Merge pull request #317 from fatestigma/development

Remove warning for no ontology input
usc-isi-i2 · Jul 13, 2018 · e74b055 · e74b055
2 parents 2babf32 + 10f3709
commit e74b055
Show file tree

Hide file tree

Showing 3 changed files with 16 additions and 9 deletions.
diff --git a/etk/document.py b/etk/document.py
@@ -44,8 +44,6 @@ def __init__(self, etk, cdr_document: Dict, mime_type, url, doc_id=None) -> None
             self.kg = None
             if not self.etk.kg_schema:
                 self.etk.log("Schema not found.", "warning", self.doc_id, self.url)
-        if not self.etk.ontology:
-            self.etk.log("Ontology not found.", "warning", self.doc_id, self.url)
         self._provenance_id_index = 0
         self._provenances = dict()
         self._jsonpath_provenances = dict()

diff --git a/etk/ontology_api.py b/etk/ontology_api.py
@@ -687,7 +687,7 @@ def __is_valid_uri_resolve(self, field_name: str, context: Optional[dict]) -> UR
 
 def rdf_generation(kg_object) -> str:
     """
-    Covert input knowledge graph object into n-triples RDF
+    Convert input knowledge graph object into n-triples RDF
 
     :param kg_object: str, dict, or json object
     :return: n-triples RDF in str

diff --git a/examples/ontology_api/create_kg_with_ontology.py b/examples/ontology_api/create_kg_with_ontology.py
@@ -4,7 +4,7 @@
 from etk.etk import ETK
 from etk.knowledge_graph import KGSchema
 from etk.etk_module import ETKModule
-from etk.ontology_api import Ontology
+from etk.ontology_api import Ontology, rdf_generation
 from rdflib.namespace import RDF
 
 
@@ -13,23 +13,29 @@ def __init__(self, etk):
         ETKModule.__init__(self, etk)
 
     def process_document(self, doc):
-        return [self.__create_document(data) for data in doc.cdr_document.get('data', [])]
+        docs = list()
+        for data in doc.cdr_document.get('data', []):
+            docs.extend(self.__create_document(data))
+        return docs
 
     def __create_document(self, data):
         type_ = data.get(RDF.type.toPython(), list())
         type_.extend(data.get('@type', list()))
         doc = self.etk.create_document(dict(), doc_id=data['@id'], type_=type_)
+        docs = [doc]
         for key in data:
             if key in {'@id', '@type', RDF.type.toPython()}:
                 continue
             field_name = doc.kg.context_resolve(key)
             for value in data[key]:
                 if isinstance(value, dict):
+                    # at most nest 1 level, no need to worry what it returns
                     subdoc = self.__create_document(value)
-                    doc.kg.add_value(field_name, subdoc.kg._kg)
+                    docs.append(subdoc[0])
+                    doc.kg.add_value(field_name, subdoc[0].doc_id)
                 else:
                     doc.kg.add_value(field_name, value)
-        return doc
+        return docs
 
 
 if __name__ == "__main__":
@@ -114,6 +120,9 @@ def __create_document(self, data):
     input_data = {'doc_id': '1', 'data': json.loads(sample_input)}
     doc = etk.create_document(input_data)
     docs = etk.process_ems(doc)
-    for doc in docs[1:]:
-        print(json.dumps(doc.kg.value, indent=2))
+    kgs = [json.dumps(doc.kg.value) for doc in docs[1:]]
+    with open('output.jsonl', 'w') as f:
+        f.write('\n'.join(kgs))
+    with open('output.nt', 'w') as f:
+        f.writelines(map(rdf_generation, kgs))