mirror of
https://github.com/Sea-Haven-Industries/open-swe.git
synced 2026-09-30 17:23:15 +00:00
Adam Moussa
ae1f883b4c
Applies the plan's C5 step: git mv every test per the domain-reorg
move-map (movemap-m50.txt) into tests/{agent,analyzer,auth,dashboard,
github,middleware,models,reviewer,sandbox,slack,tools,webhooks}/, plus
the 13 fork-only placements from the scoping report §2c (Atlassian
webhook tests -> tests/webhooks/, test_atlassian_connect.py and
test_auth_error_leak.py -> tests/auth/, jira/confluence util tests ->
tests/tools/, test_repo_binding_isolation.py -> tests/sandbox/,
bot-identity/autofix tests -> tests/github/).
Path-only move: the only content edits are parents[1] -> parents[2]
fixes in test_e2b_integration.py and test_daytona_integration.py,
required because their __file__-relative ROOT path gained one more
directory level in the move.
Monkeypatch retargets for these files were already completed in C4;
none remained outstanding here.
231 lines
8.5 KiB
Python
231 lines
8.5 KiB
Python
from unittest.mock import AsyncMock, MagicMock, patch
|
|
|
|
import pytest
|
|
from langgraph.graph.state import RunnableConfig
|
|
|
|
from agent.server import get_agent
|
|
|
|
|
|
class _DummyAgent:
|
|
def with_config(self, config: RunnableConfig) -> "_DummyAgent":
|
|
self.config = config
|
|
return self
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_agent_uses_profile_subagent_model_override() -> None:
|
|
config: RunnableConfig = {
|
|
"configurable": {
|
|
"__is_for_execution__": True,
|
|
"thread_id": "thread-123",
|
|
"github_login": "octocat",
|
|
},
|
|
"metadata": {},
|
|
}
|
|
main_model = MagicMock(name="main_model")
|
|
subagent_model = MagicMock(name="subagent_model")
|
|
captured: dict[str, object] = {}
|
|
|
|
def fake_create_deep_agent(**kwargs: object) -> _DummyAgent:
|
|
captured.update(kwargs)
|
|
return _DummyAgent()
|
|
|
|
with (
|
|
patch(
|
|
"agent.server.resolve_github_token",
|
|
new_callable=AsyncMock,
|
|
return_value=("ghp", None),
|
|
),
|
|
patch("agent.server.resolve_triggering_user_identity", return_value=None),
|
|
patch(
|
|
"agent.server.ensure_sandbox_for_thread",
|
|
new_callable=AsyncMock,
|
|
return_value=MagicMock(),
|
|
),
|
|
patch(
|
|
"agent.server.aresolve_sandbox_work_dir",
|
|
new_callable=AsyncMock,
|
|
return_value="/workspace",
|
|
),
|
|
patch(
|
|
"agent.server.get_team_default_model_pair",
|
|
new_callable=AsyncMock,
|
|
return_value=(
|
|
("bedrock_converse:us.anthropic.claude-opus-4-8", "medium"),
|
|
("fireworks:accounts/fireworks/models/deepseek-v4-pro", "low"),
|
|
),
|
|
),
|
|
patch(
|
|
"agent.server.load_profile",
|
|
new_callable=AsyncMock,
|
|
return_value={
|
|
"default_model": "bedrock_converse:us.anthropic.claude-opus-4-8",
|
|
"reasoning_effort": "high",
|
|
"default_subagent_model": "fireworks:accounts/fireworks/models/deepseek-v4-pro",
|
|
"subagent_reasoning_effort": "xhigh",
|
|
},
|
|
),
|
|
patch("agent.server.fallback_model_id_for", return_value=None),
|
|
patch(
|
|
"agent.utils.deferred_model.make_model", side_effect=[main_model, subagent_model]
|
|
) as make_model,
|
|
patch("agent.server.construct_system_prompt", return_value="prompt"),
|
|
patch("agent.server.create_deep_agent", side_effect=fake_create_deep_agent),
|
|
):
|
|
await get_agent(config)
|
|
|
|
assert captured["model"] is main_model
|
|
subagents = captured["subagents"]
|
|
assert isinstance(subagents, list)
|
|
assert subagents[0]["name"] == "general-purpose"
|
|
assert subagents[0]["model"] is subagent_model
|
|
|
|
main_call = make_model.call_args_list[0]
|
|
assert main_call.args == ("bedrock_converse:us.anthropic.claude-opus-4-8",)
|
|
assert main_call.kwargs["additional_model_request_fields"] == {
|
|
"thinking": {"type": "adaptive", "display": "summarized"},
|
|
"output_config": {"effort": "high"},
|
|
}
|
|
|
|
subagent_call = make_model.call_args_list[1]
|
|
assert subagent_call.args == ("fireworks:accounts/fireworks/models/deepseek-v4-pro",)
|
|
assert subagent_call.kwargs["model_kwargs"] == {"reasoning_effort": "xhigh"}
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_agent_subagent_inherits_profile_model_override_without_explicit_pair() -> None:
|
|
config: RunnableConfig = {
|
|
"configurable": {
|
|
"__is_for_execution__": True,
|
|
"thread_id": "thread-123",
|
|
"github_login": "octocat",
|
|
},
|
|
"metadata": {},
|
|
}
|
|
main_model = MagicMock(name="main_model")
|
|
