chenmingxuan
fac4ccb760
Add dark theme, richer Markdown and readable plugin results. Improve Android input evidence, response integrity and diagnostics. Unify lifecycle, guidance receipts and evidence sources across clients. Keep direct action dispatch without an added AI quota approval layer. Align version metadata at 2.2.0 and bundle Helper 1.2.2. Validation: 3063 backend tests before the version bump; 44 after. Frontend: 317 tests and production build passed.
182 lines
7.2 KiB
Python
182 lines
7.2 KiB
Python
from types import SimpleNamespace
|
|
from unittest.mock import AsyncMock, Mock
|
|
|
|
import pytest
|
|
from pydantic import ValidationError
|
|
|
|
from artemis.agents.flash.progress import TaskProgress
|
|
from artemis.agents.flash.runner import FlashRunner, _TurnRecord
|
|
|
|
|
|
def plan(phone="active", pc="pending"):
|
|
return {
|
|
"reason": "按用户要求先手机页脚，再电脑独立检查",
|
|
"stages": [
|
|
{
|
|
"id": "phone",
|
|
"scope": "Android 主体、页脚及最后滚动",
|
|
"resources": ["android"],
|
|
"status": phone,
|
|
"evidence": "步骤 12 页脚及 13 滚动截图" if phone in {"done", "blocked"} else "",
|
|
},
|
|
{
|
|
"id": "pc",
|
|
"scope": "电脑检查",
|
|
"resources": ["cua-driver"],
|
|
"status": pc,
|
|
"evidence": "步骤 20 电脑截图" if pc == "done" else "",
|
|
},
|
|
],
|
|
}
|
|
|
|
|
|
def test_ordered_stages_require_evidence_and_keep_outstanding_scope(tmp_path):
|
|
progress = TaskProgress()
|
|
progress.update(plan(), resources={"android", "cua-driver"}, directory=tmp_path)
|
|
assert progress.guard("cua-driver")
|
|
assert progress.guard("android") is None
|
|
assert progress.guard(None) is None # read-only preflight
|
|
assert progress.guard(None, completed=True)
|
|
assert (tmp_path / "notes/task_progress.md").exists()
|
|
with pytest.raises(ValidationError):
|
|
progress.update(plan("pending", "active"), resources={"android", "cua-driver"})
|
|
omitted = plan()
|
|
omitted["stages"] = omitted["stages"][1:]
|
|
with pytest.raises(ValueError, match="不能删除"):
|
|
progress.update(omitted, resources={"android", "cua-driver"})
|
|
progress.update(plan("done", "active"), resources={"android", "cua-driver"})
|
|
assert progress.guard("cua-driver") is None
|
|
assert progress.guard("android")
|
|
progress.update(plan("done", "done"), resources={"android", "cua-driver"})
|
|
assert progress.guard(None, completed=True) is None
|
|
|
|
|
|
def test_pc_first_and_intentional_interleaving_are_supported():
|
|
progress = TaskProgress()
|
|
value = {
|
|
"reason": "路由器调整后在手机验证",
|
|
"stages": [
|
|
{
|
|
"id": "mixed",
|
|
"scope": "电脑改参数，手机核对，循环验证",
|
|
"resources": ["android", "cua-driver"],
|
|
"status": "active",
|
|
}
|
|
],
|
|
}
|
|
progress.update(value, resources={"android", "cua-driver"})
|
|
assert progress.guard("cua-driver") is None
|
|
assert progress.guard("android") is None
|
|
value["stages"][0]["resources"] = ["cua-driver"]
|
|
progress.update(value, resources={"cua-driver"})
|
|
assert progress.guard("cua-driver") is None
|
|
|
|
|
|
def test_stage_updates_are_atomic_and_preserve_unmentioned_scope_and_order(tmp_path):
|
|
progress = TaskProgress()
|
|
resources = {"android", "cua-driver"}
|
|
progress.update(plan(), resources=resources, directory=tmp_path)
|
|
before = progress.value.model_dump()
|
|
saved = (tmp_path / "notes/task_progress.md").read_bytes()
|
|
with pytest.raises(ValidationError, match="完成或阻塞阶段必须记录"):
|
|
progress.update(
|
|
{"reason": "无证据不能完成", "stage_updates": [{"id": "phone", "status": "done"}]},
|
|
resources=resources,
|
|
directory=tmp_path,
|
|
)
|
|
assert progress.value.model_dump() == before
|
|
assert (tmp_path / "notes/task_progress.md").read_bytes() == saved
