def test_dataset_update_status_works(fill_db, sm_config, ds_config):
db = DB(sm_config['db'])
es_mock = MagicMock(spec=ESExporter)
status_queue_mock = MagicMock(spec=QueuePublisher)
upload_dt = datetime.now()
ds_id = '2000-01-01'
ds = Dataset(ds_id,
'ds_name',
'input_path',
upload_dt, {},
ds_config,
DatasetStatus.INDEXING,
mol_dbs=['HMDB'],
adducts=['+H'])
ds.set_status(db, es_mock, status_queue_mock, DatasetStatus.FINISHED)
assert DatasetStatus.FINISHED == Dataset.load(db, ds_id).status
status_queue_mock.publish.assert_called_once_with({
'ds_id':
ds_id,
'status':
DatasetStatus.FINISHED
})
def index(self, ds: Dataset):
"""Re-index all search results for the dataset.
Args:
ds: dataset to index
"""
self._es.delete_ds(ds.id, delete_dataset=False)
job_docs = self._db.select_with_fields(
'SELECT id, moldb_id FROM job WHERE ds_id = %s', params=(ds.id,)
)
moldb_ids = ds.config['database_ids']
for job_doc in job_docs:
moldb = molecular_db.find_by_id(job_doc['moldb_id'])
if job_doc['moldb_id'] not in moldb_ids:
self._db.alter('DELETE FROM job WHERE id = %s', params=(job_doc['id'],))
else:
isocalc = IsocalcWrapper(ds.config)
self._es.index_ds(ds_id=ds.id, moldb=moldb, isocalc=isocalc)
ds.set_status(self._db, self._es, DatasetStatus.FINISHED)