def setUp(self):
"""Setup function for each unit test."""
super(GoogleAnalyticsHookTest, self).setUp()
self.test_tracking_id = 'UA-12323-4'
self.payload_builder = ga_hook.PayloadBuilder(self.test_tracking_id)
self.event_test_data = {
'ec': 'ClientID',
'ea': 'test_event_action',
'el': '20190423',
'ev': 1,
'cid': '12345.456789'
}
self.small_event = {
'cid': '12345.67890',
'ec': 'ClientID',
'ea': 'test_event_action',
'el': '20190423',
'ev': 1,
'z': '1558517072202080'
}
# Both of the below are approx 4K of data
self.medium_event = {**self.small_event, 'ea': 'x' * 3800}
self.utf8_event = {**self.small_event, 'ea': b'\xf0\xa9\xb8\xbd' * 320}
self.test_hook = ga_hook.GoogleAnalyticsHook(self.test_tracking_id,
self.event_test_data)
self.test_hook._send_http_request = mock.MagicMock(autospec=True)
self.test_hook._send_http_request.return_value = mock.Mock(ok=True)
self.test_hook._send_http_request.return_value.status = 200
def __init__(self,
*args,
gcs_bucket: Text,
gcs_prefix: Text,
gcs_content_type: Text,
ga_tracking_id: Text,
ga_base_params: Optional[Mapping[Text, Any]] = None,
ga_dry_run: Optional[bool] = False,
**kwargs) -> None:
"""Initializes the DataConnectorOperator with an input and output hooks.
Args:
*args: arguments for the operator.
gcs_bucket: Unique name of the bucket holding the target blob.
gcs_prefix: The path to a location within the bucket.
gcs_content_type: Blob's content type. Either 'JSON' or 'CSV'.
ga_tracking_id: Google Analytics' tracking id to identify a property.
ga_base_params: Default parameters that serve as the base on which to
build the Measurement Protocol payload.
ga_dry_run: If True, this will not send real hits to the endpoint.
**kwargs: Other arguments to pass through to the operator or hooks.
"""
self.gcs_hook = gcs_hook.GoogleCloudStorageHook(
bucket=gcs_bucket,
prefix=gcs_prefix,
content_type=gcs_content_type)
self.ga_hook = ga_hook.GoogleAnalyticsHook(tracking_id=ga_tracking_id,
base_params=ga_base_params,
dry_run=ga_dry_run)
super(GoogleCloudStorageToGoogleAnalyticsOperator,
self).__init__(self.gcs_hook, self.ga_hook, *args, **kwargs)
def __init__(self, *args,
bq_conn_id: Text,
bq_dataset_id: Text,
bq_table_id: Text,
bq_selected_fields: Optional[Text] = None,
ga_tracking_id: Text,
ga_base_params: Optional[Mapping[Text, Any]] = None,
ga_dry_run: Optional[bool] = False,
**kwargs) -> None:
"""Initializes the DataConnectorOperator with an input and output hooks.
Args:
*args: arguments for the operator.
bq_conn_id: Connection id passed to airflow's BigQueryHook.
bq_dataset_id: Dataset id of the target table.
bq_table_id: Table name of the target table.
bq_selected_fields: Subset of fields to return (e.g. 'field_1,field_2').
ga_tracking_id: Google Analytics' tracking id to identify a property.
ga_base_params: Default parameters that serve as the base on which to
build the Measurement Protocol payload.
ga_dry_run: If True, this will not send real hits to the endpoint.
**kwargs: Other arguments to pass through to the operator or hooks.
"""
self.bq_hook = bq_hook.BigQueryHook(conn_id=bq_conn_id,
dataset_id=bq_dataset_id,
table_id=bq_table_id,
selected_fields=bq_selected_fields)
self.ga_hook = ga_hook.GoogleAnalyticsHook(tracking_id=ga_tracking_id,
base_params=ga_base_params,
dry_run=ga_dry_run)
super(BigQueryToGoogleAnalyticsOperator, self).__init__(
self.bq_hook, self.ga_hook, *args, **kwargs)
def test_ga_hook_send_batch_hit_with_return_error_status_code(self, r_code):
"""Test GoogleAnalyticsHook sends batch hit with error response."""
test_hook = ga_hook.GoogleAnalyticsHook(self.test_tracking_id,
self.event_test_data,
False)
test_payload = self.payload_builder.generate_batch_payload(
'event', [self.event_test_data])
with mock.patch('urllib.request.urlopen') as urlopen_mock:
mock_response = urlopen_mock.return_value.__enter__.return_value
mock_response.status = r_code
with self.assertRaises(errors.DataOutConnectorSendUnsuccessfulError):
test_hook.send_hit(test_payload, send_type='batch')
def test_ga_hook_send_single_hit(self):
"""Test GoogleAnalyticsHook sends single hit."""
test_hook = ga_hook.GoogleAnalyticsHook(self.test_tracking_id,
self.event_test_data,
False)
test_payload = self.payload_builder.generate_single_payload(
'event', self.event_test_data)
with mock.patch('urllib.request.urlopen') as urlopen_mock:
mock_response = urlopen_mock.return_value.__enter__.return_value
mock_response.status = 200
test_hook.send_hit(test_payload)
urlopen_mock.assert_called_once()
def test_ga_hook_send_batch_hit_with_retrys_on_retriable_error(self):
"""Test GoogleAnalyticsHook retries when retriable error occures."""
test_hook = ga_hook.GoogleAnalyticsHook(self.test_tracking_id,
self.event_test_data,
False)
test_payload = self.payload_builder.generate_batch_payload(
'event', [self.event_test_data])
with mock.patch('urllib.request.urlopen') as urlopen_mock:
urlopen_mock.return_value.__enter__.return_value.status = 429
try:
test_hook.send_hit(test_payload)
except errors.DataOutConnectorSendUnsuccessfulError:
pass
self.assertEqual(urlopen_mock.call_count,
retry_utils._RETRY_UTILS_MAX_RETRIES)
def test_ga_hook_get_invalid_tracking_id(self):
"""Test GoogleAnalyticsHook with invalid tracking id."""
with self.assertRaises(errors.DataOutConnectorValueError):
ga_hook.GoogleAnalyticsHook('UA-123-b', self.event_test_data)
