插件配置持久化:
- plugin_settings 表新增 config_json 字段,支持存储每个插件的自定义配置
- BaseCrawlerPlugin 新增 default_config 属性和 update_config 方法
- PluginSettingsRepository 新增 get_config / set_config 方法
- PluginService 新增 get_plugin_config 和 update_plugin_config
- api/routes/plugins.py 新增 GET /{id}/config 和 POST /{id}/config 接口
- 前端 Plugins.vue 增加配置编辑对话框,支持动态渲染数字/布尔/字符串类型配置
- ip3366 插件示例化:增加 max_pages 配置项,验证配置生效后会动态更新爬取 URL
任务队列持久化:
- 新建 validation_tasks 表:id, ip, port, protocol, status, result, response_time_ms, created_at, updated_at
- 新建 ValidationTaskRepository,提供 insert_batch / acquire_pending / complete_task / reset_processing 等方法
- ValidationQueue 重构:
- submit() 时把任务写入数据库并唤醒 Worker
- Worker 通过 acquire_pending 原子取任务并验证
- 验证完成后更新任务状态并入库有效代理
- 启动时自动恢复之前中断的 processing 任务为 pending
- 支持 drain() 等待所有 pending 完成
- 调度器验证流程同样自动持久化到任务表
其他适配:
- 更新 api/deps.py 和 api/lifespan.py,移除对已删除 settings_service 的残留引用
- 更新前端 pluginService.js 和 api/index.js 增加配置相关 API
140 lines
5.3 KiB
Python
140 lines
5.3 KiB
Python
"""插件业务服务"""
|
|
from datetime import datetime
|
|
from typing import List, Optional
|
|
from core.db import get_db
|
|
from core.plugin_system.registry import registry
|
|
from core.plugin_system.base import BaseCrawlerPlugin
|
|
from repositories.settings_repo import PluginSettingsRepository
|
|
from models.domain import PluginInfo, ProxyRaw
|
|
from core.log import logger
|
|
|
|
|
|
class PluginService:
|
|
"""插件业务服务:管理插件生命周期、执行爬取、配置管理"""
|
|
|
|
def __init__(self):
|
|
self.plugin_settings_repo = PluginSettingsRepository()
|
|
self._stats: dict[str, dict] = {}
|
|
|
|
async def list_plugins(self) -> List[PluginInfo]:
|
|
"""获取所有插件信息(合并持久化状态和配置)"""
|
|
async with get_db() as db:
|
|
db_states = await self.plugin_settings_repo.list_all(db)
|
|
|
|
result = []
|
|
for plugin in registry.list_plugins():
|
|
# 合并持久化状态
|
|
state = db_states.get(plugin.name, {})
|
|
if "enabled" in state:
|
|
plugin.enabled = state["enabled"]
|
|
if "config" in state and isinstance(state["config"], dict):
|
|
plugin.update_config(state["config"])
|
|
|
|
stat = self._stats.get(plugin.name, {
|
|
"success_count": 0,
|
|
"failure_count": 0,
|
|
"last_run": None,
|
|
})
|
|
result.append(PluginInfo(
|
|
id=plugin.name,
|
|
name=plugin.name,
|
|
display_name=plugin.display_name or plugin.name,
|
|
description=plugin.description or f"从 {plugin.name} 爬取代理",
|
|
enabled=plugin.enabled,
|
|
last_run=stat.get("last_run"),
|
|
success_count=stat.get("success_count", 0),
|
|
failure_count=stat.get("failure_count", 0),
|
|
))
|
|
return result
|
|
|
|
async def toggle_plugin(self, plugin_id: str, enabled: bool) -> bool:
|
|
plugin = registry.get(plugin_id)
|
|
if not plugin:
|
|
return False
|
|
async with get_db() as db:
|
|
success = await self.plugin_settings_repo.set_enabled(db, plugin_id, enabled)
|
|
if success:
|
|
plugin.enabled = enabled
|
|
logger.info(f"Plugin {plugin_id} toggled to {enabled}")
|
|
return success
|
|
|
|
async def get_plugin_config(self, plugin_id: str) -> Optional[dict]:
|
|
"""获取插件当前配置(合并默认值和持久化值)"""
|
|
plugin = registry.get(plugin_id)
|
|
if not plugin:
|
|
return None
|
|
async with get_db() as db:
|
|
saved = await self.plugin_settings_repo.get_config(db, plugin_id)
|
|
config = dict(plugin.default_config)
|
|
if saved:
|
|
config.update(saved)
|
|
return config
|
|
|
|
async def update_plugin_config(self, plugin_id: str, config: dict) -> bool:
|
|
"""更新插件配置(只保存已存在于 default_config 中的键)"""
|
|
plugin = registry.get(plugin_id)
|
|
if not plugin:
|
|
return False
|
|
# 过滤非法键
|
|
safe_config = {k: v for k, v in config.items() if k in plugin.default_config}
|
|
if not safe_config:
|
|
return False
|
|
plugin.update_config(safe_config)
|
|
async with get_db() as db:
|
|
return await self.plugin_settings_repo.set_config(db, plugin_id, plugin.config)
|
|
|
|
def get_plugin(self, plugin_id: str) -> Optional[BaseCrawlerPlugin]:
|
|
return registry.get(plugin_id)
|
|
|
|
async def run_plugin(self, plugin_id: str) -> List[ProxyRaw]:
|
|
"""执行单个插件爬取"""
|
|
plugin = self.get_plugin(plugin_id)
|
|
if not plugin:
|
|
raise ValueError(f"Plugin {plugin_id} not found")
|
|
if not plugin.enabled:
|
|
logger.warning(f"Plugin {plugin_id} is disabled, skip crawl")
|
|
return []
|
|
|
|
try:
|
|
results = await plugin.crawl()
|
|
self._record_stat(plugin_id, success=len(results))
|
|
logger.info(f"Plugin {plugin_id} crawled {len(results)} proxies")
|
|
return results
|
|
except Exception as e:
|
|
self._record_stat(plugin_id, failure=1)
|
|
logger.error(f"Plugin {plugin_id} crawl failed: {e}")
|
|
return []
|
|
|
|
async def run_all_plugins(self) -> List[ProxyRaw]:
|
|
"""执行所有启用插件的爬取"""
|
|
all_results: List[ProxyRaw] = []
|
|
for plugin in registry.list_plugins():
|
|
if not plugin.enabled:
|
|
continue
|
|
try:
|
|
results = await self.run_plugin(plugin.name)
|
|
all_results.extend(results)
|
|
except Exception as e:
|
|
logger.error(f"Run all plugins error at {plugin.name}: {e}")
|
|
# 去重
|
|
seen = set()
|
|
unique = []
|
|
for p in all_results:
|
|
key = (p.ip, p.port, p.protocol)
|
|
if key not in seen:
|
|
seen.add(key)
|
|
unique.append(p)
|
|
return unique
|
|
|
|
def _record_stat(self, plugin_id: str, success: int = 0, failure: int = 0):
|
|
if plugin_id not in self._stats:
|
|
self._stats[plugin_id] = {
|
|
"success_count": 0,
|
|
"failure_count": 0,
|
|
"last_run": None,
|
|
}
|
|
self._stats[plugin_id]["success_count"] += success
|
|
self._stats[plugin_id]["failure_count"] += failure
|
|
if success or failure:
|
|
self._stats[plugin_id]["last_run"] = datetime.now()
|