Protect null embeddings - v2.0 (#627)

* Don't emit graph embeddings if there aren't any.

* Don't store graph embeddings in a knowledge store if there's an empty list.

* Translate between Cassandra's 'null' representing an empty list and an
  empty list which is what the surrounding code wants (and stored in the
  first place).

* Avoid emitting empty embedding lists

* Avoid output empty triple lists

* Fix tests

tests/integration/test_kg_extract_store_integration.py

@@ -16,7 +16,7 @@ from trustgraph.extract.kg.definitions.extract import Processor as DefinitionsPr
 from trustgraph.extract.kg.relationships.extract import Processor as RelationshipsProcessor
 from trustgraph.storage.knowledge.store import Processor as KnowledgeStoreProcessor
 from trustgraph.schema import Chunk, Triple, Triples, Metadata, Term, Error, IRI, LITERAL
-from trustgraph.schema import EntityContext, EntityContexts, GraphEmbeddings
+from trustgraph.schema import EntityContext, EntityContexts, GraphEmbeddings, EntityEmbeddings
 from trustgraph.rdf import TRUSTGRAPH_ENTITIES, DEFINITION, RDF_LABEL, SUBJECT_OF
 
 
@@ -405,7 +405,12 @@ class TestKnowledgeGraphPipelineIntegration:
                 collection="test_collection",
                 metadata=[]
             ),
-            entities=[]
+            entities=[
+                EntityEmbeddings(
+                    entity=Term(type=IRI, iri="http://example.org/entity"),
+                    vectors=[[0.1, 0.2, 0.3]]
+                )
+            ]
         )
 
         mock_msg = MagicMock()
@@ -496,12 +501,12 @@ class TestKnowledgeGraphPipelineIntegration:
         await definitions_processor.on_message(mock_msg, mock_consumer, mock_flow_context)
 
         # Assert
-        # Should still call producers but with empty results
+        # Should NOT call producers with empty results (avoids Cassandra NULL issues)
         triples_producer = mock_flow_context("triples")
         entity_contexts_producer = mock_flow_context("entity-contexts")
 
-        triples_producer.send.assert_called_once()
+        triples_producer.send.assert_not_called()
-        entity_contexts_producer.send.assert_called_once()
+        entity_contexts_producer.send.assert_not_called()
 
     @pytest.mark.asyncio
     async def test_invalid_extraction_format_handling(self, definitions_processor, mock_flow_context, sample_chunk):

trustgraph-flow/trustgraph/embeddings/graph_embeddings/embeddings.py

@@ -73,6 +73,7 @@ class Processor(FlowProcessor):
                     )
                 )
 
+            if entities:
                 r = GraphEmbeddings(
                     metadata=v.metadata,
                     entities=entities,

trustgraph-flow/trustgraph/extract/kg/definitions/extract.py

@@ -168,6 +168,7 @@ class Processor(FlowProcessor):
 
                 entities.append(ec)
 
+            if triples:
                 await self.emit_triples(
                     flow("triples"),
                     Metadata(
@@ -179,6 +180,7 @@ class Processor(FlowProcessor):
                     triples
                 )
 
+            if entities:
                 await self.emit_ecs(
                     flow("entity-contexts"),
                     Metadata(

trustgraph-flow/trustgraph/extract/kg/ontology/extract.py

@@ -282,17 +282,6 @@ class Processor(FlowProcessor):
 
             if not ontology_subsets:
                 logger.warning("No relevant ontology elements found for chunk")
-                # Emit empty outputs
-                await self.emit_triples(
-                    flow("triples"),
-                    v.metadata,
-                    []
-                )
-                await self.emit_entity_contexts(
-                    flow("entity-contexts"),
-                    v.metadata,
-                    []
-                )
                 return
 
             # Merge subsets if multiple ontologies matched
@@ -327,6 +316,7 @@ class Processor(FlowProcessor):
             entity_contexts = self.build_entity_contexts(all_triples)
 
             # Emit all triples (extracted + ontology definitions)
+            if all_triples:
                 await self.emit_triples(
                     flow("triples"),
                     v.metadata,
@@ -334,6 +324,7 @@ class Processor(FlowProcessor):
                 )
 
             # Emit entity contexts
+            if entity_contexts:
                 await self.emit_entity_contexts(
                     flow("entity-contexts"),
                     v.metadata,
@@ -345,17 +336,6 @@ class Processor(FlowProcessor):
 
         except Exception as e:
             logger.error(f"OntoRAG extraction exception: {e}", exc_info=True)
-            # Emit empty outputs on error
-            await self.emit_triples(
-                flow("triples"),
-                v.metadata,
-                []
-            )
-            await self.emit_entity_contexts(
-                flow("entity-contexts"),
-                v.metadata,
-                []
-            )
 
     async def extract_with_simplified_format(
         self,

trustgraph-flow/trustgraph/extract/kg/relationships/extract.py

@@ -181,6 +181,7 @@ class Processor(FlowProcessor):
                         o=Term(type=IRI, iri=v.metadata.id)
                     ))
 
+            if triples:
                 await self.emit_triples(
                     flow("triples"),
                     Metadata(

trustgraph-flow/trustgraph/storage/knowledge/store.py

@@ -64,11 +64,13 @@ class Processor(FlowProcessor):
     async def on_triples(self, msg, consumer, flow):
 
         v = msg.value()
+        if v.triples:
             await self.table_store.add_triples(v)
 
     async def on_graph_embeddings(self, msg, consumer, flow):
 
         v = msg.value()
+        if v.entities:
             await self.table_store.add_graph_embeddings(v)
 
     @staticmethod

trustgraph-flow/trustgraph/tables/knowledge.py

@@ -435,6 +435,7 @@ class KnowledgeTableStore:
             else:
                 metadata = []
 
+            if row[3]:
                 triples = [
                     Triple(
                         s = tuple_to_term(elt[0], elt[1]),
@@ -443,6 +444,8 @@ class KnowledgeTableStore:
                     )
                     for elt in row[3]
                 ]
+            else:
+                triples = []
 
             await receiver(
                 Triples(
@@ -491,6 +494,7 @@ class KnowledgeTableStore:
             else:
                 metadata = []
 
+            if row[3]:
                 entities = [
                     EntityEmbeddings(
                         entity = tuple_to_term(ent[0][0], ent[0][1]),
@@ -498,6 +502,8 @@ class KnowledgeTableStore:
                     )
                     for ent in row[3]
                 ]
+            else:
+                entities = []
 
             await receiver(
                 GraphEmbeddings(