task_handler.py 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442
  1. from datetime import datetime
  2. from typing import Callable, Dict, Optional
  3. from app.core.config import GlobalConfigSettings
  4. from app.domains.analysis_task import CrawlerDetailDeal
  5. from app.domains.analysis_task import AccountPositionReadRateAvg
  6. from app.domains.analysis_task import AccountPositionReadAvg
  7. from app.domains.analysis_task import AccountPositionOpenRateAvg
  8. from app.domains.analysis_task import RateLimitedArticleFilter
  9. from app.domains.algorithm_tasks import AccountCategoryAnalysis
  10. from app.domains.cold_start_tasks import ArticlePoolColdStart
  11. from app.domains.crawler_tasks import CrawlerToutiao
  12. from app.domains.crawler_tasks import WeixinAccountManager
  13. from app.domains.crawler_tasks import CrawlerGzhAccountArticles
  14. from app.domains.crawler_tasks import CrawlerGzhSearchArticles
  15. from app.domains.data_recycle_tasks import ArticleDetailStat
  16. from app.domains.data_recycle_tasks import RecycleDailyPublishArticlesTask
  17. from app.domains.data_recycle_tasks import RecycleOutsideAccountArticlesTask
  18. from app.domains.data_recycle_tasks import CheckDailyPublishArticlesTask
  19. from app.domains.data_recycle_tasks import UpdateRootSourceIdAndUpdateTimeTask
  20. from app.domains.data_recycle_tasks import RecycleFwhDailyPublishArticlesTask
  21. from app.domains.data_recycle_tasks import RecycleMiniProgramDetailTask
  22. from app.domains.data_recycle_tasks import (
  23. UpdateOutsideRootSourceIdAndUpdateTimeTask,
  24. )
  25. from app.domains.llm_tasks import TitleRewrite
  26. from app.domains.llm_tasks import ArticlePoolCategoryGeneration
  27. from app.domains.llm_tasks import CandidateAccountQualityScoreRecognizer
  28. from app.domains.llm_tasks import ExtractTitleFeatures
  29. from app.domains.monitor_tasks import AutoReplyCardsMonitor
  30. from app.domains.monitor_tasks import check_kimi_balance
  31. from app.domains.monitor_tasks import GetOffVideos
  32. from app.domains.monitor_tasks import CheckVideoAuditStatus
  33. from app.domains.monitor_tasks import CooperateAccountsMonitorTask
  34. from app.domains.monitor_tasks import InnerGzhArticlesMonitor
  35. from app.domains.monitor_tasks import OutsideGzhArticlesMonitor
  36. from app.domains.monitor_tasks import OutsideGzhArticlesCollector
  37. from app.domains.monitor_tasks import TaskProcessingMonitor
  38. from app.domains.monitor_tasks import LimitedAccountAnalysisTask
  39. from app.jobs.task_config import TaskStatus
  40. from app.jobs.task_utils import TaskValidationError
  41. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  42. def register(task_name: str):
  43. def decorator(func):
  44. _TASK_HANDLER_REGISTRY[task_name] = func
  45. return func
  46. return decorator
  47. class TaskHandler:
  48. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  49. # 任务注册表
  50. _handlers = _TASK_HANDLER_REGISTRY
  51. def __init__(
  52. self,
  53. data: dict,
  54. log_service,
  55. db_client,
  56. trace_id: str,
  57. config: GlobalConfigSettings,
  58. ):
  59. self.data = data
  60. self.log_client = log_service
  61. self.db_client = db_client
  62. self.trace_id = trace_id
  63. self.config = config
  64. @classmethod
  65. def get_handler(cls, task_name: str) -> Optional[Callable]:
  66. """获取任务处理器"""
  67. return cls._handlers.get(task_name)
  68. @classmethod
  69. def list_registered_tasks(cls) -> list:
  70. """列出所有已注册的任务"""
  71. return list(cls._handlers.keys())
  72. async def _log_task_event(self, event_type: str, **kwargs):
  73. """统一的任务日志记录"""
  74. log_data = {
  75. "timestamp": datetime.now().isoformat(),
  76. "trace_id": self.trace_id,
  77. "event_type": event_type,
  78. "task": self.data.get("task_name"),
  79. **kwargs,
  80. }
  81. await self.log_client.log(contents=log_data)
  82. # ==================== 监控类任务 ====================
  83. @register("check_kimi_balance")
