fix: 全面修复代码问题并优化架构

修复问题:
- 添加缺失的 httpx 依赖到 requirements.txt
- 修复前端批量删除参数格式与后端不匹配(数组->对象数组)
- 移除 app/api/main.py 中重复创建 app 的冗余代码
- 修复 Plugins.vue v-model 直接修改 store 状态的 Vue 警告
- 修复 README 端口/启动命令文档与实际配置不一致
- 修正 pytest.ini 过时配置 (asyncio_default_fixture_loop_scope)
- 修复 WebUI index.html 语言设置为 zh-CN
- 修复 .gitignore 错误忽略 tests/ 目录

后端优化:
- 修复调度器默认间隔从 5 秒改为 30 分钟,避免无节制验证
- 修复 validate_all_now 在调度器停止时无法执行的 bug
- 设置保存后热更新运行中调度器的验证间隔
- 将 update_score 优化为原子单事务 SQL,消除并发竞态
- 导出功能改为真正的流式分批读取(iter_batches),降低大导出内存占用
- ProxyResponse Schema 补齐 response_time_ms 字段
- 日志级别改为从配置动态读取,不再硬编码 INFO
- 清理 validator_service 中的冗余 try/finally 代码

插件健壮性:
- 修复 ip3366/ip89/kuaidaili/proxylist_download/speedx/yundaili/proxyscrape
  的端口范围检查和 IPv6 地址解析问题(改用 rsplit + 1-65535 校验)
- 修复 PluginService.list_plugins 并发竞争条件
- 修复 run_all_plugins 去重逻辑与数据库 UNIQUE 约束保持一致
- 修复 proxyscrape 异常时错误跳过 fallback 的 bug

测试:
- 新增 7 个插件解析单元测试
- 新增 update_score 自动删除和 iter_batches 流式读取测试
- 全部 74 个测试通过
This commit is contained in:
祀梦
2026-04-04 21:03:43 +08:00
parent 875e61f17e
commit 4ef7931941
27 changed files with 212 additions and 113 deletions

View File

@@ -24,10 +24,9 @@ class PluginService:
result = []
for plugin in registry.list_plugins():
# 合并持久化状态
# 合并持久化状态(不修改全局实例,避免并发竞争)
state = db_states.get(plugin.name, {})
if "enabled" in state:
plugin.enabled = state["enabled"]
enabled = state.get("enabled", plugin.enabled)
if "config" in state and isinstance(state["config"], dict):
plugin.update_config(state["config"])
@@ -50,7 +49,7 @@ class PluginService:
name=plugin.name,
display_name=plugin.display_name or plugin.name,
description=plugin.description or f"{plugin.name} 爬取代理",
enabled=plugin.enabled,
enabled=enabled,
last_run=stat.get("last_run"),
success_count=stat.get("success_count", 0),
failure_count=stat.get("failure_count", 0),
@@ -133,11 +132,11 @@ class PluginService:
logger.error(f"Run all plugins error: {results}")
continue
all_results.extend(results)
# 去重
# 去重(与数据库 UNIQUE(ip, port) 约束保持一致)
seen = set()
unique = []
for p in all_results:
key = (p.ip, p.port, p.protocol)
key = (p.ip, p.port)
if key not in seen:
seen.add(key)
unique.append(p)

View File

@@ -61,28 +61,41 @@ class ProxyService:
protocol: Optional[str] = None,
limit: int = 10000,
) -> AsyncIterator[str]:
async with get_db() as db:
proxies = await self.proxy_repo.list_all(db, protocol=protocol, limit=limit)
if fmt == "csv":
yield "IP,Port,Protocol,Score,Last Check\n"
for p in proxies:
yield f"{p.ip},{p.port},{p.protocol},{p.score},{self._fmt_time(p.last_check)}\n"
yield "\ufeffIP,Port,Protocol,Score,Last Check\n"
elif fmt == "txt":
for p in proxies:
yield f"{p.ip}:{p.port}\n"
pass
elif fmt == "json":
data = [
{
"ip": p.ip,
"port": p.port,
"protocol": p.protocol,
"score": p.score,
"last_check": self._fmt_time(p.last_check),
}
for p in proxies
]
yield json.dumps(data, ensure_ascii=False, indent=2)
yield "["
first = True
exported = 0
async with get_db() as db:
async for batch in self.proxy_repo.iter_batches(db, protocol=protocol, batch_size=1000):
for p in batch:
if exported >= limit:
break
if fmt == "csv":
yield f"{p.ip},{p.port},{p.protocol},{p.score},{self._fmt_time(p.last_check)}\n"
elif fmt == "txt":
yield f"{p.ip}:{p.port}\n"
elif fmt == "json":
item = {
"ip": p.ip,
"port": p.port,
"protocol": p.protocol,
"score": p.score,
"last_check": self._fmt_time(p.last_check),
}
prefix = "" if first else ","
yield prefix + json.dumps(item, ensure_ascii=False)
first = False
exported += 1
if exported >= limit:
break
if fmt == "json":
yield "]"
@staticmethod
def _fmt_time(dt: Optional[datetime]) -> str:

