def parse_args(): """ This function expands the default argument list defined in cdr_cleaner.args_parser :return: an expanded argument list object """ import cdr_cleaner.args_parser as parser argument_parser = parser.get_argument_parser() argument_parser.add_argument( '-y', '--year_threshold', dest='year_threshold', action='store', help='The year threshold applied to domain tables except observation', required=False, default=DEFAULT_YEAR_THRESHOLD) argument_parser.add_argument('-o', '--observation_year_threshold', dest='observation_year_threshold', action='store', help='The threshold applied to observation', required=False, default=OBSERVATION_DEFAULT_YEAR_THRESHOLD) return argument_parser.parse_args()
def parse_args(): """ Expands the default argument list defined in cdr_cleaner.args_parser :return: an expanded argument list object """ import cdr_cleaner.args_parser as parser argument_parser = parser.get_argument_parser() argument_parser.add_argument( '-c', '--combined_dataset_id', dest='combined_dataset_id', action='store', help='The combined dataset used to generate the deid dataset', required=True) return argument_parser.parse_args()
def parse_args(): """ Expands the default argument list defined in cdr_cleaner.args_parser :return: an expanded argument list object """ import cdr_cleaner.args_parser as parser argument_parser = parser.get_argument_parser() argument_parser.add_argument( '-m', '--mapping_dataset_id', dest='mapping_dataset_id', action='store', help= 'The dataset containing mapping tables, typically the combined_dataset', required=True) return argument_parser.parse_args()
def get_parser(): """ Create a parser which raises invalid enum errors :return: parser """ from cdr_cleaner import args_parser engine_parser = args_parser.get_argument_parser() engine_parser.add_argument( '-a', '--data_stage', required=True, dest='data_stage', action='store', type=DataStage, choices=list([s for s in DataStage if s is not DataStage.UNSPECIFIED]), help='Specify the dataset') return engine_parser
if add_handler: handler = logging.StreamHandler() handler.setLevel(logging.INFO) formatter = logging.Formatter( '%(asctime)s - %(levelname)s - %(name)s - %(message)s') handler.setFormatter(formatter) logging.getLogger('').addHandler(handler) if __name__ == '__main__': import cdr_cleaner.args_parser as ap # import cdr_cleaner.clean_cdr_engine as clean_engine from constants.cdr_cleaner.clean_cdr_engine import FILENAME from utils import bq parser = ap.get_argument_parser() parser.add_argument( '--cope_survey_dataset', action='store', dest='cope_survey_dataset_id', help=('Dataset containing the mapping table provided by RDR team. ' 'These maps questionnaire_response_ids to cope_months.'), required=True) parser.add_argument( '--cope_survey_table', action='store', dest='cope_survey_table', required=True, help='Name of the table cotaining the cope survey mapping information') parser.add_argument( '--mapping_dataset',
raise NotImplementedError("Please fix me.") def get_sandbox_tablenames(self): """ Generates list of sandbox table names created by this rule. """ return [ self.sandbox_table_for(table) for table in self.affected_tables ] if __name__ == '__main__': import cdr_cleaner.args_parser as parser import cdr_cleaner.clean_cdr_engine as clean_engine ext_parser = parser.get_argument_parser() ARGS = ext_parser.parse_args() if ARGS.list_queries: clean_engine.add_console_logging() query_list = clean_engine.get_query_list( ARGS.project_id, ARGS.dataset_id, ARGS.sandbox_dataset_id, [(RemoveEhrDataWithoutConsent, )]) for query in query_list: LOGGER.info(query) else: clean_engine.add_console_logging(ARGS.console_log) clean_engine.clean_dataset(ARGS.project_id, ARGS.dataset_id, ARGS.sandbox_dataset_id, ARGS.cutoff_date, [(RemoveEhrDataWithoutConsent, )])
select_query[cdr_consts.DESTINATION_DATASET] = sandbox_dataset_id queries.append(select_query) update_query = dict() update_query[cdr_consts.QUERY] = UPDATE_NEGATIVE_PPI_QUERY.format( project_id=project_id, dataset_id=dataset_id) queries.append(update_query) return queries if __name__ == '__main__': import cdr_cleaner.args_parser as parser import cdr_cleaner.clean_cdr_engine as clean_engine ARGS = parser.get_argument_parser().parse_args() if ARGS.list_queries: clean_engine.add_console_logging() query_list = clean_engine.get_query_list(ARGS.project_id, ARGS.dataset_id, ARGS.sandbox_dataset_id, [(get_update_ppi_queries,)]) for query in query_list: LOGGER.info(query) else: clean_engine.add_console_logging(ARGS.console_log) clean_engine.clean_dataset(ARGS.project_id, ARGS.dataset_id, ARGS.sandbox_dataset_id, [(get_update_ppi_queries,)])