feat: Enable ingestion of precomputed embeddings (#2317)

tests/integration/vector_io/test_vector_io.py

@@ -120,3 +120,37 @@ def test_insert_chunks(client_with_empty_registry, embedding_model_id, sample_ch
     top_match = response.chunks[0]
     assert top_match is not None
     assert top_match.metadata["document_id"] == expected_doc_id, f"Query '{query}' should match {expected_doc_id}"
+
+
+def test_insert_chunks_with_precomputed_embeddings(client_with_empty_registry, embedding_model_id):
+    vector_db_id = "test_precomputed_embeddings_db"
+    client_with_empty_registry.vector_dbs.register(
+        vector_db_id=vector_db_id,
+        embedding_model=embedding_model_id,
+        embedding_dimension=384,
+    )
+
+    chunks_with_embeddings = [
+        Chunk(
+            content="This is a test chunk with precomputed embedding.",
+            metadata={"document_id": "doc1", "source": "precomputed"},
+            embedding=[0.1] * 384,
+        ),
+    ]
+
+    client_with_empty_registry.vector_io.insert(
+        vector_db_id=vector_db_id,
+        chunks=chunks_with_embeddings,
+    )
+
+    # Query for the first document
+    response = client_with_empty_registry.vector_io.query(
+        vector_db_id=vector_db_id,
+        query="precomputed embedding test",
+    )
+
+    # Verify the top result is the expected document
+    assert response is not None
+    assert len(response.chunks) > 0
+    assert response.chunks[0].metadata["document_id"] == "doc1"
+    assert response.chunks[0].metadata["source"] == "precomputed"