  84. async def _check_kimi_balance_handler(self) -> int:
  85. """检查 Kimi 余额"""
  86. response = await check_kimi_balance()
  87. await self._log_task_event("kimi_balance_checked", data=response)
  88. return TaskStatus.SUCCESS
  89. @register("get_off_videos")
  90. async def _get_off_videos_task_handler(self) -> int:
  91. """视频下架任务"""
  92. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  93. return await sub_task.deal()
  94. @register("check_publish_video_audit_status")
  95. async def _check_video_audit_status_handler(self) -> int:
  96. """检查视频审核状态"""
  97. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  98. return await sub_task.deal()
  99. @register("task_processing_monitor")
  100. async def _task_processing_monitor_handler(self) -> int:
  101. """任务处理监控"""
  102. sub_task = TaskProcessingMonitor(self.db_client)
  103. await sub_task.deal()
  104. return TaskStatus.SUCCESS
  105. @register("inner_article_monitor")
  106. async def _inner_gzh_articles_monitor_handler(self) -> int:
  107. """内部公众号文章监控"""
  108. sub_task = InnerGzhArticlesMonitor(self.db_client)
  109. return await sub_task.deal()
  110. @register("outside_article_monitor")
  111. async def _outside_monitor_handler(self) -> int:
  112. """外部文章监控"""
  113. collector = OutsideGzhArticlesCollector(self.db_client)
  114. await collector.deal()
  115. monitor = OutsideGzhArticlesMonitor(self.db_client)
  116. return await monitor.deal()
  117. @register("cooperate_accounts_monitor")
  118. async def _cooperate_accounts_monitor_handler(self) -> int:
  119. """合作账号文章监测"""
  120. task = CooperateAccountsMonitorTask(
  121. pool=self.db_client, log_client=self.log_client
  122. )
  123. await task.deal(task_name="save_articles")
  124. return TaskStatus.SUCCESS
  125. @register("cooperate_accounts_detail")
  126. async def _cooperate_accounts_detail_handler(self) -> int:
  127. """合作账号文章详情更新"""
  128. task = CooperateAccountsMonitorTask(
  129. pool=self.db_client, log_client=self.log_client
  130. )
  131. await task.deal(task_name="get_detail")
  132. return TaskStatus.SUCCESS
  133. # ==================== 爬虫类任务 ====================
  134. @register("crawler_toutiao")
  135. async def _crawler_toutiao_handler(self) -> int:
  136. """头条文章/视频抓取"""
  137. sub_task = CrawlerToutiao(
  138. self.db_client, self.log_client, self.trace_id, self.config
  139. )
  140. method = self.data.get("method", "account")
  141. media_type = self.data.get("media_type", "article")
  142. category_list = self.data.get("category_list", [])
  143. match method:
  144. case "account":
  145. await sub_task.crawler_task(media_type=media_type)
  146. case "recommend":
  147. await sub_task.crawl_toutiao_recommend_task(category_list)
  148. case "search":
  149. await sub_task.search_candidate_accounts()
  150. case _:
  151. raise TaskValidationError(f"Unsupported method: {method}")
  152. return TaskStatus.SUCCESS
  153. @register("crawler_gzh_articles")
  154. async def _crawler_gzh_article_handler(self) -> int:
  155. """抓取公众号文章"""
  156. account_method = self.data.get("account_method")
  157. crawl_mode = self.data.get("crawl_mode")
  158. strategy = self.data.get("strategy")
  159. match crawl_mode:
  160. case "account":
  161. task = CrawlerGzhAccountArticles(
  162. self.db_client, self.log_client, self.trace_id, self.config
  163. )
  164. await task.deal(account_method, strategy)
  165. case "search":
  166. task = CrawlerGzhSearchArticles(
  167. self.db_client, self.log_client, self.trace_id, self.config
  168. )
  169. await task.deal(strategy)
  170. case _:
  171. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  172. return TaskStatus.SUCCESS
  173. @register("crawler_account_manager")
  174. async def _crawler_account_manager_handler(self) -> int:
  175. """账号管理任务"""
  176. platform = self.data.get("platform", "weixin")
  177. account_id_list = self.data.get("account_id_list")
  178. match platform:
  179. case "weixin":
