task_handler.py 17 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476
  1. from datetime import datetime
  2. from typing import Callable, Dict, Optional
  3. from app.core.config import GlobalConfigSettings
  4. from app.jobs.domains import *
  5. from app.jobs.task_config import TaskStatus
  6. from app.jobs.task_utils import TaskValidationError
  7. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  8. def register(task_name: str):
  9. def decorator(func):
  10. _TASK_HANDLER_REGISTRY[task_name] = func
  11. return func
  12. return decorator
  13. class TaskHandler:
  14. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  15. # 任务注册表
  16. _handlers = _TASK_HANDLER_REGISTRY
  17. def __init__(
  18. self,
  19. data: dict,
  20. log_service,
  21. db_client,
  22. trace_id: str,
  23. config: GlobalConfigSettings,
  24. ):
  25. self.data = data
  26. self.log_client = log_service
  27. self.db_client = db_client
  28. self.trace_id = trace_id
  29. self.config = config
  30. @classmethod
  31. def get_handler(cls, task_name: str) -> Optional[Callable]:
  32. """获取任务处理器"""
  33. return cls._handlers.get(task_name)
  34. @classmethod
  35. def list_registered_tasks(cls) -> list:
  36. """列出所有已注册的任务"""
  37. return list(cls._handlers.keys())
  38. async def _log_task_event(self, event_type: str, **kwargs):
  39. """统一的任务日志记录"""
  40. log_data = {
  41. "timestamp": datetime.now().isoformat(),
  42. "trace_id": self.trace_id,
  43. "event_type": event_type,
  44. "task": self.data.get("task_name"),
  45. **kwargs,
  46. }
  47. await self.log_client.log(contents=log_data)
  48. # ==================== 监控类任务 ====================
  49. @register("rank_log_monitor")
  50. async def _rank_log_monitor_handler(self) -> int:
  51. task = RankLogMonitor(self.db_client, self.log_client)
  52. await task.deal()
  53. return TaskStatus.SUCCESS
  54. @register("check_kimi_balance")
  55. async def _check_kimi_balance_handler(self) -> int:
  56. """检查 Kimi 余额"""
  57. response = await check_kimi_balance()
  58. await self._log_task_event("kimi_balance_checked", data=response)
  59. return TaskStatus.SUCCESS
  60. @register("get_off_videos")
  61. async def _get_off_videos_task_handler(self) -> int:
  62. """视频下架任务"""
  63. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  64. return await sub_task.deal()
  65. @register("check_publish_video_audit_status")
  66. async def _check_video_audit_status_handler(self) -> int:
  67. """检查视频审核状态"""
  68. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  69. return await sub_task.deal()
  70. @register("task_processing_monitor")
  71. async def _task_processing_monitor_handler(self) -> int:
  72. """任务处理监控"""
  73. sub_task = TaskProcessingMonitor(self.db_client)
  74. await sub_task.deal()
  75. return TaskStatus.SUCCESS
  76. @register("inner_article_monitor")
  77. async def _inner_gzh_articles_monitor_handler(self) -> int:
  78. """内部公众号文章监控"""
  79. sub_task = InnerArticleMonitorTask(self.db_client)
  80. return await sub_task.deal()
  81. @register("outside_article_monitor")
  82. async def _outside_monitor_handler(self) -> int:
  83. """外部文章监控"""
  84. collector = OutsideGzhArticlesCollector(self.db_client)
  85. await collector.deal()
  86. monitor = OutsideGzhArticlesMonitor(self.db_client)
  87. return await monitor.deal()
  88. @register("cooperate_accounts_monitor")
  89. async def _cooperate_accounts_monitor_handler(self) -> int:
  90. """合作账号文章监测"""
  91. task = CooperateAccountsMonitorTask(
  92. pool=self.db_client, log_client=self.log_client
  93. )
  94. await task.deal(task_name="save_articles")
  95. return TaskStatus.SUCCESS
  96. @register("cooperate_accounts_detail")
  97. async def _cooperate_accounts_detail_handler(self) -> int:
  98. """合作账号文章详情更新"""
  99. task = CooperateAccountsMonitorTask(
  100. pool=self.db_client, log_client=self.log_client
  101. )
  102. await task.deal(task_name="get_detail")
  103. return TaskStatus.SUCCESS
  104. @register("ad_platform_accounts_crawler")
