def test_replace_nested_properties_with_underscores_for_default_column_name(
self):
self.assertEqual(
'parent_child_grandchild',
Property('parent',
Property('child', Property('grandchild',
'VARCHAR(MAX)'))).column_name)
def test_bulk_copy_for_hourly_time_series(self):
paths = ['test.table/2014-11-10/00', 'test.table/2014-11-11/01']
properties = Property('userid',
'VARCHAR(36)'), Property('timestamp',
'TIMESTAMP')
series_column = 'test'
max_error = 20
import_pipeline = Mock()
sources = Mock()
sources.get = Mock(return_value=paths)
time_series_import = SqlTimeSeriesImport('test_table',
'2014-11-10 00:00:00',
sources, *properties)
time_series_import.bulk_copy(import_pipeline, '', max_error,
series_column)
first_bulk_copy, second_bulk_copy = \
[c[1] for c in import_pipeline.bulk_copy.call_args_list]
self.__assert_bulk_copy(
first_bulk_copy, '', paths[0],
self.__expected_schema('test_table_2014_11_10_00'), max_error)
self.__assert_bulk_copy(
second_bulk_copy, '', paths[1],
self.__expected_schema('test_table_2014_11_11_01'), max_error)
self.assertEqual(import_pipeline.bulk_copy.call_count, 2)
import_pipeline.sql.assert_called_once_with(
*self.expected_hourly_update_sql(series_column))
def test_have_column_name_for_nested_properties(self):
self.assertEqual(
'my_column_name',
Property(
'parent',
Property(
'child',
Property('grandchild', 'VARCHAR(MAX)',
'my_column_name'))).column_name)
def test_stage_update_when_column_name_defined(self):
with patch.object(Database, 'execute') as execute:
schema = JsonObject(
TABLE_NAME, Property('property1', 'VARCHAR(10)', 'someColumn'),
Property('property2', 'TIMESTAMP', 'anotherColumn'))
table = TargetTable(schema, Database(Mock()))
table.stage_update()
expected_sql = 'CREATE TABLE {0}_update (someColumn VARCHAR(10), ' \
'anotherColumn TIMESTAMP)'.format(TABLE_NAME)
execute.assert_called_once_with(expected_sql)
def test_create_when_column_name_not_defined_for_nested_property(self):
with patch.object(Database, 'execute') as execute:
schema = JsonObject(
TABLE_NAME, Property('property1', 'VARCHAR(10)'),
Property('property2', Property('timestamp', 'TIMESTAMP')))
table = TargetTable(schema, Database(Mock()))
table.create()
expected_sql = 'CREATE TABLE {0} (property1 VARCHAR(10), ' \
'property2_timestamp TIMESTAMP)'.format(TABLE_NAME)
execute.assert_called_once_with(expected_sql)
def setUp(self):
self.schema = JsonObject(TABLE_NAME, Property('id', 'VARCHAR(36)'))
self.database = Database(psycopg2.connect(CONNECTION))
self.database.open()
table = TargetTable(self.schema, self.database)
table.create()
self.database.commit()
def test_add_step(self):
schema = JsonObject(TABLE_NAME, Property('id', 'VARCHAR(36)'))
bucket = Mock()
database = create_autospec(Database)
expected = BulkCopyFromS3JsonStep(
metadata='',
source='',
schema=schema,
aws_access_key_id=AWS_ACCESS_KEY_ID,
aws_secret_access_key=AWS_SECRET_ACCESS_KEY,
bucket=bucket,
table=TargetTable(schema, database))
pipeline = S3BulkCopyPipeline(AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY,
bucket, database)
pipeline.step(metadata='', source='', schema=schema)
step = pipeline.steps()[0]
self.assertEqual(expected.metadata, step.metadata)
self.assertEqual(expected.source, step.source)
self.assertEqual(expected.schema, step.schema)
self.assertEqual(expected.aws_access_key_id, step.aws_access_key_id)
self.assertEqual(expected.aws_secret_access_key,
step.aws_secret_access_key)
self.assertEqual(expected.bucket, step.bucket)
self.assertEqual(expected.table.schema, step.table.schema)
self.assertEqual(expected.table.database, step.table.database)
def test_have_paths_for_nested_objects(self):
schema = JsonObject(
TABLE_NAME, Property('property1', 'VARCHAR(10)'),
Property('property2', 'TIMESTAMP'),
Property('property3.dottedName', 'DOUBLE PRECISION'),
Property('property4',
Property('child', Property('subchild', 'BOOLEAN'))))
self.assertEqual(
{
'jsonpaths': [
"$['property1']", "$['property2']",
"$['property3.dottedName']",
"$['property4']['child']['subchild']"
]
}, schema.paths())
def setUp(self):
self.schema = JsonObject(TABLE_NAME, Property('id', 'VARCHAR(36)'))
By default the name of the JSON property is used as the column, but can be set
to a custom column name.