subagent_model = MagicMock(name="subagent_model")
|
|
captured: dict[str, object] = {}
|
|
|
|
def fake_create_deep_agent(**kwargs: object) -> _DummyAgent:
|
|
captured.update(kwargs)
|
|
return _DummyAgent()
|
|
|
|
with (
|
|
patch(
|
|
"agent.server.resolve_github_token",
|
|
new_callable=AsyncMock,
|
|
return_value=("ghp", None),
|
|
),
|
|
patch("agent.server.resolve_triggering_user_identity", return_value=None),
|
|
patch(
|
|
"agent.server.ensure_sandbox_for_thread",
|
|
new_callable=AsyncMock,
|
|
return_value=MagicMock(),
|
|
),
|
|
patch(
|
|
"agent.server.aresolve_sandbox_work_dir",
|
|
new_callable=AsyncMock,
|
|
return_value="/workspace",
|
|
),
|
|
patch(
|
|
"agent.server.get_team_default_model_pair",
|
|
new_callable=AsyncMock,
|
|
return_value=(
|
|
("bedrock_converse:us.anthropic.claude-opus-4-8", "medium"),
|
|
("fireworks:accounts/fireworks/models/deepseek-v4-pro", "low"),
|
|
),
|
|
),
|
|
patch(
|
|
"agent.server.load_profile",
|
|
new_callable=AsyncMock,
|
|
return_value={
|
|
"default_model": "bedrock_converse:us.anthropic.claude-opus-4-8",
|
|
"reasoning_effort": "high",
|
|
},
|
|
),
|
|
patch("agent.server.fallback_model_id_for", return_value=None),
|
|
patch(
|
|
"agent.utils.deferred_model.make_model", side_effect=[main_model, subagent_model]
|
|
) as make_model,
|
|
patch("agent.server.construct_system_prompt", return_value="prompt"),
|
|
patch("agent.server.create_deep_agent", side_effect=fake_create_deep_agent),
|
|
):
|
|
await get_agent(config)
|
|
|
|
subagents = captured["subagents"]
|
|
assert isinstance(subagents, list)
|
|
assert subagents[0]["model"] is subagent_model
|
|
assert make_model.call_args_list[0].args == ("bedrock_converse:us.anthropic.claude-opus-4-8",)
|
|
assert make_model.call_args_list[1].args == ("bedrock_converse:us.anthropic.claude-opus-4-8",)
|
|
assert make_model.call_args_list[1].kwargs["additional_model_request_fields"] == {
|
|
"thinking": {"type": "adaptive", "display": "summarized"},
|
|
"output_config": {"effort": "high"},
|
|
}
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_agent_gate_swaps_disabled_fable_profile_to_opus() -> None:
|
|
config: RunnableConfig = {
|
|
"configurable": {
|
|
"__is_for_execution__": True,
|
|
"thread_id": "thread-123",
|
|
"github_login": "octocat",
|
|
},
|
|
"metadata": {},
|
|
}
|
|
main_model = MagicMock(name="main_model")
|
|
subagent_model = MagicMock(name="subagent_model")
|
|
captured: dict[str, object] = {}
|
|
|
|
def fake_create_deep_agent(**kwargs: object) -> _DummyAgent:
|
|
captured.update(kwargs)
|
|
return _DummyAgent()
|
|
|
|
with (
|
|
patch(
|
|
"agent.server.resolve_github_token", new_callable=AsyncMock, return_value=("ghp", None)
|
|
),
|
|
patch("agent.server.resolve_triggering_user_identity", return_value=None),
|
|
patch(
|
|
"agent.server.ensure_sandbox_for_thread",
|
|
new_callable=AsyncMock,
|
|
return_value=MagicMock(),
|
|
),
|
|
patch(
|
|
"agent.server.aresolve_sandbox_work_dir",
|
|
new_callable=AsyncMock,
|
|
return_value="/workspace",
|
|
),
|
|
patch(
|
|
"agent.server.get_team_default_model_pair",
|
|
new_callable=AsyncMock,
|
|
return_value=(
|
|
("bedrock_converse:us.anthropic.claude-opus-4-8", "medium"),
|
|
("bedrock_converse:us.anthropic.claude-opus-4-8", "low"),
|
|
),
|
|
),
|
|
# Profile selected Fable back when it was allowed; it's now disabled.
|
|
patch(
|
|
"agent.server.load_profile",
|
|
new_callable=AsyncMock,
|
|
return_value={
|
|
"default_model": "bedrock_converse:us.anthropic.claude-fable-5",
|
|
"reasoning_effort": "high",
|
|
},
|
|
),
|
|
patch("agent.server.get_team_fable_enabled", new_callable=AsyncMock, return_value=False),
|
|
patch("agent.server.fallback_model_id_for", return_value=None),
|
|
patch(
|
|
"agent.utils.deferred_model.make_model", side_effect=[main_model, subagent_model]
|
|
) as make_model,
|
|
patch("agent.server.construct_system_prompt", return_value="prompt"),
|
|
patch("agent.server.create_deep_agent", side_effect=fake_create_deep_agent),
|
|
):
|
|
await get_agent(config)
|
|
|
|
# Fable was scrubbed to Opus for both main and subagent; effort preserved.
|
|
assert make_model.call_args_list[0].args == ("bedrock_converse:us.anthropic.claude-opus-4-8",)
|
|
assert make_model.call_args_list[0].kwargs["additional_model_request_fields"][
|
|
"output_config"
|
|
] == {"effort": "high"}
|
|
assert make_model.call_args_list[1].args == ("bedrock_converse:us.anthropic.claude-opus-4-8",)
|