def parse_string(self, content):
class Sink(object):
def triple(sink, s, p, o):
self.triples.add((s, p, o))
p = ntriples.NTriplesParser(sink=Sink())
p.parsestring(content)
def parse(self, uri):
class Sink(object):
def triple(sink, s, p, o):
self.triples.add((s, p, o))
p = ntriples.NTriplesParser(sink=Sink())
u = urllib.urlopen(uri)
p.parse(u)
u.close()
def _import_data(src, graph_uri):
"""Imports RDF/NTriples from a single source, either local file or via URL."""
nt_parser = ntriples.NTriplesParser(
sink=HBaseSink(server_port=HBASE_THRIFT_PORT, graph_uri=graph_uri))
sink = nt_parser.parse(src)
sink.wrapup(
) # needed as the number of triples can be smaller than batch size (!)
src.close()
return sink.length
def import_dataset(self, uri): parser = ntriples.NTriplesParser(sink=GKSink()) u = urllib.urlopen(uri) s = parser.parse(u) u.close() for ng in s.ng: print ng print s.ng[ng] self.put_ng(ng, s.ng[ng])
def ntriples_to_dict(file):
class TripleDict(dict):
def triple(self, s, p, o):
if isinstance(s, ntriples.URI):
subject = unicode(s)
elif isinstance(s, ntriples.bNode):
subject = u'_:%s' % s
else:
raise ValueError('Unknown subject type: %s' % type(s))
predicate = unicode(p)
predicates = self.get(subject, {})
values = predicates.get(predicate, [])
value = {'value' : unicode(o)}
if isinstance(o, ntriples.URI):
value['type'] = u'uri'
elif isinstance(o, ntriples.bNode):
value['type'] = u'bnode'
elif isinstance(o, ntriples.Literal):
lang, dtype, literal = o.split(' ', 2)
value['type'] = u'literal'
value['value'] = literal
if lang != 'None':
value['lang'] = lang
elif dtype != 'None':
value['datatype'] = dtype
else:
raise ValueError('Unknown object type: %s' % type(o))
values.append(value)
self[subject] = predicates
predicates[predicate] = values
parser = ntriples.NTriplesParser(TripleDict())
result = parser.parse(file)
return dict(result)