Merge pull request #207 from IFB-ElixirFr/bug-json-ld

fixed JSON-LD parsing
IFB-ElixirFr · Oct 31, 2023 · 7e857bf · 7e857bf
2 parents 5cbb2dd + 7a499cf
commit 7e857bf
Show file tree

Hide file tree

Showing 2 changed files with 10 additions and 8 deletions.
diff --git a/metrics/WebResource.py b/metrics/WebResource.py
@@ -488,36 +488,31 @@ def html_to_rdf_extruct(self, html_source) -> ConjunctiveGraph:
         )
 
         kg_jsonld = ConjunctiveGraph()
-
         if "json-ld" in data.keys():
             for md in data["json-ld"]:
-                kg_jsonld.parse(
+                kg_jsonld += kg_jsonld.parse(
                     data=json.dumps(md, ensure_ascii=False),
                     format="json-ld",
                     publicID=self.url,
                 )
 
         kg_rdfa = ConjunctiveGraph()
-
         if "rdfa" in data.keys():
             for md in data["rdfa"]:
-                kg_rdfa.parse(
+                kg_rdfa += kg_rdfa.parse(
                     data=json.dumps(md, ensure_ascii=False),
                     format="json-ld",
                     publicID=self.url,
                 )
-                # print(len(kg_rdfa))
 
         kg_microdata = ConjunctiveGraph()
-
         if "microdata" in data.keys():
             for md in data["microdata"]:
-                kg_microdata.parse(
+                kg_microdata += kg_microdata.parse(
                     data=json.dumps(md, ensure_ascii=False),
                     format="json-ld",
                     publicID=self.url,
                 )
-                # print(len(kg_microdata))
 
         kg_extruct = kg_jsonld + kg_rdfa + kg_microdata
 

diff --git a/tests/test_web_resource.py b/tests/test_web_resource.py
@@ -138,6 +138,13 @@ def test_turtle(self):
         logging.info(f"{len(turtle_WR.get_rdf())} loaded RDF triples")
         # self.assertEqual(60, len(turtle_WR.get_rdf()))
 
+    def test_MassBank(self):
+        mb = WebResource("https://massbank.eu/MassBank/RecordDisplay?id=MSBNK-RIKEN_IMS-LQB00001")
+        kg = mb.get_rdf()
+        #print(kg.serialize(format="turtle"))
+        #logging.info(f"{len(kg)} loaded RDF triples")
+        self.assertGreater(len(kg), 70)
+
     def test_n3(self):
         n3_WR = WebResource("https://www.w3.org/2002/11/rddl/ex1.n3")
         logging.info(f"{len(n3_WR.get_rdf())} loaded RDF triples")