def sync_data(self): table = self.__class__.TABLE endpoints = { 'sent': FuelSDK.ET_SentEvent, 'click': FuelSDK.ET_ClickEvent, 'open': FuelSDK.ET_OpenEvent, 'bounce': FuelSDK.ET_BounceEvent, 'unsub': FuelSDK.ET_UnsubEvent } for event_name, selector in endpoints.items(): search_filter = None start = get_last_record_value_for_table(self.state, event_name) if start is None: start = self.config.get('start_date') if start is None: raise RuntimeError('start_date not defined!') pagination_unit = self.config.get( 'pagination__{}_interval_unit'.format(event_name), 'minutes') pagination_quantity = self.config.get( 'pagination__{}_interval_quantity'.format(event_name), 10) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) while before_now(start): LOGGER.info("Fetching {} from {} to {}" .format(event_name, start, end)) search_filter = get_date_page('EventDate', start, unit) stream = request(event_name, selector, self.auth_stub, search_filter) for event in stream: event = self.filter_keys_and_parse(event) self.state = incorporate(self.state, event_name, 'EventDate', event.get('EventDate')) singer.write_records(table, [event]) self.state = incorporate(self.state, event_name, 'EventDate', start) save_state(self.state) start = end end = increment_date(start, unit)
def sync_data(self): table = self.__class__.TABLE subscriber_dao = SubscriberDataAccessObject(self.config, self.state, self.auth_stub, self.subscriber_catalog) # pass config to return start date if not bookmark is found start = get_last_record_value_for_table(self.state, table, self.config) pagination_unit = self.config.get( 'pagination__list_subscriber_interval_unit', 'days') pagination_quantity = self.config.get( 'pagination__list_subscriber_interval_quantity', 1) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) all_subscribers_list = self._get_all_subscribers_list() while before_now(start): stream = request('ListSubscriber', FuelSDK.ET_List_Subscriber, self.auth_stub, _get_list_subscriber_filter( all_subscribers_list, start, unit), batch_size=self.batch_size) batch_size = 100 if self.replicate_subscriber: subscriber_dao.write_schema() catalog_copy = copy.deepcopy(self.catalog) for list_subscribers_batch in partition_all(stream, batch_size): for list_subscriber in list_subscribers_batch: list_subscriber = self.filter_keys_and_parse( list_subscriber) if list_subscriber.get('ModifiedDate'): self.state = incorporate( self.state, table, 'ModifiedDate', list_subscriber.get('ModifiedDate')) self.write_records_with_transform(list_subscriber, catalog_copy, table) if self.replicate_subscriber: # make the list of subscriber keys subscriber_keys = list( map(_get_subscriber_key, list_subscribers_batch)) # pass the list of 'subscriber_keys' to fetch subscriber details subscriber_dao.pull_subscribers_batch(subscriber_keys) save_state(self.state) start = end end = increment_date(start, unit)
def test_increment_date(self): # verify that if there is no 'unit' mentioned, then the date should be incremented by '1 day' self.assertEqual(increment_date("2015-09-28T10:05:53Z"), "2015-09-29T10:05:53Z") # verify that the 'increment_date' correctly increments the date by sepcified 'unit'(here: 1 hour) self.assertEqual(increment_date("2015-09-28T10:05:53Z", {'hours': 1}), "2015-09-28T11:05:53Z")
def test_increment_date(self): self.assertEqual( increment_date("2015-09-28T10:05:53Z"), "2015-09-29T10:05:53Z") self.assertEqual( increment_date("2015-09-28T10:05:53Z", {'hours': 1}), "2015-09-28T11:05:53Z")
def sync_data(self): table = self.__class__.TABLE subscriber_dao = SubscriberDataAccessObject(self.config, self.state, self.auth_stub, self.subscriber_catalog) start = get_last_record_value_for_table(self.state, table, self.config.get('start_date')) if start is None: start = self.config.get('start_date') pagination_unit = self.config.get( 'pagination__list_subscriber_interval_unit', 'days') pagination_quantity = self.config.get( 'pagination__list_subsctiber_interval_quantity', 1) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) all_subscribers_list = self._get_all_subscribers_list() while before_now(start): stream = request( 'ListSubscriber', FuelSDK.ET_List_Subscriber, self.auth_stub, _get_list_subscriber_filter(all_subscribers_list, start, unit)) batch_size = 100 if self.replicate_subscriber: subscriber_dao.write_schema() for list_subscribers_batch in partition_all(stream, batch_size): for list_subscriber in list_subscribers_batch: list_subscriber = self.filter_keys_and_parse( list_subscriber) if list_subscriber.get('ModifiedDate'): self.state = incorporate( self.state, table, 'ModifiedDate', list_subscriber.get('ModifiedDate')) list_subscriber = self.remove_sensitive_data( list_subscriber) singer.write_records(table, [list_subscriber]) if self.replicate_subscriber: subscriber_keys = list( map(_get_subscriber_key, list_subscribers_batch)) subscriber_dao.pull_subscribers_batch(subscriber_keys) save_state(self.state) start = end end = increment_date(start, unit)
