def _upsert_samples_from_csv(csv_reader):
"""Inserts/updates BiobankStoredSamples from a csv.DictReader."""
missing_cols = set(CsvColumns.ALL) - set(csv_reader.fieldnames)
if missing_cols:
raise DataError(
'CSV is missing columns %s, had columns %s.' % (missing_cols, csv_reader.fieldnames))
samples_dao = BiobankStoredSampleDao()
biobank_id_prefix = get_biobank_id_prefix()
written = 0
try:
samples = []
with ParticipantDao().session() as session:
for row in csv_reader:
sample = _create_sample_from_row(row, biobank_id_prefix)
if sample:
# DA-601 - Ensure biobank_id exists before accepting a sample record.
if session.query(Participant).filter(
Participant.biobankId == sample.biobankId).count() < 1:
logging.error('Bio bank Id ({0}) does not exist in the Participant table.'.
format(sample.biobankId))
continue
samples.append(sample)
if len(samples) >= _BATCH_SIZE:
written += samples_dao.upsert_all(samples)
samples = []
if samples:
written += samples_dao.upsert_all(samples)
return written
except ValueError, e:
raise DataError(e)
def _upsert_samples_from_csv(csv_reader):
"""Inserts/updates BiobankStoredSamples from a csv.DictReader."""
missing_cols = _Columns.ALL - set(csv_reader.fieldnames)
if missing_cols:
raise DataError('CSV is missing columns %s, had columns %s.' %
(missing_cols, csv_reader.fieldnames))
samples_dao = BiobankStoredSampleDao()
biobank_id_prefix = get_biobank_id_prefix()
written = 0
try:
samples = []
for row in csv_reader:
sample = _create_sample_from_row(row, biobank_id_prefix)
if sample:
samples.append(sample)
if len(samples) >= _BATCH_SIZE:
written += samples_dao.upsert_all(samples)
samples = []
if samples:
written += samples_dao.upsert_all(samples)
return written
except ValueError, e:
raise DataError(e)