Consider it done. The open-source AI agent that works out of the box · 想到,就能做到。开源、开箱即用的 AI Agent。
TypeScript agentai-agentai-assistantandroidclaude-codecodexelectroniosllmmacosreact-nativetypescriptwindows
"""No compositor connection: exercise ownership, validation and rollback."""
import copy
import importlib.util
import json
import pathlib
import sys
import unittest
sys.dont_write_bytecode = True
spec = importlib.util.spec_from_file_location(
"viewer_display", pathlib.Path(__file__).with_name("linux-viewer-display.py"))
module = importlib.util.module_from_spec(spec)
spec.loader.exec_module(module)
class ViewerDisplayTests(unittest.TestCase):
def fixture(self):
source = {"name": "eDP-2", "width": 1600, "height": 1000, "refreshRate": 60,
"x": 0, "y": 0, "scale": 1, "transform": 0, "mirrorOf": "none"}
monitors = {source["name"]: source}
calls = []
def run(*args):
calls.append(args)
if args == ("-j", "monitors", "all"):
return json.dumps(list(monitors.values()))
if args == ("-j", "workspaces"):
return "[]"
if args[:2] == ("output", "create"):
monitors[args[-1]] = dict(source, name=args[-1])
if args[:2] == ("output", "remove"):
monitors.pop(args[-1], None)
return "ok"
owner = module.ViewerDisplay(source["name"], run)
def rule(value, mirror=None):
monitors[value["name"]] = dict(value, mirrorOf=mirror or "none")
owner.rule = rule
return owner, monitors, calls
def test_rejects_invalid_sizes_before_mutating_any_output(self):
owner, _, calls = self.fixture()
for width, height in [(True, 600), (319, 600), (4096, 4096), (800, "600")]:
with self.assertRaises(ValueError):
owner.resize(width, height)
self.assertFalse(any(c[0] == "output" for c in calls))
def test_moves_numbered_and_named_workspaces_in_both_directions(self):
for lua in (True, False):
owner, _, _ = self.fixture()
owner.lua = lua
workspaces = [
{"id": 2, "name": "2", "monitor": owner.source},
{"id": -1337, "name": '工作"区', "monitor": owner.source},
{"id": -99, "name": "special:scratch", "monitor": owner.source},
{"id": 3, "name": "3", "monitor": "other"},
]
calls = []
def run(*args):
if args == ("-j", "workspaces"):
return json.dumps(workspaces)
calls.append(args)
return "ok"
owner.run = run
for source, target in ((owner.source, owner.name), (owner.name, owner.source)):
owner.move_workspaces(source, target)
self.assertEqual(len(calls), 2)
if lua:
self.assertIn("workspace=2,", calls[0][1])
self.assertIn('workspace=' + json.dumps('name:工作"区', ensure_ascii=False), calls[1][1])
self.assertTrue(all('monitor=' + json.dumps(target) in c[1] for c in calls))
else:
self.assertEqual(calls, [("dispatch", "moveworkspacetomonitor", "2 " + target),
("dispatch", "moveworkspacetomonitor", 'name:工作"区 ' + target)])
for workspace in workspaces[:2]:
workspace["monitor"] = target
calls.clear()
def test_restores_exact_source_and_removes_only_owned_output(self):
owner, monitors, _ = self.fixture()
before = copy.deepcopy(monitors)
result = owner.resize(800, 600)
self.assertEqual(result["width"], 800)
self.assertEqual(monitors["eDP-2"]["mirrorOf"], owner.name)
owner.resize(600, 800)
owner.restore()
self.assertEqual(monitors, before)
owner.restore() # Idempotent cleanup.
def test_cleanup_keeps_snapshot_after_a_temporary_restore_failure(self):
owner, monitors, _ = self.fixture()
owner.resize(800, 600)
rule = owner.rule
owner.rule = lambda *_: (_ for _ in ()).throw(RuntimeError("temporary"))
with self.assertRaises(RuntimeError):
owner.restore()
self.assertTrue(owner.created)
self.assertTrue(owner.mirrored)
owner.rule = rule
owner.restore()
self.assertEqual(set(monitors), {"eDP-2"})
def test_unplugged_source_does_not_block_owned_output_removal(self):
owner, monitors, _ = self.fixture()
owner.resize(800, 600)
del monitors["eDP-2"]
owner.restore()
self.assertFalse(monitors)
if __name__ == "__main__":
unittest.main()