task_handler.py 16 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376377378379380381382383384385386387388389390391392393394395396397398399400401402403404405406407408409
  1. from datetime import datetime
  2. from typing import Callable, Awaitable, Dict, Any, Optional
  3. from applications.tasks.analysis_task import CrawlerDetailDeal
  4. from applications.tasks.analysis_task import AccountPositionReadRateAvg
  5. from applications.tasks.analysis_task import AccountPositionReadAvg
  6. from applications.tasks.analysis_task import AccountPositionOpenRateAvg
  7. from applications.tasks.algorithm_tasks import AccountCategoryAnalysis
  8. from applications.tasks.cold_start_tasks import ArticlePoolColdStart
  9. from applications.tasks.crawler_tasks import CrawlerToutiao
  10. from applications.tasks.crawler_tasks import WeixinAccountManager
  11. from applications.tasks.crawler_tasks import CrawlerGzhAccountArticles
  12. from applications.tasks.crawler_tasks import CrawlerGzhSearchArticles
  13. from applications.tasks.data_recycle_tasks import RecycleDailyPublishArticlesTask
  14. from applications.tasks.data_recycle_tasks import RecycleOutsideAccountArticlesTask
  15. from applications.tasks.data_recycle_tasks import CheckDailyPublishArticlesTask
  16. from applications.tasks.data_recycle_tasks import UpdateRootSourceIdAndUpdateTimeTask
  17. from applications.tasks.data_recycle_tasks import RecycleFwhDailyPublishArticlesTask
  18. from applications.tasks.data_recycle_tasks import RecycleMiniProgramDetailTask
  19. from applications.tasks.data_recycle_tasks import (
  20. UpdateOutsideRootSourceIdAndUpdateTimeTask,
  21. )
  22. from applications.tasks.llm_tasks import TitleRewrite
  23. from applications.tasks.llm_tasks import ArticlePoolCategoryGeneration
  24. from applications.tasks.llm_tasks import CandidateAccountQualityScoreRecognizer
  25. from applications.tasks.llm_tasks import ExtractTitleFeatures
  26. from applications.tasks.monitor_tasks import AutoReplyCardsMonitor
  27. from applications.tasks.monitor_tasks import check_kimi_balance
  28. from applications.tasks.monitor_tasks import GetOffVideos
  29. from applications.tasks.monitor_tasks import CheckVideoAuditStatus
  30. from applications.tasks.monitor_tasks import CooperateAccountsMonitorTask
  31. from applications.tasks.monitor_tasks import InnerGzhArticlesMonitor
  32. from applications.tasks.monitor_tasks import OutsideGzhArticlesMonitor
  33. from applications.tasks.monitor_tasks import OutsideGzhArticlesCollector
  34. from applications.tasks.monitor_tasks import TaskProcessingMonitor
  35. from applications.tasks.monitor_tasks import LimitedAccountAnalysisTask
  36. from applications.tasks.task_config import TaskStatus
  37. from applications.tasks.task_utils import TaskValidationError
  38. _TASK_HANDLER_REGISTRY: Dict[str, Callable] = {}
  39. def register(task_name: str):
  40. def decorator(func):
  41. _TASK_HANDLER_REGISTRY[task_name] = func
  42. return func
  43. return decorator
  44. class TaskHandler:
  45. """任务处理器基类 - 使用装饰器模式自动注册任务"""
  46. # 任务注册表
  47. _handlers = _TASK_HANDLER_REGISTRY
  48. def __init__(self, data: dict, log_service, db_client, trace_id: str):
  49. self.data = data
  50. self.log_client = log_service
  51. self.db_client = db_client
  52. self.trace_id = trace_id
  53. @classmethod
  54. def get_handler(cls, task_name: str) -> Optional[Callable]:
  55. """获取任务处理器"""
  56. return cls._handlers.get(task_name)
  57. @classmethod
  58. def list_registered_tasks(cls) -> list:
  59. """列出所有已注册的任务"""
  60. return list(cls._handlers.keys())
  61. async def _log_task_event(self, event_type: str, **kwargs):
  62. """统一的任务日志记录"""
  63. log_data = {
  64. "timestamp": datetime.now().isoformat(),
  65. "trace_id": self.trace_id,
  66. "event_type": event_type,
  67. "task": self.data.get("task_name"),
  68. **kwargs,
