f96edfe5ca
* fix: exclude deepseek from local tool-calling keyword list
deepseek-r1 on Ollama returns HTTP 400 when tool schemas are sent.
The cloud API (api.deepseek.com) is already caught by the _API_HOSTS
check, so the generic 'deepseek' keyword match was only causing false
positives for local Ollama-served models.
* fix: add model no-tools blocklist and regression tests for deepseek-r1
The previous fix removed 'deepseek' from the keyword allow-list, but
_is_api_model is still True for localhost endpoints because 'localhost'
appears in _API_HOSTS — so the keyword change had no effect for Ollama.
Proper fix: add an explicit _model_no_tools blocklist ('deepseek-r1')
that overrides the endpoint URL check. The endpoint's supports_tools DB
flag still takes priority either way (True forces tools on, False forces
them off), so users can override per-endpoint when needed.
Also refined the deepseek allow-list: 'deepseek-v' and 'deepseek-chat'
cover the cloud models (v2, v3, chat) that do support tools, without
matching deepseek-r1 variants.
13 regression tests cover:
- deepseek-r1 on localhost/docker: no tools (was HTTP 400)
- deepseek-v3/chat on api.deepseek.com: tools enabled (no regression)
- endpoint_supports=True/False overrides both lists
- qwen/llama on localhost: unaffected
107 lines
4.0 KiB
Python
107 lines
4.0 KiB
Python
"""Regression tests for the tool-support heuristic in stream_agent_loop.
|
|
|
|
Verifies two critical cases:
|
|
1. deepseek-r1 on a local Ollama endpoint must NOT enable native tool schemas
|
|
(Ollama returns HTTP 400 for these models when tools are sent).
|
|
2. api.deepseek.com must still be treated as tool-capable via the host
|
|
allow-list (_API_HOSTS), so cloud deepseek users keep working.
|
|
"""
|
|
import pytest
|
|
from src.agent_loop import _API_HOSTS
|
|
|
|
|
|
def _compute_is_api_model(model: str, endpoint_url: str, endpoint_supports=None) -> bool:
|
|
"""Replicate the heuristic from stream_agent_loop without side effects."""
|
|
model_lc = model.lower()
|
|
|
|
model_supports_tools = any(kw in model_lc for kw in (
|
|
"gpt-4", "gpt-5", "gpt-o", "claude", "gemini", "gemma",
|
|
"qwen3", "qwen2.5", "mixtral", "mistral", "llama-3.1", "llama-3.2",
|
|
"llama-3.3", "llama-4",
|
|
"minimax", "kimi", "yi-", "phi-3", "phi-4", "command-r",
|
|
"glm-4", "internlm", "hermes",
|
|
"deepseek-v", "deepseek-chat",
|
|
))
|
|
model_no_tools = any(kw in model_lc for kw in (
|
|
"deepseek-r1",
|
|
))
|
|
|
|
if endpoint_supports is True:
|
|
return True
|
|
if endpoint_supports is False or model_no_tools:
|
|
return False
|
|
return any(h in endpoint_url for h in _API_HOSTS) or model_supports_tools
|
|
|
|
|
|
class TestDeepSeekToolSupport:
|
|
# --- local Ollama cases (must NOT get tool schemas) ---
|
|
|
|
def test_deepseek_r1_7b_local_ollama_no_tools(self):
|
|
result = _compute_is_api_model(
|
|
"deepseek-r1:7b", "http://localhost:11434/v1"
|
|
)
|
|
assert result is False, (
|
|
"deepseek-r1:7b on Ollama must not enable tool schemas "
|
|
"(Ollama returns HTTP 400 for this model)"
|
|
)
|
|
|
|
def test_deepseek_r1_14b_local_no_tools(self):
|
|
assert _compute_is_api_model("deepseek-r1:14b", "http://localhost:11434/v1") is False
|
|
|
|
def test_deepseek_r1_70b_local_no_tools(self):
|
|
assert _compute_is_api_model("deepseek-r1:70b", "http://127.0.0.1:11434/v1") is False
|
|
|
|
def test_deepseek_r1_via_docker_no_tools(self):
|
|
assert _compute_is_api_model(
|
|
"deepseek-r1:7b", "http://host.docker.internal:11434/v1"
|
|
) is False
|
|
|
|
# --- cloud API cases (must still get tool schemas) ---
|
|
|
|
def test_deepseek_cloud_api_gets_tools(self):
|
|
result = _compute_is_api_model(
|
|
"deepseek-chat", "https://api.deepseek.com/v1"
|
|
)
|
|
assert result is True, (
|
|
"api.deepseek.com must be treated as tool-capable via _API_HOSTS"
|
|
)
|
|
|
|
def test_deepseek_v3_cloud_gets_tools(self):
|
|
assert _compute_is_api_model("deepseek-v3", "https://api.deepseek.com/v1") is True
|
|
|
|
def test_deepseek_v2_cloud_gets_tools(self):
|
|
assert _compute_is_api_model("deepseek-v2.5", "https://api.deepseek.com/v1") is True
|
|
|
|
# --- endpoint_supports override takes priority ---
|
|
|
|
def test_endpoint_supports_true_overrides_blocklist(self):
|
|
"""A user who explicitly sets supports_tools=True on their endpoint
|
|
can force tool schemas even for deepseek-r1 (e.g. custom server)."""
|
|
result = _compute_is_api_model(
|
|
"deepseek-r1:7b", "http://localhost:11434/v1", endpoint_supports=True
|
|
)
|
|
assert result is True
|
|
|
|
def test_endpoint_supports_false_overrides_cloud(self):
|
|
"""supports_tools=False on an endpoint gates even cloud APIs."""
|
|
result = _compute_is_api_model(
|
|
"deepseek-chat", "https://api.deepseek.com/v1", endpoint_supports=False
|
|
)
|
|
assert result is False
|
|
|
|
# --- other local models unaffected ---
|
|
|
|
def test_qwen_local_still_gets_tools(self):
|
|
assert _compute_is_api_model("qwen2.5:14b", "http://localhost:11434/v1") is True
|
|
|
|
def test_llama_local_gets_tools_via_host(self):
|
|
assert _compute_is_api_model("llama3.2:3b", "http://localhost:11434/v1") is True
|
|
|
|
|
|
class TestApiHostsContainsDeepSeek:
|
|
def test_api_deepseek_com_in_api_hosts(self):
|
|
assert "api.deepseek.com" in _API_HOSTS
|
|
|
|
def test_deepseek_com_in_api_hosts(self):
|
|
assert "deepseek.com" in _API_HOSTS
|