  105. async def _ad_platform_accounts_crawler_handler(self) -> int:
  106. """广告平台账号监测"""
  107. task = AdPlatformAccountsMonitorTask(
  108. pool=self.db_client, log_service=self.log_client
  109. )
  110. # 抓账号内所有的文章
  111. await task.deal(task_name="crawl_articles")
  112. # 抓完之后, 计算账号的阅读量中位数
  113. await task.deal(task_name="cal_read_median")
  114. return TaskStatus.SUCCESS
  115. @register("ad_platform_article_detail")
  116. async def _ad_platform_article_detail_handler(self) -> int:
  117. """广告平台文章详情更新"""
  118. task = AdPlatformAccountsMonitorTask(
  119. pool=self.db_client, log_service=self.log_client
  120. )
  121. await task.deal(task_name="get_detail")
  122. return TaskStatus.SUCCESS
  123. # ==================== 爬虫类任务 ====================
  124. @register("crawler_toutiao")
  125. async def _crawler_toutiao_handler(self) -> int:
  126. """头条文章/视频抓取"""
  127. sub_task = CrawlerToutiao(
  128. self.db_client, self.log_client, self.trace_id, self.config
  129. )
  130. method = self.data.get("method", "account")
  131. media_type = self.data.get("media_type", "article")
  132. category_list = self.data.get("category_list", [])
  133. match method:
  134. case "account":
  135. await sub_task.crawler_task(media_type=media_type)
  136. case "recommend":
  137. await sub_task.crawl_toutiao_recommend_task(category_list)
  138. case "search":
  139. await sub_task.search_candidate_accounts()
  140. case _:
  141. raise TaskValidationError(f"Unsupported method: {method}")
  142. return TaskStatus.SUCCESS
  143. @register("crawler_gzh_articles")
  144. async def _crawler_gzh_article_handler(self) -> int:
  145. """抓取公众号文章"""
  146. account_method = self.data.get("account_method")
  147. crawl_mode = self.data.get("crawl_mode")
  148. strategy = self.data.get("strategy")
  149. match crawl_mode:
  150. case "account":
  151. task = CrawlerGzhAccountArticles(
  152. self.db_client, self.log_client, self.trace_id, self.config
  153. )
  154. await task.deal(account_method, strategy)
  155. case "search":
  156. task = CrawlerGzhSearchArticles(
  157. self.db_client, self.log_client, self.trace_id, self.config
  158. )
  159. await task.deal(strategy)
  160. case _:
  161. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  162. return TaskStatus.SUCCESS
  163. @register("crawler_account_manager")
  164. async def _crawler_account_manager_handler(self) -> int:
  165. """账号管理任务"""
  166. platform = self.data.get("platform", "weixin")
  167. account_id_list = self.data.get("account_id_list")
  168. match platform:
  169. case "weixin":
  170. task = WeixinAccountManager(
  171. self.db_client, self.log_client, self.trace_id
  172. )
  173. await task.deal(platform=platform, account_id_list=account_id_list)
  174. case _:
  175. raise TaskValidationError(f"Unsupported platform: {platform}")
  176. return TaskStatus.SUCCESS
  177. @register("crawler_detail_analysis")
  178. async def _crawler_article_analysis_handler(self) -> int:
  179. """抓取视频/文章详情分析统计"""
  180. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  181. await task.deal(params=self.data)
  182. return TaskStatus.SUCCESS
  183. # ==================== 数据回收类任务 ====================
  184. @register("article_detail_stat")
  185. async def _article_detail_stat_handler(self) -> int:
  186. """文章详情统计"""
  187. task = ArticleDetailStat(self.db_client, self.log_client)
  188. await task.deal()
  189. return TaskStatus.SUCCESS
  190. @register("daily_publish_articles_recycle")
  191. async def _recycle_article_data_handler(self) -> int:
  192. """每日发文数据回收"""
  193. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  194. recycle = RecycleDailyPublishArticlesTask(
  195. self.db_client, self.log_client, date_str
  196. )
  197. await recycle.deal()
  198. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  199. await check.deal()
  200. return TaskStatus.SUCCESS
  201. @register("update_root_source_id")
  202. async def _update_root_source_id_handler(self) -> int:
  203. """更新 root_source_id"""
