test(relationships): add test for standalone file yielding no extra chunks on expansion

Author: Ahmath-Gadji

openrag/components/indexer/vectordb/utils.py

@@ -578,6 +578,7 @@ def get_file_ancestors(self, partition: str, file_id: str, max_ancestor_depth: i
                     **(row.file_metadata or {}),
                 }
                 for row in result
+                if row.relationship_id is not None  # Only include files that are part of a relationship
             ]
 
     def get_ancestor_file_ids(self, partition: str, file_id: str, max_ancestor_depth: int | None = None) -> list[str]:

openrag/components/test_relationships.py

@@ -580,6 +580,50 @@ def test_get_ancestor_file_ids_with_max_ancestor_depth(self, file_manager):
         assert ancestor_ids == ["chain_2", "chain_3"]
 
 
+class TestStandaloneFileNoExpansion:
+    """Test that a file indexed without relationship_id yields no additional chunks
+    when include_related and include_ancestors are both active."""
+
+    def test_no_extra_chunks_for_file_without_relationship_id(self, file_manager):
+        """A standalone file (no relationship_id, no parent_id) must not bring
+        additional files when both include_related and include_ancestors are activated.
+
+        Mirrors the logic in _expand_with_related_chunks:
+        - include_related: the guard `metadata.get("relationship_id")` is falsy,
+          so no related lookup is issued and the related task set stays empty.
+        - include_ancestors: get_file_ancestors returns only the file itself when
+          there is no parent, so it is already in seen_ids — nothing new is added.
+        """
+        file_manager.add_file_to_partition(
+            partition="test_partition",
+            file_id="standalone",
+            file_metadata={"filename": "standalone.pdf"},
+            # No relationship_id, no parent_id
+        )
+
+        # Verify the file has no relationship_id (the falsy guard that prevents
+        # the include_related lookup from being issued at all).
+        files = file_manager.get_files_by_relationship(
+            partition="test_partition",
+            relationship_id="standalone",  # non-existent → empty
+        )
+        assert files == [], "No files should share a relationship with a standalone file"
+
+        with file_manager.Session() as session:
+            row = session.execute(text("SELECT relationship_id FROM files WHERE file_id = 'standalone'")).fetchone()
+            assert not row[0], "relationship_id must be falsy so include_related is skipped"
+
+        # include_ancestors path: returns only the file itself → nothing new to add
+        ancestors = file_manager.get_file_ancestors(
+            partition="test_partition",
+            file_id="standalone",
+        )
+        assert len(ancestors) == 1
+        assert ancestors[0]["file_id"] == "standalone", (
+            "Only the file itself should be returned — no ancestors to expand with"
+        )
+
+
 class TestFileModelFields:
     """Test that File model correctly handles relationship fields."""