feat(tests): make inference_recorder into api_recorder (include tool_invoke)

tests/integration/responses/helpers.py

@@ -7,7 +7,7 @@
 import time
 
 
-def new_vector_store(openai_client, name):
+def new_vector_store(openai_client, name, embedding_model, embedding_dimension):
     """Create a new vector store, cleaning up any existing one with the same name."""
     # Ensure we don't reuse an existing vector store
     vector_stores = openai_client.vector_stores.list()
@@ -16,7 +16,21 @@ def new_vector_store(openai_client, name):
             openai_client.vector_stores.delete(vector_store_id=vector_store.id)
 
     # Create a new vector store
-    vector_store = openai_client.vector_stores.create(name=name)
+    # OpenAI SDK client uses extra_body for non-standard parameters
+    from openai import OpenAI
+
+    if isinstance(openai_client, OpenAI):
+        # OpenAI SDK client - use extra_body
+        vector_store = openai_client.vector_stores.create(
+            name=name,
+            extra_body={"embedding_model": embedding_model, "embedding_dimension": embedding_dimension},
+        )
+    else:
+        # LlamaStack client - direct parameter
+        vector_store = openai_client.vector_stores.create(
+            name=name, embedding_model=embedding_model, embedding_dimension=embedding_dimension
+        )
+
     return vector_store
 
 

tests/integration/responses/test_extra_body_shields.py

@@ -16,6 +16,7 @@ import pytest
 from llama_stack_client import APIStatusError
 
 
+@pytest.mark.xfail(reason="Shields are not yet implemented inside responses")
 def test_shields_via_extra_body(compat_client, text_model_id):
     """Test that shields parameter is received by the server and raises NotImplementedError."""
 

tests/integration/responses/test_file_search.py

@@ -47,12 +47,14 @@ def test_response_text_format(compat_client, text_model_id, text_format):
 
 
 @pytest.fixture
-def vector_store_with_filtered_files(compat_client, text_model_id, tmp_path_factory):
-    """Create a vector store with multiple files that have different attributes for filtering tests."""
+def vector_store_with_filtered_files(compat_client, embedding_model_id, embedding_dimension, tmp_path_factory):
+    # """Create a vector store with multiple files that have different attributes for filtering tests."""
     if isinstance(compat_client, LlamaStackAsLibraryClient):
-        pytest.skip("Responses API file search is not yet supported in library client.")
+        pytest.skip("upload_file() is not yet supported in library client somehow?")
 
-    vector_store = new_vector_store(compat_client, "test_vector_store_with_filters")
+    vector_store = new_vector_store(
+        compat_client, "test_vector_store_with_filters", embedding_model_id, embedding_dimension
+    )
     tmp_path = tmp_path_factory.mktemp("filter_test_files")
 
     # Create multiple files with different attributes

tests/integration/responses/test_tool_responses.py

@@ -46,11 +46,13 @@ def test_response_non_streaming_web_search(compat_client, text_model_id, case):
 
 
 @pytest.mark.parametrize("case", file_search_test_cases)
-def test_response_non_streaming_file_search(compat_client, text_model_id, tmp_path, case):
+def test_response_non_streaming_file_search(
+    compat_client, text_model_id, embedding_model_id, embedding_dimension, tmp_path, case
+):
     if isinstance(compat_client, LlamaStackAsLibraryClient):
         pytest.skip("Responses API file search is not yet supported in library client.")
 
-    vector_store = new_vector_store(compat_client, "test_vector_store")
+    vector_store = new_vector_store(compat_client, "test_vector_store", embedding_model_id, embedding_dimension)
 
     if case.file_content:
         file_name = "test_response_non_streaming_file_search.txt"
@@ -101,11 +103,13 @@ def test_response_non_streaming_file_search(compat_client, text_model_id, tmp_pa
     assert case.expected.lower() in response.output_text.lower().strip()
 
 
-def test_response_non_streaming_file_search_empty_vector_store(compat_client, text_model_id):
+def test_response_non_streaming_file_search_empty_vector_store(
+    compat_client, text_model_id, embedding_model_id, embedding_dimension
+):
     if isinstance(compat_client, LlamaStackAsLibraryClient):
         pytest.skip("Responses API file search is not yet supported in library client.")
 
-    vector_store = new_vector_store(compat_client, "test_vector_store")
+    vector_store = new_vector_store(compat_client, "test_vector_store", embedding_model_id, embedding_dimension)
 
     # Create the response request, which should query our vector store
     response = compat_client.responses.create(
@@ -127,12 +131,14 @@ def test_response_non_streaming_file_search_empty_vector_store(compat_client, te
     assert response.output_text
 
 
-def test_response_sequential_file_search(compat_client, text_model_id, tmp_path):
+def test_response_sequential_file_search(
+    compat_client, text_model_id, embedding_model_id, embedding_dimension, tmp_path
+):
     """Test file search with sequential responses using previous_response_id."""
     if isinstance(compat_client, LlamaStackAsLibraryClient):
         pytest.skip("Responses API file search is not yet supported in library client.")
 
-    vector_store = new_vector_store(compat_client, "test_vector_store")
+    vector_store = new_vector_store(compat_client, "test_vector_store", embedding_model_id, embedding_dimension)
 
     # Create a test file with content
     file_content = "The Llama 4 Maverick model has 128 experts in its mixture of experts architecture."