def delete_table_not_in_metastore(schema_id, table_names, session=None):
db_tables = get_table_by_schema_id(schema_id, session=session)
with session.no_autoflush:
for data_table in db_tables:
if data_table.name not in table_names:
table_id = data_table.id
delete_table(table_id=table_id, commit=False, session=session)
delete_es_table_by_id(table_id)
LOG.info(f"deleted table {table_id}")
session.commit()
def delete_schema_not_in_metastore(metastore_id, schema_names, session=None):
for data_schema in iterate_data_schema(metastore_id, session=session):
LOG.info("checking schema %d" % data_schema.id)
if data_schema.name not in schema_names:
# We delete table 1 by 1 since we need to delete it for elasticsearch
# Maybe we can optimize it to allow batch deletion
for table in data_schema.tables:
table_id = table.id
delete_table(table_id=table_id, commit=False, session=session)
delete_es_table_by_id(table_id)
delete_schema(id=data_schema.id, commit=False, session=session)
LOG.info("deleted schema %d" % data_schema.id)
session.commit()
def sync_delete_table(self, schema_name, table_name, session=None):
"""Given a full qualified table name,
Remove the table if it exists
Arguments:
schema_name {str} -- the schema name
table_name {str} -- the table name
"""
# Double check if the table is indeed removed
table_exists = self.check_if_table_exists(schema_name, table_name)
# If no schema, then table doesn't exist
schema = get_schema_by_name(schema_name, self.metastore_id, session=session)
if not table_exists and schema is not None:
table = get_table_by_schema_id_and_name(
schema.id, table_name, session=session
)
if table:
delete_table(table_id=table.id, session=session)
