def get_scigraph_nodes(id_list) -> Iterator[Dict]:
"""
Queries scigraph neighbors to get a list of nodes back
We use the scigraph neighbors function because ids can be sent in batch
which is faster than iteratively querying solr search
or the scigraph graph/id function
:return: json decoded result from scigraph_ontology._neighbors_graph
:raises ValueError: If id is not in scigraph
"""
scigraph = RemoteScigraphOntology('scigraph:data')
chunks = [id_list[i:i + 100] for i in range(0, len(list(id_list)), 100)]
for chunk in chunks:
params = {'id': chunk, 'depth': 0}
try:
result_graph = scigraph._neighbors_graph(**params)
for node in result_graph['nodes']:
yield node
except JSONDecodeError as exception:
# Assume json decode is due to an incorrect class ID
# Should we handle this?
raise ValueError(exception.doc)
def create_ontology(handle=None, **args):
ont = None
logging.info("Determining strategy to load '{}' into memory...".format(handle))
if handle.find("+") > -1:
handles = handle.split("+")
onts = [create_ontology(ont) for ont in handles]
ont = onts.pop()
ont.merge(onts)
return ont
# TODO: consider replacing with plugin architecture
if handle.find(".") > 0 and os.path.isfile(handle):
logging.info("Fetching obograph-json file from filesystem")
ont = translate_file_to_ontology(handle, **args)
elif handle.startswith("obo:"):
logging.info("Fetching from OBO PURL")
if handle.find(".") == -1:
handle += '.owl'
fn = '/tmp/'+handle
if not os.path.isfile(fn):
url = handle.replace("obo:","http://purl.obolibrary.org/obo/")
cmd = ['owltools',url,'-o','-f','json',fn]
cp = subprocess.run(cmd, check=True)
logging.info(cp)
else:
logging.info("using cached file: "+fn)
g = obograph_util.convert_json_file(fn)
ont = Ontology(handle=handle, payload=g)
elif handle.startswith("wdq:"):
from ontobio.sparql.wikidata_ontology import EagerWikidataOntology
logging.info("Fetching from Wikidata")
ont = EagerWikidataOntology(handle=handle)
elif handle.startswith("scigraph:"):
from ontobio.neo.scigraph_ontology import RemoteScigraphOntology
logging.info("Fetching from SciGraph")
ont = RemoteScigraphOntology(handle=handle)
elif handle.startswith("http:"):
logging.info("Fetching from Web PURL: "+handle)
encoded = hashlib.sha256(handle.encode()).hexdigest()
#encoded = binascii.hexlify(bytes(handle, 'utf-8'))
#base64.b64encode(bytes(handle, 'utf-8'))
logging.info(" encoded: "+str(encoded))
fn = '/tmp/'+encoded
if not os.path.isfile(fn):
cmd = ['owltools',handle,'-o','-f','json',fn]
cp = subprocess.run(cmd, check=True)
logging.info(cp)
else:
logging.info("using cached file: "+fn)
g = obograph_util.convert_json_file(fn)
ont = Ontology(handle=handle, payload=g)
else:
logging.info("Fetching from SPARQL")
ont = EagerRemoteSparqlOntology(handle=handle)
#g = get_digraph(handle, None, True)
return ont