add a new channel with no user. no_user_channel: Dict[str, Any] = { "_id": "rand0mID", "ts": datetime.datetime(2021, 7, 15, 10, 58, 23, 647000), "t": "c", "name": "no-user-channel", } room_id_to_room_map[no_user_channel["_id"]] = no_user_channel zerver_stream = convert_channel_data( room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, stream_id_mapper=stream_id_mapper, realm_id=realm_id, ) convert_stream_subscription_data( user_id_to_user_map=user_id_to_user_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, zerver_stream=zerver_stream, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, subscriber_handler=subscriber_handler, ) self.assert_length( subscriber_handler.get_users(stream_id=zerver_stream[6]["id"]), 0) self.assertTrue(zerver_stream[6]["deactivated"])
def test_process_message_attachments(self) -> None: mattermost_data_dir = self.fixture_file_name( "", "mattermost_fixtures/direct_channel") output_dir = self.make_import_output_dir("mattermost") fixture_file_name = self.fixture_file_name( "export.json", "mattermost_fixtures/direct_channel") mattermost_data = mattermost_data_file_to_dict(fixture_file_name) username_to_user = create_username_to_user_mapping( mattermost_data["user"]) reset_mirror_dummy_users(username_to_user) user_handler = UserHandler() user_id_mapper = IdMapper() team_name = "gryffindor" convert_user_data( user_handler=user_handler, user_id_mapper=user_id_mapper, user_data_map=username_to_user, realm_id=3, team_name=team_name, ) zerver_attachments: List[ZerverFieldsT] = [] uploads_list: List[ZerverFieldsT] = [] process_message_attachments( attachments=mattermost_data["post"]["direct_post"][0] ["attachments"], realm_id=3, message_id=1, user_id=2, user_handler=user_handler, zerver_attachment=zerver_attachments, uploads_list=uploads_list, mattermost_data_dir=mattermost_data_dir, output_dir=output_dir, ) self.assert_length(zerver_attachments, 1) self.assertEqual(zerver_attachments[0]["file_name"], "harry-ron.jpg") self.assertEqual(zerver_attachments[0]["owner"], 2) self.assertEqual( user_handler.get_user(zerver_attachments[0]["owner"])["email"], "*****@*****.**") # TODO: Assert this for False after fixing the file permissions in PMs self.assertTrue(zerver_attachments[0]["is_realm_public"]) self.assert_length(uploads_list, 1) self.assertEqual(uploads_list[0]["user_profile_email"], "*****@*****.**") attachment_path = self.fixture_file_name( mattermost_data["post"]["direct_post"][0]["attachments"][0] ["path"], "mattermost_fixtures/direct_channel/data", ) attachment_out_path = os.path.join(output_dir, "uploads", zerver_attachments[0]["path_id"]) self.assertTrue(os.path.exists(attachment_out_path)) self.assertTrue(filecmp.cmp(attachment_path, attachment_out_path))
def test_map_receiver_id_to_recipient_id(self) -> None: fixture_dir_name = self.fixture_file_name("", "rocketchat_fixtures") rocketchat_data = rocketchat_data_to_dict(fixture_dir_name) realm_id = 3 domain_name = "zulip.com" user_handler = UserHandler() subscriber_handler = SubscriberHandler() user_id_mapper = IdMapper() stream_id_mapper = IdMapper() huddle_id_mapper = IdMapper() user_id_to_user_map = map_user_id_to_user(rocketchat_data["user"]) process_users( user_id_to_user_map=user_id_to_user_map, realm_id=realm_id, domain_name=domain_name, user_handler=user_handler, user_id_mapper=user_id_mapper, ) room_id_to_room_map: Dict[str, Dict[str, Any]] = {} team_id_to_team_map: Dict[str, Dict[str, Any]] = {} dsc_id_to_dsc_map: Dict[str, Dict[str, Any]] = {} direct_id_to_direct_map: Dict[str, Dict[str, Any]] = {} huddle_id_to_huddle_map: Dict[str, Dict[str, Any]] = {} categorize_channels_and_map_with_id( channel_data=rocketchat_data["room"], room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, ) zerver_stream = convert_channel_data( room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, stream_id_mapper=stream_id_mapper, realm_id=realm_id, ) zerver_huddle = convert_huddle_data( huddle_id_to_huddle_map=huddle_id_to_huddle_map, huddle_id_mapper=huddle_id_mapper, user_id_mapper=user_id_mapper, subscriber_handler=subscriber_handler, ) all_users = user_handler.get_all_users() zerver_recipient = build_recipients( zerver_userprofile=all_users, zerver_stream=zerver_stream, zerver_huddle=zerver_huddle, ) stream_id_to_recipient_id: Dict[int, int] = {} user_id_to_recipient_id: Dict[int, int] = {} huddle_id_to_recipient_id: Dict[int, int] = {} map_receiver_id_to_recipient_id( zerver_recipient=zerver_recipient, stream_id_to_recipient_id=stream_id_to_recipient_id, user_id_to_recipient_id=user_id_to_recipient_id, huddle_id_to_recipient_id=huddle_id_to_recipient_id, ) # 6 for streams and 6 for users. self.assert_length(zerver_recipient, 13) self.assert_length(stream_id_to_recipient_id, 6) self.assert_length(user_id_to_recipient_id, 6) self.assert_length(huddle_id_to_recipient_id, 1) # First user recipients are built, followed by stream recipients in `build_recipients`. self.assertEqual( user_id_to_recipient_id[zerver_recipient[0]["type_id"]], zerver_recipient[0]["id"]) self.assertEqual( user_id_to_recipient_id[zerver_recipient[1]["type_id"]], zerver_recipient[1]["id"]) self.assertEqual( stream_id_to_recipient_id[zerver_recipient[6]["type_id"]], zerver_recipient[6]["id"]) self.assertEqual( stream_id_to_recipient_id[zerver_recipient[7]["type_id"]], zerver_recipient[7]["id"]) self.assertEqual( huddle_id_to_recipient_id[zerver_recipient[12]["type_id"]], zerver_recipient[12]["id"])
def test_convert_user_data(self) -> None: user_id_mapper = IdMapper() realm_id = 3 fixture_file_name = self.fixture_file_name("export.json", "mattermost_fixtures") mattermost_data = mattermost_data_file_to_dict(fixture_file_name) username_to_user = create_username_to_user_mapping( mattermost_data["user"]) reset_mirror_dummy_users(username_to_user) team_name = "gryffindor" user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, username_to_user, realm_id, team_name) self.assert_length(user_handler.get_all_users(), 2) self.assertTrue(user_id_mapper.has("harry")) self.assertTrue(user_id_mapper.has("ron")) self.assertEqual( user_handler.get_user(user_id_mapper.get("harry"))["full_name"], "Harry Potter") self.assertEqual( user_handler.get_user(user_id_mapper.get("ron"))["full_name"], "Ron Weasley") team_name = "slytherin" user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, username_to_user, realm_id, team_name) self.assert_length(user_handler.get_all_users(), 3) self.assertTrue(user_id_mapper.has("malfoy")) self.assertTrue(user_id_mapper.has("pansy")) self.assertTrue(user_id_mapper.has("snape")) team_name = "gryffindor" # Snape is a mirror dummy user in Harry's team. label_mirror_dummy_users(2, team_name, mattermost_data, username_to_user) user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, username_to_user, realm_id, team_name) self.assert_length(user_handler.get_all_users(), 3) self.assertTrue(user_id_mapper.has("snape")) team_name = "slytherin" user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, username_to_user, realm_id, team_name) self.assert_length(user_handler.get_all_users(), 3)
def test_convert_channel_data(self) -> None: fixture_file_name = self.fixture_file_name("export.json", "mattermost_fixtures") mattermost_data = mattermost_data_file_to_dict(fixture_file_name) username_to_user = create_username_to_user_mapping( mattermost_data["user"]) reset_mirror_dummy_users(username_to_user) user_handler = UserHandler() subscriber_handler = SubscriberHandler() stream_id_mapper = IdMapper() user_id_mapper = IdMapper() team_name = "gryffindor" convert_user_data( user_handler=user_handler, user_id_mapper=user_id_mapper, user_data_map=username_to_user, realm_id=3, team_name=team_name, ) zerver_stream = convert_channel_data( channel_data=mattermost_data["channel"], user_data_map=username_to_user, subscriber_handler=subscriber_handler, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, realm_id=3, team_name=team_name, ) self.assert_length(zerver_stream, 3) self.assertEqual(zerver_stream[0]["name"], "Gryffindor common room") self.assertEqual(zerver_stream[0]["invite_only"], False) self.assertEqual(zerver_stream[0]["description"], "A place for talking about Gryffindor common room") self.assertEqual(zerver_stream[0]["rendered_description"], "") self.assertEqual(zerver_stream[0]["realm"], 3) self.assertEqual(zerver_stream[1]["name"], "Gryffindor quidditch team") self.assertEqual(zerver_stream[1]["invite_only"], False) self.assertEqual( zerver_stream[1]["description"], "A place for talking about Gryffindor quidditch team") self.assertEqual(zerver_stream[1]["rendered_description"], "") self.assertEqual(zerver_stream[1]["realm"], 3) self.assertEqual(zerver_stream[2]["name"], "Dumbledores army") self.assertEqual(zerver_stream[2]["invite_only"], True) self.assertEqual(zerver_stream[2]["description"], "A place for talking about Dumbledores army") self.assertEqual(zerver_stream[2]["rendered_description"], "") self.assertEqual(zerver_stream[2]["realm"], 3) self.assertTrue(stream_id_mapper.has("gryffindor-common-room")) self.assertTrue(stream_id_mapper.has("gryffindor-quidditch-team")) self.assertTrue(stream_id_mapper.has("dumbledores-army")) # TODO: Add ginny ron_id = user_id_mapper.get("ron") harry_id = user_id_mapper.get("harry") self.assertEqual({ron_id, harry_id}, {1, 2}) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("gryffindor-common-room")), {ron_id, harry_id}, ) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("gryffindor-quidditch-team")), {ron_id, harry_id}, ) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("dumbledores-army")), {ron_id, harry_id}, ) # Converting channel data when a user's `teams` value is `null`. username_to_user["ron"].update(teams=None) zerver_stream = convert_channel_data( channel_data=mattermost_data["channel"], user_data_map=username_to_user, subscriber_handler=subscriber_handler, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, realm_id=3, team_name=team_name, ) harry_id = user_id_mapper.get("harry") self.assertIn(harry_id, {1, 2}) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("gryffindor-common-room")), {harry_id}, ) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("gryffindor-quidditch-team")), {harry_id}, ) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("dumbledores-army")), {harry_id}, ) team_name = "slytherin" zerver_stream = convert_channel_data( channel_data=mattermost_data["channel"], user_data_map=username_to_user, subscriber_handler=subscriber_handler, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, realm_id=4, team_name=team_name, ) malfoy_id = user_id_mapper.get("malfoy") pansy_id = user_id_mapper.get("pansy") snape_id = user_id_mapper.get("snape") self.assertEqual({malfoy_id, pansy_id, snape_id}, {3, 4, 5}) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("slytherin-common-room")), {malfoy_id, pansy_id, snape_id}, ) self.assertEqual( subscriber_handler.get_users( stream_id=stream_id_mapper.get("slytherin-quidditch-team")), {malfoy_id, pansy_id}, )
