def _get_existing_transfer(self,
data_source_id: str,
destination_dataset_id: str = None,
params: Dict[str, str] = None,
name: str = None) -> bool:
"""Gets data transfer if it already exists.
Args:
data_source_id: Data source id.
destination_dataset_id: BigQuery dataset id.
params: Data transfer specific parameters.
Returns:
Data Transfer if the transfer already exists.
None otherwise.
"""
parent = self.client.location_path(
self.project_id, config_parser.get_dataset_location())
for transfer_config in self.client.list_transfer_configs(parent):
if transfer_config.data_source_id != data_source_id:
continue
if destination_dataset_id and transfer_config.destination_dataset_id != destination_dataset_id:
continue
# If the transfer config is in Failed state, we should ignore.
is_valid_state = transfer_config.state in (_PENDING_STATE,
_RUNNING_STATE,
_SUCCESS_STATE)
params_match = self._check_params_match(transfer_config, params)
name_matches = name is None or name == transfer_config.display_name
if params_match and is_valid_state and name_matches:
return transfer_config
return None
def create_merchant_center_transfer(
self, merchant_id: str, destination_dataset: str,
enable_market_insights: bool) -> types.TransferConfig:
"""Creates a new merchant center transfer.
Merchant center allows retailers to store product info into Google. This
method creates a data transfer config to copy the product data to BigQuery.
Args:
merchant_id: Google Merchant Center(GMC) account id.
destination_dataset: BigQuery dataset id.
enable_market_insights: Whether to deploy market insights solution.
Returns:
Transfer config.
"""
logging.info('Creating Merchant Center Transfer.')
parameters = struct_pb2.Struct()
parameters['merchant_id'] = merchant_id
parameters['export_products'] = True
if enable_market_insights:
parameters['export_price_benchmarks'] = True
parameters['export_best_sellers'] = True
data_transfer_config = self._get_existing_transfer(
_MERCHANT_CENTER_ID, destination_dataset, parameters)
if data_transfer_config:
logging.info(
'Data transfer for merchant id %s to destination dataset %s '
'already exists.', merchant_id, destination_dataset)
return self._update_existing_transfer(data_transfer_config,
parameters)
logging.info(
'Creating data transfer for merchant id %s to destination dataset %s',
merchant_id, destination_dataset)
has_valid_credentials = self._check_valid_credentials(
_MERCHANT_CENTER_ID)
authorization_code = None
if not has_valid_credentials:
authorization_code = self._get_authorization_code(
_MERCHANT_CENTER_ID)
dataset_location = config_parser.get_dataset_location()
parent = self.client.location_path(self.project_id, dataset_location)
transfer_config_input = {
'display_name': f'Merchant Center Transfer - {merchant_id}',
'data_source_id': _MERCHANT_CENTER_ID,
'destination_dataset_id': destination_dataset,
'params': parameters,
'data_refresh_window_days': 0,
}
transfer_config = self.client.create_transfer_config(
parent, transfer_config_input, authorization_code)
logging.info(
'Data transfer created for merchant id %s to destination dataset %s',
merchant_id, destination_dataset)
return transfer_config
def _get_data_source(self, data_source_id: str) -> types.DataSource:
"""Returns data source.
Args:
data_source_id: Data source id.
"""
dataset_location = config_parser.get_dataset_location()
name = self.client.location_data_source_path(self.project_id,
dataset_location,
data_source_id)
return self.client.get_data_source(name)
def _check_valid_credentials(self, data_source_id: str) -> bool:
"""Returns true if valid credentials exist for the given data source.
Args:
data_source_id: Data source id.
"""
dataset_location = config_parser.get_dataset_location()
name = self.client.location_data_source_path(self.project_id,
dataset_location,
data_source_id)
response = self.client.check_valid_creds(name)
return response.has_valid_creds
def schedule_query(self, name: str,
query_string: str) -> types.TransferConfig:
"""Schedules query to run every day.
Args:
name: Name of the scheduled query.
query_string: The query to be run.
