task_handler.py 18 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466
  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.domains.monitor_tasks import AdPlatformAccountsMonitorTask
  40. from app.jobs.task_config import TaskStatus
  41. from app.jobs.task_utils import TaskValidationError
  42. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  43. def register(task_name: str):
  44. def decorator(func):
  45. _TASK_HANDLER_REGISTRY[task_name] = func
  46. return func
  47. return decorator
  48. class TaskHandler:
  49. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  50. # 任务注册表
  51. _handlers = _TASK_HANDLER_REGISTRY
  52. def __init__(
  53. self,
  54. data: dict,
  55. log_service,
  56. db_client,
  57. trace_id: str,
  58. config: GlobalConfigSettings,
  59. ):
  60. self.data = data
  61. self.log_client = log_service
  62. self.db_client = db_client
  63. self.trace_id = trace_id
  64. self.config = config
  65. @classmethod
  66. def get_handler(cls, task_name: str) -> Optional[Callable]:
  67. """获取任务处理器"""
  68. return cls._handlers.get(task_name)
  69. @classmethod
  70. def list_registered_tasks(cls) -> list:
  71. """列出所有已注册的任务"""
  72. return list(cls._handlers.keys())
  73. async def _log_task_event(self, event_type: str, **kwargs):
  74. """统一的任务日志记录"""
  75. log_data = {
  76. "timestamp": datetime.now().isoformat(),
  77. "trace_id": self.trace_id,
  78. "event_type": event_type,
  79. "task": self.data.get("task_name"),
  80. **kwargs,
  81. }
  82. await self.log_client.log(contents=log_data)
  83. # ==================== 监控类任务 ====================
  84. @register("check_kimi_balance")
  85. async def _check_kimi_balance_handler(self) -> int:
  86. """检查 Kimi 余额"""
  87. response = await check_kimi_balance()
  88. await self._log_task_event("kimi_balance_checked", data=response)
  89. return TaskStatus.SUCCESS
  90. @register("get_off_videos")
  91. async def _get_off_videos_task_handler(self) -> int:
  92. """视频下架任务"""
  93. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  94. return await sub_task.deal()
  95. @register("check_publish_video_audit_status")
  96. async def _check_video_audit_status_handler(self) -> int:
  97. """检查视频审核状态"""
  98. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  99. return await sub_task.deal()
  100. @register("task_processing_monitor")
  101. async def _task_processing_monitor_handler(self) -> int:
  102. """任务处理监控"""
  103. sub_task = TaskProcessingMonitor(self.db_client)
  104. await sub_task.deal()
  105. return TaskStatus.SUCCESS
  106. @register("inner_article_monitor")
  107. async def _inner_gzh_articles_monitor_handler(self) -> int:
  108. """内部公众号文章监控"""
  109. sub_task = InnerGzhArticlesMonitor(self.db_client)
  110. return await sub_task.deal()
  111. @register("outside_article_monitor")
  112. async def _outside_monitor_handler(self) -> int:
  113. """外部文章监控"""
  114. collector = OutsideGzhArticlesCollector(self.db_client)
  115. await collector.deal()
  116. monitor = OutsideGzhArticlesMonitor(self.db_client)
  117. return await monitor.deal()
  118. @register("cooperate_accounts_monitor")
  119. async def _cooperate_accounts_monitor_handler(self) -> int:
  120. """合作账号文章监测"""
  121. task = CooperateAccountsMonitorTask(
  122. pool=self.db_client, log_client=self.log_client
  123. )
  124. await task.deal(task_name="save_articles")
  125. return TaskStatus.SUCCESS
  126. @register("cooperate_accounts_detail")
  127. async def _cooperate_accounts_detail_handler(self) -> int:
  128. """合作账号文章详情更新"""
  129. task = CooperateAccountsMonitorTask(
  130. pool=self.db_client, log_client=self.log_client
  131. )
  132. await task.deal(task_name="get_detail")
  133. return TaskStatus.SUCCESS
  134. @register("ad_platform_accounts_crawler")
  135. async def _ad_platform_accounts_crawler_handler(self) -> int:
  136. """广告平台账号监测"""
  137. task = AdPlatformAccountsMonitorTask(
  138. pool=self.db_client, log_service=self.log_client
  139. )
  140. # 抓账号内所有的文章
  141. await task.deal(task_name="crawl_articles")
  142. # 抓完之后, 计算账号的阅读量中位数
  143. await task.deal(task_name="cal_read_median")
  144. return TaskStatus.SUCCESS
  145. @register("ad_platform_article_detail")
  146. async def _ad_platform_article_detail_handler(self) -> int:
  147. """广告平台文章详情更新"""
  148. task = AdPlatformAccountsMonitorTask(
  149. pool=self.db_client, log_service=self.log_client
  150. )
  151. await task.deal(task_name="get_detail")
  152. return TaskStatus.SUCCESS
  153. # ==================== 爬虫类任务 ====================
  154. @register("crawler_toutiao")
  155. async def _crawler_toutiao_handler(self) -> int:
  156. """头条文章/视频抓取"""
  157. sub_task = CrawlerToutiao(
  158. self.db_client, self.log_client, self.trace_id, self.config
  159. )