def do_convert_data(rocketchat_data_dir: str, output_dir: str) -> None: # Get all required exported data in a dictionary rocketchat_data = rocketchat_data_to_dict(rocketchat_data_dir) # Subdomain is set by the user while running the import command realm_subdomain = "" realm_id = 0 domain_name = settings.EXTERNAL_HOST realm = make_realm(realm_id, realm_subdomain, domain_name, rocketchat_data["instance"][0]) user_id_to_user_map: Dict[str, Dict[str, Any]] = map_user_id_to_user( rocketchat_data["user"]) username_to_user_id_map: Dict[str, str] = map_username_to_user_id( user_id_to_user_map) user_handler = UserHandler() subscriber_handler = SubscriberHandler() user_id_mapper = IdMapper() stream_id_mapper = IdMapper() huddle_id_mapper = IdMapper() process_users( user_id_to_user_map=user_id_to_user_map, realm_id=realm_id, domain_name=domain_name, user_handler=user_handler, user_id_mapper=user_id_mapper, ) room_id_to_room_map: Dict[str, Dict[str, Any]] = {} team_id_to_team_map: Dict[str, Dict[str, Any]] = {} dsc_id_to_dsc_map: Dict[str, Dict[str, Any]] = {} direct_id_to_direct_map: Dict[str, Dict[str, Any]] = {} huddle_id_to_huddle_map: Dict[str, Dict[str, Any]] = {} categorize_channels_and_map_with_id( channel_data=rocketchat_data["room"], room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, ) zerver_stream = convert_channel_data( room_id_to_room_map=room_id_to_room_map, team_id_to_team_map=team_id_to_team_map, stream_id_mapper=stream_id_mapper, realm_id=realm_id, ) realm["zerver_stream"] = zerver_stream # Add stream subscription data to `subscriber_handler` convert_stream_subscription_data( user_id_to_user_map=user_id_to_user_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, zerver_stream=zerver_stream, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, subscriber_handler=subscriber_handler, ) zerver_huddle = convert_huddle_data( huddle_id_to_huddle_map=huddle_id_to_huddle_map, huddle_id_mapper=huddle_id_mapper, user_id_mapper=user_id_mapper, subscriber_handler=subscriber_handler, ) realm["zerver_huddle"] = zerver_huddle all_users = user_handler.get_all_users() zerver_recipient = build_recipients( zerver_userprofile=all_users, zerver_stream=zerver_stream, zerver_huddle=zerver_huddle, ) realm["zerver_recipient"] = zerver_recipient stream_subscriptions = build_stream_subscriptions( get_users=subscriber_handler.get_users, zerver_recipient=zerver_recipient, zerver_stream=zerver_stream, ) huddle_subscriptions = build_huddle_subscriptions( get_users=subscriber_handler.get_users, zerver_recipient=zerver_recipient, zerver_huddle=zerver_huddle, ) personal_subscriptions = build_personal_subscriptions( zerver_recipient=zerver_recipient, ) zerver_subscription = personal_subscriptions + stream_subscriptions + huddle_subscriptions realm["zerver_subscription"] = zerver_subscription zerver_realmemoji = build_custom_emoji( realm_id=realm_id, custom_emoji_data=rocketchat_data["custom_emoji"], output_dir=output_dir, ) realm["zerver_realmemoji"] = zerver_realmemoji subscriber_map = make_subscriber_map( zerver_subscription=zerver_subscription, ) stream_id_to_recipient_id: Dict[int, int] = {} huddle_id_to_recipient_id: Dict[int, int] = {} user_id_to_recipient_id: Dict[int, int] = {} map_receiver_id_to_recipient_id( zerver_recipient=zerver_recipient, stream_id_to_recipient_id=stream_id_to_recipient_id, huddle_id_to_recipient_id=huddle_id_to_recipient_id, user_id_to_recipient_id=user_id_to_recipient_id, ) channel_messages: List[Dict[str, Any]] = [] private_messages: List[Dict[str, Any]] = [] separate_channel_and_private_messages( messages=rocketchat_data["message"], direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, channel_messages=channel_messages, private_messages=private_messages, ) total_reactions: List[ZerverFieldsT] = [] uploads_list: List[ZerverFieldsT] = [] zerver_attachment: List[ZerverFieldsT] = [] upload_id_to_upload_data_map = map_upload_id_to_upload_data( rocketchat_data["upload"]) # Process channel messages process_messages( realm_id=realm_id, messages=channel_messages, subscriber_map=subscriber_map, is_pm_data=False, username_to_user_id_map=username_to_user_id_map, user_id_mapper=user_id_mapper, user_handler=user_handler, user_id_to_recipient_id=user_id_to_recipient_id, stream_id_mapper=stream_id_mapper, stream_id_to_recipient_id=stream_id_to_recipient_id, huddle_id_mapper=huddle_id_mapper, huddle_id_to_recipient_id=huddle_id_to_recipient_id, room_id_to_room_map=room_id_to_room_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, zerver_realmemoji=zerver_realmemoji, total_reactions=total_reactions, uploads_list=uploads_list, zerver_attachment=zerver_attachment, upload_id_to_upload_data_map=upload_id_to_upload_data_map, output_dir=output_dir, ) # Process