"""
data_transfer_config = self._get_existing_transfer('scheduled_query',
name=name)
parameters = struct_pb2.Struct()
parameters['query'] = query_string
if data_transfer_config:
logging.info(
'Data transfer for scheduling query "%s" already exists.',
name)
updated_transfer_config = self._update_existing_transfer(
data_transfer_config, parameters)
logging.info('Data transfer for scheduling query "%s" updated.',
name)
start_time_pb = timestamp_pb2.Timestamp()
start_time = datetime.datetime.now(tz=pytz.utc)
start_time_pb.FromDatetime(start_time)
self.client.start_manual_transfer_runs(
parent=updated_transfer_config.name,
requested_run_time=start_time_pb)
logging.info(
'One time manual run started. It might take upto 1 hour for performance data'
' to reflect on the dash.')
return updated_transfer_config
dataset_location = config_parser.get_dataset_location()
parent = self.client.location_path(self.project_id, dataset_location)
params = {
'query': query_string,
}
transfer_config_input = google.protobuf.json_format.ParseDict(
{
'display_name': name,
'data_source_id': 'scheduled_query',
'params': params,
'schedule': 'every 24 hours',
},
bigquery_datatransfer_v1.types.TransferConfig(),
)
has_valid_credentials = self._check_valid_credentials(
'scheduled_query')
authorization_code = ''
if not has_valid_credentials:
authorization_code = self._get_authorization_code(
'scheduled_query')
transfer_config = self.client.create_transfer_config(
parent, transfer_config_input, authorization_code)
return transfer_config
def wait_for_transfer_completion(self, transfer_config: Dict[str,
Any]) -> None:
"""Waits for the completion of data transfer operation.
This method retrieves data transfer operation and checks for its status. If
the operation is not completed, then the operation is re-checked after
`_SLEEP_SECONDS` seconds.
Args:
transfer_config: Resource representing data transfer.
Raises:
DataTransferError: If the data transfer is not successfully completed.
"""
# TODO: Use exponential back-off for polling.
transfer_config_name = transfer_config.name
transfer_config_id = transfer_config_name.split('/')[-1]
poll_counter = 0 # Counter to keep polling count.
while True:
transfer_config_path = self.client.location_transfer_config_path(
self.project_id, config_parser.get_dataset_location(),
transfer_config_id)
response = self.client.list_transfer_runs(transfer_config_path)
latest_transfer = None
for transfer in response:
latest_transfer = transfer
break
if not latest_transfer:
return
if latest_transfer.state == _SUCCESS_STATE:
logging.info('Transfer %s was successful.',
transfer_config_name)
return
if (latest_transfer.state == _FAILED_STATE
or latest_transfer.state == _CANCELLED_STATE):
error_message = (
f'Transfer {transfer_config_name} was not successful. '
f'Error - {latest_transfer.error_status}')
logging.error(error_message)
raise DataTransferError(error_message)
logging.info(
'Transfer %s still in progress. Sleeping for %s seconds before '
'checking again.', transfer_config_name, _SLEEP_SECONDS)
time.sleep(_SLEEP_SECONDS)
poll_counter += 1
if poll_counter >= _MAX_POLL_COUNTER:
error_message = (
f'Transfer {transfer_config_name} is taking too long'
' to finish. Hence failing the request.')
logging.error(error_message)
raise DataTransferError(error_message)
def create_dataset_if_not_exists(project_id: str, dataset_id: str) -> None:
"""Creates BigQuery dataset if it doesn't exists.
Args:
project_id: A cloud project id.
dataset_id: BigQuery dataset id.
"""
# Construct a BigQuery client object.
client = bigquery.Client(project=project_id)
fully_qualified_dataset_id = f'{project_id}.{dataset_id}'
try:
client.get_dataset(fully_qualified_dataset_id)
logging.info('Dataset %s already exists.', fully_qualified_dataset_id)
except exceptions.NotFound:
logging.info('Dataset %s is not found.', fully_qualified_dataset_id)
dataset = bigquery.Dataset(fully_qualified_dataset_id)
dataset.location = config_parser.get_dataset_location()
client.create_dataset(dataset)
logging.info('Dataset %s created.', fully_qualified_dataset_id)
def execute_queries(project_id: str, dataset_id: str, merchant_id: str,
customer_id: str, enable_market_insights: bool) -> None:
"""Executes list of queries."""
