def setUp(self):
args = {'owner': 'airflow', 'start_date': DEFAULT_DATE}
dag = DAG('test_dag_id', default_args=args)
self.dag = dag
self.sql = 'SELECT 1'
self.hook = AwsDynamoDBHook(aws_conn_id='aws_default',
region_name='us-east-1')
def execute(self, context):
table = AwsDynamoDBHook().get_conn().Table(self.dynamodb_table_name)
scan_kwargs = copy(
self.dynamodb_scan_kwargs) if self.dynamodb_scan_kwargs else {}
err = None
f = NamedTemporaryFile()
try:
f = self._scan_dynamodb_and_upload_to_s3(f, scan_kwargs, table)
except Exception as e:
err = e
raise e
finally:
if err is None:
_upload_file_to_s3(f, self.s3_bucket_name, self.s3_key_prefix)
f.close()
def test_insert_batch_items_dynamodb_table(self):
hook = AwsDynamoDBHook(aws_conn_id='aws_default',
table_name='test_airflow',
table_keys=['id'],
region_name='us-east-1')
# this table needs to be created in production
table = hook.get_conn().create_table(TableName='test_airflow',
KeySchema=[
{
'AttributeName': 'id',
'KeyType': 'HASH'
},
],
AttributeDefinitions=[{
'AttributeName':
'id',
'AttributeType':
'S'
}],
ProvisionedThroughput={
'ReadCapacityUnits': 10,
'WriteCapacityUnits': 10
})
table = hook.get_conn().Table('test_airflow')
items = [{
'id': str(uuid.uuid4()),
'name': 'airflow'
} for _ in range(10)]
hook.write_batch_data(items)
table.meta.client.get_waiter('table_exists').wait(
TableName='test_airflow')
self.assertEqual(table.item_count, 10)
def execute(self, context):
hive = HiveServer2Hook(hiveserver2_conn_id=self.hiveserver2_conn_id)
self.log.info('Extracting data from Hive')
self.log.info(self.sql)
data = hive.get_pandas_df(self.sql, schema=self.schema)
dynamodb = AwsDynamoDBHook(aws_conn_id=self.aws_conn_id,
table_name=self.table_name,
table_keys=self.table_keys,
region_name=self.region_name)
self.log.info('Inserting rows into dynamodb')
if self.pre_process is None:
dynamodb.write_batch_data(
json.loads(data.to_json(orient='records')))
else:
dynamodb.write_batch_data(
self.pre_process(data=data,
args=self.pre_process_args,
kwargs=self.pre_process_kwargs))
self.log.info('Done.')
def test_get_conn_returns_a_boto3_connection(self):
hook = AwsDynamoDBHook(aws_conn_id='aws_default')
self.assertIsNotNone(hook.get_conn())