  69. }
  70. await self.log_client.log(contents=log_data)
  71. # ==================== 监控类任务 ====================
  72. @register("check_kimi_balance")
  73. async def _check_kimi_balance_handler(self) -> int:
  74. """检查 Kimi 余额"""
  75. response = await check_kimi_balance()
  76. await self._log_task_event("kimi_balance_checked", data=response)
  77. return TaskStatus.SUCCESS
  78. @register("get_off_videos")
  79. async def _get_off_videos_task_handler(self) -> int:
  80. """视频下架任务"""
  81. sub_task = GetOffVideos(self.db_client, self.log_client, self.trace_id)
  82. return await sub_task.deal()
  83. @register("check_publish_video_audit_status")
  84. async def _check_video_audit_status_handler(self) -> int:
  85. """检查视频审核状态"""
  86. sub_task = CheckVideoAuditStatus(self.db_client, self.log_client, self.trace_id)
  87. return await sub_task.deal()
  88. @register("task_processing_monitor")
  89. async def _task_processing_monitor_handler(self) -> int:
  90. """任务处理监控"""
  91. sub_task = TaskProcessingMonitor(self.db_client)
  92. await sub_task.deal()
  93. return TaskStatus.SUCCESS
  94. @register("inner_article_monitor")
  95. async def _inner_gzh_articles_monitor_handler(self) -> int:
  96. """内部公众号文章监控"""
  97. sub_task = InnerGzhArticlesMonitor(self.db_client)
  98. return await sub_task.deal()
  99. @register("outside_article_monitor")
  100. async def _outside_monitor_handler(self) -> int:
  101. """外部文章监控"""
  102. collector = OutsideGzhArticlesCollector(self.db_client)
  103. await collector.deal()
  104. monitor = OutsideGzhArticlesMonitor(self.db_client)
  105. return await monitor.deal()
  106. @register("cooperate_accounts_monitor")
  107. async def _cooperate_accounts_monitor_handler(self) -> int:
  108. """合作账号文章监测"""
  109. task = CooperateAccountsMonitorTask(
  110. pool=self.db_client, log_client=self.log_client
  111. )
  112. await task.deal(task_name="save_articles")
  113. return TaskStatus.SUCCESS
  114. @register("cooperate_accounts_detail")
  115. async def _cooperate_accounts_detail_handler(self) -> int:
  116. """合作账号文章详情更新"""
  117. task = CooperateAccountsMonitorTask(
  118. pool=self.db_client, log_client=self.log_client
  119. )
  120. await task.deal(task_name="get_detail")
  121. return TaskStatus.SUCCESS
  122. # ==================== 爬虫类任务 ====================
  123. @register("crawler_toutiao")
  124. async def _crawler_toutiao_handler(self) -> int:
  125. """头条文章/视频抓取"""
  126. sub_task = CrawlerToutiao(self.db_client, self.log_client, self.trace_id)
  127. method = self.data.get("method", "account")
  128. media_type = self.data.get("media_type", "article")
  129. category_list = self.data.get("category_list", [])
  130. match method:
  131. case "account":
  132. await sub_task.crawler_task(media_type=media_type)
  133. case "recommend":
  134. await sub_task.crawl_toutiao_recommend_task(category_list)
  135. case "search":
  136. await sub_task.search_candidate_accounts()
  137. case _:
  138. raise TaskValidationError(f"Unsupported method: {method}")
  139. return TaskStatus.SUCCESS
  140. @register("crawler_gzh_articles")
  141. async def _crawler_gzh_article_handler(self) -> int:
  142. """抓取公众号文章"""
  143. account_method = self.data.get("account_method")
  144. crawl_mode = self.data.get("crawl_mode")
  145. strategy = self.data.get("strategy")
  146. match crawl_mode:
  147. case "account":
  148. task = CrawlerGzhAccountArticles(
  149. self.db_client, self.log_client, self.trace_id
  150. )
  151. await task.deal(account_method, strategy)
  152. case "search":
  153. task = CrawlerGzhSearchArticles(
  154. self.db_client, self.log_client, self.trace_id
  155. )
  156. await task.deal(strategy)
  157. case _:
  158. raise TaskValidationError(f"Unsupported crawl mode: {crawl_mode}")
  159. return TaskStatus.SUCCESS
  160. @register("crawler_account_manager")
