[Blocked by #127] GH-125: Update assumptions and correct type handling.

src/ipyradiant/rdf2nx/converter.py

@@ -125,7 +125,8 @@ def transform_nodes(
 
         if node_iris is None:
             # TODO paginate (LIMIT+OFFSET) for batch processing?
-            node_iris = list(cls.query_manager(cls.node_iris, rdf_graph)["iri"])
+            # node_iris = list(cls.query_manager(cls.node_iris, rdf_graph)["iri"])
+            node_iris = list(rdf_graph.subjects())
 
         for node_iri in node_iris:
             # TODO this isn't actually used (should it be?)

src/ipyradiant/rdf2nx/edges.py

@@ -25,7 +25,7 @@ class RelationTypes(SPARQLQueryFramer):
           # Plain relations, non-reified
             ?source ?predicate ?target.
 
-            FILTER ( isIRI ( ?target ) ).  # prevent bad things
+            FILTER ( !isLiteral( ?target ) ).  # prevent bad things
 
             # Plain relations must get their fictitious IRI from constructs like this.
             # Using the triple components ensures a unique IRI identifying the triple.

src/ipyradiant/visualization/cytoscape/interactive.py

@@ -8,6 +8,7 @@
 from pandas import DataFrame
 from rdflib.graph import Graph as RDFGraph
 from rdflib.term import URIRef
+from rdflib.namespace import OWL
 
 from ipyradiant.basic_tools.custom_uri_ref import CustomURIRef
 from ipyradiant.rdf2nx.converter import RDF2NX
@@ -56,6 +57,9 @@ def get_color_list_css(color_list):
 
 def get_desc(uri, namespaces, count=None):
     """Get a shorthand way to describe a URI and its counts."""
+    if type(uri) is not URIRef:
+        # just return the string
+        return str(uri)
 
     shorthand = str(CustomURIRef(uri, namespaces=namespaces))
     if count:
@@ -67,9 +71,11 @@ def get_type_counts(klass, graph: NXGraph) -> DataFrame:
     type_dict = {}
     for node, data in graph.nodes(data=True):
         # node_type can be a list (make all list)
-        type_attr = data.get("rdf:type")
+        type_attr = data.get("rdf:type") or data.get("rdfs:subClassOf")
         if not type_attr:
-            raise ValueError(f"Node has no 'rdf:type': {data.keys()}")
+            # TODO logging
+            # raise ValueError(f"Node has no 'rdf:type': {data.keys()}")
+            type_attr = "Blank Node"
 
         if not isinstance(type_attr, (list, tuple)):
             node_types = [
@@ -189,7 +195,7 @@ def assign_css_classes(self):
         color_list = COLOR_LIST.copy()
         n_to_add = len(self.uri_to_string_type.keys()) - len(color_list)
         if n_to_add > 0 and self.allow_large_graphs:
-            color_list.extend([(255, 255, 255)] * n_to_add)
+            color_list.extend([(220, 220, 220)] * n_to_add)
         elif n_to_add > 0:
             raise ValueError(
                 f"Cannot render more than {len(COLOR_LIST)} visually distinct colors."
@@ -244,7 +250,8 @@ def update_classes(self, change):
 
         if change_type in {"node_type", "both"}:
             for node in self.viewer.cytoscape_widget.graph.nodes:
-                raw_types = node.data["rdf:type"]
+                # TODO improve coloring after this
+                raw_types = node.data.get("rdf:type") or node.data.get("rdfs:subClassOf")
                 types = raw_types if type(raw_types) is tuple else (raw_types,)
                 if not any([_type in visible_node_types for _type in types]):
                     node.classes = "invisible"
@@ -279,10 +286,19 @@ def apply_node_styling(self, change):
         # assign CSS classes to nodes based on their rdf:type
         # TODO add types instead of replacing once we figure out how to make partial matches of css classes in ipycytoscape
         for node in self.viewer.cytoscape_widget.graph.nodes:
-            node_types = node.data.get("rdf:type", [])
+            node_types = (
+                node.data.get("rdf:type") or 
+                node.data.get("rdfs:subClassOf") or 
+                []
+            )
             if type(node_types) == URIRef:
                 node_types = (node_types,)
 
+            # Remove instance type for coloring
+            node_types_pruned = set(node_types)
+            node_types_pruned.discard(OWL.NamedIndividual)
+            node_types = tuple(node_types_pruned,)
+
             if len(node_types) == 1:
                 # assign specific class to node
                 assert node_types[0] in self.uri_to_string_type
@@ -386,7 +402,8 @@ def update_rdf_graph(self, change):
             uri: str(CustomURIRef(uri, namespaces=rdf_graph.namespace_manager)).replace(
                 ":", "-"
             )
-            for uri in type_count.type_
+            if type(uri) is URIRef else str(uri)
+            for uri in type_count.type_ 
         }
         self.uri_to_string_type["multi-type"] = "multi-type"