def test_check_project_and_sample(self):
sa = SampleAdaptor(**{'session_class': self.session_class})
sample_data = [
{
'sample_igf_id': 'IGFS001',
'library_id': 'IGFS001',
'project_igf_id': 'IGFP0001_test_22-8-2017_rna',
},
{
'sample_igf_id': 'IGFS002',
'library_id': 'IGFS002',
'project_igf_id': 'IGFP0001_test_22-8-2017_rna',
},
{
'sample_igf_id': 'IGFS003',
'library_id': 'IGFS003',
'project_igf_id': 'IGFP0001_test_22-8-2017_rna',
},
{
'sample_igf_id': 'IGFS004',
'library_id': 'IGFS004',
'project_igf_id': 'IGFP0001_test_22-8-2017_rna',
},
]
sa.start_session()
sa.store_sample_and_attribute_data(data=sample_data)
sa1=sa.check_project_and_sample(project_igf_id='IGFP0001_test_22-8-2017_rna',\
sample_igf_id='IGFS001')
self.assertEqual(sa1, True)
sa2=sa.check_project_and_sample(project_igf_id='IGFP0001_test_22-8-2017_rna',\
sample_igf_id='IGFS0011')
self.assertEqual(sa2, False)
sa.close_session()
def _check_existing_data(self,
data,
dbsession,
table_name,
check_column='EXISTS'):
'''
An internal function for checking and registering project info
:param data: A pandas data series
:param dbsession: A sqlalchemy database session object
:param table_name: A database table name
:param check_column: Column name for existing data
'''
try:
if not isinstance(data, pd.Series):
raise ValueError('Expecting a data series and got {0}'.format(
type(data)))
if table_name == 'project':
if self.project_lookup_column in data and \
not pd.isnull(data[self.project_lookup_column]):
project_igf_id = data[self.project_lookup_column]
pa = ProjectAdaptor(**{'session': dbsession
}) # connect to project adaptor
project_exists = pa.check_project_records_igf_id(
project_igf_id)
if project_exists: # store data only if project is not existing
data[check_column] = True
else:
data[check_column] = False
return data
else:
raise ValueError('Missing or empty required column {0}'.\
format(self.project_lookup_column))
elif table_name == 'user':
if self.user_lookup_column in data and \
not pd.isnull(data[self.user_lookup_column]):
user_email = data[self.user_lookup_column]
ua = UserAdaptor(**{'session':
dbsession}) # connect to user adaptor
user_exists = ua.check_user_records_email_id(
email_id=user_email)
if user_exists: # store data only if user is not existing
data[check_column] = True
else:
data[check_column] = False
return data
else:
raise ValueError('Missing or empty required column {0}'.\
format(self.user_lookup_column))
elif table_name == 'sample':
if self.sample_lookup_column in data and \
not pd.isnull(data[self.sample_lookup_column]):
project_igf_id = data[self.project_lookup_column]
sample_igf_id = data[self.sample_lookup_column]
sa = SampleAdaptor(**{'session': dbsession
}) # connect to sample adaptor
sample_project_exists=sa.check_project_and_sample(project_igf_id=project_igf_id,\
sample_igf_id=sample_igf_id) # check for existing sample_id and project-id combination
if sample_project_exists: # store data only if sample is not existing
data[check_column] = True
else:
sample_exists = sa.check_sample_records_igf_id(
sample_igf_id) # check for existing sample
if sample_exists:
raise ValueError('Sample {0} exists in database but not associated with project {1}'.\
format(sample_igf_id,project_igf_id)) # inconsistency in sample project combination
data[check_column] = False
return data
else:
raise ValueError('Missing or empty required column {0}'.\
format(self.sample_lookup_column))
elif table_name == 'project_user':
if self.user_lookup_column in data and \
not pd.isnull(data[self.user_lookup_column]) and \
self.project_lookup_column in data and \
not pd.isnull(data[self.project_lookup_column]):
project_igf_id = data[self.project_lookup_column]
user_email = data[self.user_lookup_column]
pa = ProjectAdaptor(**{'session': dbsession
}) # connect to project adaptor
project_user_exists=pa.check_existing_project_user(project_igf_id,\
email_id=user_email)
if user_email != self.default_user_email and \
(self.data_authority_column not in data or \
pd.isnull(data[self.data_authority_column])):
data[
self.
data_authority_column] = True # set user as data authority, filter default user
if project_user_exists: # store data only if sample is not existing
data[check_column] = True
else:
data[check_column] = False
return data
else:
raise ValueError('Missing or empty required column {0}, {1}'.\
format(self.project_lookup_column,\
self.user_lookup_column))
else:
raise ValueError('table {0} not supported'.format(table_name))
except:
raise