task_handler.py 19 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420421422423424425426427428429430431432433434435436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490491492493494495496497
  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.decode_task import AdPlatformArticlesDecodeTask
  27. from app.domains.llm_tasks import TitleRewrite
  28. from app.domains.llm_tasks import ArticlePoolCategoryGeneration
  29. from app.domains.llm_tasks import CandidateAccountQualityScoreRecognizer
  30. from app.domains.llm_tasks import ExtractTitleFeatures
  31. from app.domains.monitor_tasks import AutoReplyCardsMonitor
  32. from app.domains.monitor_tasks import check_kimi_balance
  33. from app.domains.monitor_tasks import GetOffVideos
  34. from app.domains.monitor_tasks import CheckVideoAuditStatus
  35. from app.domains.monitor_tasks import CooperateAccountsMonitorTask
  36. from app.domains.monitor_tasks import InnerGzhArticlesMonitor
  37. from app.domains.monitor_tasks import OutsideGzhArticlesMonitor
  38. from app.domains.monitor_tasks import OutsideGzhArticlesCollector
  39. from app.domains.monitor_tasks import TaskProcessingMonitor
  40. from app.domains.monitor_tasks import LimitedAccountAnalysisTask
  41. from app.domains.monitor_tasks import AdPlatformAccountsMonitorTask
  42. from app.jobs.task_config import TaskStatus
  43. from app.jobs.task_utils import TaskValidationError
  44. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  45. def register(task_name: str):
  46. def decorator(func):
  47. _TASK_HANDLER_REGISTRY[task_name] = func
  48. return func
  49. return decorator
  50. class TaskHandler:
  51. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  52. # 任务注册表
  53. _handlers = _TASK_HANDLER_REGISTRY
  54. def __init__(
  55. self,
  56. data: dict,
  57. log_service,
  58. db_client,
  59. trace_id: str,
  60. config: GlobalConfigSettings,
  61. ):
  62. self.data = data
  63. self.log_client = log_service
  64. self.db_client = db_client
  65. self.trace_id = trace_id
  66. self.config = config
  67. @classmethod
  68. def get_handler(cls, task_name: str) -> Optional[Callable]:
  69. """获取任务处理器"""
  70. return cls._handlers.get(task_name)
  71. @classmethod
  72. def list_registered_tasks(cls) -> list:
  73. """列出所有已注册的任务"""
  74. return list(cls._handlers.keys())
  75. async def _log_task_event(self, event_type: str, **kwargs):
  76. """统一的任务日志记录"""
  77. log_data = {
  78. "timestamp": datetime.now().isoformat(),
  79. "trace_id": self.trace_id,
  80. "event_type": event_type,
  81. "task": self.data.get("task_name"),
  82. **kwargs,
  83. }
  84. await self.log_client.log(contents=log_data)
  85. # ==================== 监控类任务 ====================
  86. @register("check_kimi_balance")
  87. async def _check_kimi_balance_handler(self) -> int:
  88. """检查 Kimi 余额"""
  89. response = await check_kimi_balance()
  90. await self._log_task_event("kimi_balance_checked", data=response)
  91. return TaskStatus.SUCCESS
  92. @register("get_off_videos")
  93. async def _get_off_videos_task_handler(self) -> int:
  94. """视频下架任务"""
  95. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  96. return await sub_task.deal()
  97. @register("check_publish_video_audit_status")
  98. async def _check_video_audit_status_handler(self) -> int:
  99. """检查视频审核状态"""
  100. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  101. return await sub_task.deal()
  102. @register("task_processing_monitor")
  103. async def _task_processing_monitor_handler(self) -> int:
  104. """任务处理监控"""
  105. sub_task = TaskProcessingMonitor(self.db_client)
  106. await sub_task.deal()
  107. return TaskStatus.SUCCESS
  108. @register("inner_article_monitor")
  109. async def _inner_gzh_articles_monitor_handler(self) -> int:
  110. """内部公众号文章监控"""
  111. sub_task = InnerGzhArticlesMonitor(self.db_client)
  112. return await sub_task.deal()
  113. @register("outside_article_monitor")
  114. async def _outside_monitor_handler(self) -> int:
  115. """外部文章监控"""
  116. collector = OutsideGzhArticlesCollector(self.db_client)
  117. await collector.deal()
