task_handler.py 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497498499500501502503504505506507508509510511512
  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.cold_start_tasks import AdPlatformArticlePublishTask
  12. from app.domains.crawler_tasks import CrawlerToutiao
  13. from app.domains.crawler_tasks import WeixinAccountManager
  14. from app.domains.crawler_tasks import CrawlerGzhAccountArticles
  15. from app.domains.crawler_tasks import CrawlerGzhSearchArticles
  16. from app.domains.data_recycle_tasks import ArticleDetailStat
  17. from app.domains.data_recycle_tasks import RecycleDailyPublishArticlesTask
  18. from app.domains.data_recycle_tasks import RecycleOutsideAccountArticlesTask
  19. from app.domains.data_recycle_tasks import CheckDailyPublishArticlesTask
  20. from app.domains.data_recycle_tasks import UpdateRootSourceIdAndUpdateTimeTask
  21. from app.domains.data_recycle_tasks import RecycleFwhDailyPublishArticlesTask
  22. from app.domains.data_recycle_tasks import RecycleMiniProgramDetailTask
  23. from app.domains.data_recycle_tasks import (
  24. UpdateOutsideRootSourceIdAndUpdateTimeTask,
  25. )
  26. from app.domains.llm_tasks.aigc_decode_task import CreateAdPlatformArticlesDecodeTask
  27. from app.domains.llm_tasks.aigc_decode_task import CreateInnerArticlesDecodeTask
  28. from app.domains.llm_tasks.aigc_decode_task import FetchDecodeResults
  29. from app.domains.llm_tasks.aigc_decode_task import ExtractDecodeTaskDetail
  30. from app.domains.llm_tasks import TitleRewrite
  31. from app.domains.llm_tasks import ArticlePoolCategoryGeneration
  32. from app.domains.llm_tasks import CandidateAccountQualityScoreRecognizer
  33. from app.domains.llm_tasks import ExtractTitleFeatures
  34. from app.domains.monitor_tasks import AutoReplyCardsMonitor
  35. from app.domains.monitor_tasks import check_kimi_balance
  36. from app.domains.monitor_tasks import GetOffVideos
  37. from app.domains.monitor_tasks import CheckVideoAuditStatus
  38. from app.domains.monitor_tasks import CooperateAccountsMonitorTask
  39. from app.domains.monitor_tasks import InnerGzhArticlesMonitor
  40. from app.domains.monitor_tasks import OutsideGzhArticlesMonitor
  41. from app.domains.monitor_tasks import OutsideGzhArticlesCollector
  42. from app.domains.monitor_tasks import TaskProcessingMonitor
  43. from app.domains.monitor_tasks import LimitedAccountAnalysisTask
  44. from app.domains.monitor_tasks import AdPlatformAccountsMonitorTask
  45. from app.domains.monitor_tasks import RankLogMonitor
  46. from app.jobs.task_config import TaskStatus
  47. from app.jobs.task_utils import TaskValidationError
  48. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  49. def register(task_name: str):
  50. def decorator(func):
  51. _TASK_HANDLER_REGISTRY[task_name] = func
  52. return func
  53. return decorator
  54. class TaskHandler:
  55. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  56. # 任务注册表
  57. _handlers = _TASK_HANDLER_REGISTRY
  58. def __init__(
  59. self,
  60. data: dict,
  61. log_service,
  62. db_client,
  63. trace_id: str,
  64. config: GlobalConfigSettings,
  65. ):
  66. self.data = data
  67. self.log_client = log_service
  68. self.db_client = db_client
  69. self.trace_id = trace_id
  70. self.config = config
  71. @classmethod
  72. def get_handler(cls, task_name: str) -> Optional[Callable]:
  73. """获取任务处理器"""
  74. return cls._handlers.get(task_name)
  75. @classmethod
  76. def list_registered_tasks(cls) -> list:
  77. """列出所有已注册的任务"""
  78. return list(cls._handlers.keys())
  79. async def _log_task_event(self, event_type: str, **kwargs):
  80. """统一的任务日志记录"""
  81. log_data = {
  82. "timestamp": datetime.now().isoformat(),
  83. "trace_id": self.trace_id,
  84. "event_type": event_type,
  85. "task": self.data.get("task_name"),
  86. **kwargs,
  87. }
  88. await self.log_client.log(contents=log_data)
  89. # ==================== 监控类任务 ====================
  90. @register("rank_log_monitor")
  91. async def _rank_log_monitor_handler(self) -> int:
  92. """迁移排序日志"""
  93. sub_task = RankLogMonitor(self.db_client, self.log_client)
  94. return await sub_task.deal()
  95. @register("check_kimi_balance")
  96. async def _check_kimi_balance_handler(self) -> int:
  97. """检查 Kimi 余额"""
  98. response = await check_kimi_balance()
  99. await self._log_task_event("kimi_balance_checked", data=response)
  100. return TaskStatus.SUCCESS
  101. @register("get_off_videos")
  102. async def _get_off_videos_task_handler(self) -> int:
