def convert_user_data(user_handler: UserHandler, user_id_mapper: IdMapper, user_data_map: Dict[str, Dict[str, Any]], realm_id: int, team_name: str) -> None: user_data_list = [] for username in user_data_map: user = user_data_map[username] if check_user_in_team(user, team_name) or user["is_mirror_dummy"]: user_data_list.append(user) for raw_item in user_data_list: user = process_user(raw_item, realm_id, team_name, user_id_mapper) user_handler.add_user(user)
def test_convert_huddle_data(self) -> None: 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() subscriber_handler = SubscriberHandler() huddle_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_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=3, team_name=team_name, ) self.assertEqual(len(zerver_huddle), 1) huddle_members = mattermost_data["direct_channel"][1]["members"] huddle_name = generate_huddle_name(huddle_members) self.assertTrue(huddle_id_mapper.has(huddle_name)) self.assertEqual( subscriber_handler.get_users( huddle_id=huddle_id_mapper.get(huddle_name)), {1, 2, 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.assertEqual(len(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 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.assertEqual(len(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.assertEqual(len(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.assertEqual(len(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.assertEqual(len(user_handler.get_all_users()), 3)
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
def test_convert_channel_data(self) -> None: 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=self.username_to_user, realm_id=3, team_name=team_name, ) zerver_stream = convert_channel_data( channel_data=self.mattermost_data["channel"], user_data_map=self.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.assertEqual(len(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 self.assertEqual( subscriber_handler.get_users( stream_id_mapper.get("gryffindor-common-room")), {1, 2}) self.assertEqual( subscriber_handler.get_users( stream_id_mapper.get("gryffindor-quidditch-team")), {1, 2}) self.assertEqual( subscriber_handler.get_users( stream_id_mapper.get("dumbledores-army")), {1, 2}) team_name = "slytherin" zerver_stream = convert_channel_data( channel_data=self.mattermost_data["channel"], user_data_map=self.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, ) self.assertEqual( subscriber_handler.get_users( stream_id_mapper.get("slytherin-common-room")), {3, 4, 5}) self.assertEqual( subscriber_handler.get_users( stream_id_mapper.get("slytherin-quidditch-team")), {3, 4})
def test_convert_user_data(self) -> None: user_id_mapper = IdMapper() realm_id = 3 team_name = "gryffindor" user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, self.username_to_user, realm_id, team_name) 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, self.username_to_user, realm_id, team_name) self.assertEqual(len(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(team_name, self.mattermost_data, self.username_to_user) user_handler = UserHandler() convert_user_data(user_handler, user_id_mapper, self.username_to_user, realm_id, team_name) self.assertEqual(len(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, self.username_to_user, realm_id, team_name) self.assertEqual(len(user_handler.get_all_users()), 3)
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))