# Sql files to be executed in a specific order. The prefix "scripts" should be omitted.
sql_files = [
'1_product_view.sql',
'targeted_products/targeted_product_ddl.sql',
'targeted_products/construct_parsed_criteria.sql',
'2_product_metrics_view.sql',
'3_customer_view.sql',
'4_product_detailed_view.sql',
'materialize_product_detailed.sql',
'materialize_product_historical.sql',
]
if enable_market_insights:
market_insights_sql_files = [
'market_insights/snapshot_view.sql',
'market_insights/historical_view.sql'
]
sql_files.extend(market_insights_sql_files)
prefix = 'scripts'
query_params = {
'project_id': project_id,
'dataset': dataset_id,
'merchant_id': merchant_id,
'external_customer_id': customer_id
}
location = config_parser.get_dataset_location()
client = bigquery.Client(project=project_id)
for sql_file in sql_files:
try:
query = configure_sql(os.path.join(prefix, sql_file), query_params)
query_job = client.query(query, location=location)
query_job.result()
except:
logging.exception('Error in %s', sql_file)
raise
def create_google_ads_transfer(
self,
customer_id: str,
destination_dataset: str,
backfill_days: int = 30) -> types.TransferConfig:
"""Creates a new Google Ads transfer.
This method creates a data transfer config to copy Google Ads data to
BigQuery dataset.
Args:
customer_id: Google Ads customer id.
destination_dataset: BigQuery dataset id.
backfill_days: Number of days to backfill.
Returns:
Transfer config.
"""
logging.info('Creating Google Ads Transfer.')
parameters = struct_pb2.Struct()
parameters['customer_id'] = customer_id
data_transfer_config = self._get_existing_transfer(
_GOOGLE_ADS_ID, destination_dataset, parameters)
if data_transfer_config:
logging.info(
'Data transfer for Google Ads customer id %s to destination dataset '
'%s already exists.', customer_id, destination_dataset)
return data_transfer_config
logging.info(
'Creating data transfer for Google Ads customer id %s to destination '
'dataset %s', customer_id, destination_dataset)
has_valid_credentials = self._check_valid_credentials(_GOOGLE_ADS_ID)
authorization_code = None
if not has_valid_credentials:
authorization_code = self._get_authorization_code(_GOOGLE_ADS_ID)
dataset_location = config_parser.get_dataset_location()
parent = self.client.location_path(self.project_id, dataset_location)
transfer_config_input = {
'display_name': f'Google Ads Transfer - {customer_id}',
'data_source_id': _GOOGLE_ADS_ID,
'destination_dataset_id': destination_dataset,
'params': parameters,
'data_refresh_window_days': 1,
}
transfer_config = self.client.create_transfer_config(
parent, transfer_config_input, authorization_code)
logging.info(
'Data transfer created for Google Ads customer id %s to destination '
'dataset %s', customer_id, destination_dataset)
if backfill_days:
transfer_config_name = transfer_config.name
transfer_config_id = transfer_config_name.split('/')[-1]
start_time = datetime.datetime.now(
tz=pytz.utc) - datetime.timedelta(days=backfill_days)
end_time = datetime.datetime.now(tz=pytz.utc)
start_time = start_time.replace(hour=0,
minute=0,
second=0,
microsecond=0)
end_time = end_time.replace(hour=0,
minute=0,
second=0,
microsecond=0)
parent = self.client.location_transfer_config_path(
self.project_id, dataset_location, transfer_config_id)
start_time_pb = timestamp_pb2.Timestamp()
end_time_pb = timestamp_pb2.Timestamp()
start_time_pb.FromDatetime(start_time)
end_time_pb.FromDatetime(end_time)
self.client.schedule_transfer_runs(parent, start_time_pb,
end_time_pb)
return transfer_config
