[Frontend] Update OpenAI error response to upstream format (#22099)

Signed-off-by: Moritz Sanft <58110325+msanft@users.noreply.github.com>
This commit is contained in:
Moritz Sanft
2025-08-06 23:06:00 -07:00
committed by GitHub
parent cbc8457b26
commit 370661856b
10 changed files with 73 additions and 67 deletions
@@ -121,8 +121,7 @@ def test_invalid_truncate_prompt_tokens_error(server: RemoteOpenAIServer,
error = classification_response.json()
assert classification_response.status_code == 400
assert error["object"] == "error"
assert "truncate_prompt_tokens" in error["message"]
assert "truncate_prompt_tokens" in error["error"]["message"]
@pytest.mark.parametrize("model_name", [MODEL_NAME])
@@ -137,7 +136,7 @@ def test_empty_input_error(server: RemoteOpenAIServer, model_name: str):
error = classification_response.json()
assert classification_response.status_code == 400
assert error["object"] == "error"
assert "error" in error
@pytest.mark.parametrize("model_name", [MODEL_NAME])
@@ -160,8 +160,8 @@ async def test_serving_completion_resolver_not_found(mock_serving_setup,
mock_engine.generate.assert_not_called()
assert isinstance(response, ErrorResponse)
assert response.code == HTTPStatus.NOT_FOUND.value
assert non_existent_model in response.message
assert response.error.code == HTTPStatus.NOT_FOUND.value
assert non_existent_model in response.error.message
@pytest.mark.asyncio
@@ -190,8 +190,8 @@ async def test_serving_completion_resolver_add_lora_fails(
# Assert the correct error response
assert isinstance(response, ErrorResponse)
assert response.code == HTTPStatus.BAD_REQUEST.value
assert invalid_model in response.message
assert response.error.code == HTTPStatus.BAD_REQUEST.value
assert invalid_model in response.error.message
@pytest.mark.asyncio
@@ -66,8 +66,8 @@ async def test_load_lora_adapter_missing_fields():
request = LoadLoRAAdapterRequest(lora_name="", lora_path="")
response = await serving_models.load_lora_adapter(request)
assert isinstance(response, ErrorResponse)
assert response.type == "InvalidUserInput"
assert response.code == HTTPStatus.BAD_REQUEST
assert response.error.type == "InvalidUserInput"
assert response.error.code == HTTPStatus.BAD_REQUEST
@pytest.mark.asyncio
@@ -84,8 +84,8 @@ async def test_load_lora_adapter_duplicate():
lora_path="/path/to/adapter1")
response = await serving_models.load_lora_adapter(request)
assert isinstance(response, ErrorResponse)
assert response.type == "InvalidUserInput"
assert response.code == HTTPStatus.BAD_REQUEST
assert response.error.type == "InvalidUserInput"
assert response.error.code == HTTPStatus.BAD_REQUEST
assert len(serving_models.lora_requests) == 1
@@ -110,8 +110,8 @@ async def test_unload_lora_adapter_missing_fields():
request = UnloadLoRAAdapterRequest(lora_name="", lora_int_id=None)
response = await serving_models.unload_lora_adapter(request)
assert isinstance(response, ErrorResponse)
assert response.type == "InvalidUserInput"
assert response.code == HTTPStatus.BAD_REQUEST
assert response.error.type == "InvalidUserInput"
assert response.error.code == HTTPStatus.BAD_REQUEST
@pytest.mark.asyncio
@@ -120,5 +120,5 @@ async def test_unload_lora_adapter_not_found():
request = UnloadLoRAAdapterRequest(lora_name="nonexistent_adapter")
response = await serving_models.unload_lora_adapter(request)
assert isinstance(response, ErrorResponse)
assert response.type == "NotFoundError"
assert response.code == HTTPStatus.NOT_FOUND
assert response.error.type == "NotFoundError"
assert response.error.code == HTTPStatus.NOT_FOUND
@@ -116,8 +116,10 @@ async def test_non_asr_model(winning_call):
file=winning_call,
language="en",
temperature=0.0)
assert res.code == 400 and not res.text
assert res.message == "The model does not support Transcriptions API"
err = res.error
assert err["code"] == 400 and not res.text
assert err[
"message"] == "The model does not support Transcriptions API"
@pytest.mark.asyncio
@@ -133,12 +135,15 @@ async def test_completion_endpoints():
"role": "system",
"content": "You are a helpful assistant."
}])
assert res.code == 400
assert res.message == "The model does not support Chat Completions API"
err = res.error
assert err["code"] == 400
assert err[
"message"] == "The model does not support Chat Completions API"
res = await client.completions.create(model=model_name, prompt="Hello")
assert res.code == 400
assert res.message == "The model does not support Completions API"
err = res.error
assert err["code"] == 400
assert err["message"] == "The model does not support Completions API"
@pytest.mark.asyncio
@@ -73,8 +73,9 @@ async def test_non_asr_model(foscolo):
res = await client.audio.translations.create(model=model_name,
file=foscolo,
temperature=0.0)
assert res.code == 400 and not res.text
assert res.message == "The model does not support Translations API"
err = res.error
assert err["code"] == 400 and not res.text
assert err["message"] == "The model does not support Translations API"
@pytest.mark.asyncio