  103. """视频下架任务"""
  104. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  105. return await sub_task.deal()
  106. @register("check_publish_video_audit_status")
  107. async def _check_video_audit_status_handler(self) -> int:
  108. """检查视频审核状态"""
  109. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  110. return await sub_task.deal()
  111. @register("task_processing_monitor")
  112. async def _task_processing_monitor_handler(self) -> int:
  113. """任务处理监控"""
  114. sub_task = TaskProcessingMonitor(self.db_client)
  115. await sub_task.deal()
  116. return TaskStatus.SUCCESS
  117. @register("inner_article_monitor")
  118. async def _inner_gzh_articles_monitor_handler(self) -> int:
  119. """内部公众号文章监控"""
  120. sub_task = InnerGzhArticlesMonitor(self.db_client)
  121. return await sub_task.deal()
  122. @register("outside_article_monitor")
  123. async def _outside_monitor_handler(self) -> int:
  124. """外部文章监控"""
  125. collector = OutsideGzhArticlesCollector(self.db_client)
  126. await collector.deal()
  127. monitor = OutsideGzhArticlesMonitor(self.db_client)
  128. return await monitor.deal()
  129. @register("cooperate_accounts_monitor")
  130. async def _cooperate_accounts_monitor_handler(self) -> int:
  131. """合作账号文章监测"""
  132. task = CooperateAccountsMonitorTask(
  133. pool=self.db_client, log_client=self.log_client
  134. )
  135. await task.deal(task_name="save_articles")
  136. return TaskStatus.SUCCESS
  137. @register("cooperate_accounts_detail")
  138. async def _cooperate_accounts_detail_handler(self) -> int:
  139. """合作账号文章详情更新"""
  140. task = CooperateAccountsMonitorTask(
  141. pool=self.db_client, log_client=self.log_client
  142. )
  143. await task.deal(task_name="get_detail")
  144. return TaskStatus.SUCCESS
  145. @register("ad_platform_accounts_crawler")
  146. async def _ad_platform_accounts_crawler_handler(self) -> int:
  147. """广告平台账号监测"""
  148. task = AdPlatformAccountsMonitorTask(
  149. pool=self.db_client, log_service=self.log_client
  150. )
  151. # 抓账号内所有的文章
  152. await task.deal(task_name="crawl_articles")
  153. # 抓完之后, 计算账号的阅读量中位数
  154. await task.deal(task_name="cal_read_median")
  155. return TaskStatus.SUCCESS
  156. @register("ad_platform_article_detail")
  157. async def _ad_platform_article_detail_handler(self) -> int:
  158. """广告平台文章详情更新"""
  159. task = AdPlatformAccountsMonitorTask(
  160. pool=self.db_client, log_service=self.log_client
  161. )
  162. await task.deal(task_name="get_detail")
  163. return TaskStatus.SUCCESS
  164. # ==================== 爬虫类任务 ====================
  165. @register("crawler_toutiao")
  166. async def _crawler_toutiao_handler(self) -> int:
  167. """头条文章/视频抓取"""
  168. sub_task = CrawlerToutiao(
  169. self.db_client, self.log_client, self.trace_id, self.config
  170. )
  171. method = self.data.get("method", "account")
  172. media_type = self.data.get("media_type", "article")
  173. category_list = self.data.get("category_list", [])
  174. match method:
  175. case "account":
  176. await sub_task.crawler_task(media_type=media_type)
  177. case "recommend":
  178. await sub_task.crawl_toutiao_recommend_task(category_list)
  179. case "search":
  180. await sub_task.search_candidate_accounts()
  181. case _:
  182. raise TaskValidationError(f"Unsupported method: {method}")
  183. return TaskStatus.SUCCESS
  184. @register("crawler_gzh_articles")
  185. async def _crawler_gzh_article_handler(self) -> int:
  186. """抓取公众号文章"""
  187. account_method = self.data.get("account_method")
  188. crawl_mode = self.data.get("crawl_mode")
  189. strategy = self.data.get("strategy")
  190. match crawl_mode:
  191. case "account":
  192. task = CrawlerGzhAccountArticles(
  193. self.db_client, self.log_client, self.trace_id, self.config
  194. )
  195. await task.deal(account_method, strategy)
  196. case "search":
  197. task = CrawlerGzhSearchArticles(
  198. self.db_client, self.log_client, self.trace_id, self.config
  199. )
  200. await task.deal(strategy)
  201. case _:
  202. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  203. return TaskStatus.SUCCESS
  204. @register("crawler_account_manager")
  205. async def _crawler_account_manager_handler(self) -> int:
  206. """账号管理任务"""
  207. platform = self.data.get("platform", "weixin")
  208. account_id_list = self.data.get("account_id_list")
  209. match platform:
  210. case "weixin":
  211. task = WeixinAccountManager(
  212. self.db_client, self.log_client, self.trace_id
