task_handler.py 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409410411412413414415416417418419420
  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.algorithm_tasks import AccountCategoryAnalysis
  9. from app.domains.cold_start_tasks import ArticlePoolColdStart
  10. from app.domains.crawler_tasks import CrawlerToutiao
  11. from app.domains.crawler_tasks import WeixinAccountManager
  12. from app.domains.crawler_tasks import CrawlerGzhAccountArticles
  13. from app.domains.crawler_tasks import CrawlerGzhSearchArticles
  14. from app.domains.data_recycle_tasks import ArticleDetailStat
  15. from app.domains.data_recycle_tasks import RecycleDailyPublishArticlesTask
  16. from app.domains.data_recycle_tasks import RecycleOutsideAccountArticlesTask
  17. from app.domains.data_recycle_tasks import CheckDailyPublishArticlesTask
  18. from app.domains.data_recycle_tasks import UpdateRootSourceIdAndUpdateTimeTask
  19. from app.domains.data_recycle_tasks import RecycleFwhDailyPublishArticlesTask
  20. from app.domains.data_recycle_tasks import RecycleMiniProgramDetailTask
  21. from app.domains.data_recycle_tasks import (
  22. UpdateOutsideRootSourceIdAndUpdateTimeTask,
  23. )
  24. from app.domains.llm_tasks import TitleRewrite
  25. from app.domains.llm_tasks import ArticlePoolCategoryGeneration
  26. from app.domains.llm_tasks import CandidateAccountQualityScoreRecognizer
  27. from app.domains.llm_tasks import ExtractTitleFeatures
  28. from app.domains.monitor_tasks import AutoReplyCardsMonitor
  29. from app.domains.monitor_tasks import check_kimi_balance
  30. from app.domains.monitor_tasks import GetOffVideos
  31. from app.domains.monitor_tasks import CheckVideoAuditStatus
  32. from app.domains.monitor_tasks import CooperateAccountsMonitorTask
  33. from app.domains.monitor_tasks import InnerGzhArticlesMonitor
  34. from app.domains.monitor_tasks import OutsideGzhArticlesMonitor
  35. from app.domains.monitor_tasks import OutsideGzhArticlesCollector
  36. from app.domains.monitor_tasks import TaskProcessingMonitor
  37. from app.domains.monitor_tasks import LimitedAccountAnalysisTask
  38. from app.jobs.task_config import TaskStatus
  39. from app.jobs.task_utils import TaskValidationError
  40. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  41. def register(task_name: str):
  42. def decorator(func):
  43. _TASK_HANDLER_REGISTRY[task_name] = func
  44. return func
  45. return decorator
  46. class TaskHandler:
  47. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  48. # 任务注册表
  49. _handlers = _TASK_HANDLER_REGISTRY
  50. def __init__(self, data: dict, log_service, db_client, trace_id: str, config: GlobalConfigSettings):
  51. self.data = data
  52. self.log_client = log_service
  53. self.db_client = db_client
  54. self.trace_id = trace_id
  55. self.config = config
  56. @classmethod
  57. def get_handler(cls, task_name: str) -> Optional[Callable]:
  58. """获取任务处理器"""
  59. return cls._handlers.get(task_name)
  60. @classmethod
  61. def list_registered_tasks(cls) -> list:
  62. """列出所有已注册的任务"""
  63. return list(cls._handlers.keys())
  64. async def _log_task_event(self, event_type: str, **kwargs):
  65. """统一的任务日志记录"""
  66. log_data = {
  67. "timestamp": datetime.now().isoformat(),
  68. "trace_id": self.trace_id,
  69. "event_type": event_type,
  70. "task": self.data.get("task_name"),
  71. **kwargs,
  72. }
  73. await self.log_client.log(contents=log_data)
  74. # ==================== 监控类任务 ====================
  75. @register("check_kimi_balance")
  76. async def _check_kimi_balance_handler(self) -> int:
  77. """检查 Kimi 余额"""
  78. response = await check_kimi_balance()
  79. await self._log_task_event("kimi_balance_checked", data=response)
  80. return TaskStatus.SUCCESS
  81. @register("get_off_videos")
  82. async def _get_off_videos_task_handler(self) -> int:
  83. """视频下架任务"""
  84. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  85. return await sub_task.deal()
  86. @register("check_publish_video_audit_status")
  87. async def _check_video_audit_status_handler(self) -> int:
  88. """检查视频审核状态"""
  89. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  90. return await sub_task.deal()
  91. @register("task_processing_monitor")
  92. async def _task_processing_monitor_handler(self) -> int:
  93. """任务处理监控"""
  94. sub_task = TaskProcessingMonitor(self.db_client)
  95. await sub_task.deal()
  96. return TaskStatus.SUCCESS
  97. @register("inner_article_monitor")
  98. async def _inner_gzh_articles_monitor_handler(self) -> int:
  99. """内部公众号文章监控"""
  100. sub_task = InnerGzhArticlesMonitor(self.db_client)
  101. return await sub_task.deal()
  102. @register("outside_article_monitor")
  103. async def _outside_monitor_handler(self) -> int:
  104. """外部文章监控"""
  105. collector = OutsideGzhArticlesCollector(self.db_client)
  106. await collector.deal()
  107. monitor = OutsideGzhArticlesMonitor(self.db_client)
  108. return await monitor.deal()
  109. @register("cooperate_accounts_monitor")
  110. async def _cooperate_accounts_monitor_handler(self) -> int:
  111. """合作账号文章监测"""
  112. task = CooperateAccountsMonitorTask(
  113. pool=self.db_client, log_client=self.log_client
  114. )