  204. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  205. await sub_task.deal()
  206. return TaskStatus.SUCCESS
  207. @register("recycle_outside_account_articles")
  208. async def _recycle_outside_account_article_handler(self) -> int:
  209. """回收外部账号文章"""
  210. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  211. task = RecycleOutsideAccountArticlesTask(
  212. pool=self.db_client, log_client=self.log_client, date_string=date_str
  213. )
  214. await task.deal()
  215. return TaskStatus.SUCCESS
  216. @register("update_outside_account_article_root_source_id")
  217. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  218. """更新外部账号文章的 root_source_id"""
  219. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  220. pool=self.db_client, log_client=self.log_client
  221. )
  222. await task.deal()
  223. return TaskStatus.SUCCESS
  224. @register("fwh_daily_recycle")
  225. async def _recycle_fwh_article_handler(self) -> int:
  226. """回收服务号文章"""
  227. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  228. await task.deal()
  229. return TaskStatus.SUCCESS
  230. # ==================== 算法类任务 ====================
  231. @register("article_pool_cold_start")
  232. async def _article_pool_cold_start_handler(self) -> int:
  233. """文章池冷启动"""
  234. cold_start = ArticlePoolColdStart(
  235. self.db_client, self.log_client, self.trace_id, self.config
  236. )
  237. platform = self.data.get("platform", "weixin")
  238. crawler_methods = self.data.get("crawler_methods", [])
  239. category_list = self.data.get("category_list", [])
  240. strategy = self.data.get("strategy", "strategy_v1")
  241. await cold_start.deal(
  242. platform=platform,
  243. crawl_methods=crawler_methods,
  244. category_list=category_list,
  245. strategy=strategy,
  246. )
  247. return TaskStatus.SUCCESS
  248. @register("ad_platform_article_publish")
  249. async def _ad_platform_article_publish(self) -> int:
  250. """文章池冷启动"""
  251. task = AdPlatformArticlePublishTask(self.db_client, self.log_client)
  252. await task.deal()
  253. return TaskStatus.SUCCESS
  254. @register("candidate_account_quality_analysis")
  255. async def _candidate_account_quality_score_handler(self) -> int:
  256. """候选账号质量分析"""
  257. task = CandidateAccountQualityScoreRecognizer(
  258. self.db_client, self.log_client, self.trace_id
  259. )
  260. await task.deal()
  261. return TaskStatus.SUCCESS
  262. @register("article_pool_category_generation")
  263. async def _article_pool_category_generation_handler(self) -> int:
  264. """文章池品类生成"""
  265. task = ArticlePoolCategoryGeneration(
  266. self.db_client, self.log_client, self.trace_id
  267. )
  268. limit_num = self.data.get("limit")
  269. await task.deal(limit=limit_num)
  270. return TaskStatus.SUCCESS
  271. @register("account_category_analysis")
  272. async def _account_category_analysis_handler(self) -> int:
  273. """账号品类分析"""
  274. task = AccountCategoryAnalysis(
  275. pool=self.db_client,
  276. log_client=self.log_client,
  277. trace_id=self.trace_id,
  278. data=self.data,
  279. date_string=None,
  280. )
  281. await task.deal()
  282. return TaskStatus.SUCCESS
  283. @register("update_limited_account_info")
  284. async def _update_limited_account_info_handler(self) -> int:
  285. """更新限流账号信息"""
  286. task = LimitedAccountAnalysisTask(
  287. pool=self.db_client, log_client=self.log_client
  288. )
  289. await task.deal(date_string=self.data.get("date_string"))
  290. return TaskStatus.SUCCESS
  291. # ==================== LLM 类任务 ====================
  292. @register("title_rewrite")
  293. async def _title_rewrite_handler(self) -> int:
  294. """标题重写"""
  295. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  296. return await sub_task.deal()
  297. @register("extract_title_features")
  298. async def _extract_title_features_handler(self) -> int:
  299. """提取标题特征"""
  300. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  301. await task.deal(data=self.data)
  302. return TaskStatus.SUCCESS
  303. # ==================== 统计分析类任务 ====================