|
|
progress.update(
|
|
{
|
|
"reason": "手机已检查完",
|
|
"stage_updates": [
|
|
{"id": "phone", "status": "done", "evidence": "实际步骤记录 real-step"}
|
|
],
|
|
},
|
|
resources=resources,
|
|
directory=tmp_path,
|
|
)
|
|
assert progress.value.stages[1].model_dump() == before["stages"][1]
|
|
assert [s.id for s in progress.value.stages] == ["phone", "pc"]
|
|
assert progress.value.stages[0].scope == before["stages"][0]["scope"]
|
|
progress.update(
|
|
{"reason": "开始电脑检查", "stage_updates": [{"id": "pc", "status": "active"}]},
|
|
resources=resources,
|
|
)
|
|
assert progress.guard("cua-driver") is None
|
|
with pytest.raises(ValueError, match="未知 id"):
|
|
progress.update(
|
|
{"reason": "错拼不能覆盖", "stage_updates": [{"id": "typo", "status": "done"}]},
|
|
resources=resources,
|
|
)
|
|
with pytest.raises(ValidationError, match="只能提供一种"):
|
|
progress.update(
|
|
{**plan(), "stage_updates": [{"id": "pc", "status": "active"}]}, resources=resources
|
|
)
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_flash_preflight_does_not_authorize_pc_mutation(monkeypatch):
|
|
plugins = SimpleNamespace(
|
|
packages={"cua-driver": None},
|
|
bindings={"plug_pc_look": ("cua-driver", None), "plug_pc_click": ("cua-driver", None)},
|
|
tool_names={"plug_pc_look", "plug_pc_click"},
|
|
is_read_only=lambda name: name == "plug_pc_look",
|
|
)
|
|
monkeypatch.setattr("artemis.plugins.runtime.get_plugin_runtime", lambda _: plugins)
|
|
runner = FlashRunner.__new__(FlashRunner)
|
|
runner.ctx = SimpleNamespace(data_engine=None, execution_target="android")
|
|
runner.executor = SimpleNamespace(action_tool_names={"click"})
|
|
runner._progress = TaskProgress()
|
|
runner._progress.update(plan(), resources={"android", "cua-driver"})
|
|
runner._execute_and_record_action = AsyncMock(return_value=(None, "", 0))
|
|
runner._refresh_observation = AsyncMock(return_value=(None, ""))
|
|
messages = []
|
|
await runner._process_tool_calls(
|
|
[
|
|
{"name": "plug_pc_look", "id": "look", "args": {}},
|
|
{"name": "plug_pc_click", "id": "click", "args": {}},
|
|
],
|
|
SimpleNamespace(indexed_elements=[]),
|
|
messages,
|
|
"预检",
|
|
{},
|
|
None,
|
|
"",
|
|
0,
|
|
_TurnRecord(),
|
|
)
|
|
runner._execute_and_record_action.assert_awaited_once()
|
|
assert runner._execute_and_record_action.call_args.args[0] == "plug_pc_look"
|
|
assert messages[-1].tool_call_id == "click" and messages[-1].status == "error"
|
|
runner._refresh_observation.assert_not_awaited()
|
|
|
|
|
|
@pytest.mark.asyncio
|
|
async def test_final_report_reuses_exact_saved_body_and_rejects_missing_note(tmp_path):
|
|
runner = FlashRunner.__new__(FlashRunner)
|
|
runner.ctx = SimpleNamespace(
|
|
data_engine=SimpleNamespace(base_dir=tmp_path, current_step_id="s", record_step=Mock())
|
|
)
|
|
runner.summarizer = None
|
|
runner.final_report_received = False
|
|
from artemis.utils.notes import save_verified_note
|
|
|
|
body = "# 报告\nPC 是深色；手机检查未全部覆盖。"
|
|
save_verified_note(tmp_path, "report", body)
|
|
messages = []
|
|
args = {"status": "completed", "explanation": "全部完成", "report_note_key": "missing"}
|
|
result = await runner._finalize_task_report(
|
|
"report_task_status", args, "r", "", {}, None, "", messages
|
|
)
|
|
assert result is None and not runner.final_report_received
|
|
args["report_note_key"] = "report"
|
|
result = await runner._finalize_task_report(
|
|
"report_task_status", args, "r2", "", {}, None, "", messages
|
|
)
|
|
assert result["explanation"] == body
|
|
assert (
|
|
runner.ctx.data_engine.record_step.call_args.kwargs["action_taken"]["args"]["explanation"]
|
|
== body
|
|
)
|