  180. task = WeixinAccountManager(
  181. self.db_client, self.log_client, self.trace_id
  182. )
  183. await task.deal(platform=platform, account_id_list=account_id_list)
  184. case _:
  185. raise TaskValidationError(f"Unsupported platform: {platform}")
  186. return TaskStatus.SUCCESS
  187. @register("crawler_detail_analysis")
  188. async def _crawler_article_analysis_handler(self) -> int:
  189. """抓取视频/文章详情分析统计"""
  190. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  191. await task.deal(params=self.data)
  192. return TaskStatus.SUCCESS
  193. # ==================== 数据回收类任务 ====================
  194. @register("article_detail_stat")
  195. async def _article_detail_stat_handler(self) -> int:
  196. """文章详情统计"""
  197. task = ArticleDetailStat(self.db_client, self.log_client)
  198. await task.deal()
  199. return TaskStatus.SUCCESS
  200. @register("daily_publish_articles_recycle")
  201. async def _recycle_article_data_handler(self) -> int:
  202. """每日发文数据回收"""
  203. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  204. recycle = RecycleDailyPublishArticlesTask(
  205. self.db_client, self.log_client, date_str
  206. )
  207. await recycle.deal()
  208. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  209. await check.deal()
  210. return TaskStatus.SUCCESS
  211. @register("update_root_source_id")
  212. async def _update_root_source_id_handler(self) -> int:
  213. """更新 root_source_id"""
  214. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  215. await sub_task.deal()
  216. return TaskStatus.SUCCESS
  217. @register("recycle_outside_account_articles")
  218. async def _recycle_outside_account_article_handler(self) -> int:
  219. """回收外部账号文章"""
  220. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  221. task = RecycleOutsideAccountArticlesTask(
  222. pool=self.db_client, log_client=self.log_client, date_string=date_str
  223. )
  224. await task.deal()
  225. return TaskStatus.SUCCESS
  226. @register("update_outside_account_article_root_source_id")
  227. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  228. """更新外部账号文章的 root_source_id"""
  229. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  230. pool=self.db_client, log_client=self.log_client
  231. )
  232. await task.deal()
  233. return TaskStatus.SUCCESS
  234. @register("fwh_daily_recycle")
  235. async def _recycle_fwh_article_handler(self) -> int:
  236. """回收服务号文章"""
  237. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  238. await task.deal()
  239. return TaskStatus.SUCCESS
  240. # ==================== 算法类任务 ====================
  241. @register("article_pool_cold_start")
  242. async def _article_pool_cold_start_handler(self) -> int:
  243. """文章池冷启动"""
  244. cold_start = ArticlePoolColdStart(
  245. self.db_client, self.log_client, self.trace_id, self.config
  246. )
  247. platform = self.data.get("platform", "weixin")
  248. crawler_methods = self.data.get("crawler_methods", [])
  249. category_list = self.data.get("category_list", [])
  250. strategy = self.data.get("strategy", "strategy_v1")
  251. await cold_start.deal(
  252. platform=platform,
  253. crawl_methods=crawler_methods,
  254. category_list=category_list,
  255. strategy=strategy,
  256. )
  257. return TaskStatus.SUCCESS
  258. @register("candidate_account_quality_analysis")
  259. async def _candidate_account_quality_score_handler(self) -> int:
  260. """候选账号质量分析"""
  261. task = CandidateAccountQualityScoreRecognizer(
  262. self.db_client, self.log_client, self.trace_id
  263. )
  264. await task.deal()
  265. return TaskStatus.SUCCESS
  266. @register("article_pool_category_generation")
  267. async def _article_pool_category_generation_handler(self) -> int:
  268. """文章池品类生成"""
  269. task = ArticlePoolCategoryGeneration(
  270. self.db_client, self.log_client, self.trace_id
  271. )
  272. limit_num = self.data.get("limit")
  273. await task.deal(limit=limit_num)
  274. return TaskStatus.SUCCESS
  275. @register("account_category_analysis")