  304. @register("update_account_read_rate_avg")
  305. async def _update_account_read_rate_avg_handler(self) -> int:
  306. """更新账号阅读率均值"""
  307. task = AccountPositionReadRateAvg(
  308. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  309. )
  310. await task.deal(end_date=self.data.get("end_date"))
  311. return TaskStatus.SUCCESS
  312. @register("update_account_read_avg")
  313. async def _update_account_read_avg_handler(self) -> int:
  314. """更新账号阅读均值"""
  315. task = AccountPositionReadAvg(
  316. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  317. )
  318. await task.deal(end_date=self.data.get("end_date"))
  319. return TaskStatus.SUCCESS
  320. @register("update_account_open_rate_avg")
  321. async def _update_account_open_rate_avg_handler(self) -> int:
  322. """更新账号打开率均值"""
  323. task = AccountPositionOpenRateAvg(
  324. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  325. )
  326. await task.deal(date_string=self.data.get("date_string"))
  327. return TaskStatus.SUCCESS
  328. # ==================== 自动化任务 ====================
  329. @register("auto_follow_account")
  330. async def _auto_follow_account_handler(self) -> int:
  331. """自动关注公众号"""
  332. task = AutoReplyCardsMonitor(
  333. pool=self.db_client, log_service=self.log_client, config=self.config
  334. )
  335. await task.deal(task_name="follow_gzh_task")
  336. return TaskStatus.SUCCESS
  337. @register("get_follow_result")
  338. async def _get_follow_result_handler(self) -> int:
  339. """获取自动关注回复"""
  340. task = AutoReplyCardsMonitor(
  341. pool=self.db_client, log_service=self.log_client, config=self.config
  342. )
  343. await task.deal(task_name="get_auto_reply_task")
  344. return TaskStatus.SUCCESS
  345. @register("extract_reply_result")
  346. async def _extract_reply_result_handler(self) -> int:
  347. """解析自动回复结果"""
  348. task = AutoReplyCardsMonitor(
  349. pool=self.db_client, log_service=self.log_client, config=self.config
  350. )
  351. await task.deal(task_name="extract_task")
  352. return TaskStatus.SUCCESS
  353. # ==================== 其他任务 ====================
  354. @register("mini_program_detail_process")
  355. async def _mini_program_detail_handler(self) -> int:
  356. """更新小程序裂变信息"""
  357. task = RecycleMiniProgramDetailTask(
  358. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  359. )
  360. await task.deal(params=self.data)
  361. return TaskStatus.SUCCESS
  362. @register("rate_limited_article_filter")
  363. async def _rate_limited_article_filter(self) -> int:
  364. """限流文章删除"""
  365. task = RateLimitedArticleFilter(
  366. pool=self.db_client,
  367. config=self.config,
  368. log_service=self.log_client,
  369. )
  370. await task.deal()
  371. return TaskStatus.SUCCESS
  372. @register("create_ad_platform_accounts_decode_task")
  373. async def _create_ad_platform_accounts_decode_task(self) -> int:
  374. """创建解构任务"""
  375. task = CreateAdPlatformArticlesDecodeTask(
  376. pool=self.db_client, log_service=self.log_client
  377. )
  378. await task.deal()
  379. return TaskStatus.SUCCESS
  380. @register("fetch_decode_result")
  381. async def _fetch_decode_result(self) -> int:
  382. """获取解构任务结果"""
  383. task = FetchDecodeResults(
  384. pool=self.db_client, log_service=self.log_client
  385. )
  386. await task.deal()
  387. return TaskStatus.SUCCESS
  388. @register("extract_decode_result")
  389. async def _extract_ad_platform_accounts_decode_result(self) -> int:
  390. """提取解构任务结果"""
  391. task = ExtractDecodeTaskDetail(
  392. pool=self.db_client, log_service=self.log_client
  393. )
  394. await task.deal()
  395. return TaskStatus.SUCCESS
  396. # ====================== Recommend Tasks=====================
  397. @register("i2i_recommend_data_sync")
  398. async def _i2i_recommend_data_sync_handler(self) -> int:
  399. task = I2IRecommendDataSyncTask(
  400. pool=self.db_client, log_service=self.log_client
  401. )
  402. await task.deal()
  403. return TaskStatus.SUCCESS
  404. __all__ = ["TaskHandler"]