  213. )
  214. await task.deal(platform=platform, account_id_list=account_id_list)
  215. case _:
  216. raise TaskValidationError(f"Unsupported platform: {platform}")
  217. return TaskStatus.SUCCESS
  218. @register("crawler_detail_analysis")
  219. async def _crawler_article_analysis_handler(self) -> int:
  220. """抓取视频/文章详情分析统计"""
  221. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  222. await task.deal(params=self.data)
  223. return TaskStatus.SUCCESS
  224. # ==================== 数据回收类任务 ====================
  225. @register("article_detail_stat")
  226. async def _article_detail_stat_handler(self) -> int:
  227. """文章详情统计"""
  228. task = ArticleDetailStat(self.db_client, self.log_client)
  229. await task.deal()
  230. return TaskStatus.SUCCESS
  231. @register("daily_publish_articles_recycle")
  232. async def _recycle_article_data_handler(self) -> int:
  233. """每日发文数据回收"""
  234. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  235. recycle = RecycleDailyPublishArticlesTask(
  236. self.db_client, self.log_client, date_str
  237. )
  238. await recycle.deal()
  239. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  240. await check.deal()
  241. return TaskStatus.SUCCESS
  242. @register("update_root_source_id")
  243. async def _update_root_source_id_handler(self) -> int:
  244. """更新 root_source_id"""
  245. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  246. await sub_task.deal()
  247. return TaskStatus.SUCCESS
  248. @register("recycle_outside_account_articles")
  249. async def _recycle_outside_account_article_handler(self) -> int:
  250. """回收外部账号文章"""
  251. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  252. task = RecycleOutsideAccountArticlesTask(
  253. pool=self.db_client, log_client=self.log_client, date_string=date_str
  254. )
  255. await task.deal()
  256. return TaskStatus.SUCCESS
  257. @register("update_outside_account_article_root_source_id")
  258. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  259. """更新外部账号文章的 root_source_id"""
  260. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  261. pool=self.db_client, log_client=self.log_client
  262. )
  263. await task.deal()
  264. return TaskStatus.SUCCESS
  265. @register("fwh_daily_recycle")
  266. async def _recycle_fwh_article_handler(self) -> int:
  267. """回收服务号文章"""
  268. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  269. await task.deal()
  270. return TaskStatus.SUCCESS
  271. # ==================== 算法类任务 ====================
  272. @register("article_pool_cold_start")
  273. async def _article_pool_cold_start_handler(self) -> int:
  274. """文章池冷启动"""
  275. cold_start = ArticlePoolColdStart(
  276. self.db_client, self.log_client, self.trace_id, self.config
  277. )
  278. platform = self.data.get("platform", "weixin")
  279. crawler_methods = self.data.get("crawler_methods", [])
  280. category_list = self.data.get("category_list", [])
  281. strategy = self.data.get("strategy", "strategy_v1")
  282. await cold_start.deal(
  283. platform=platform,
  284. crawl_methods=crawler_methods,
  285. category_list=category_list,
  286. strategy=strategy,
  287. )
  288. return TaskStatus.SUCCESS
  289. @register("ad_platform_article_publish")
  290. async def _ad_platform_article_publish(self) -> int:
  291. """文章池冷启动"""
  292. task = AdPlatformArticlePublishTask(self.db_client, self.log_client)
  293. await task.deal()
  294. return TaskStatus.SUCCESS
  295. @register("candidate_account_quality_analysis")
  296. async def _candidate_account_quality_score_handler(self) -> int:
  297. """候选账号质量分析"""
  298. task = CandidateAccountQualityScoreRecognizer(
  299. self.db_client, self.log_client, self.trace_id
  300. )
  301. await task.deal()
  302. return TaskStatus.SUCCESS
  303. @register("article_pool_category_generation")
  304. async def _article_pool_category_generation_handler(self) -> int:
  305. """文章池品类生成"""
  306. task = ArticlePoolCategoryGeneration(
  307. self.db_client, self.log_client, self.trace_id
  308. )
  309. limit_num = self.data.get("limit")
  310. await task.deal(limit=limit_num)
  311. return TaskStatus.SUCCESS
  312. @register("account_category_analysis")
  313. async def _account_category_analysis_handler(self) -> int:
  314. """账号品类分析"""
  315. task = AccountCategoryAnalysis(
  316. pool=self.db_client,
  317. log_client=self.log_client,
  318. trace_id=self.trace_id,
  319. data=self.data,
  320. date_string=None,
  321. )
  322. await task.deal()
  323. return TaskStatus.SUCCESS
