get_off_videos.py 8.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251
  1. import time
  2. import traceback
  3. from typing import List, Optional
  4. from tqdm import tqdm
  5. from applications.api import change_video_audit_status
  6. from applications.api import fetch_piaoquan_video_list_detail
  7. from applications.api import feishu_robot
  8. class GetOffVideosConst:
  9. EXPIRE_TIME = 3 * 24 * 3600
  10. EARLIEST_TIME = 7 * 24 * 3600
  11. VIDEO_AVAILABLE_STATUS = 1
  12. VIDEO_DISABLE_STATUS = 0
  13. # VIDEO_AUDIT
  14. VIDEO_AUDIT_FAIL_STATUS = 2
  15. VIDEO_AUDIT_SUCCESS_STATUS = 5
  16. # Task code
  17. TASK_SUCCESS_STATUS = 2
  18. TASK_FAILED_STATUS = 99
  19. # check status
  20. CHECK_INIT_STATUS = 0
  21. CHECK_FINISHED_STATUS = 1
  22. # table
  23. table = "get_off_videos"
  24. class GetOffVideos(GetOffVideosConst):
  25. def __init__(self, db_client, log_client, trace_id):
  26. self.db_client = db_client
  27. self.log_client = log_client
  28. self.trace_id = trace_id
  29. async def get_task_list(
  30. self, earliest_timestamp_threshold: int, expire_timestamp_threshold: int
  31. ):
  32. """get videos which need get off"""
  33. query = f"""
  34. select video_id from {self.table} where video_status = %s and publish_time between %s and %s;
  35. """
  36. video_list = await self.db_client.async_fetch(
  37. query,
  38. params=(
  39. self.VIDEO_AVAILABLE_STATUS,
  40. earliest_timestamp_threshold,
  41. expire_timestamp_threshold,
  42. ),
  43. )
  44. return video_list
  45. async def update_video_status(self, video_id):
  46. query = f"""
  47. update {self.table} set video_status = %s, get_off_time = %s where video_id = %s;
  48. """
  49. return await self.db_client.async_save(
  50. query, params=(self.VIDEO_DISABLE_STATUS, int(time.time()), video_id)
  51. )
  52. async def update_video_audit_status(self, video_id):
  53. """use pq api to update video status"""
  54. response = await change_video_audit_status(
  55. video_id, self.VIDEO_AUDIT_FAIL_STATUS
  56. )
  57. await self.update_video_status(video_id)
  58. return response
  59. async def get_off_job(self):
  60. """get off videos out of expire time"""
  61. expire_timestamp_threshold = int(time.time()) - self.EXPIRE_TIME
  62. earliest_timestamp_threshold = int(time.time()) - self.EARLIEST_TIME
  63. task_list = await self.get_task_list(
  64. earliest_timestamp_threshold, expire_timestamp_threshold
  65. )
  66. await self.log_client.log(
  67. contents={
  68. "task": "get_off_videos",
  69. "trace_id": self.trace_id,
  70. "message": f"获取{len(task_list)}条待下架视频",
  71. "data": task_list,
  72. }
  73. )
  74. success_count = 0
  75. failed_count = 0
  76. for task in tqdm(task_list):
  77. video_id = task["video_id"]
  78. if video_id in (0, 1):
  79. continue
  80. try:
  81. await self.update_video_audit_status(video_id)
  82. success_count += 1
  83. except Exception as e:
  84. await self.log_client.log(
  85. contents={
  86. "task": "get_off_videos",
  87. "function": "get_off_job",
  88. "status": "fail",
  89. "message": "get off video fail",
  90. "trace_id": self.trace_id,
  91. "data": {
  92. "video_id": video_id,
  93. "error": str(e),
  94. "traceback": traceback.format_exc(),
  95. },
  96. }
  97. )
  98. failed_count += 1
  99. if success_count or failed_count:
  100. await feishu_robot.bot(
  101. title="自动下架任务完成",
  102. detail={
  103. "成功下架视频数量": success_count,
  104. "失败数量": failed_count,
  105. },
  106. mention=False,
  107. )
  108. async def check(self):
  109. earliest_timestamp = int(time.time()) - self.EARLIEST_TIME
  110. expire_timestamp = int(time.time()) - self.EXPIRE_TIME
  111. task_list = await self.get_task_list(earliest_timestamp, expire_timestamp)
  112. if task_list:
  113. await feishu_robot.bot(
