mirror of
https://github.com/trustgraph-ai/trustgraph.git
synced 2026-04-25 08:26:21 +02:00
38 lines
713 B
Text
38 lines
713 B
Text
|
#!/usr/bin/env python3
|
||
|
|
|
||
|
|
from trustgraph.trustgraph import TrustGraph
|
||
|
|
import rdflib
|
||
|
|
import sys
|
||
|
|
import io
|
||
|
|
|
||
|
|
t = TrustGraph()
|
||
|
|
|
||
|
|
g = rdflib.Graph()
|
||
|
|
|
||
|
|
rows = t.get_all(limit=100_000_000)
|
||
|
|
for s, p, o in rows:
|
||
|
|
|
||
|
|
# print(s, p, o)
|
||
|
|
sv = rdflib.term.URIRef(s)
|
||
|
|
pv = rdflib.term.URIRef(p)
|
||
|
|
|
||
|
|
if o.startswith("https://") or o.startswith("http://"):
|
||
|
|
|
||
|
|
# Skip malformed URLs with spaces in
|
||
|
|
if " " in o:
|
||
|
|
continue
|
||
|
|
|
||
|
|
ov = rdflib.term.URIRef(o)
|
||
|
|
else:
|
||
|
|
ov = rdflib.term.Literal(o)
|
||
|
|
|
||
|
|
g.add((sv, pv, ov))
|
||
|
|
|
||
|
|
g.serialize(destination="output.ttl", format="turtle")
|
||
|
|
|
||
|
|
buf = io.BytesIO()
|
||
|
|
|
||
|
|
g.serialize(destination=buf, format="turtle")
|
||
|
|
|
||
|
|
sys.stdout.write(buf.getvalue().decode("utf-8"))
|