def query(yaml: str,
output_dir: str,
query_key: str = 'query',
endpoint_key: str = 'endpoint',
outfile_ext: str = ".tsv") -> None:
"""Perform a query of knowledge graph using a class contained in query_utils
Args:
yaml: A YAML file containing a SPARQL query (see queries/sparql/ for examples)
output_dir: Directory to output results of query
query_key: the key in the yaml file containing the query string
endpoint_key: the key in the yaml file containing the sparql endpoint URL
outfile_ext: file extension for output file [.tsv]
Returns:
None.
"""
query = parse_query_yaml(yaml)
result_dict = run_query(query=query[query_key],
endpoint=query[endpoint_key])
if not os.path.exists(output_dir):
os.makedirs(output_dir)
outfile = os.path.join(
output_dir,
os.path.splitext(os.path.basename(yaml))[0] + outfile_ext)
result_dict_to_tsv(result_dict, outfile)
def test_result_dict_to_tsv_makes_correct_file(self):
result_dict = load_obj(self.test_result_dict_file)
result_dict_to_tsv(result_dict, self.tempfile)
df = pd.read_csv(self.tempfile, sep="\t")
self.assertEqual((18, 2), df.shape)
self.assertEqual(['v1', 'v0'], list(df.columns))
self.assertEqual([10384, 'human_phenotype'], list(df.iloc[1]))
def query(yaml: str, output_dir: str,
query_key: str='query', endpoint_key: str='endpoint',
outfile_ext: str=".tsv") -> None:
"""Perform a query of knowledge graph using a class contained in query_utils
Args:
yaml: A rq file containing a SPARQL query in grlc format:
https://github.com/CLARIAH/grlc/blob/master/README.md
output_dir: Directory to output results of query
query_key: the key in the yaml file containing the query string
endpoint_key: the key in the yaml file containing the sparql endpoint URL
outfile_ext: file extension for output file [.tsv]
Returns:
None.
"""
query = parse_query_rq(yaml)
result_dict = run_query(query=query[query_key], endpoint=query[endpoint_key])
if not os.path.exists(output_dir):
os.makedirs(output_dir)
outfile = os.path.join(output_dir, os.path.splitext(os.path.basename(yaml))[0] +
outfile_ext)
result_dict_to_tsv(result_dict, outfile)
def test_result_dict_to_tsv_makes_file(self):
result_dict = load_obj(self.test_result_dict_file)
result_dict_to_tsv(result_dict, self.tempfile)
self.assertTrue(os.path.isfile(self.tempfile))