  276. async def _account_category_analysis_handler(self) -> int:
  277. """账号品类分析"""
  278. task = AccountCategoryAnalysis(
  279. pool=self.db_client,
  280. log_client=self.log_client,
  281. trace_id=self.trace_id,
  282. data=self.data,
  283. date_string=None,
  284. )
  285. await task.deal()
  286. return TaskStatus.SUCCESS
  287. @register("update_limited_account_info")
  288. async def _update_limited_account_info_handler(self) -> int:
  289. """更新限流账号信息"""
  290. task = LimitedAccountAnalysisTask(
  291. pool=self.db_client, log_client=self.log_client
  292. )
  293. await task.deal(date_string=self.data.get("date_string"))
  294. return TaskStatus.SUCCESS
  295. # ==================== LLM 类任务 ====================
  296. @register("title_rewrite")
  297. async def _title_rewrite_handler(self) -> int:
  298. """标题重写"""
  299. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  300. return await sub_task.deal()
  301. @register("extract_title_features")
  302. async def _extract_title_features_handler(self) -> int:
  303. """提取标题特征"""
  304. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  305. await task.deal(data=self.data)
  306. return TaskStatus.SUCCESS
  307. # ==================== 统计分析类任务 ====================
  308. @register("update_account_read_rate_avg")
  309. async def _update_account_read_rate_avg_handler(self) -> int:
  310. """更新账号阅读率均值"""
  311. task = AccountPositionReadRateAvg(
  312. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  313. )
  314. await task.deal(end_date=self.data.get("end_date"))
  315. return TaskStatus.SUCCESS
  316. @register("update_account_read_avg")
  317. async def _update_account_read_avg_handler(self) -> int:
  318. """更新账号阅读均值"""
  319. task = AccountPositionReadAvg(
  320. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  321. )
  322. await task.deal(end_date=self.data.get("end_date"))
  323. return TaskStatus.SUCCESS
  324. @register("update_account_open_rate_avg")
  325. async def _update_account_open_rate_avg_handler(self) -> int:
  326. """更新账号打开率均值"""
  327. task = AccountPositionOpenRateAvg(
  328. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  329. )
  330. await task.deal(date_string=self.data.get("date_string"))
  331. return TaskStatus.SUCCESS
  332. # ==================== 自动化任务 ====================
  333. @register("auto_follow_account")
  334. async def _auto_follow_account_handler(self) -> int:
  335. """自动关注公众号"""
  336. task = AutoReplyCardsMonitor(
  337. pool=self.db_client, log_service=self.log_client, config=self.config
  338. )
  339. await task.deal(task_name="follow_gzh_task")
  340. return TaskStatus.SUCCESS
  341. @register("get_follow_result")
  342. async def _get_follow_result_handler(self) -> int:
  343. """获取自动关注回复"""
  344. task = AutoReplyCardsMonitor(
  345. pool=self.db_client, log_service=self.log_client, config=self.config
  346. )
  347. await task.deal(task_name="get_auto_reply_task")
  348. return TaskStatus.SUCCESS
  349. @register("extract_reply_result")
  350. async def _extract_reply_result_handler(self) -> int:
  351. """解析自动回复结果"""
  352. task = AutoReplyCardsMonitor(
  353. pool=self.db_client, log_service=self.log_client, config=self.config
  354. )
  355. await task.deal(task_name="extract_task")
  356. return TaskStatus.SUCCESS
  357. # ==================== 其他任务 ====================
  358. @register("mini_program_detail_process")
  359. async def _mini_program_detail_handler(self) -> int:
  360. """更新小程序裂变信息"""
  361. task = RecycleMiniProgramDetailTask(
  362. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  363. )
  364. await task.deal(params=self.data)
  365. return TaskStatus.SUCCESS
  366. @register("rate_limited_article_filter")
  367. async def _rate_limited_article_filter(self) -> int:
  368. """限流文章删除"""
  369. task = RateLimitedArticleFilter(pool=self.db_client, config=self.config)
  370. await task.deal()
  371. return TaskStatus.SUCCESS
  372. __all__ = ["TaskHandler"]