def test__load_genomic_json_dir(self):
self._reset(do_reset_patient=True)
# load clinical docs
args = Namespace(genomic=None,
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='json',
clinical='matchengine/tests/data/clinical_json/',
upsert_fields='')
load(args)
# load genomic docs
args = Namespace(genomic='matchengine/tests/data/genomic_json/',
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='json',
clinical=None,
upsert_fields='')
load(args)
clinical = list(self.db_ro.clinical.find({}))
genomic = list(self.db_ro.genomic.find({}))
assert len(clinical) == 2
assert len(genomic) == 2
assert genomic[0]['CLINICAL_ID'] == clinical[0]['_id']
assert genomic[1]['CLINICAL_ID'] == clinical[1]['_id']
def test__load_genomic_csv(self):
self._reset(do_reset_patient=True)
# load clinical doc
args = Namespace(genomic=None,
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='json',
clinical='matchengine/tests/data/clinical_json/',
upsert_fields='')
load(args)
# load genomic doc
args = Namespace(
genomic='matchengine/tests/data/genomic_csv/test_patients.csv',
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='csv',
clinical=None,
upsert_fields='')
load(args)
clinical = list(self.db_ro.clinical.find({}))
genomic = list(self.db_ro.genomic.find({}))
assert len(clinical) == 2
assert len(genomic) == 2
track = defaultdict(lambda: 1)
for clinical_doc in clinical:
track[clinical_doc['_id']] -= 1
for genomic_doc in genomic:
assert track[genomic_doc['CLINICAL_ID']] == 0
def test__load_trial_single_yaml(self):
self._reset(do_reset_trials=True)
args = Namespace(clinical=None,
genomic=None, db_name='integration_load',
plugin_dir='plugins',
trial_format='yaml',
trial='matchengine/tests/data/yaml/11-111.yaml',
upsert_fields='')
load(args)
assert len(list(self.db_ro.trial.find({}))) == 1
def test__load_clinical_single_csv_file(self):
self._reset(do_reset_patient=True)
args = Namespace(genomic=None,
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='csv',
clinical='matchengine/tests/data/clinical_csv/test_patients.csv',
upsert_fields='')
load(args)
assert len(list(self.db_ro.clinical.find({}))) == 2
def test__load_trials_single_json_multiple_trials(self):
"""mongoexport by default creates 'json' objects separated by new lines."""
self._reset(do_reset_trials=True)
args = Namespace(clinical=None,
genomic=None,
db_name='integration_load',
plugin_dir='plugins',
trial_format='json',
trial='matchengine/tests/data/trials/two_trials_one_doc.json',
upsert_fields='')
load(args)
assert len(list(self.db_ro.trial.find({}))) == 2
def test__load_trials_json_array(self):
"""mongoexport also allows exporting of trials as an array of json objects."""
self._reset(do_reset_trials=True)
args = Namespace(
clinical=None,
genomic=None,
db_name='integration_load',
plugin_dir='plugins',
trial_format='json',
trial='matchengine/tests/data/trials/trials_json_array.json',
upsert_fields='')
load(args)
assert len(list(self.db_ro.trial.find({}))) == 2
def test__load_clinical_json_dir(self):
self._reset(do_reset_patient=True)
args = Namespace(genomic=None,
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='json',
clinical='matchengine/tests/data/clinical_json/',
upsert_fields='')
load(args)
assert len(list(self.db_ro.clinical.find({}))) == 2
assert isinstance(
list(self.db_ro.clinical.find({}))[0]['BIRTH_DATE'], datetime.date)
def test__load_clinical_json_dir(self):
self._reset(do_reset_patient=True)
args = Namespace(genomic=None,
trial=None,
db_name='integration_load',
plugin_dir='plugins',
patient_format='json',
clinical='matchengine/tests/data/clinical_json/',
upsert_fields='')
load(args)
assert len(list(self.db_ro.clinical.find({}))) == 2
date_class = list(self.db_ro.clinical.find(
{}))[0]['BIRTH_DATE'].__class__
assert date_class is _scope_handler[
'datetime'] or date_class is _scope_handler['old_datetime']
