Merge branch 'main' into vest_search

.github/workflows/test-external-providers.yml

@@ -45,20 +45,22 @@ jobs:
 
       - name: Build distro from config file
         run: |
-          USE_COPY_NOT_MOUNT=true LLAMA_STACK_DIR=. uv run llama stack build --config tests/external-provider/llama-stack-provider-ollama/custom-distro.yaml
+          USE_COPY_NOT_MOUNT=true LLAMA_STACK_DIR=. llama stack build --config tests/external-provider/llama-stack-provider-ollama/custom-distro.yaml
 
       - name: Start Llama Stack server in background
         if: ${{ matrix.image-type }} == 'venv'
         env:
           INFERENCE_MODEL: "meta-llama/Llama-3.2-3B-Instruct"
         run: |
-          uv run pip list
-          nohup uv run --active llama stack run tests/external-provider/llama-stack-provider-ollama/run.yaml --image-type ${{ matrix.image-type }} > server.log 2>&1 &
+          # Use the virtual environment created by the build step (name comes from build config)
+          source ci-test/bin/activate
+          uv pip list
+          nohup llama stack run tests/external-provider/llama-stack-provider-ollama/run.yaml --image-type ${{ matrix.image-type }} > server.log 2>&1 &
 
       - name: Wait for Llama Stack server to be ready
         run: |
           for i in {1..30}; do
-            if ! grep -q "remote::custom_ollama from /home/runner/.llama/providers.d/remote/inference/custom_ollama.yaml" server.log; then
+            if ! grep -q "Successfully loaded external provider remote::custom_ollama" server.log; then
               echo "Waiting for Llama Stack server to load the provider..."
               sleep 1
             else

llama_stack/distribution/distribution.py

@@ -180,6 +180,7 @@ def get_provider_registry(
                         if provider_type_key in ret[api]:
                             logger.warning(f"Overriding already registered provider {provider_type_key} for {api.name}")
                         ret[api][provider_type_key] = spec
+                        logger.info(f"Successfully loaded external provider {provider_type_key}")
                     except yaml.YAMLError as yaml_err:
                         logger.error(f"Failed to parse YAML file {spec_path}: {yaml_err}")
                         raise yaml_err

llama_stack/log.py

@@ -115,7 +115,7 @@ def parse_environment_config(env_config: str) -> dict[str, int]:
 
 class CustomRichHandler(RichHandler):
     def __init__(self, *args, **kwargs):
-        kwargs["console"] = Console(width=120)
+        kwargs["console"] = Console(width=150)
         super().__init__(*args, **kwargs)
 
     def emit(self, record):

llama_stack/providers/remote/vector_io/weaviate/weaviate.py

@@ -76,7 +76,7 @@ class WeaviateIndex(EmbeddingIndex):
                 continue
 
             chunks.append(chunk)
-            scores.append(1.0 / doc.metadata.distance)
+            scores.append(1.0 / doc.metadata.distance if doc.metadata.distance != 0 else float("inf"))
 
         return QueryChunksResponse(chunks=chunks, scores=scores)
 

tests/integration/vector_io/test_vector_io.py

@@ -154,3 +154,36 @@ def test_insert_chunks_with_precomputed_embeddings(client_with_empty_registry, e
     assert len(response.chunks) > 0
     assert response.chunks[0].metadata["document_id"] == "doc1"
     assert response.chunks[0].metadata["source"] == "precomputed"
+
+
+def test_query_returns_valid_object_when_identical_to_embedding_in_vdb(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="duplicate",
+            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,
+    )
+
+    response = client_with_empty_registry.vector_io.query(
+        vector_db_id=vector_db_id,
+        query="duplicate",
+    )
+
+    # 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"