private messages process_messages( realm_id=realm_id, messages=private_messages, subscriber_map=subscriber_map, is_pm_data=True, username_to_user_id_map=username_to_user_id_map, user_id_mapper=user_id_mapper, user_handler=user_handler, user_id_to_recipient_id=user_id_to_recipient_id, stream_id_mapper=stream_id_mapper, stream_id_to_recipient_id=stream_id_to_recipient_id, huddle_id_mapper=huddle_id_mapper, huddle_id_to_recipient_id=huddle_id_to_recipient_id, room_id_to_room_map=room_id_to_room_map, dsc_id_to_dsc_map=dsc_id_to_dsc_map, direct_id_to_direct_map=direct_id_to_direct_map, huddle_id_to_huddle_map=huddle_id_to_huddle_map, zerver_realmemoji=zerver_realmemoji, total_reactions=total_reactions, uploads_list=uploads_list, zerver_attachment=zerver_attachment, upload_id_to_upload_data_map=upload_id_to_upload_data_map, output_dir=output_dir, ) realm["zerver_reaction"] = total_reactions realm["zerver_userprofile"] = user_handler.get_all_users() realm["sort_by_date"] = True create_converted_data_files(realm, output_dir, "/realm.json") # TODO: Add support for importing avatars create_converted_data_files([], output_dir, "/avatars/records.json") # Import attachments attachment: Dict[str, List[Any]] = {"zerver_attachment": zerver_attachment} create_converted_data_files(attachment, output_dir, "/attachment.json") create_converted_data_files(uploads_list, output_dir, "/uploads/records.json") logging.info("Start making tarball") subprocess.check_call( ["tar", "-czf", output_dir + ".tar.gz", output_dir, "-P"]) logging.info("Done making tarball")
def process_users( user_id_to_user_map: Dict[str, Dict[str, Any]], realm_id: int, domain_name: str, user_handler: UserHandler, user_id_mapper: IdMapper, ) -> None: realm_owners: List[int] = [] bots: List[int] = [] for rc_user_id in user_id_to_user_map: user_dict = user_id_to_user_map[rc_user_id] is_mirror_dummy = False is_bot = False is_active = True # Rocket.Chat has three user types: # "user": This is a regular user of the system. # "bot": A special user types for bots. # "unknown": This usually represents a livechat guest. if user_dict["type"] != "user": is_active = False if user_dict["type"] == "bot": is_bot = True else: is_mirror_dummy = True if not user_dict.get("emails"): user_dict["emails"] = [{ "address": "{}-{}@{}".format(user_dict["username"], user_dict["type"], domain_name) }] # TODO: Change this to use actual exported avatar avatar_source = "G" full_name = user_dict["name"] id = user_id_mapper.get(rc_user_id) delivery_email = user_dict["emails"][0]["address"] email = user_dict["emails"][0]["address"] short_name = user_dict["username"] date_joined = float(user_dict["createdAt"].timestamp()) timezone = "UTC" role = UserProfile.ROLE_MEMBER if "admin" in user_dict["roles"]: role = UserProfile.ROLE_REALM_OWNER realm_owners.append(id) elif "guest" in user_dict["roles"]: role = UserProfile.ROLE_GUEST if is_bot: bots.append(id) user = build_user_profile( avatar_source=avatar_source, date_joined=date_joined, delivery_email=delivery_email, email=email, full_name=full_name, id=id, is_active=is_active, role=role, is_mirror_dummy=is_mirror_dummy, realm_id=realm_id, short_name=short_name, timezone=timezone, is_bot=is_bot, bot_type=1 if is_bot else None, ) user_handler.add_user(user) # Set the first realm_owner as the owner of # all the bots. if realm_owners: for bot_id in bots: bot_user = user_handler.get_user(user_id=bot_id) bot_user["bot_owner"] = realm_owners[0]
