[Bugfix] Route Step3p5 forced tool choices through XML parser (#51810)

Signed-off-by: taking-lying-flat <1615405@qq.com>
Signed-off-by: Flora Feng <4florafeng@gmail.com>
Co-authored-by: Flora Feng <4florafeng@gmail.com>
This commit is contained in:
cherry77-cloud
2026-09-29 21:58:46 +00:00
committed by GitHub
co-authored by Flora Feng
parent 4e7af3b0de
commit 5e887a078b
3 changed files with 83 additions and 82 deletions
+25 -21
View File
@@ -5,6 +5,7 @@ import json
from collections.abc import Generator
import pytest
from openai.types.responses import FunctionTool
from vllm.entrypoints.generate.base.protocol import (
DeltaMessage,
@@ -386,26 +387,29 @@ TX
assert extracted_tool_calls.tool_calls[0].function.name == "get_current_weather"
def test_extract_tool_calls_type_conversion(step3p5_tokenizer):
"""Test parameter type conversion based on tool schema."""
tools = [
ChatCompletionToolsParam(
type="function",
function={
"name": "test_types",
"parameters": {
"type": "object",
"properties": {
"int_param": {"type": "integer"},
"float_param": {"type": "float"},
"bool_param": {"type": "boolean"},
"str_param": {"type": "string"},
"obj_param": {"type": "object"},
},
},
},
)
]
@pytest.mark.parametrize("api", ["chat", "responses"])
def test_extract_tool_calls_type_conversion(step3p5_tokenizer, api):
"""Test parameter type conversion based on Chat and Responses tool schemas."""
parameters = {
"type": "object",
"properties": {
"int_param": {"type": "integer"},
"float_param": {"type": "float"},
"bool_param": {"type": "boolean"},
"str_param": {"type": "string"},
"obj_param": {"type": "object"},
},
}
tools: list[ChatCompletionToolsParam | FunctionTool] = (
[
ChatCompletionToolsParam(
type="function",
function={"name": "test_types", "parameters": parameters},
)
]
if api == "chat"
else [FunctionTool(type="function", name="test_types", parameters=parameters)]
)
model_output = """<tool_call>
<function=test_types>
@@ -428,7 +432,7 @@ hello world
</tool_call>"""
parser = Step3p5ToolParser(step3p5_tokenizer, tools=tools)
request = ChatCompletionRequest(model=MODEL, messages=[], tools=tools)
request = ChatCompletionRequest(model=MODEL, messages=[])
extracted_tool_calls = parser.extract_tool_calls(model_output, request=request)
args = json.loads(extracted_tool_calls.tool_calls[0].function.arguments)