  160. method = self.data.get("method", "account")
  161. media_type = self.data.get("media_type", "article")
  162. category_list = self.data.get("category_list", [])
  163. match method:
  164. case "account":
  165. await sub_task.crawler_task(media_type=media_type)
  166. case "recommend":
  167. await sub_task.crawl_toutiao_recommend_task(category_list)
  168. case "search":
  169. await sub_task.search_candidate_accounts()
  170. case _:
  171. raise TaskValidationError(f"Unsupported method: {method}")
  172. return TaskStatus.SUCCESS
  173. @register("crawler_gzh_articles")
  174. async def _crawler_gzh_article_handler(self) -> int:
  175. """抓取公众号文章"""
  176. account_method = self.data.get("account_method")
  177. crawl_mode = self.data.get("crawl_mode")
  178. strategy = self.data.get("strategy")
  179. match crawl_mode:
  180. case "account":
  181. task = CrawlerGzhAccountArticles(
  182. self.db_client, self.log_client, self.trace_id, self.config
  183. )
  184. await task.deal(account_method, strategy)
  185. case "search":
  186. task = CrawlerGzhSearchArticles(
  187. self.db_client, self.log_client, self.trace_id, self.config
  188. )
  189. await task.deal(strategy)
  190. case _:
  191. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  192. return TaskStatus.SUCCESS
  193. @register("crawler_account_manager")
  194. async def _crawler_account_manager_handler(self) -> int:
  195. """账号管理任务"""
  196. platform = self.data.get("platform", "weixin")
  197. account_id_list = self.data.get("account_id_list")
  198. match platform:
  199. case "weixin":
  200. task = WeixinAccountManager(
  201. self.db_client, self.log_client, self.trace_id
  202. )
  203. await task.deal(platform=platform, account_id_list=account_id_list)
  204. case _:
  205. raise TaskValidationError(f"Unsupported platform: {platform}")
  206. return TaskStatus.SUCCESS
  207. @register("crawler_detail_analysis")
  208. async def _crawler_article_analysis_handler(self) -> int:
  209. """抓取视频/文章详情分析统计"""
  210. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  211. await task.deal(params=self.data)
  212. return TaskStatus.SUCCESS
  213. # ==================== 数据回收类任务 ====================
  214. @register("article_detail_stat")
  215. async def _article_detail_stat_handler(self) -> int:
  216. """文章详情统计"""
  217. task = ArticleDetailStat(self.db_client, self.log_client)
  218. await task.deal()
  219. return TaskStatus.SUCCESS
  220. @register("daily_publish_articles_recycle")
  221. async def _recycle_article_data_handler(self) -> int:
  222. """每日发文数据回收"""
  223. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  224. recycle = RecycleDailyPublishArticlesTask(
  225. self.db_client, self.log_client, date_str
  226. )
  227. await recycle.deal()
  228. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  229. await check.deal()
  230. return TaskStatus.SUCCESS
  231. @register("update_root_source_id")
  232. async def _update_root_source_id_handler(self) -> int:
  233. """更新 root_source_id"""
  234. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  235. await sub_task.deal()
  236. return TaskStatus.SUCCESS
  237. @register("recycle_outside_account_articles")
  238. async def _recycle_outside_account_article_handler(self) -> int:
  239. """回收外部账号文章"""
  240. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  241. task = RecycleOutsideAccountArticlesTask(
  242. pool=self.db_client, log_client=self.log_client, date_string=date_str
  243. )
  244. await task.deal()
  245. return TaskStatus.SUCCESS
  246. @register("update_outside_account_article_root_source_id")
  247. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  248. """更新外部账号文章的 root_source_id"""
  249. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  250. pool=self.db_client, log_client=self.log_client
  251. )
  252. await task.deal()
  253. return TaskStatus.SUCCESS
  254. @register("fwh_daily_recycle")
  255. async def _recycle_fwh_article_handler(self) -> int:
  256. """回收服务号文章"""
  257. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  258. await task.deal()
  259. return TaskStatus.SUCCESS
  260. # ==================== 算法类任务 ====================
  261. @register("article_pool_cold_start")
  262. async def _article_pool_cold_start_handler(self) -> int:
  263. """文章池冷启动"""
  264. cold_start = ArticlePoolColdStart(
  265. self.db_client, self.log_client, self.trace_id, self.config
  266. )
  267. platform = self.data.get("platform", "weixin")
  268. crawler_methods = self.data.get("crawler_methods", [])
  269. category_list = self.data.get("category_list", [])
  270. strategy = self.data.get("strategy", "strategy_v1")
  271. await cold_start.deal(
  272. platform=platform,
  273. crawl_methods=crawler_methods,
  274. category_list=category_list,