  115. await task.deal(task_name="save_articles")
  116. return TaskStatus.SUCCESS
  117. @register("cooperate_accounts_detail")
  118. async def _cooperate_accounts_detail_handler(self) -> int:
  119. """合作账号文章详情更新"""
  120. task = CooperateAccountsMonitorTask(
  121. pool=self.db_client, log_client=self.log_client
  122. )
  123. await task.deal(task_name="get_detail")
  124. return TaskStatus.SUCCESS
  125. # ==================== 爬虫类任务 ====================
  126. @register("crawler_toutiao")
  127. async def _crawler_toutiao_handler(self) -> int:
  128. """头条文章/视频抓取"""
  129. sub_task = CrawlerToutiao(self.db_client, self.log_client, self.trace_id)
  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
  153. )
  154. await task.deal(account_method, strategy)
  155. case "search":
  156. task = CrawlerGzhSearchArticles(
  157. self.db_client, self.log_client, self.trace_id
  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.cold_start
  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("candidate_account_quality_analysis")
  249. async def _candidate_account_quality_score_handler(self) -> int:
  250. """候选账号质量分析"""
  251. task = CandidateAccountQualityScoreRecognizer(
  252. self.db_client, self.log_client, self.trace_id
  253. )
  254. await task.deal()
  255. return TaskStatus.SUCCESS
  256. @register("article_pool_category_generation")
  257. async def _article_pool_category_generation_handler(self) -> int:
  258. """文章池品类生成"""
  259. task = ArticlePoolCategoryGeneration(
  260. self.db_client, self.log_client, self.trace_id
  261. )
  262. limit_num = self.data.get("limit")
  263. await task.deal(limit=limit_num)
  264. return TaskStatus.SUCCESS
  265. @register("account_category_analysis")
  266. async def _account_category_analysis_handler(self) -> int:
  267. """账号品类分析"""
  268. task = AccountCategoryAnalysis(
  269. pool=self.db_client,
  270. log_client=self.log_client,
  271. trace_id=self.trace_id,
  272. data=self.data,
  273. date_string=None,
  274. )
  275. await task.deal()
  276. return TaskStatus.SUCCESS
  277. @register("update_limited_account_info")
  278. async def _update_limited_account_info_handler(self) -> int:
  279. """更新限流账号信息"""
  280. task = LimitedAccountAnalysisTask(
  281. pool=self.db_client, log_client=self.log_client
  282. )
  283. await task.deal(date_string=self.data.get("date_string"))
  284. return TaskStatus.SUCCESS
  285. # ==================== LLM 类任务 ====================
  286. @register("title_rewrite")
  287. async def _title_rewrite_handler(self) -> int:
  288. """标题重写"""
  289. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  290. return await sub_task.deal()
  291. @register("extract_title_features")
  292. async def _extract_title_features_handler(self) -> int:
  293. """提取标题特征"""
  294. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  295. await task.deal(data=self.data)
  296. return TaskStatus.SUCCESS
  297. # ==================== 统计分析类任务 ====================
  298. @register("update_account_read_rate_avg")
  299. async def _update_account_read_rate_avg_handler(self) -> int:
  300. """更新账号阅读率均值"""
  301. task = AccountPositionReadRateAvg(
  302. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  303. )
  304. await task.deal(end_date=self.data.get("end_date"))
  305. return TaskStatus.SUCCESS
  306. @register("update_account_read_avg")
  307. async def _update_account_read_avg_handler(self) -> int:
  308. """更新账号阅读均值"""
  309. task = AccountPositionReadAvg(
  310. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  311. )
  312. await task.deal(end_date=self.data.get("end_date"))
  313. return TaskStatus.SUCCESS
  314. @register("update_account_open_rate_avg")
  315. async def _update_account_open_rate_avg_handler(self) -> int:
  316. """更新账号打开率均值"""
  317. task = AccountPositionOpenRateAvg(
  318. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  319. )
  320. await task.deal(date_string=self.data.get("date_string"))
  321. return TaskStatus.SUCCESS
  322. # ==================== 自动化任务 ====================
  323. @register("auto_follow_account")
  324. async def _auto_follow_account_handler(self) -> int:
  325. """自动关注公众号"""
  326. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  327. await task.deal(task_name="follow_gzh_task")
  328. return TaskStatus.SUCCESS
  329. @register("get_follow_result")
  330. async def _get_follow_result_handler(self) -> int:
  331. """获取自动关注回复"""
  332. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  333. await task.deal(task_name="get_auto_reply_task")
  334. return TaskStatus.SUCCESS
  335. @register("extract_reply_result")
  336. async def _extract_reply_result_handler(self) -> int:
  337. """解析自动回复结果"""
  338. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  339. await task.deal(task_name="extract_task")
  340. return TaskStatus.SUCCESS
  341. # ==================== 其他任务 ====================
  342. @register("mini_program_detail_process")
  343. async def _mini_program_detail_handler(self) -> int:
  344. """更新小程序裂变信息"""
  345. task = RecycleMiniProgramDetailTask(
  346. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  347. )
  348. await task.deal(params=self.data)
  349. return TaskStatus.SUCCESS
  350. __all__ = ["TaskHandler"]