refactor(platform): Reject client_name at _acompletion.

src/any_llm/providers/platform/platform.py

@@ -104,8 +104,12 @@ async def _acompletion(
         **kwargs: Any,
     ) -> ChatCompletion | AsyncIterator[ChatCompletionChunk]:
         client_name = kwargs.pop("client_name", None)
-        if self.client_name is None:
-            self.client_name = client_name
+        if client_name is not None:
+            msg = (
+                "Passing client_name at request time is not supported for PlatformProvider. "
+                "Set client_name when creating the provider (for example, AnyLLM.create(..., client_name=...))."
+            )
+            raise ValueError(msg)
 
         start_time = time.perf_counter()
 

tests/unit/providers/test_platform_provider.py

@@ -1163,14 +1163,14 @@ async def test_usage_event_includes_version_header(
 @pytest.mark.asyncio
 @patch("any_llm_platform_client.AnyLLMPlatformClient.get_decrypted_provider_key")
 @patch("any_llm.providers.platform.platform.post_completion_usage_event")
-async def test_acompletion_handles_client_name_in_kwargs(
+async def test_acompletion_rejects_client_name_in_kwargs_when_not_initialized(
     mock_post_usage: AsyncMock,
     mock_get_decrypted_provider_key: Mock,
     any_llm_key: str,
     mock_decrypted_provider_key: DecryptedProviderKey,
     mock_completion: ChatCompletion,
 ) -> None:
-    """Test that _acompletion correctly handles client_name passed in kwargs."""
+    """Test that _acompletion rejects request-time client_name when provider was not initialized with it."""
     mock_get_decrypted_provider_key.return_value = mock_decrypted_provider_key
 
     provider_instance = PlatformProvider(api_key=any_llm_key)
@@ -1186,29 +1186,24 @@ async def test_acompletion_handles_client_name_in_kwargs(
 
     client_name = "test-client-from-kwargs"
 
-    # Call _acompletion with client_name in kwargs
-    await provider_instance._acompletion(params, client_name=client_name)
+    with pytest.raises(ValueError, match="Passing client_name at request time is not supported"):
+        await provider_instance._acompletion(params, client_name=client_name)
 
-    # Verify self.client_name was updated
-    assert provider_instance.client_name == client_name
-
-    # Verify post_completion_usage_event was called with the correct client_name
-    mock_post_usage.assert_called_once()
-    call_args = mock_post_usage.call_args
-    assert call_args.kwargs["client_name"] == client_name
+    provider_instance.provider._acompletion.assert_not_called()
+    mock_post_usage.assert_not_called()
 
 
 @pytest.mark.asyncio
 @patch("any_llm_platform_client.AnyLLMPlatformClient.get_decrypted_provider_key")
 @patch("any_llm.providers.platform.platform.post_completion_usage_event")
-async def test_acompletion_does_not_overwrite_existing_client_name(
+async def test_acompletion_rejects_changing_existing_client_name(
     mock_post_usage: AsyncMock,
     mock_get_decrypted_provider_key: Mock,
     any_llm_key: str,
     mock_decrypted_provider_key: DecryptedProviderKey,
     mock_completion: ChatCompletion,
 ) -> None:
-    """Test that _acompletion does not overwrite an existing client_name if one is already set."""
+    """Test that _acompletion rejects a request-time client_name that differs from configured client_name."""
     mock_get_decrypted_provider_key.return_value = mock_decrypted_provider_key
 
     initial_client_name = "initial-client"
@@ -1225,13 +1220,39 @@ async def test_acompletion_does_not_overwrite_existing_client_name(
 
     new_client_name = "new-client-from-kwargs"
 
-    # Call _acompletion with a new client_name in kwargs
-    await provider_instance._acompletion(params, client_name=new_client_name)
+    with pytest.raises(ValueError, match="Passing client_name at request time is not supported"):
+        await provider_instance._acompletion(params, client_name=new_client_name)
 
-    # Verify self.client_name was NOT updated
-    assert provider_instance.client_name == initial_client_name
+    provider_instance.provider._acompletion.assert_not_called()
+    mock_post_usage.assert_not_called()
 
-    # Verify post_completion_usage_event was called with the INITIAL client_name
-    mock_post_usage.assert_called_once()
-    call_args = mock_post_usage.call_args
-    assert call_args.kwargs["client_name"] == initial_client_name
+
+@pytest.mark.asyncio
+@patch("any_llm_platform_client.AnyLLMPlatformClient.get_decrypted_provider_key")
+@patch("any_llm.providers.platform.platform.post_completion_usage_event")
+async def test_acompletion_rejects_same_client_name_in_kwargs(
+    mock_post_usage: AsyncMock,
+    mock_get_decrypted_provider_key: Mock,
+    any_llm_key: str,
+    mock_decrypted_provider_key: DecryptedProviderKey,
+    mock_completion: ChatCompletion,
+) -> None:
+    """Test that _acompletion rejects request-time client_name even when it matches configured value."""
+    mock_get_decrypted_provider_key.return_value = mock_decrypted_provider_key
+
+    client_name = "configured-client"
+    provider_instance = PlatformProvider(api_key=any_llm_key, client_name=client_name)
+    provider_instance.provider = OpenaiProvider
+    provider_instance.provider._acompletion = AsyncMock(return_value=mock_completion)  # type: ignore[method-assign]
+
+    params = CompletionParams(
+        model_id="gpt-4",
+        messages=[{"role": "user", "content": "Hello"}],
+        stream=False,
+    )
+
+    with pytest.raises(ValueError, match="Passing client_name at request time is not supported"):
+        await provider_instance._acompletion(params, client_name=client_name)
+
+    provider_instance.provider._acompletion.assert_not_called()
+    mock_post_usage.assert_not_called()