get_off_videos.py 8.6 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252
  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. and video_id > 1;
  36. """
  37. video_list = await self.db_client.async_fetch(
  38. query,
  39. params=(
  40. self.VIDEO_AVAILABLE_STATUS,
  41. earliest_timestamp_threshold,
  42. expire_timestamp_threshold,
  43. ),
  44. )
  45. return video_list
  46. async def update_video_status(self, video_id):
  47. query = f"""
  48. update {self.table} set video_status = %s, get_off_time = %s where video_id = %s;
  49. """
  50. return await self.db_client.async_save(
  51. query, params=(self.VIDEO_DISABLE_STATUS, int(time.time()), video_id)
  52. )
  53. async def update_video_audit_status(self, video_id):
  54. """use pq api to update video status"""
  55. response = await change_video_audit_status(
  56. video_id, self.VIDEO_AUDIT_FAIL_STATUS
  57. )
  58. await self.update_video_status(video_id)
  59. return response
  60. async def get_off_job(self):
  61. """get off videos out of expire time"""
  62. expire_timestamp_threshold = int(time.time()) - self.EXPIRE_TIME
  63. earliest_timestamp_threshold = int(time.time()) - self.EARLIEST_TIME
  64. task_list = await self.get_task_list(
  65. earliest_timestamp_threshold, expire_timestamp_threshold
  66. )
  67. await self.log_client.log(
  68. contents={
  69. "task": "get_off_videos",
  70. "trace_id": self.trace_id,
  71. "message": f"获取{len(task_list)}条待下架视频",
  72. "data": task_list,
  73. }
  74. )
  75. success_count = 0
  76. failed_count = 0
  77. for task in tqdm(task_list):
  78. video_id = task["video_id"]
  79. if video_id in (0, 1):
  80. continue
  81. try:
  82. await self.update_video_audit_status(video_id)
  83. success_count += 1
  84. except Exception as e:
  85. await self.log_client.log(
  86. contents={
  87. "task": "get_off_videos",
  88. "function": "get_off_job",
  89. "status": "fail",
  90. "message": "get off video fail",
  91. "trace_id": self.trace_id,
  92. "data": {
  93. "video_id": video_id,
  94. "error": str(e),
  95. "traceback": traceback.format_exc(),
  96. },
  97. }
  98. )
  99. failed_count += 1
  100. if success_count or failed_count:
  101. await feishu_robot.bot(
  102. title="自动下架任务完成",
  103. detail={
  104. "成功下架视频数量": success_count,
  105. "失败数量": failed_count,
  106. },
  107. mention=False,
  108. )
  109. async def check(self):
  110. earliest_timestamp = int(time.time()) - self.EARLIEST_TIME
  111. expire_timestamp = int(time.time()) - self.EXPIRE_TIME
  112. task_list = await self.get_task_list(earliest_timestamp, expire_timestamp)
  113. if task_list:
  114. await feishu_robot.bot(
  115. title="自动下架视频失败",
  116. detail={
  117. "total_video": len(task_list),
  118. "video_list": [i["video_id"] for i in task_list],
  119. },
  120. mention=False,
  121. )
  122. return self.TASK_FAILED_STATUS
  123. else:
  124. return self.TASK_SUCCESS_STATUS
  125. async def deal(self):
  126. await self.get_off_job()
  127. task_status = await self.check()
  128. await self.log_client.log(
  129. contents={
  130. "task": "get_off_videos",
  131. "function": "deal",
  132. "trace_id": self.trace_id,
  133. "message": "任务执行完成",
  134. }
  135. )
  136. return task_status
  137. class CheckVideoAuditStatus(GetOffVideosConst):
  138. def __init__(self, db_client, log_client, trace_id):
  139. self.db_client = db_client
  140. self.log_client = log_client
  141. self.trace_id = trace_id
  142. async def get_video_list_status(self, video_list: List[int]):
  143. response = await fetch_piaoquan_video_list_detail(video_list)
  144. video_detail_list = response.get("data", [])
  145. if video_detail_list:
  146. bad_video_list = [
  147. i["id"]
  148. for i in video_detail_list
  149. if i["auditStatus"] != self.VIDEO_AUDIT_SUCCESS_STATUS
  150. ]
  151. else:
  152. bad_video_list = []
  153. return bad_video_list
  154. async def get_unchecked_video_list(self) -> Optional[List[int]]:
  155. """find unchecked videos"""
  156. query = f"""
  157. select video_id from {self.table} where check_status = %s and video_status = %s and video_id > 1 limit 1000;
  158. """
  159. video_id_list = await self.db_client.async_fetch(
  160. query, params=(self.CHECK_INIT_STATUS, self.VIDEO_AVAILABLE_STATUS)
  161. )
  162. if video_id_list:
  163. return [i["video_id"] for i in video_id_list]
  164. else:
  165. return None
  166. async def update_check_status(self, video_list: List[int]):
  167. query = f"""update {self.table} set check_status = %s where video_id in %s;"""
  168. return await self.db_client.async_save(
  169. query, params=(self.CHECK_FINISHED_STATUS, tuple(video_list))
  170. )
  171. async def deal(self):
  172. def chuck_iterator(arr, chunk_size):
  173. for i in range(0, len(arr), chunk_size):
  174. yield arr[i : i + chunk_size]
  175. video_id_list = await self.get_unchecked_video_list()
  176. if video_id_list:
  177. await self.log_client.log(
  178. contents={
  179. "task": "check_video_audit_status",
  180. "function": "deal",
  181. "trace_id": self.trace_id,
  182. "message": f"一共获取{len(video_id_list)}条视频",
  183. }
  184. )
  185. else:
  186. return self.TASK_SUCCESS_STATUS
  187. video_chunks = chuck_iterator(video_id_list, 10)
  188. bad_videos_count = 0
  189. fail_list = []
  190. batch = 0
  191. for video_chunk in video_chunks:
  192. batch += 1
  193. bad_video_id_list = await self.get_video_list_status(video_chunk)
  194. if bad_video_id_list:
  195. bad_videos_count += len(bad_video_id_list)
  196. await self.log_client.log(
  197. contents={
  198. "task": "check_video_audit_status",
  199. "function": "deal",
  200. "trace_id": self.trace_id,
  201. "message": f"batch: {batch} has {len(bad_video_id_list)} bad videos",
  202. "data": bad_video_id_list,
  203. }
  204. )
  205. for bad_video_id in tqdm(bad_video_id_list):
  206. response = await change_video_audit_status(bad_video_id)
  207. if not response:
  208. fail_list.append(bad_video_id)
  209. await self.update_check_status(video_chunk)
  210. await self.log_client.log(
  211. contents={
  212. "task": "check_video_audit_status",
  213. "function": "deal",
  214. "trace_id": self.trace_id,
  215. "message": f"finish process batch: {batch}",
  216. }
  217. )
  218. if fail_list:
  219. await feishu_robot.bot(
  220. title="校验已发布视频状态出现错误", detail=fail_list, mention=False
  221. )
  222. return self.TASK_FAILED_STATUS
  223. else:
  224. return self.TASK_SUCCESS_STATUS