Populate name fields in nodes (#499)

Author: vish-cs

pipeline/util/src/main/java/org/datacommons/ingestion/util/GraphReader.java

@@ -43,7 +43,7 @@ public static List<Node> graphToNodes(McfGraph graph, Counter mcfNodesWithoutTyp
         // Generate corresponding node
         Map<String, McfGraph.Values> pv = pvs.getPvsMap();
         Node.Builder node = Node.builder();
-        String dcid = GraphUtils.getPropertyValue(pv, "dcid");
+        String dcid = GraphUtils.getPropVal(pvs, GraphUtils.Property.dcid.name());
         String subjectId = !dcid.isEmpty() ? dcid : McfUtil.stripNamespace(nodeEntry.getKey());
         node.subjectId(subjectId);
 
@@ -54,7 +54,8 @@ public static List<Node> graphToNodes(McfGraph graph, Counter mcfNodesWithoutTyp
           mcfNodesWithoutTypeCounter.inc();
         }
         node.value(subjectId);
-        node.name(GraphUtils.getPropertyValue(pv, "name"));
+        String name = GraphUtils.getPropVal(pvs, GraphUtils.Property.name.name());
+        node.name(!name.isEmpty() ? name : subjectId);
         node.types(types);
         nodes.add(node.build());
 
@@ -92,7 +93,7 @@ public static List<Edge> graphToEdges(McfGraph graph, String provenance) {
       PropertyValues pvs = nodeEntry.getValue();
       if (!GraphUtils.isObservation(pvs)) {
         Map<String, McfGraph.Values> pv = pvs.getPvsMap();
-        String dcid = GraphUtils.getPropertyValue(pv, "dcid");
+        String dcid = GraphUtils.getPropVal(pvs, GraphUtils.Property.dcid.name());
         String subjectId = !dcid.isEmpty() ? dcid : McfUtil.stripNamespace(nodeEntry.getKey());
         for (Map.Entry<String, McfGraph.Values> entry : pv.entrySet()) {
           if (entry.getKey().equals("dcid")) {

pipeline/util/src/test/java/org/datacommons/ingestion/util/GraphReaderTest.java

@@ -117,7 +117,12 @@ public void testGraphToNodes() {
                 .name("Node One")
                 .types(List.of("Property"))
                 .build(),
-            Node.builder().subjectId("dcid2").value("dcid2").types(List.of("Thing")).build(),
+            Node.builder()
+                .subjectId("dcid2")
+                .value("dcid2")
+                .name("dcid2")
+                .types(List.of("Thing"))
+                .build(),
             Node.builder()
                 .subjectId("Node Zero:kUyRupzrJkxe/HIOIctxlJX4woEGeOTtlVwqyXYnfDE=")
                 .value("Node Zero")

util/src/main/java/org/datacommons/util/GraphUtils.java

@@ -32,6 +32,7 @@ public enum Property {
     measurementMethod,
     unit,
     value,
+    name,
     scalingFactor,
     dcid;
   }
@@ -98,8 +99,7 @@ public static boolean useDcidForLocalNodeIdInOptimizedMcf(String dcid, String lo
    * @return True if the property values indicate a StatVarObservation, false otherwise.
    */
   public static boolean isObservation(PropertyValues pvs) {
-    return GraphUtils.getPropertyValue(pvs.getPvsMap(), GraphUtils.Property.typeOf.name())
-        .contains(GraphUtils.STAT_VAR_OB);
+    return getPropVal(pvs, GraphUtils.Property.typeOf.name()).contains(GraphUtils.STAT_VAR_OB);
   }
 
   /**
@@ -129,11 +129,7 @@ public static String getPropVal(McfGraph.PropertyValues node, String prop) {
     McfGraph.Values v = node.getPvsMap().get(prop);
     if (v != null && v.getTypedValuesCount() > 0) {
       String val = v.getTypedValues(0).getValue();
-      if (val.startsWith("dcid:") || val.startsWith("dcs:") || val.startsWith("schema:")) {
-        return val.substring(val.indexOf(':') + 1);
-      } else {
-        return val;
-      }
+      return McfUtil.stripNamespace(val);
     }
     return "";
   }