def test_generate_sample_id_with_file_path(self):
hash_code = hashing_util.generate_sample_id(
'Sample1', 'gs://bucket1/dir1/file1.vcf')
self.assertEqual(hash_code, 7715696391291253656)
hash_code = hashing_util.generate_sample_id(
'Sample2', 'gs://bucket1/dir1/file1.vcf')
self.assertEqual(hash_code, 5682150464643626236)
hash_code = hashing_util.generate_sample_id(
'Sample1', 'gs://bucket1/dir1/file2.vcf')
self.assertEqual(hash_code, 668336000922978678)
hash_code = hashing_util.generate_sample_id(
'Sample2', 'gs://bucket1/dir1/file2.vcf')
self.assertEqual(hash_code, 5498327443813165683)
def _lookup_encoded_sample_name(self, sample_name):
sample_id = self._encoded_sample_names.get(sample_name)
if not sample_id:
if self._sample_name_encoding == SampleNameEncoding.WITH_FILE_PATH:
sample_id = hashing_util.generate_sample_id(sample_name,
self._file_name)
elif self._sample_name_encoding == SampleNameEncoding.WITHOUT_FILE_PATH:
sample_id = hashing_util.generate_sample_id(sample_name)
elif self._sample_name_encoding == SampleNameEncoding.NONE:
sample_id = sample_name
else:
raise ValueError('Unknown Sample Name Encoding supplied: {}'.format(
self._sample_name_encoding))
self._encoded_sample_names[sample_name] = sample_id
return sample_id
def process(self, vcf_header):
# type: (vcf_header_io.VcfHeader, bool) -> Dict[str, Union[int, str]]
current_minute = self._get_now_to_minute()
for sample in vcf_header.samples:
if self._sample_name_encoding == SampleNameEncoding.WITH_FILE_PATH:
sample = hashing_util.create_composite_sample_name(sample,
vcf_header.file_path)
sample_id = hashing_util.generate_sample_id(sample)
row = {
sample_info_table_schema_generator.SAMPLE_ID: sample_id,
sample_info_table_schema_generator.SAMPLE_NAME: sample,
sample_info_table_schema_generator.FILE_PATH: vcf_header.file_path,
sample_info_table_schema_generator.INGESTION_DATETIME: current_minute
}
yield row
def hash_name(sample_name, file_name=''):
return hashing_util.generate_sample_id(sample_name, file_name)
def test_generate_sample_id_without_file_path(self):
hash_code = hashing_util.generate_sample_id('Sample1')
self.assertEqual(hash_code, 6365297890523177914)
hash_code = hashing_util.generate_sample_id('Sample2')
self.assertEqual(hash_code, 8341768597576477893)