  118. monitor = OutsideGzhArticlesMonitor(self.db_client)
  119. return await monitor.deal()
  120. @register("cooperate_accounts_monitor")
  121. async def _cooperate_accounts_monitor_handler(self) -> int:
  122. """合作账号文章监测"""
  123. task = CooperateAccountsMonitorTask(
  124. pool=self.db_client, log_client=self.log_client
  125. )
  126. await task.deal(task_name="save_articles")
  127. return TaskStatus.SUCCESS
  128. @register("cooperate_accounts_detail")
  129. async def _cooperate_accounts_detail_handler(self) -> int:
  130. """合作账号文章详情更新"""
  131. task = CooperateAccountsMonitorTask(
  132. pool=self.db_client, log_client=self.log_client
  133. )
  134. await task.deal(task_name="get_detail")
  135. return TaskStatus.SUCCESS
  136. @register("ad_platform_accounts_crawler")
  137. async def _ad_platform_accounts_crawler_handler(self) -> int:
  138. """广告平台账号监测"""
  139. task = AdPlatformAccountsMonitorTask(
  140. pool=self.db_client, log_service=self.log_client
  141. )
  142. # 抓账号内所有的文章
  143. await task.deal(task_name="crawl_articles")
  144. # 抓完之后, 计算账号的阅读量中位数
  145. await task.deal(task_name="cal_read_median")
  146. return TaskStatus.SUCCESS
  147. @register("ad_platform_article_detail")
  148. async def _ad_platform_article_detail_handler(self) -> int:
  149. """广告平台文章详情更新"""
  150. task = AdPlatformAccountsMonitorTask(
  151. pool=self.db_client, log_service=self.log_client
  152. )
  153. await task.deal(task_name="get_detail")
  154. return TaskStatus.SUCCESS
  155. # ==================== 爬虫类任务 ====================
  156. @register("crawler_toutiao")
  157. async def _crawler_toutiao_handler(self) -> int:
  158. """头条文章/视频抓取"""
  159. sub_task = CrawlerToutiao(
  160. self.db_client, self.log_client, self.trace_id, self.config
  161. )
  162. method = self.data.get("method", "account")
  163. media_type = self.data.get("media_type", "article")
  164. category_list = self.data.get("category_list", [])
  165. match method:
  166. case "account":
  167. await sub_task.crawler_task(media_type=media_type)
  168. case "recommend":
  169. await sub_task.crawl_toutiao_recommend_task(category_list)
  170. case "search":
  171. await sub_task.search_candidate_accounts()
  172. case _:
  173. raise TaskValidationError(f"Unsupported method: {method}")
  174. return TaskStatus.SUCCESS
  175. @register("crawler_gzh_articles")
  176. async def _crawler_gzh_article_handler(self) -> int:
  177. """抓取公众号文章"""
  178. account_method = self.data.get("account_method")
  179. crawl_mode = self.data.get("crawl_mode")
  180. strategy = self.data.get("strategy")
  181. match crawl_mode:
  182. case "account":
  183. task = CrawlerGzhAccountArticles(
  184. self.db_client, self.log_client, self.trace_id, self.config
  185. )
  186. await task.deal(account_method, strategy)
  187. case "search":
  188. task = CrawlerGzhSearchArticles(
  189. self.db_client, self.log_client, self.trace_id, self.config
  190. )
  191. await task.deal(strategy)
  192. case _:
  193. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  194. return TaskStatus.SUCCESS
  195. @register("crawler_account_manager")
  196. async def _crawler_account_manager_handler(self) -> int:
  197. """账号管理任务"""
  198. platform = self.data.get("platform", "weixin")
  199. account_id_list = self.data.get("account_id_list")
  200. match platform:
  201. case "weixin":
  202. task = WeixinAccountManager(
  203. self.db_client, self.log_client, self.trace_id
  204. )
  205. await task.deal(platform=platform, account_id_list=account_id_list)
  206. case _:
  207. raise TaskValidationError(f"Unsupported platform: {platform}")
  208. return TaskStatus.SUCCESS
  209. @register("crawler_detail_analysis")
  210. async def _crawler_article_analysis_handler(self) -> int:
  211. """抓取视频/文章详情分析统计"""
  212. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  213. await task.deal(params=self.data)
  214. return TaskStatus.SUCCESS
  215. # ==================== 数据回收类任务 ====================
  216. @register("article_detail_stat")
  217. async def _article_detail_stat_handler(self) -> int:
  218. """文章详情统计"""