  161. async def _crawler_account_manager_handler(self) -> int:
  162. """账号管理任务"""
  163. platform = self.data.get("platform", "weixin")
  164. account_id_list = self.data.get("account_id_list")
  165. match platform:
  166. case "weixin":
  167. task = WeixinAccountManager(
  168. self.db_client, self.log_client, self.trace_id
  169. )
  170. await task.deal(platform=platform, account_id_list=account_id_list)
  171. case _:
  172. raise TaskValidationError(f"Unsupported platform: {platform}")
  173. return TaskStatus.SUCCESS
  174. @register("crawler_detail_analysis")
  175. async def _crawler_article_analysis_handler(self) -> int:
  176. """抓取视频/文章详情分析统计"""
  177. task = CrawlerDetailDeal(pool=self.db_client, trace_id=self.trace_id)
  178. await task.deal(params=self.data)
  179. return TaskStatus.SUCCESS
  180. # ==================== 数据回收类任务 ====================
  181. @register("daily_publish_articles_recycle")
  182. async def _recycle_article_data_handler(self) -> int:
  183. """每日发文数据回收"""
  184. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  185. recycle = RecycleDailyPublishArticlesTask(
  186. self.db_client, self.log_client, date_str
  187. )
  188. await recycle.deal()
  189. check = CheckDailyPublishArticlesTask(self.db_client, self.log_client, date_str)
  190. await check.deal()
  191. return TaskStatus.SUCCESS
  192. @register("update_root_source_id")
  193. async def _update_root_source_id_handler(self) -> int:
  194. """更新 root_source_id"""
  195. sub_task = UpdateRootSourceIdAndUpdateTimeTask(self.db_client, self.log_client)
  196. await sub_task.deal()
  197. return TaskStatus.SUCCESS
  198. @register("recycle_outside_account_articles")
  199. async def _recycle_outside_account_article_handler(self) -> int:
  200. """回收外部账号文章"""
  201. date_str = self.data.get("date_string") or datetime.now().strftime("%Y-%m-%d")
  202. task = RecycleOutsideAccountArticlesTask(
  203. pool=self.db_client, log_client=self.log_client, date_string=date_str
  204. )
  205. await task.deal()
  206. return TaskStatus.SUCCESS
  207. @register("update_outside_account_article_root_source_id")
  208. async def _update_outside_account_article_root_source_id_handler(self) -> int:
  209. """更新外部账号文章的 root_source_id"""
  210. task = UpdateOutsideRootSourceIdAndUpdateTimeTask(
  211. pool=self.db_client, log_client=self.log_client
  212. )
  213. await task.deal()
  214. return TaskStatus.SUCCESS
  215. @register("fwh_daily_recycle")
  216. async def _recycle_fwh_article_handler(self) -> int:
  217. """回收服务号文章"""
  218. task = RecycleFwhDailyPublishArticlesTask(self.db_client, self.log_client)
  219. await task.deal()
  220. return TaskStatus.SUCCESS
  221. # ==================== 算法类任务 ====================
  222. @register("article_pool_cold_start")
  223. async def _article_pool_cold_start_handler(self) -> int:
  224. """文章池冷启动"""
  225. cold_start = ArticlePoolColdStart(
  226. self.db_client, self.log_client, self.trace_id
  227. )
  228. platform = self.data.get("platform", "weixin")
  229. crawler_methods = self.data.get("crawler_methods", [])
  230. category_list = self.data.get("category_list", [])
  231. strategy = self.data.get("strategy", "strategy_v1")
  232. await cold_start.deal(
  233. platform=platform,
  234. crawl_methods=crawler_methods,
  235. category_list=category_list,
  236. strategy=strategy,
  237. )
  238. return TaskStatus.SUCCESS
  239. @register("candidate_account_quality_analysis")
  240. async def _candidate_account_quality_score_handler(self) -> int:
  241. """候选账号质量分析"""
  242. task = CandidateAccountQualityScoreRecognizer(
  243. self.db_client, self.log_client, self.trace_id
  244. )
  245. await task.deal()
  246. return TaskStatus.SUCCESS
  247. @register("article_pool_category_generation")
  248. async def _article_pool_category_generation_handler(self) -> int:
  249. """文章池品类生成"""