View File

@@ -8,6 +8,7 @@ from app.core.tasks.queue import ValidationQueue
from app.core.config import settings as app_settings
from app.core.log import logger
from app.models.domain import ProxyRaw
from app.services.task_service import task_service
class SchedulerService:
@@ -58,16 +59,15 @@ class SchedulerService:
"""立即执行一次全量验证(后台运行,不阻塞)"""
if self._validate_task and not self._validate_task.done():
return
self._validate_task = asyncio.create_task(self._do_validate_all())
self._validate_task = asyncio.create_task(self._do_validate_all(from_loop=False))
async def _run_loop(self):
"""定时循环"""
while self.running:
try:
# 清理过期任务,防止内存无限增长
from app.services.task_service import task_service
task_service.cleanup_old_tasks()
await self._do_validate_all()
await self._do_validate_all(from_loop=True)
except Exception as e:
logger.error(f"Scheduler loop error: {e}", exc_info=True)
# 等待下一次
@@ -76,10 +76,17 @@ class SchedulerService:
except asyncio.TimeoutError:
pass
async def _do_validate_all(self):
async def _do_validate_all(self, from_loop: bool = True):
"""验证数据库中所有存量代理"""
queue_started_here = False
try:
logger.info("Starting scheduled validation for all proxies")
# 如果队列未运行,临时启动它(适用于 validate_all_now 在调度器停止时调用)
if not self.validation_queue._running:
await self.validation_queue.start()
queue_started_here = True
async with get_db() as db:
# 清理 7 天前的验证任务记录,防止表无限增长
cleaned = await ValidationTaskRepository.cleanup_old(db, days=7)
@@ -95,7 +102,7 @@ class SchedulerService:
batch_size = 100
total_batches = (len(proxies) - 1) // batch_size + 1
for i in range(0, len(proxies), batch_size):
if not self.running:
if from_loop and not self.running:
break
batch = proxies[i : i + batch_size]
await self.validation_queue.submit([
@@ -106,3 +113,6 @@ class SchedulerService:
logger.info("Scheduled validation batches submitted")
except Exception as e:
logger.error(f"Scheduled validation error: {e}", exc_info=True)
finally:
if queue_started_here:
await self.validation_queue.stop()

View File

@@ -95,17 +95,13 @@ class ValidatorService:
timeout = aiohttp.ClientTimeout(total=self.timeout, connect=self.connect_timeout)
test_url = self._get_test_url("http")
try:
async with aiohttp.ClientSession(connector=connector, timeout=timeout) as session:
async with session.get(test_url, allow_redirects=True) as response:
if response.status in (200, 301, 302):
latency = round((time.time() - start) * 1000, 2)
logger.info(f"SOCKS valid: {ip}:{port} ({protocol}) {latency}ms")
return True, latency
return False, 0.0
finally:
# ClientSession 的 async with 退出时会自动关闭 connector无需手动重复关闭
pass
async with aiohttp.ClientSession(connector=connector, timeout=timeout) as session:
async with session.get(test_url, allow_redirects=True) as response:
if response.status in (200, 301, 302):
latency = round((time.time() - start) * 1000, 2)
logger.info(f"SOCKS valid: {ip}:{port} ({protocol}) {latency}ms")
return True, latency
return False, 0.0
async def close(self):
"""关闭共享的 HTTP ClientSession"""