JiuwenSwarm is an intelligent AI Agent built on openJiuwen. It extends the powerful capabilities of large language models directly to your fingertips through various communication apps you use daily.
Python
# Copyright (c) Huawei Technologies Co., Ltd. 2026. All rights reserved.
"""Workflow checkpoint persistence unit tests."""
from __future__ import annotations
import asyncio
from typing import Any, AsyncIterator
import pytest
from jiuwenswarm.agents.harness.team.handlers.workflow_state import (
WorkflowRunState,
WorkflowProgress,
)
from jiuwenswarm.agents.harness.team.handlers.workflow_monitor_handler import WorkflowMonitorHandler
_DEFAULT_RUN_ID = "wf_testrun00001"
def _make_progress(kind: str, **kwargs: Any) -> WorkflowProgress:
if "run_id" not in kwargs:
kwargs["run_id"] = _DEFAULT_RUN_ID
return WorkflowProgress(kind=kind, **kwargs)
class _FakeTeamMonitor:
"""Minimal TeamMonitor stand-in for unit tests."""
def __init__(self) -> None:
self._queue: list[object] = []
self.started = False
self.stopped = False
async def start(self) -> None:
self.started = True
async def stop(self) -> None:
self.stopped = True
self._queue.append(None)
async def workflow_events(self) -> AsyncIterator[object]:
while self._queue:
event = self._queue.pop(0)
if event is None:
break
yield event
def put_event(self, event: object) -> None:
self._queue.append(event)
async def drain(self) -> None:
"""Wait until all injected events have been consumed."""
while self._queue:
await asyncio.sleep(0)
# ---------------------------------------------------------------------------
# Round-trip tests for WorkflowRunState model_dump / model_validate
# ---------------------------------------------------------------------------
class TestWorkflowCheckpointRoundTrip:
@staticmethod
def test_model_dump_and_restore_preserves_state():
state = WorkflowRunState()
state.apply(_make_progress("workflow_started", workflow_name="test"))
state.apply(_make_progress("agent_started", phase="Phase 1", label="agent-a", prompt="prompt"))
runs_data = {state.id: state.model_dump()}
restored = {
k: WorkflowRunState.model_validate(v)
for k, v in runs_data.items()
}
assert len(restored) == 1
r = restored[state.id]
assert r.id == state.id
assert r.name == "test"
assert r.status == "running"
assert len(r.phases) == 1
assert r.phases[0].name == "Phase 1"
@staticmethod
def test_terminal_state_survives_round_trip():
state = WorkflowRunState()
state.apply(_make_progress("workflow_started", workflow_name="test"))
state.apply(_make_progress("workflow_completed", text="done"))
runs_data = {state.id: state.model_dump()}
restored = {
k: WorkflowRunState.model_validate(v)
for k, v in runs_data.items()
}
r = restored[state.id]
assert r.status == "completed"
assert r.is_terminal is True
# ---------------------------------------------------------------------------
# WorkflowMonitorHandler initial_runs (interrupt recovery)
# ---------------------------------------------------------------------------
class TestWorkflowMonitorHandlerInitialRuns:
@staticmethod
def test_handler_starts_with_initial_runs() -> None:
"""Handler initialised with restored runs preserves them."""
state = WorkflowRunState()
state.apply(_make_progress("workflow_started", workflow_name="restored-flow"))
state.apply(_make_progress("agent_started", phase="Planning", label="agent-a"))
monitor = _FakeTeamMonitor()
handler = WorkflowMonitorHandler(
monitor=monitor, session_id="sess-1",
initial_runs={state.id: state},
)
snapshot = handler.get_workflow_snapshot()
assert len(snapshot) == 1
assert snapshot[0]["name"] == "restored-flow"
assert snapshot[0]["status"] == "running"
assert len(snapshot[0]["phases"]) == 1
@staticmethod
def test_handler_without_initial_runs_starts_empty() -> None:
monitor = _FakeTeamMonitor()
handler = WorkflowMonitorHandler(monitor=monitor, session_id="sess-1")
assert handler.get_workflow_snapshot() == []
@pytest.mark.anyio
async def test_initial_runs_plus_new_events_merge(self) -> None:
"""Restored runs + new events from stream produce combined snapshot."""
state = WorkflowRunState()
state.apply(_make_progress("workflow_started", workflow_name="old-flow"))
state.apply(_make_progress("agent_started", phase="Planning", label="agent-a"))
monitor = _FakeTeamMonitor()
handler = WorkflowMonitorHandler(
monitor=monitor, session_id="sess-1",
initial_runs={state.id: state},
)
await handler.start()
# Push a new agent_started event for the existing run (phase via agent event)
from types import SimpleNamespace
new_progress = WorkflowProgress(
kind="agent_started",
run_id=state.id,
phase="NewPhase",
label="agent-b",
)
monitor.put_event(SimpleNamespace(
event_type=SimpleNamespace(value="workflow_progress"),
sender_id="swarmflow",
payload=new_progress,
get_payload=lambda: new_progress,
))
await monitor.drain()
await monitor.stop()
await handler.stop()
snapshot = handler.get_workflow_snapshot()
assert len(snapshot) == 1
phase_names = {p["name"] for p in snapshot[0]["phases"]}
assert "Planning" in phase_names
assert "NewPhase" in phase_names