fix: harden scheduling and deployment flow

This commit is contained in:
Rixuan Shao
2026-07-11 14:58:05 +08:00
parent c496d90039
commit e3142cabbf
28 changed files with 552 additions and 1750 deletions
@@ -0,0 +1,37 @@
import json
import os
import tempfile
import unittest
from pathlib import Path
from unittest.mock import patch
from core import tasks
from utils import config as config_module
class ConfigContractTests(unittest.TestCase):
def test_default_config_matches_public_example(self):
example_path = Path(config_module.__file__).resolve().parents[1] / "config.example.json"
example = json.loads(example_path.read_text(encoding="utf-8"))
self.assertEqual(example, config_module.DEFAULT_CONFIG)
def test_missing_runtime_config_is_created_from_safe_defaults(self):
with tempfile.TemporaryDirectory() as temp_dir:
path = Path(temp_dir) / "config.json"
loaded = config_module._load_json_file(path, config_module.DEFAULT_CONFIG)
self.assertEqual(config_module.DEFAULT_CONFIG, loaded)
self.assertEqual(
config_module.DEFAULT_CONFIG,
json.loads(path.read_text(encoding="utf-8")),
)
self.assertFalse(loaded["useProtocolSender"])
self.assertTrue(loaded["persistentBrowserProfiles"]["enabled"])
def test_profile_root_environment_override_wins(self):
with patch.dict(os.environ, {"SPARKFLOW_BROWSER_PROFILE_ROOT": "/tmp/sparkflow-profiles"}):
normalized = tasks._normalize_persistent_profile_config(config_module.DEFAULT_CONFIG)
self.assertEqual("/tmp/sparkflow-profiles", normalized["root"])
if __name__ == "__main__":
unittest.main()
@@ -0,0 +1,103 @@
import unittest
from pathlib import Path
REPO_ROOT = Path(__file__).resolve().parents[2]
SOURCE_ROOT = REPO_ROOT / "DouYinSparkFlow"
class DeploymentContractTests(unittest.TestCase):
def test_github_workflow_is_at_repository_root(self):
workflow = REPO_ROOT / ".github" / "workflows" / "schedule.yml"
self.assertTrue(workflow.is_file())
self.assertFalse((SOURCE_ROOT / ".github" / "workflows" / "schedule.yml").exists())
text = workflow.read_text(encoding="utf-8")
self.assertIn("working-directory: DouYinSparkFlow", text)
self.assertIn("SPARKFLOW_BROWSER_PROFILE_ROOT", text)
self.assertIn("SPARKFLOW_MANUAL_RUN", text)
self.assertIn("path: DouYinSparkFlow/logs/", text)
def test_github_actions_are_pinned_to_commit_shas(self):
import re
workflow = (REPO_ROOT / ".github" / "workflows" / "schedule.yml").read_text(encoding="utf-8")
uses_values = re.findall(r"^\s*-?\s*uses:\s*([^#\s]+)", workflow, flags=re.MULTILINE)
self.assertTrue(uses_values)
for value in uses_values:
self.assertRegex(value, r"^[^@]+@[0-9a-f]{40}$")
def test_runtime_config_is_not_tracked_as_the_template(self):
self.assertTrue((SOURCE_ROOT / "config.example.json").is_file())
self.assertIn("config.json", (SOURCE_ROOT / ".gitignore").read_text(encoding="utf-8"))
def test_compose_runtime_mounts_follow_least_privilege(self):
text = (REPO_ROOT / "docker-compose.yml").read_text(encoding="utf-8")
web = text.split(" web:", 1)[1].split("\n login-desktop:", 1)[0]
scheduler = text.split(" scheduler:", 1)[1].split("\n task:", 1)[0]
task = text.split(" task:", 1)[1]
self.assertIn("/var/run/docker.sock:/var/run/docker.sock", web)
self.assertIn(".:/opt/douyin-sparkflow", web)
self.assertIn("/var/run/docker.sock:/var/run/docker.sock", scheduler)
self.assertNotIn(".:/opt/douyin-sparkflow", scheduler)
self.assertNotIn("/var/run/docker.sock:/var/run/docker.sock", task)
self.assertNotIn(".:/opt/douyin-sparkflow", task)
for service in (scheduler, task):
self.assertIn(
"./state/browser-profiles:/opt/douyin-sparkflow/state/browser-profiles",
service,
)
def test_cron_reader_accepts_windows_utf8_bom(self):
import tempfile
from scripts.cron_runner import read_crontab
