示例#1
0
    def sync_data(self):
        table = self.__class__.TABLE
        selector = FuelSDK.ET_List

        search_filter = None
        retrieve_all_since = get_last_record_value_for_table(self.state, table)

        if retrieve_all_since is not None:
            search_filter = {
                'Property': 'ModifiedDate',
                'SimpleOperator': 'greaterThan',
                'Value': retrieve_all_since
            }

        stream = request('List', selector, self.auth_stub, search_filter)

        for _list in stream:
            _list = self.filter_keys_and_parse(_list)

            self.state = incorporate(self.state, table, 'ModifiedDate',
                                     _list.get('ModifiedDate'))

            singer.write_records(table, [_list])

        save_state(self.state)
示例#2
0
    def sync_data(self):
        table = self.__class__.TABLE
        selector = FuelSDK.ET_ContentArea

        search_filter = None

        # pass config to return start date if not bookmark is found
        retrieve_all_since = get_last_record_value_for_table(
            self.state, table, self.config)

        if retrieve_all_since is not None:
            search_filter = {
                'Property': 'ModifiedDate',
                'SimpleOperator': 'greaterThan',
                'Value': retrieve_all_since
            }

        stream = request('ContentAreaDataAccessObject',
                         selector,
                         self.auth_stub,
                         search_filter,
                         batch_size=self.batch_size)

        catalog_copy = copy.deepcopy(self.catalog)

        for content_area in stream:
            content_area = self.filter_keys_and_parse(content_area)

            self.state = incorporate(self.state, table, 'ModifiedDate',
                                     content_area.get('ModifiedDate'))

            self.write_records_with_transform(content_area, catalog_copy,
                                              table)

        save_state(self.state)
    def sync_data(self):
        table = self.__class__.TABLE
        selector = FuelSDK.ET_ContentArea

        search_filter = None
        retrieve_all_since = get_last_record_value_for_table(self.state, table)

        if retrieve_all_since is not None:
            search_filter = {
                'Property': 'ModifiedDate',
                'SimpleOperator': 'greaterThan',
                'Value': retrieve_all_since
            }

        stream = request('ContentAreaDataAccessObject', selector,
                         self.auth_stub, search_filter)

        for content_area in stream:
            content_area = self.filter_keys_and_parse(content_area)

            self.state = incorporate(self.state, table, 'ModifiedDate',
                                     content_area.get('ModifiedDate'))

            singer.write_records(table, [content_area])

        save_state(self.state)
示例#4
0
    def sync_data(self):
        table = self.__class__.TABLE
        selector = FuelSDK.ET_Folder

        search_filter = None

        retrieve_all_since = get_last_record_value_for_table(
            self.state, table, self.config.get('start_date'))

        if retrieve_all_since is not None:
            search_filter = {
                'Property': 'ModifiedDate',
                'SimpleOperator': 'greaterThan',
                'Value': retrieve_all_since
            }

        stream = request('Folder', selector, self.auth_stub, search_filter)

        for folder in stream:
            folder = self.filter_keys_and_parse(folder)

            self.state = incorporate(self.state, table, 'ModifiedDate',
                                     folder.get('ModifiedDate'))

            singer.write_records(table, [folder])

        save_state(self.state)
示例#5
0
    def test_incorporate(self):
        self.assertEqual(
            incorporate({}, 'table', 'modifieddate', '2017-11-01'),
            {
                'bookmarks': {
                    'table': {
                        'last_record': '2017-11-01T00:00:00Z',
                        'field': 'modifieddate'
                    }
                }
            })

        self.assertEqual(
            incorporate({
                'bookmarks': {
                    'table': {
                        'last_record': '2017-01-01T00:00:00Z',
                        'field': 'modifieddate'
                    }
                }
            }, 'table', 'modifieddate', '2017-11-01'),
            {
                'bookmarks': {
                    'table': {
                        'last_record': '2017-11-01T00:00:00Z',
                        'field': 'modifieddate'
                    }
                }
            })

        self.assertEqual(
            incorporate({
                'bookmarks': {
                    'table': {
                        'last_record': '2017-11-01T00:00:00Z',
                        'field': 'modifieddate'
                    }
                }
            }, 'table', 'modifieddate', '2017-01-01'),
            {
                'bookmarks': {
                    'table': {
                        'last_record': '2017-11-01T00:00:00Z',
                        'field': 'modifieddate'
                    }
                }
            })
示例#6
0
    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 _replicate(self, customer_key, keys,
                   parent_category_id, table,
                   partial=False, start=None,
                   end=None, unit=None, replication_key=None):
        if partial:
            LOGGER.info("Fetching {} from {} to {}"
                        .format(table, start, end))

        cursor = FuelSDK.ET_DataExtension_Row()
        cursor.auth_stub = self.auth_stub
        cursor.CustomerKey = customer_key
        cursor.props = keys

        if partial:
            cursor.search_filter = get_date_page(replication_key,
                                                 start,
                                                 unit)

        batch_size = int(self.config.get('batch_size', 2500))
        result = request_from_cursor('DataExtensionObject', cursor,
                                     batch_size=batch_size)

        for row in result:
            row = self.filter_keys_and_parse(row)
            row['CategoryID'] = parent_category_id

            self.state = incorporate(self.state,
                                     table,
                                     replication_key,
                                     row.get(replication_key))

            singer.write_records(table, [row])

        if partial:
            self.state = incorporate(self.state,
                                     table,
                                     replication_key,
                                     start)

            save_state(self.state)
示例#8
0
    def sync_data(self):
        table = self.__class__.TABLE
        selector = FuelSDK.ET_Send
        list_sends_dao = ListSendDataAccessObject(self.config, self.state,
                                                  self.auth_stub,
                                                  self.listsend_catalog)
        link_sends_dao = LinkSendDataAccessObject(self.config, self.state,
                                                  self.auth_stub,
                                                  self.linksend_catalog)
        search_filter = None
        retrieve_all_since = get_last_record_value_for_table(
            self.state, table, self.config.get('start_date'))
        if self.REPLICATION_METHOD == "FULL_TABLE":
            retrieve_all_since = datetime.datetime.strptime(
                self.config.get('start_date'), "%Y-%m-%dT%H:%M:%SZ")

        if retrieve_all_since is not None:
            search_filter = {
                'Property': 'CreatedDate',
                'SimpleOperator': 'greaterThan',
                'Value': retrieve_all_since
            }

        stream = request('Send', selector, self.auth_stub, search_filter)

        if self.replicate_listsend:
            list_sends_dao.write_schema()

        if self.replicate_linksend:
            link_sends_dao.write_schema()

        for send in stream:
            send = self.filter_keys_and_parse(send)
            if self.replicate_listsend:
                list_sends_dao.sync_data_by_sendID(send.get('ID'))
            if self.replicate_linksend:
                link_sends_dao.sync_data_by_sendID(send.get('ID'))
            if retrieve_all_since.strftime('%Y-%m-%d') < send.get(
                    'CreatedDate'
            )[:
              10] and self.REPLICATION_METHOD == 'INCREMENTAL' or self.REPLICATION_METHOD == 'FULL_TABLE':
                self.state = incorporate(self.state, table, 'CreatedDate',
                                         send.get('CreatedDate'))

                singer.write_records(table, [send])

        save_state(self.state)
示例#9
0
    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'))

                    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

                    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):
        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)