"""
if __name__ == '__main__':
pipeline = S3CopyPipeline(
aws_access_key_id=env('AWS_ACCESS_KEY_ID'),
aws_secret_access_key=env('AWS_SECRET_ACCESS_KEY'),
bucket=env('BUCKET_NAME'),
db_connection=psycopg2.connect(env('REDSHIFT_CONNECTION')))
pipeline.bulk_copy(metadata='path_to_save_pipeline_metadata',
source='path_of_source_data',
schema=JsonObject(
'destination_table_name',
Property('id', 'VARCHAR(36)'),
Property('someNumber', 'INTEGER',
'custom_column_name')))
pipeline.manifest_copy(metadata='path_to_save_pipeline_metadata',
source='path_of_incremental_source_data',
schema=JsonObject(
'incremental_destination_table_name',
Property('id', 'VARCHAR(36)'),
Property('someNumber', 'INTEGER',
'custom_column_name')))
pipeline.sql(('SELECT someNumber + %s '
'INTO some_olap_table FROM destination_table_name', 1),
('SELECT * INTO destination_table_name_copy '
'FROM destination_table_name'))
def __expected_schema(table):
return JsonObject(table, Property('userid', 'VARCHAR(36)'),
Property('timestamp', 'TIMESTAMP'))
def setUp(self):
self.schema = JsonObject(TABLE_NAME, Property('id', 'VARCHAR(36)'))
self.bucket = Bucket(AWS_ACCESS_KEY_ID, AWS_SECRET_ACCESS_KEY,
BUCKET_NAME, Mock())
self.bucket.save = Mock()
self.database = create_autospec(Database)
self.key_names = [
'object_path/00c68a1e-85f2-49e5-9d07-6922046dbc5a',
'object_path/19440481-7766-4061-bd42-4a54fa0aac7c',
'object_path/2014-09-02/19440481-7766-4061-bd42-4a54fa0aac7c',
'object_path/282e6063-ecef-4e45-bdfb-9fdfb39840cd',
'object_path/35cbf09a-b2dc-43f2-96f6-7d7573906268',
'object_path/80536e83-6bbe-4a42-ade1-533d99321a6c',
'object_path/cf00b394-3ff3-4418-b244-2ccf104fcc40',
'object_path/e822e2ae-61f5-4be0-aacd-ca6de70faad1'
]
self.bucket.list = Mock(
return_value=[self.mock_key(key) for key in self.key_names])
self.manifest = SqlManifest(metadata='',
source='',
schema=self.schema,
bucket=self.bucket,
db_connection=self.database)
self.expected_manifest = {
'entries': [{
'url':
's3://{0}/object_path/00c68a1e-85f2-49e5-9d07-6922046dbc5a'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/19440481-7766-4061-bd42-4a54fa0aac7c'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/2014-09-02/19440481-7766-4061-bd42-4a54fa0aac7c'
.format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/282e6063-ecef-4e45-bdfb-9fdfb39840cd'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/35cbf09a-b2dc-43f2-96f6-7d7573906268'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/80536e83-6bbe-4a42-ade1-533d99321a6c'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/cf00b394-3ff3-4418-b244-2ccf104fcc40'.
format(BUCKET_NAME),
'mandatory':
True
}, {
'url':
's3://{0}/object_path/e822e2ae-61f5-4be0-aacd-ca6de70faad1'.
format(BUCKET_NAME),
'mandatory':
True
}]
}
def test_preserve_periods_for_default_column_name(self):
self.assertEqual(
'some.column.name',
Property('some.column.name', 'VARCHAR(MAX)').column_name)
def test_have_column_name(self):
self.assertEqual(
'my_column_name',
Property('some.column.name', 'VARCHAR(MAX)',
'my_column_name').column_name)