with tempfile.TemporaryDirectory() as temp_dir:
path = Path(temp_dir) / "root"
path.write_text(
"*/20 10-17 * * * cd /app && python main.py --doTask\n",
encoding="utf-8-sig",
)
lines = read_crontab(path)
self.assertEqual(len(lines), 1)
self.assertTrue(lines[0].startswith("*/20 "))
def test_sensitive_ports_bind_to_loopback_by_default(self):
text = (REPO_ROOT / "docker-compose.yml").read_text(encoding="utf-8")
self.assertIn("${PROXY_BIND_ADDRESS:-127.0.0.1}:${PROXY_HTTP_PORT:-7890}:7890", text)
self.assertIn(
"${LOGIN_DESKTOP_BIND_ADDRESS:-127.0.0.1}:${LOGIN_DESKTOP_WEB_PORT:-8788}:6080",
text,
)
def test_container_login_api_and_public_url_are_wired(self):
text = (REPO_ROOT / "docker-compose.yml").read_text(encoding="utf-8")
self.assertIn("SPARKFLOW_LOGIN_DESKTOP_API_URL: http://login-desktop:18090", text)
self.assertIn("SPARKFLOW_LOGIN_DESKTOP_PUBLIC_URL", text)
def test_installers_preserve_runtime_config_and_do_not_require_bash_on_windows(self):
server = (REPO_ROOT / "deploy" / "install-server.sh").read_text(encoding="utf-8")
windows = (REPO_ROOT / "deploy" / "install-local.ps1").read_text(encoding="utf-8")
self.assertIn("runtime_config_backup", server)
self.assertIn("Restored runtime config.json", server)
self.assertNotIn("bash ./refresh_proxy.sh", windows)
self.assertIn("Initialize-ProxyConfig", windows)
def test_playwright_base_image_argument_is_used(self):
dockerfile = (SOURCE_ROOT / "Dockerfile.server").read_text(encoding="utf-8")
self.assertTrue(dockerfile.startswith("ARG PLAYWRIGHT_BASE_IMAGE="))
self.assertIn("FROM ${PLAYWRIGHT_BASE_IMAGE}", dockerfile)
self.assertIn("docker.io", dockerfile)
self.assertIn("node --version", dockerfile)
self.assertNotIn("github.com/docker/compose", dockerfile)
def test_legacy_unused_entrypoints_are_removed(self):
self.assertFalse((SOURCE_ROOT / "webui" / "login_sessions.py").exists())
self.assertFalse((SOURCE_ROOT / "relogin_worker.py").exists())
self.assertFalse((SOURCE_ROOT / "docker-compose.example.yml").exists())
if __name__ == "__main__":
unittest.main()
@@ -104,6 +104,53 @@ class WebUiSafetyTests(unittest.TestCase):
self.assertEqual(200, response.status_code, path)
self.assertEqual("no-store", response.headers["cache-control"])
def test_login_desktop_urls_honor_container_environment(self):
with (
patch.dict(
os.environ,
{
"SPARKFLOW_LOGIN_DESKTOP_API_URL": "http://login-desktop:18090",
"SPARKFLOW_LOGIN_DESKTOP_PUBLIC_URL": "http://127.0.0.1:8788/vnc.html",
},
),
patch.object(app_module, "get_app_settings", return_value={}),
):
self.assertEqual("http://login-desktop:18090", app_module.login_desktop_api_url())
request = type("Request", (), {"url": type("Url", (), {"hostname": "example", "scheme": "http"})()})()
self.assertEqual(
"http://127.0.0.1:8788/vnc.html",
app_module.login_desktop_public_url(request),
)
def test_schedule_sync_writes_configured_window_to_shared_spool(self):
with tempfile.TemporaryDirectory() as temp_dir:
cron_path = Path(temp_dir) / "root"
with (
patch.object(ops, "HOST_CRONTAB_PATH", cron_path),
patch.object(ops, "running_in_container", return_value=True),
patch.object(ops, "read_crontab", return_value=""),
patch.object(
ops,
"get_config",
return_value={
"dailySendWindow": {
"enabled": True,
"startHour": 10,
"endHour": 18,
"scheduleIntervalMinutes": 20,
}
},
),
):
result = ops.sync_daily_schedule_from_config()
self.assertEqual(0, result.returncode)
text = cron_path.read_text(encoding="utf-8")
self.assertIn("*/20 10-17 * * *", text)
self.assertIn("0 18 * * *", text)
self.assertIn("20 18 * * *", text)
self.assertIn("docker exec", text)
def test_overview_api_requires_authentication_and_disables_cache(self):
client = TestClient(app_module.app)
response = client.get("/api/ops/overview")