def upload_artifacts_for_prefix(*, prefix: str, bucket: str, s3_client=None):
"""Upload compiled parts for the given prefix to AWS."""
if s3_client is None:
s3_client = boto3.client("s3")
logger.info("[%s] getting id->name mapping", prefix)
get_id_name_mapping(prefix)
id_name_path = prefix_cache_join(prefix, name="names.tsv", version=get_version(prefix))
if not id_name_path.exists():
raise FileNotFoundError
id_name_key = os.path.join(prefix, "cache", "names.tsv")
logger.info("[%s] uploading id->name mapping", prefix)
upload_file(path=id_name_path, bucket=bucket, key=id_name_key, s3_client=s3_client)
logger.info("[%s] getting id->synonyms mapping", prefix)
get_id_synonyms_mapping(prefix)
id_synonyms_path = prefix_cache_join(prefix, name="synonyms.tsv", version=get_version(prefix))
if not id_synonyms_path.exists():
raise FileNotFoundError
id_synonyms_key = os.path.join(prefix, "cache", "synonyms.tsv")
logger.info("[%s] uploading id->synonyms mapping", prefix)
upload_file(path=id_synonyms_path, bucket=bucket, key=id_synonyms_key, s3_client=s3_client)
logger.info("[%s] getting xrefs", prefix)
get_xrefs_df(prefix)
xrefs_path = prefix_cache_join(prefix, name="xrefs.tsv", version=get_version(prefix))
if not xrefs_path.exists():
raise FileNotFoundError
xrefs_key = os.path.join(prefix, "cache", "xrefs.tsv")
logger.info("[%s] uploading xrefs", prefix)
upload_file(path=xrefs_path, bucket=bucket, key=xrefs_key, s3_client=s3_client)
logger.info("[%s] getting relations", prefix)
get_relations_df(prefix)
relations_path = prefix_cache_join(prefix, name="relations.tsv", version=get_version(prefix))
if not relations_path.exists():
raise FileNotFoundError
relations_key = os.path.join(prefix, "cache", "relations.tsv")
logger.info("[%s] uploading relations", prefix)
upload_file(path=relations_path, bucket=bucket, key=relations_key, s3_client=s3_client)
logger.info("[%s] getting properties", prefix)
get_properties_df(prefix)
properties_path = prefix_cache_join(prefix, name="properties.tsv", version=get_version(prefix))
if not properties_path.exists():
raise FileNotFoundError
properties_key = os.path.join(prefix, "cache", "properties.tsv")
logger.info("[%s] uploading properties", prefix)
upload_file(path=properties_path, bucket=bucket, key=properties_key, s3_client=s3_client)
logger.info("[%s] getting alternative identifiers", prefix)
get_id_to_alts(prefix)
alts_path = prefix_cache_join(prefix, name="alt_ids.tsv", version=get_version(prefix))
if not alts_path.exists():
raise FileNotFoundError
alts_key = os.path.join(prefix, "cache", "alt_ids.tsv")
logger.info("[%s] uploading alternative identifiers", prefix)
upload_file(path=alts_path, bucket=bucket, key=alts_key)
def test_get_xrefs(self):
"""Test getting xrefs."""
with chebi_patch:
df = get_xrefs_df("chebi")
self.assertIsInstance(df, pd.DataFrame)
self.assertEqual(["chebi_id", TARGET_PREFIX, TARGET_ID],
list(df.columns))
def test_get_xrefs(self):
"""Test getting xrefs."""
df = get_xrefs_df('chebi', url=TEST_CHEBI_OBO_PATH, local=True)
self.assertIsInstance(df, pd.DataFrame)
for key, value in df[['source_ns', 'source_id']].values: # no need for targets since are external
self.assertFalse(value.startswith(key))
self.assertFalse(value.lower().startswith(key.lower()), msg=f'Bad value: {value}')
self.assertFalse(value.startswith(f'{key}:'))
self.assertFalse(value.lower().startswith(f'{key.lower()}:'))