  219. task = ArticleDetailStat(self.db_client, self.log_client)
  220. await task.deal()
  221. return TaskStatus.SUCCESS
  222. @register("daily_publish_articles_recycle")
  223. async def _recycle_article_data_handler(self) -> int:
  224. """每日发文数据回收"""
  225. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  226. recycle = RecycleDailyPublishArticlesTask(
  227. self.db_client, self.log_client, date_str
  228. )
  229. await recycle.deal()
  230. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  231. await check.deal()
  232. return TaskStatus.SUCCESS
  233. @register("update_root_source_id")
  234. async def _update_root_source_id_handler(self) -> int:
  235. """更新 root_source_id"""
  236. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  237. await sub_task.deal()
  238. return TaskStatus.SUCCESS
  239. @register("recycle_outside_account_articles")
  240. async def _recycle_outside_account_article_handler(self) -> int:
  241. """回收外部账号文章"""
  242. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  243. task = RecycleOutsideAccountArticlesTask(
  244. pool=self.db_client, log_client=self.log_client, date_string=date_str
  245. )
  246. await task.deal()
  247. return TaskStatus.SUCCESS
  248. @register("update_outside_account_article_root_source_id")
  249. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  250. """更新外部账号文章的 root_source_id"""
  251. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  252. pool=self.db_client, log_client=self.log_client
  253. )
  254. await task.deal()
  255. return TaskStatus.SUCCESS
  256. @register("fwh_daily_recycle")
  257. async def _recycle_fwh_article_handler(self) -> int:
  258. """回收服务号文章"""
  259. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  260. await task.deal()
  261. return TaskStatus.SUCCESS
  262. # ==================== 算法类任务 ====================
  263. @register("article_pool_cold_start")
  264. async def _article_pool_cold_start_handler(self) -> int:
  265. """文章池冷启动"""
  266. cold_start = ArticlePoolColdStart(
  267. self.db_client, self.log_client, self.trace_id, self.config
  268. )
  269. platform = self.data.get("platform", "weixin")
  270. crawler_methods = self.data.get("crawler_methods", [])
  271. category_list = self.data.get("category_list", [])
  272. strategy = self.data.get("strategy", "strategy_v1")
  273. await cold_start.deal(
  274. platform=platform,
  275. crawl_methods=crawler_methods,
  276. category_list=category_list,
  277. strategy=strategy,
  278. )
  279. return TaskStatus.SUCCESS
  280. @register("ad_platform_article_publish")
  281. async def _ad_platform_article_publish(self) -> int:
  282. """文章池冷启动"""
  283. task = AdPlatformArticlePublishTask(
  284. self.db_client, self.log_client
  285. )
  286. await task.deal()
  287. return TaskStatus.SUCCESS
  288. @register("candidate_account_quality_analysis")
  289. async def _candidate_account_quality_score_handler(self) -> int:
  290. """候选账号质量分析"""
  291. task = CandidateAccountQualityScoreRecognizer(
  292. self.db_client, self.log_client, self.trace_id
  293. )
  294. await task.deal()
  295. return TaskStatus.SUCCESS
  296. @register("article_pool_category_generation")
  297. async def _article_pool_category_generation_handler(self) -> int:
  298. """文章池品类生成"""
  299. task = ArticlePoolCategoryGeneration(
  300. self.db_client, self.log_client, self.trace_id
  301. )
  302. limit_num = self.data.get("limit")
  303. await task.deal(limit=limit_num)
  304. return TaskStatus.SUCCESS
  305. @register("account_category_analysis")
  306. async def _account_category_analysis_handler(self) -> int:
  307. """账号品类分析"""
  308. task = AccountCategoryAnalysis(
  309. pool=self.db_client,
  310. log_client=self.log_client,
  311. trace_id=self.trace_id,
  312. data=self.data,
  313. date_string=None,
  314. )
  315. await task.deal()
  316. return TaskStatus.SUCCESS
  317. @register("update_limited_account_info")
  318. async def _update_limited_account_info_handler(self) -> int:
  319. """更新限流账号信息"""
  320. task = LimitedAccountAnalysisTask(
  321. pool=self.db_client, log_client=self.log_client