  275. strategy=strategy,
  276. )
  277. return TaskStatus.SUCCESS
  278. @register("candidate_account_quality_analysis")
  279. async def _candidate_account_quality_score_handler(self) -> int:
  280. """候选账号质量分析"""
  281. task = CandidateAccountQualityScoreRecognizer(
  282. self.db_client, self.log_client, self.trace_id
  283. )
  284. await task.deal()
  285. return TaskStatus.SUCCESS
  286. @register("article_pool_category_generation")
  287. async def _article_pool_category_generation_handler(self) -> int:
  288. """文章池品类生成"""
  289. task = ArticlePoolCategoryGeneration(
  290. self.db_client, self.log_client, self.trace_id
  291. )
  292. limit_num = self.data.get("limit")
  293. await task.deal(limit=limit_num)
  294. return TaskStatus.SUCCESS
  295. @register("account_category_analysis")
  296. async def _account_category_analysis_handler(self) -> int:
  297. """账号品类分析"""
  298. task = AccountCategoryAnalysis(
  299. pool=self.db_client,
  300. log_client=self.log_client,
  301. trace_id=self.trace_id,
  302. data=self.data,
  303. date_string=None,
  304. )
  305. await task.deal()
  306. return TaskStatus.SUCCESS
  307. @register("update_limited_account_info")
  308. async def _update_limited_account_info_handler(self) -> int:
  309. """更新限流账号信息"""
  310. task = LimitedAccountAnalysisTask(
  311. pool=self.db_client, log_client=self.log_client
  312. )
  313. await task.deal(date_string=self.data.get("date_string"))
  314. return TaskStatus.SUCCESS
  315. # ==================== LLM 类任务 ====================
  316. @register("title_rewrite")
  317. async def _title_rewrite_handler(self) -> int:
  318. """标题重写"""
  319. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  320. return await sub_task.deal()
  321. @register("extract_title_features")
  322. async def _extract_title_features_handler(self) -> int:
  323. """提取标题特征"""
  324. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  325. await task.deal(data=self.data)
  326. return TaskStatus.SUCCESS
  327. # ==================== 统计分析类任务 ====================
  328. @register("update_account_read_rate_avg")
  329. async def _update_account_read_rate_avg_handler(self) -> int:
  330. """更新账号阅读率均值"""
  331. task = AccountPositionReadRateAvg(
  332. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  333. )
  334. await task.deal(end_date=self.data.get("end_date"))
  335. return TaskStatus.SUCCESS
  336. @register("update_account_read_avg")
  337. async def _update_account_read_avg_handler(self) -> int:
  338. """更新账号阅读均值"""
  339. task = AccountPositionReadAvg(
  340. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  341. )
  342. await task.deal(end_date=self.data.get("end_date"))
  343. return TaskStatus.SUCCESS
  344. @register("update_account_open_rate_avg")
  345. async def _update_account_open_rate_avg_handler(self) -> int:
  346. """更新账号打开率均值"""
  347. task = AccountPositionOpenRateAvg(
  348. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  349. )
  350. await task.deal(date_string=self.data.get("date_string"))
  351. return TaskStatus.SUCCESS
  352. # ==================== 自动化任务 ====================
  353. @register("auto_follow_account")
  354. async def _auto_follow_account_handler(self) -> int:
  355. """自动关注公众号"""
  356. task = AutoReplyCardsMonitor(
  357. pool=self.db_client, log_service=self.log_client, config=self.config
  358. )
  359. await task.deal(task_name="follow_gzh_task")
  360. return TaskStatus.SUCCESS
  361. @register("get_follow_result")
  362. async def _get_follow_result_handler(self) -> int:
  363. """获取自动关注回复"""
  364. task = AutoReplyCardsMonitor(
  365. pool=self.db_client, log_service=self.log_client, config=self.config
  366. )
  367. await task.deal(task_name="get_auto_reply_task")
  368. return TaskStatus.SUCCESS
  369. @register("extract_reply_result")
  370. async def _extract_reply_result_handler(self) -> int:
  371. """解析自动回复结果"""
  372. task = AutoReplyCardsMonitor(
  373. pool=self.db_client, log_service=self.log_client, config=self.config
  374. )
  375. await task.deal(task_name="extract_task")
  376. return TaskStatus.SUCCESS
  377. # ==================== 其他任务 ====================
  378. @register("mini_program_detail_process")
  379. async def _mini_program_detail_handler(self) -> int:
  380. """更新小程序裂变信息"""
  381. task = RecycleMiniProgramDetailTask(
  382. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  383. )
  384. await task.deal(params=self.data)
  385. return TaskStatus.SUCCESS
  386. @register("rate_limited_article_filter")
  387. async def _rate_limited_article_filter(self) -> int:
  388. """限流文章删除"""
  389. task = RateLimitedArticleFilter(pool=self.db_client, config=self.config)
  390. await task.deal()
  391. return TaskStatus.SUCCESS
  392. __all__ = ["TaskHandler"]