  250. task = ArticlePoolCategoryGeneration(
  251. self.db_client, self.log_client, self.trace_id
  252. )
  253. limit_num = self.data.get("limit")
  254. await task.deal(limit=limit_num)
  255. return TaskStatus.SUCCESS
  256. @register("account_category_analysis")
  257. async def _account_category_analysis_handler(self) -> int:
  258. """账号品类分析"""
  259. task = AccountCategoryAnalysis(
  260. pool=self.db_client,
  261. log_client=self.log_client,
  262. trace_id=self.trace_id,
  263. data=self.data,
  264. date_string=None,
  265. )
  266. await task.deal()
  267. return TaskStatus.SUCCESS
  268. @register("update_limited_account_info")
  269. async def _update_limited_account_info_handler(self) -> int:
  270. """更新限流账号信息"""
  271. task = LimitedAccountAnalysisTask(
  272. pool=self.db_client, log_client=self.log_client
  273. )
  274. await task.deal(date_string=self.data.get("date_string"))
  275. return TaskStatus.SUCCESS
  276. # ==================== LLM 类任务 ====================
  277. @register("title_rewrite")
  278. async def _title_rewrite_handler(self) -> int:
  279. """标题重写"""
  280. sub_task = TitleRewrite(self.db_client, self.log_client, self.trace_id)
  281. return await sub_task.deal()
  282. @register("extract_title_features")
  283. async def _extract_title_features_handler(self) -> int:
  284. """提取标题特征"""
  285. task = ExtractTitleFeatures(self.db_client, self.log_client, self.trace_id)
  286. await task.deal(data=self.data)
  287. return TaskStatus.SUCCESS
  288. # ==================== 统计分析类任务 ====================
  289. @register("update_account_read_rate_avg")
  290. async def _update_account_read_rate_avg_handler(self) -> int:
  291. """更新账号阅读率均值"""
  292. task = AccountPositionReadRateAvg(
  293. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  294. )
  295. await task.deal(end_date=self.data.get("end_date"))
  296. return TaskStatus.SUCCESS
  297. @register("update_account_read_avg")
  298. async def _update_account_read_avg_handler(self) -> int:
  299. """更新账号阅读均值"""
  300. task = AccountPositionReadAvg(
  301. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  302. )
  303. await task.deal(end_date=self.data.get("end_date"))
  304. return TaskStatus.SUCCESS
  305. @register("update_account_open_rate_avg")
  306. async def _update_account_open_rate_avg_handler(self) -> int:
  307. """更新账号打开率均值"""
  308. task = AccountPositionOpenRateAvg(
  309. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  310. )
  311. await task.deal(date_string=self.data.get("date_string"))
  312. return TaskStatus.SUCCESS
  313. # ==================== 自动化任务 ====================
  314. @register("auto_follow_account")
  315. async def _auto_follow_account_handler(self) -> int:
  316. """自动关注公众号"""
  317. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  318. await task.deal(task_name="follow_gzh_task")
  319. return TaskStatus.SUCCESS
  320. @register("get_follow_result")
  321. async def _get_follow_result_handler(self) -> int:
  322. """获取自动关注回复"""
  323. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  324. await task.deal(task_name="get_auto_reply_task")
  325. return TaskStatus.SUCCESS
  326. @register("extract_reply_result")
  327. async def _extract_reply_result_handler(self) -> int:
  328. """解析自动回复结果"""
  329. task = AutoReplyCardsMonitor(pool=self.db_client, log_client=self.log_client)
  330. await task.deal(task_name="extract_task")
  331. return TaskStatus.SUCCESS
  332. # ==================== 其他任务 ====================
  333. @register("mini_program_detail_process")
  334. async def _mini_program_detail_handler(self) -> int:
  335. """更新小程序裂变信息"""
  336. task = RecycleMiniProgramDetailTask(
  337. pool=self.db_client, log_client=self.log_client, trace_id=self.trace_id
  338. )
  339. await task.deal(params=self.data)
  340. return TaskStatus.SUCCESS
  341. __all__ = ["TaskHandler"]