  324. @register("update_limited_account_info")
  325. async def _update_limited_account_info_handler(self) -> int:
  326. """更新限流账号信息"""
  327. task = LimitedAccountAnalysisTask(
  328. pool=self.db_client, log_client=self.log_client
  329. )
  330. await task.deal(date_string=self.data.get("date_string"))
  331. return TaskStatus.SUCCESS
  332. # ==================== LLM 类任务 ====================
  333. @register("title_rewrite")
  334. async def _title_rewrite_handler(self) -> int:
  335. """标题重写"""
  336. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  337. return await sub_task.deal()
  338. @register("extract_title_features")
  339. async def _extract_title_features_handler(self) -> int:
  340. """提取标题特征"""
  341. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  342. await task.deal(data=self.data)
  343. return TaskStatus.SUCCESS
  344. # ==================== 统计分析类任务 ====================
  345. @register("update_account_read_rate_avg")
  346. async def _update_account_read_rate_avg_handler(self) -> int:
  347. """更新账号阅读率均值"""
  348. task = AccountPositionReadRateAvg(
  349. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  350. )
  351. await task.deal(end_date=self.data.get("end_date"))
  352. return TaskStatus.SUCCESS
  353. @register("update_account_read_avg")
  354. async def _update_account_read_avg_handler(self) -> int:
  355. """更新账号阅读均值"""
  356. task = AccountPositionReadAvg(
  357. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  358. )
  359. await task.deal(end_date=self.data.get("end_date"))
  360. return TaskStatus.SUCCESS
  361. @register("update_account_open_rate_avg")
  362. async def _update_account_open_rate_avg_handler(self) -> int:
  363. """更新账号打开率均值"""
  364. task = AccountPositionOpenRateAvg(
  365. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  366. )
  367. await task.deal(date_string=self.data.get("date_string"))
  368. return TaskStatus.SUCCESS
  369. # ==================== 自动化任务 ====================
  370. @register("auto_follow_account")
  371. async def _auto_follow_account_handler(self) -> int:
  372. """自动关注公众号"""
  373. task = AutoReplyCardsMonitor(
  374. pool=self.db_client, log_service=self.log_client, config=self.config
  375. )
  376. await task.deal(task_name="follow_gzh_task")
  377. return TaskStatus.SUCCESS
  378. @register("get_follow_result")
  379. async def _get_follow_result_handler(self) -> int:
  380. """获取自动关注回复"""
  381. task = AutoReplyCardsMonitor(
  382. pool=self.db_client, log_service=self.log_client, config=self.config
  383. )
  384. await task.deal(task_name="get_auto_reply_task")
  385. return TaskStatus.SUCCESS
  386. @register("extract_reply_result")
  387. async def _extract_reply_result_handler(self) -> int:
  388. """解析自动回复结果"""
  389. task = AutoReplyCardsMonitor(
  390. pool=self.db_client, log_service=self.log_client, config=self.config
  391. )
  392. await task.deal(task_name="extract_task")
  393. return TaskStatus.SUCCESS
  394. # ==================== 其他任务 ====================
  395. @register("mini_program_detail_process")
  396. async def _mini_program_detail_handler(self) -> int:
  397. """更新小程序裂变信息"""
  398. task = RecycleMiniProgramDetailTask(
  399. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  400. )
  401. await task.deal(params=self.data)
  402. return TaskStatus.SUCCESS
  403. @register("rate_limited_article_filter")
  404. async def _rate_limited_article_filter(self) -> int:
  405. """限流文章删除"""
  406. task = RateLimitedArticleFilter(pool=self.db_client, config=self.config)
  407. await task.deal()
  408. return TaskStatus.SUCCESS
  409. @register("create_ad_platform_accounts_decode_task")
  410. async def _create_ad_platform_accounts_decode_task(self) -> int:
  411. """创建解构任务"""
  412. task = CreateAdPlatformArticlesDecodeTask(
  413. pool=self.db_client, log_service=self.log_client
  414. )
  415. await task.deal()
  416. return TaskStatus.SUCCESS
  417. @register("fetch_decode_result")
  418. async def _fetch_decode_result(self) -> int:
  419. """获取解构任务结果"""
  420. task = FetchDecodeResults(
  421. pool=self.db_client, log_service=self.log_client
  422. )
  423. await task.deal()
  424. return TaskStatus.SUCCESS
  425. @register("extract_decode_result")
  426. async def _extract_ad_platform_accounts_decode_result(self) -> int:
  427. """提取解构任务结果"""
  428. task = ExtractDecodeTaskDetail(
  429. pool=self.db_client, log_service=self.log_client
  430. )
  431. await task.deal()
  432. return TaskStatus.SUCCESS
  433. __all__ = ["TaskHandler"]