  114. title="自动下架视频失败",
  115. detail={
  116. "total_video": len(task_list),
  117. "video_list": [i["video_id"] for i in task_list],
  118. },
  119. mention=False,
  120. )
  121. return self.TASK_FAILED_STATUS
  122. else:
  123. return self.TASK_SUCCESS_STATUS
  124. async def deal(self):
  125. await self.get_off_job()
  126. task_status = await self.check()
  127. await self.log_client.log(
  128. contents={
  129. "task": "get_off_videos",
  130. "function": "deal",
  131. "trace_id": self.trace_id,
  132. "message": "任务执行完成",
  133. }
  134. )
  135. return task_status
  136. class CheckVideoAuditStatus(GetOffVideosConst):
  137. def __init__(self, db_client, log_client, trace_id):
  138. self.db_client = db_client
  139. self.log_client = log_client
  140. self.trace_id = trace_id
  141. async def get_video_list_status(self, video_list: List[int]):
  142. response = await fetch_piaoquan_video_list_detail(video_list)
  143. video_detail_list = response.get("data", [])
  144. if video_detail_list:
  145. bad_video_list = [
  146. i["id"]
  147. for i in video_detail_list
  148. if i["auditStatus"] != self.VIDEO_AUDIT_SUCCESS_STATUS
  149. ]
  150. else:
  151. bad_video_list = []
  152. return bad_video_list
  153. async def get_unchecked_video_list(self) -> Optional[List[int]]:
  154. """find unchecked videos"""
  155. query = f"""
  156. select video_id from {self.table} where check_status = %s and video_status = %s limit 1000;
  157. """
  158. video_id_list = await self.db_client.async_fetch(
  159. query, params=(self.CHECK_INIT_STATUS, self.VIDEO_AVAILABLE_STATUS)
  160. )
  161. if video_id_list:
  162. return [i["video_id"] for i in video_id_list]
  163. else:
  164. return None
  165. async def update_check_status(self, video_list: List[int]):
  166. query = f"""update {self.table} set check_status = %s where video_id in %s;"""
  167. return await self.db_client.async_save(
  168. query, params=(self.CHECK_FINISHED_STATUS, tuple(video_list))
  169. )
  170. async def deal(self):
  171. def chuck_iterator(arr, chunk_size):
  172. for i in range(0, len(arr), chunk_size):
  173. yield arr[i : i + chunk_size]
  174. video_id_list = await self.get_unchecked_video_list()
  175. if video_id_list:
  176. await self.log_client.log(
  177. contents={
  178. "task": "check_video_audit_status",
  179. "function": "deal",
  180. "trace_id": self.trace_id,
  181. "message": f"一共获取{len(video_id_list)}条视频",
  182. }
  183. )
  184. else:
  185. return self.TASK_SUCCESS_STATUS
  186. video_chunks = chuck_iterator(video_id_list, 10)
  187. bad_videos_count = 0
  188. fail_list = []
  189. batch = 0
  190. for video_chunk in video_chunks:
  191. batch += 1
  192. bad_video_id_list = await self.get_video_list_status(video_chunk)
  193. if bad_video_id_list:
  194. bad_videos_count += len(bad_video_id_list)
  195. await self.log_client.log(
  196. contents={
  197. "task": "check_video_audit_status",
  198. "function": "deal",
  199. "trace_id": self.trace_id,
  200. "message": f"batch: {batch} has {len(bad_video_id_list)} bad videos",
  201. "data": bad_video_id_list,
  202. }
  203. )
  204. for bad_video_id in tqdm(bad_video_id_list):
  205. response = await change_video_audit_status(bad_video_id)
  206. if not response:
  207. fail_list.append(bad_video_id)
  208. await self.update_check_status(video_chunk)
  209. await self.log_client.log(
  210. contents={
  211. "task": "check_video_audit_status",
  212. "function": "deal",
  213. "trace_id": self.trace_id,
  214. "message": f"finish process batch: {batch}",
  215. }
  216. )
  217. if fail_list:
  218. await feishu_robot.bot(
  219. title="校验已发布视频状态出现错误", detail=fail_list, mention=False
  220. )
  221. return self.TASK_FAILED_STATUS
  222. else:
  223. return self.TASK_SUCCESS_STATUS