  322. )
  323. await task.deal(date_string=self.data.get("date_string"))
  324. return TaskStatus.SUCCESS
  325. # ==================== LLM 类任务 ====================
  326. @register("title_rewrite")
  327. async def _title_rewrite_handler(self) -> int:
  328. """标题重写"""
  329. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  330. return await sub_task.deal()
  331. @register("extract_title_features")
  332. async def _extract_title_features_handler(self) -> int:
  333. """提取标题特征"""
  334. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  335. await task.deal(data=self.data)
  336. return TaskStatus.SUCCESS
  337. # ==================== 统计分析类任务 ====================
  338. @register("update_account_read_rate_avg")
  339. async def _update_account_read_rate_avg_handler(self) -> int:
  340. """更新账号阅读率均值"""
  341. task = AccountPositionReadRateAvg(
  342. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  343. )
  344. await task.deal(end_date=self.data.get("end_date"))
  345. return TaskStatus.SUCCESS
  346. @register("update_account_read_avg")
  347. async def _update_account_read_avg_handler(self) -> int:
  348. """更新账号阅读均值"""
  349. task = AccountPositionReadAvg(
  350. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  351. )
  352. await task.deal(end_date=self.data.get("end_date"))
  353. return TaskStatus.SUCCESS
  354. @register("update_account_open_rate_avg")
  355. async def _update_account_open_rate_avg_handler(self) -> int:
  356. """更新账号打开率均值"""
  357. task = AccountPositionOpenRateAvg(
  358. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  359. )
  360. await task.deal(date_string=self.data.get("date_string"))
  361. return TaskStatus.SUCCESS
  362. # ==================== 自动化任务 ====================
  363. @register("auto_follow_account")
  364. async def _auto_follow_account_handler(self) -> int:
  365. """自动关注公众号"""
  366. task = AutoReplyCardsMonitor(
  367. pool=self.db_client, log_service=self.log_client, config=self.config
  368. )
  369. await task.deal(task_name="follow_gzh_task")
  370. return TaskStatus.SUCCESS
  371. @register("get_follow_result")
  372. async def _get_follow_result_handler(self) -> int:
  373. """获取自动关注回复"""
  374. task = AutoReplyCardsMonitor(
  375. pool=self.db_client, log_service=self.log_client, config=self.config
  376. )
  377. await task.deal(task_name="get_auto_reply_task")
  378. return TaskStatus.SUCCESS
  379. @register("extract_reply_result")
  380. async def _extract_reply_result_handler(self) -> int:
  381. """解析自动回复结果"""
  382. task = AutoReplyCardsMonitor(
  383. pool=self.db_client, log_service=self.log_client, config=self.config
  384. )
  385. await task.deal(task_name="extract_task")
  386. return TaskStatus.SUCCESS
  387. # ==================== 其他任务 ====================
  388. @register("mini_program_detail_process")
  389. async def _mini_program_detail_handler(self) -> int:
  390. """更新小程序裂变信息"""
  391. task = RecycleMiniProgramDetailTask(
  392. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  393. )
  394. await task.deal(params=self.data)
  395. return TaskStatus.SUCCESS
  396. @register("rate_limited_article_filter")
  397. async def _rate_limited_article_filter(self) -> int:
  398. """限流文章删除"""
  399. task = RateLimitedArticleFilter(pool=self.db_client, config=self.config)
  400. await task.deal()
  401. return TaskStatus.SUCCESS
  402. @register("create_ad_platform_accounts_decode_task")
  403. async def _create_ad_platform_accounts_decode_task(self) -> int:
  404. """创建解构任务"""
  405. task = AdPlatformArticlesDecodeTask(
  406. pool=self.db_client, log_service=self.log_client
  407. )
  408. await task.deal(task_name="create_tasks")
  409. return TaskStatus.SUCCESS
  410. @register("fetch_ad_platform_accounts_decode_result")
  411. async def _fetch_ad_platform_accounts_decode_result(self) -> int:
  412. """获取解构任务结果"""
  413. task = AdPlatformArticlesDecodeTask(
  414. pool=self.db_client, log_service=self.log_client
  415. )
  416. await task.deal(task_name="fetch_results")
  417. return TaskStatus.SUCCESS
  418. __all__ = ["TaskHandler"]