def main(args):
"""Main method"""
loglevel = logging.WARNING
if args.verbose:
loglevel = logging.DEBUG
logging.basicConfig(format='%(levelname)s: %(message)s', level=loglevel)
if args.list_providers:
list_provider_classes()
return 0
schema = yaml.load(open(args.schema), Loader=yaml.FullLoader)
pg_args = get_pg_args(args)
connection = get_connection(pg_args)
if args.init_sql:
cursor = connection.cursor()
logging.info('Executing initialisation sql {}'.format(args.init_sql))
cursor.execute(args.init_sql)
cursor.close()
start_time = time.time()
truncate_tables(connection, schema.get('truncate', []))
anonymize_tables(connection, schema.get('tables', []), verbose=args.verbose, dry_run=args.dry_run)
if not args.dry_run:
connection.commit()
connection.close()
end_time = time.time()
logging.info('Anonymization took {:.2f}s'.format(end_time - start_time))
if args.dump_file:
create_database_dump(args.dump_file, pg_args)
def test(self):
mock_cursor = Mock()
connection = Mock()
connection.cursor.return_value = mock_cursor
truncate_tables(connection, ['table_a', 'table_b'])
assert mock_cursor.execute.call_args_list == [
call('TRUNCATE TABLE table_a, table_b;'),
]
mock_cursor.close.assert_called_once()
def test(self, quote_ident, tables, expected):
mock_cursor = Mock()
connection = Mock()
connection.cursor.return_value = mock_cursor
truncate_tables(connection, tables)
if tables:
connection.cursor.assert_called_once()
assert mock_cursor.execute.call_args_list == [call(expected)]
mock_cursor.close.assert_called_once()
else:
connection.cursor.assert_not_called()
mock_cursor.execute.assert_not_called()
mock_cursor.close.assert_not_called()
def main():
"""Main method"""
parser = argparse.ArgumentParser(description='Anonymize data of a PostgreSQL database')
parser.add_argument('-v', '--verbose', action='count', help='Increase verbosity')
parser.add_argument('-l', '--list-providers', action='store_true', help='Show a list of all available providers',
default=False)
parser.add_argument('--schema', help='A YAML schema file that contains the anonymization rules',
default=DEFAULT_SCHEMA_FILE)
parser.add_argument('--dbname', help='Name of the database')
parser.add_argument('--user', help='Name of the database user')
parser.add_argument('--password', default='', help='Password for the database user')
parser.add_argument('--host', help='Database hostname', default='localhost')
parser.add_argument('--port', help='Port of the database', default='5432')
parser.add_argument('--dry-run', action='store_true', help='Don\'t commit changes made on the database',
default=False)
parser.add_argument('--dump-file', help='Create a database dump file with the given name')
args = parser.parse_args()
loglevel = logging.WARNING
if args.verbose:
loglevel = logging.DEBUG
logging.basicConfig(format='%(levelname)s: %(message)s', level=loglevel)
if args.list_providers:
list_provider_classes()
sys.exit(0)
schema = yaml.load(open(args.schema), Loader=yaml.FullLoader)
pg_args = get_pg_args(args)
connection = get_connection(pg_args)
start_time = time.time()
truncate_tables(connection, schema.get('truncate', []))
anonymize_tables(connection, schema.get('tables', []), verbose=args.verbose)
if not args.dry_run:
connection.commit()
connection.close()
end_time = time.time()
logging.info('Anonymization took {:.2f}s'.format(end_time - start_time))
if args.dump_file:
create_database_dump(args.dump_file, pg_args)