[router] add basic ci tests for gpt-oss model support (#12651)
Co-authored-by: Chang Su <chang.s.su@oracle.com>
This commit is contained in:
Keyang Ru
@@ -0,0 +1,202 @@
|
||||
"""
|
||||
Base test class for function calling tests.
|
||||
|
||||
This module provides test cases for function calling functionality
|
||||
across different backends.
|
||||
"""
|
||||
|
||||
import json
|
||||
import sys
|
||||
from pathlib import Path
|
||||
|
||||
# Add current directory for local imports
|
||||
_TEST_DIR = Path(__file__).parent
|
||||
sys.path.insert(0, str(_TEST_DIR))
|
||||
|
||||
from util import CustomTestCase
|
||||
|
||||
|
||||
class ResponseAPIBaseTest(CustomTestCase):
|
||||
"""Base class for Response API tests with common utilities."""
|
||||
|
||||
# To be set by subclasses
|
||||
base_url: str = None
|
||||
api_key: str = None
|
||||
model: str = None
|
||||
|
||||
def make_request(
|
||||
self,
|
||||
endpoint: str,
|
||||
method: str = "POST",
|
||||
json_data: dict = None,
|
||||
params: dict = None,
|
||||
):
|
||||
"""
|
||||
Make HTTP request to router.
|
||||
|
||||
This is a minimal implementation - subclasses should import from basic_crud.
|
||||
"""
|
||||
import requests
|
||||
|
||||
url = f"{self.base_url}{endpoint}"
|
||||
headers = {"Content-Type": "application/json"}
|
||||
if self.api_key:
|
||||
headers["Authorization"] = f"Bearer {self.api_key}"
|
||||
|
||||
if method == "POST":
|
||||
resp = requests.post(url, json=json_data, headers=headers, params=params)
|
||||
elif method == "GET":
|
||||
resp = requests.get(url, headers=headers, params=params)
|
||||
elif method == "DELETE":
|
||||
resp = requests.delete(url, headers=headers, params=params)
|
||||
else:
|
||||
raise ValueError(f"Unsupported method: {method}")
|
||||
return resp
|
||||
|
||||
|
||||
class FunctionCallingBaseTest(ResponseAPIBaseTest):
|
||||
"""Base class for function calling tests."""
|
||||
|
||||
def test_basic_function_call(self):
|
||||
"""
|
||||
Test basic function calling workflow.
|
||||
|
||||
This test follows the pattern from function_call_test.py:
|
||||
1. Define a function tool (get_horoscope)
|
||||
2. Send user message asking for horoscope
|
||||
3. Model should return function_call
|
||||
4. Execute function locally and provide output
|
||||
5. Model should generate final response using the function output
|
||||
"""
|
||||
# 1. Define a list of callable tools for the model
|
||||
tools = [
|
||||
{
|
||||
"type": "function",
|
||||
"name": "get_horoscope",
|
||||
"description": "Get today's horoscope for an astrological sign.",
|
||||
"parameters": {
|
||||
"type": "object",
|
||||
"properties": {
|
||||
"sign": {
|
||||
"type": "string",
|
||||
"description": "An astrological sign like Taurus or Aquarius",
|
||||
},
|
||||
},
|
||||
"required": ["sign"],
|
||||
},
|
||||
},
|
||||
]
|
||||
|
||||
# Create a running input list we will add to over time
|
||||
input_list = [
|
||||
{"role": "user", "content": "What is my horoscope? I am an Aquarius."}
|
||||
]
|
||||
|
||||
# 2. Prompt the model with tools defined
|
||||
resp = self.make_request(
|
||||
"/v1/responses",
|
||||
"POST",
|
||||
{
|
||||
"model": self.model,
|
||||
"tools": tools,
|
||||
"input": input_list,
|
||||
},
|
||||
)
|
||||
|
||||
# Should successfully make the request
|
||||
self.assertEqual(resp.status_code, 200)
|
||||
|
||||
data = resp.json()
|
||||
|
||||
# Basic response structure
|
||||
self.assertIn("id", data)
|
||||
self.assertIn("status", data)
|
||||
self.assertEqual(data["status"], "completed")
|
||||
self.assertIn("output", data)
|
||||
|
||||
# Verify output array is not empty
|
||||
output = data["output"]
|
||||
self.assertIsInstance(output, list)
|
||||
self.assertGreater(len(output), 0)
|
||||
|
||||
# Check for function_call in output
|
||||
function_calls = [
|
||||
item for item in output if item.get("type") == "function_call"
|
||||
]
|
||||
self.assertGreater(
|
||||
len(function_calls), 0, "Response should contain at least one function_call"
|
||||
)
|
||||
|
||||
# Verify function_call structure
|
||||
function_call = function_calls[0]
|
||||
self.assertIn("call_id", function_call)
|
||||
self.assertIn("name", function_call)
|
||||
self.assertEqual(function_call["name"], "get_horoscope")
|
||||
self.assertIn("arguments", function_call)
|
||||
|
||||
# Parse arguments
|
||||
args = json.loads(function_call["arguments"])
|
||||
self.assertIn("sign", args)
|
||||
self.assertEqual(args["sign"].lower(), "aquarius")
|
||||
|
||||
# 3. Save function call outputs for subsequent requests
|
||||
input_list += output
|
||||
|
||||
# 4. Execute the function logic for get_horoscope
|
||||
horoscope = f"{args['sign']}: Next Tuesday you will befriend a baby otter."
|
||||
|
||||
# 5. Provide function call results to the model
|
||||
input_list.append(
|
||||
{
|
||||
"type": "function_call_output",
|
||||
"call_id": function_call["call_id"],
|
||||
"output": json.dumps({"horoscope": horoscope}),
|
||||
}
|
||||
)
|
||||
|
||||
# 6. Make second request with function output
|
||||
resp2 = self.make_request(
|
||||
"/v1/responses",
|
||||
"POST",
|
||||
{
|
||||
"model": self.model,
|
||||
"instructions": "Respond only with a horoscope generated by a tool.",
|
||||
"tools": tools,
|
||||
"input": input_list,
|
||||
},
|
||||
)
|
||||
|
||||
self.assertEqual(resp2.status_code, 200)
|
||||
|
||||
data2 = resp2.json()
|
||||
self.assertEqual(data2["status"], "completed")
|
||||
|
||||
# The model should be able to give a response using the function output
|
||||
output2 = data2["output"]
|
||||
self.assertGreater(len(output2), 0)
|
||||
|
||||
# Find message output
|
||||
messages = [item for item in output2 if item.get("type") == "message"]
|
||||
self.assertGreater(
|
||||
len(messages), 0, "Response should contain at least one message"
|
||||
)
|
||||
|
||||
# Verify message contains the horoscope
|
||||
message = messages[0]
|
||||
self.assertIn("content", message)
|
||||
content_parts = message["content"]
|
||||
self.assertGreater(len(content_parts), 0)
|
||||
|
||||
# Get text from content
|
||||
text_parts = [
|
||||
part.get("text", "")
|
||||
for part in content_parts
|
||||
if part.get("type") == "output_text"
|
||||
]
|
||||
full_text = " ".join(text_parts).lower()
|
||||
|
||||
# Should mention the horoscope or baby otter
|
||||
self.assertTrue(
|
||||
"baby otter" in full_text or "aquarius" in full_text,
|
||||
"Response should reference the horoscope content",
|
||||
)
|
||||
Reference in New Issue
Block a user