def process_message_attachment( upload: Dict[str, Any], realm_id: int, message_id: int, user_id: int, user_handler: UserHandler, zerver_attachment: List[ZerverFieldsT], uploads_list: List[ZerverFieldsT], upload_id_to_upload_data_map: Dict[str, Dict[str, Any]], output_dir: str, ) -> Tuple[str, bool]: upload_file_data = upload_id_to_upload_data_map[upload["_id"]] file_name = upload["name"] file_ext = f'.{upload["type"].split("/")[-1]}' has_image = False if file_ext.lower() in IMAGE_EXTENSIONS: has_image = True s3_path = "/".join([ str(realm_id), format(random.randint(0, 255), "x"), secrets.token_urlsafe(18), sanitize_name(file_name), ]) # Build the attachment from chunks and save it to s3_path. file_out_path = os.path.join(output_dir, "uploads", s3_path) os.makedirs(os.path.dirname(file_out_path), exist_ok=True) with open(file_out_path, "wb") as upload_file: upload_file.write(b"".join(upload_file_data["chunk"])) attachment_content = ( f'{upload_file_data["description"]}\n\n[{file_name}](/user_uploads/{s3_path})' ) fileinfo = { "name": file_name, "size": upload_file_data["size"], "created": float(upload_file_data["_updatedAt"].timestamp()), } upload = dict( path=s3_path, realm_id=realm_id, content_type=upload["type"], user_profile_id=user_id, last_modified=fileinfo["created"], user_profile_email=user_handler.get_user(user_id=user_id)["email"], s3_path=s3_path, size=fileinfo["size"], ) uploads_list.append(upload) build_attachment( realm_id=realm_id, message_ids={message_id}, user_id=user_id, fileinfo=fileinfo, s3_path=s3_path, zerver_attachment=zerver_attachment, ) return attachment_content, has_image
def do_convert_data(mattermost_data_dir: str, output_dir: str, masking_content: bool) -> None: username_to_user: Dict[str, Dict[str, Any]] = {} os.makedirs(output_dir, exist_ok=True) if os.listdir(output_dir): # nocoverage raise Exception("Output directory should be empty!") mattermost_data_file = os.path.join(mattermost_data_dir, "export.json") mattermost_data = mattermost_data_file_to_dict(mattermost_data_file) username_to_user = create_username_to_user_mapping(mattermost_data["user"]) for team in mattermost_data["team"]: realm_id = NEXT_ID("realm_id") team_name = team["name"] user_handler = UserHandler() subscriber_handler = SubscriberHandler() user_id_mapper = IdMapper() stream_id_mapper = IdMapper() huddle_id_mapper = IdMapper() print("Generating data for", team_name) realm = make_realm(realm_id, team) realm_output_dir = os.path.join(output_dir, team_name) reset_mirror_dummy_users(username_to_user) label_mirror_dummy_users(len(mattermost_data["team"]), team_name, mattermost_data, username_to_user) convert_user_data( user_handler=user_handler, user_id_mapper=user_id_mapper, user_data_map=username_to_user, realm_id=realm_id, team_name=team_name, ) zerver_stream = convert_channel_data( channel_data=mattermost_data["channel"], user_data_map=username_to_user, subscriber_handler=subscriber_handler, stream_id_mapper=stream_id_mapper, user_id_mapper=user_id_mapper, realm_id=realm_id, team_name=team_name, ) realm["zerver_stream"] = zerver_stream zerver_huddle: List[ZerverFieldsT] = [] if len(mattermost_data["team"]) == 1: zerver_huddle = convert_huddle_data( huddle_data=mattermost_data["direct_channel"], user_data_map=username_to_user, subscriber_handler=subscriber_handler, huddle_id_mapper=huddle_id_mapper, user_id_mapper=user_id_mapper, realm_id=realm_id, team_name=team_name, ) realm["zerver_huddle"] = zerver_huddle all_users = user_handler.get_all_users() zerver_recipient = build_recipients( zerver_userprofile=all_users, zerver_stream=zerver_stream, zerver_huddle=zerver_huddle, ) realm["zerver_recipient"] = zerver_recipient stream_subscriptions = build_stream_subscriptions( get_users=subscriber_handler.get_users, zerver_recipient=zerver_recipient, zerver_stream=zerver_stream, ) huddle_subscriptions = build_huddle_subscriptions( get_users=subscriber_handler.get_users, zerver_recipient=zerver_recipient, zerver_huddle=zerver_huddle, ) personal_subscriptions = build_personal_subscriptions( zerver_recipient=zerver_recipient, ) # Mattermost currently supports only exporting messages from channels. # Personal messages and huddles are not exported. zerver_subscription = personal_subscriptions + stream_subscriptions + huddle_subscriptions realm["zerver_subscription"] = zerver_subscription zerver_realmemoji = write_emoticon_data( realm_id=realm_id, custom_emoji_data=mattermost_data["emoji"], data_dir=mattermost_data_dir, output_dir=realm_output_dir, ) realm["zerver_realmemoji"] = zerver_realmemoji subscriber_map = make_subscriber_map( zerver_subscription=zerver_subscription, ) total_reactions: List[Dict[str, Any]] = [] uploads_list: List[ZerverFieldsT] = [] zerver_attachment: List[ZerverFieldsT] = [] write_message_data( num_teams=len(mattermost_data["team"]), team_name=team_name, realm_id=realm_id, post_data=mattermost_data["post"], zerver_recipient=zerver_recipient, subscriber_map=subscriber_map, output_dir=realm_output_dir, masking_content=masking_content, stream_id_mapper=stream_id_mapper, huddle_id_mapper=huddle_id_mapper, user_id_mapper=user_id_mapper, user_handler=user_handler, zerver_realmemoji=zerver_realmemoji, total_reactions=total_reactions, uploads_list=uploads_list, zerver_attachment=zerver_attachment, mattermost_data_dir=mattermost_data_dir, ) realm["zerver_reaction"] = total_reactions realm["zerver_userprofile"] = user_handler.get_all_users() realm["sort_by_date"] = True create_converted_data_files(realm, realm_output_dir, "/realm.json") # Mattermost currently doesn't support exporting avatars create_converted_data_files([], realm_output_dir, "/avatars/records.json") # Export message attachments attachment: Dict[str, List[Any]] = { "zerver_attachment": zerver_attachment } create_converted_data_files(uploads_list, realm_output_dir, "/uploads/records.json") create_converted_data_files(attachment, realm_output_dir, "/attachment.json") logging.info("Start making tarball") subprocess.check_call([ "tar", "-czf", realm_output_dir + ".tar.gz", realm_output_dir, "-P" ]) logging.info("Done making tarball")
def process_message_attachments( attachments: List[Dict[str, Any]], realm_id: int, message_id: int, user_id: int, user_handler: UserHandler, zerver_attachment: List[ZerverFieldsT], uploads_list: List[ZerverFieldsT], mattermost_data_dir: str, output_dir: str, ) -> Tuple[str, bool]: has_image = False markdown_links = [] for attachment in attachments: attachment_path = attachment["path"] attachment_full_path = os.path.join(mattermost_data_dir, "data", attachment_path) file_name = attachment_path.split("/")[-1] file_ext = f'.{file_name.split(".")[-1]}' if file_ext.lower() in IMAGE_EXTENSIONS: has_image = True s3_path = "/".join([ str(realm_id), format(random.randint(0, 255), "x"), secrets.token_urlsafe(18), sanitize_name(file_name), ]) content_for_link = f"[{file_name}](/user_uploads/{s3_path})" markdown_links.append(content_for_link) fileinfo = { "name": file_name, "size": os.path.getsize(attachment_full_path), "created": os.path.getmtime(attachment_full_path), } upload = dict( path=s3_path, realm_id=realm_id, content_type=None, user_profile_id=user_id, last_modified=fileinfo["created"], user_profile_email=user_handler.get_user(user_id=user_id)["email"], s3_path=s3_path, size=fileinfo["size"], ) uploads_list.append(upload) build_attachment( realm_id=realm_id, message_ids={message_id}, user_id=user_id, fileinfo=fileinfo, s3_path=s3_path, zerver_attachment=zerver_attachment, ) # Copy the attachment file to output_dir attachment_out_path = os.path.join(output_dir, "uploads", s3_path) os.makedirs(os.path.dirname(attachment_out_path), exist_ok=True) shutil.copyfile(attachment_full_path, attachment_out_path) content = "\n".join(markdown_links) return content, has_image