def sync_data(self): table = self.__class__.TABLE search_filter = None start = get_last_record_value_for_table(self.state, self.event_name, self.config.get('start_date')) if start is None: start = self.config.get('start_date') if start is None: raise RuntimeError('start_date not defined!') pagination_unit = self.config.get( 'pagination__{}_interval_unit'.format(self.event_name), 'minutes') pagination_quantity = self.config.get( 'pagination__{}_interval_quantity'.format(self.event_name), 10) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) while before_now(start): LOGGER.info("Fetching {} from {} to {}".format( self.event_name, start, end)) search_filter = get_date_page('EventDate', start, unit) stream = request(self.event_name, self.selector, self.auth_stub, search_filter) for event in stream: event = self.filter_keys_and_parse(event) self.state = incorporate(self.state, self.event_name, 'EventDate', event.get('EventDate')) if event.get('SubscriberKey') is None: LOGGER.info( "SubscriberKey is NULL so ignoring {} record with SendID: {} and EventDate: {}" .format(self.event_name, event.get('SendID'), event.get('EventDate'))) continue event = self.remove_sensitive_data(event) singer.write_records(table, [event]) self.state = incorporate(self.state, self.event_name, 'EventDate', start) save_state(self.state) start = end end = increment_date(start, unit)
def sync_data(self): table = self.__class__.TABLE endpoints = { 'sent': FuelSDK.ET_SentEvent, 'click': FuelSDK.ET_ClickEvent, 'open': FuelSDK.ET_OpenEvent, 'bounce': FuelSDK.ET_BounceEvent, 'unsub': FuelSDK.ET_UnsubEvent } for event_name, selector in endpoints.items(): search_filter = None # pass config to return start date if not bookmark is found start = get_last_record_value_for_table(self.state, event_name, self.config) if start is None: raise RuntimeError('start_date not defined!') pagination_unit = self.config.get( 'pagination__{}_interval_unit'.format(event_name), 'minutes') pagination_quantity = self.config.get( 'pagination__{}_interval_quantity'.format(event_name), 10) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) while before_now(start): LOGGER.info("Fetching {} from {} to {}" .format(event_name, start, end)) search_filter = get_date_page('EventDate', start, unit) stream = request(event_name, selector, self.auth_stub, search_filter, batch_size=self.batch_size) catalog_copy = copy.deepcopy(self.catalog) for event in stream: event = self.filter_keys_and_parse(event) self.state = incorporate(self.state, event_name, 'EventDate', event.get('EventDate')) if event.get('SubscriberKey') is None: LOGGER.info("SubscriberKey is NULL so ignoring {} record with SendID: {} and EventDate: {}" .format(event_name, event.get('SendID'), event.get('EventDate'))) continue self.write_records_with_transform(event, catalog_copy, table) self.state = incorporate(self.state, event_name, 'EventDate', start) save_state(self.state) start = end end = increment_date(start, unit)
def sync_data(self): tap_stream_id = self.catalog.get('tap_stream_id') table = self.catalog.get('stream') (_, customer_key) = tap_stream_id.split('.', 1) keys = self.get_catalog_keys() keys.remove('CategoryID') replication_key = None start = get_last_record_value_for_table(self.state, table) if start is None: start = self.config.get('start_date') for key in ['ModifiedDate', 'JoinDate']: if key in keys: replication_key = key pagination_unit = self.config.get( 'pagination__data_extension_interval_unit', 'days') pagination_quantity = self.config.get( 'pagination__data_extension_interval_quantity', 7) unit = {pagination_unit: int(pagination_quantity)} end = increment_date(start, unit) parent_result = None parent_extension = None parent_result = request('DataExtension', FuelSDK.ET_DataExtension, self.auth_stub, search_filter={ 'Property': 'CustomerKey', 'SimpleOperator': 'equals', 'Value': customer_key, }, props=['CustomerKey', 'CategoryID']) parent_extension = next(parent_result) parent_category_id = parent_extension.CategoryID while before_now(start) or replication_key is None: self._replicate(customer_key, keys, parent_category_id, table, partial=(replication_key is not None), start=start, end=end, unit=unit, replication_key=replication_key) if replication_key is None: return self.state = incorporate(self.state, table, replication_key, start) save_state(self.state) start = end end = increment_date(start, unit)