process_deal.py 8.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237
  1. """
  2. @author: luojunhui
  3. """
  4. import asyncio
  5. from applications.static.config import db_article
  6. from applications.schedule import search_videos
  7. class ProcessDeal(object):
  8. """
  9. 定时执行任务
  10. """
  11. def __init__(self, mysql_client):
  12. """
  13. :param mysql_client:
  14. """
  15. self.mysql_client = mysql_client
  16. async def get_task(self):
  17. """
  18. 获取任务
  19. :return:
  20. """
  21. select_sql = f"""
  22. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  23. FROM {db_article}
  24. WHERE content_status = 0 and process_times <= 5
  25. ORDER BY request_time_stamp
  26. ASC
  27. LIMIT 10;
  28. """
  29. print(select_sql)
  30. task_list = await self.mysql_client.async_select(sql=select_sql)
  31. task_obj_list = [
  32. {
  33. "trace_id": item[0],
  34. "content_id": item[1],
  35. "gh_id": item[2],
  36. "title": item[3],
  37. "text": item[4],
  38. "content_status": item[5],
  39. "process_times": item[6]
  40. } for item in task_list
  41. ]
  42. return task_obj_list
  43. async def get_history_contents(self, content_id):
  44. """
  45. check whether the content id exists
  46. :return: trace_id or None
  47. """
  48. select_sql = f"""
  49. SELECT trace_id, content_status
  50. FROM {db_article}
  51. WHERE content_id = '{content_id}'
  52. ORDER BY id DESC;
  53. """
  54. result = await self.mysql_client.async_select(select_sql)
  55. if result:
  56. for item in result:
  57. trace_id, content_status = item
  58. if content_status == 2:
  59. return trace_id
  60. else:
  61. continue
  62. return None
  63. else:
  64. return None
  65. async def judge_content_processing(self, content_id):
  66. """
  67. 判断该content_id是否在处理中
  68. :param content_id:
  69. :return:
  70. """
  71. select_sql = f"""
  72. SELECT trace_id, content_status
  73. FROM {db_article}
  74. WHERE content_id = '{content_id}'
  75. ORDER BY id DESC;
  76. """
  77. result = await self.mysql_client.async_select(select_sql)
  78. if result:
  79. for item in result:
  80. trace_id, content_status = item
  81. if content_status == 1:
  82. return False
  83. return True
  84. else:
  85. return True
  86. async def insert_history_contents_videos(self, history_trace_id, params):
  87. """
  88. 插入历史视频id
  89. :return:
  90. """
  91. select_sql = f"""
  92. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  93. FROM {db_article}
  94. WHERE trace_id = '{history_trace_id}';
  95. """
  96. info = await self.mysql_client.async_select(sql=select_sql)
  97. kimi_title, vid1, vid2, vid3 = info[0]
  98. update_sql = f"""
  99. UPDATE {db_article}
  100. SET
  101. kimi_title='{kimi_title}',
  102. recall_video_id1={vid1},
  103. recall_video_id2={"NULL" if vid2 is None else vid2},
  104. recall_video_id3={"NULL" if vid3 is None else vid3},
  105. content_status=2,
  106. process_times = {int(params['process_times']) + 1}
  107. WHERE trace_id = '{params['trace_id']}'
  108. """
  109. await self.mysql_client.async_insert(update_sql)
  110. async def process_video_id(self, title, trace_id, process_times):
  111. """
  112. 如果video_id在标题中,则做特殊处理
  113. :return:
  114. """
  115. video_id = title.split("video_id=")[-1]
  116. update_sql = f"""
  117. UPDATE
  118. {db_article}
  119. SET
  120. recall_video_id1 = '{video_id}',
  121. content_status = 2,
  122. process_times = {int(process_times) + 1}
  123. WHERE
  124. trace_id = '{trace_id}';"""
  125. await self.mysql_client.async_insert(update_sql)
  126. async def start_process(self, params):
  127. """
  128. 开始处理
  129. :param params:
  130. :return:
  131. """
  132. # 更新文章contentId为1, 说明该文章正在处理中
  133. update_sql = f"""
  134. UPDATE {db_article}
  135. SET
  136. content_status = 1
  137. WHERE
  138. trace_id = '{params["trace_id"]}'
  139. """
  140. await self.mysql_client.async_insert(sql=update_sql)
  141. try:
  142. # 判断标题中是否包含video_id
  143. if "video_id=" in params['title']:
  144. await self.process_video_id(
  145. title=params['title'],
  146. trace_id=params['trace_id'],
  147. process_times=params['process_times']
  148. )
  149. else:
  150. print("开始搜索视频")
  151. await search_videos(
  152. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  153. trace_id=params['trace_id'],
  154. gh_id=params['gh_id'],
  155. mysql_client=self.mysql_client
  156. )
  157. # 执行完成之后,判断是否存在视频id
  158. select_sql = f"""
  159. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  160. FROM {db_article}
  161. WHERE trace_id = '{params["trace_id"]}';
  162. """
  163. result = await self.mysql_client.async_select(sql=select_sql)
  164. vid1, vid2, vid3 = result[0]
  165. if vid1:
  166. update_sql2 = f"""
  167. UPDATE {db_article}
  168. SET
  169. content_status = 2,
  170. process_times = {int(params['process_times']) + 1}
  171. WHERE trace_id = '{params["trace_id"]}';
  172. """
  173. await self.mysql_client.async_insert(sql=update_sql2)
  174. print("搜索视频成功")
  175. else:
  176. print("搜索视频失败")
  177. update_sql3 = f"""
  178. UPDATE {db_article}
  179. SET
  180. content_status = 0,
  181. process_times = {int(params['process_times']) + 1}
  182. WHERE trace_id = '{params["trace_id"]}';
  183. """
  184. await self.mysql_client.async_insert(sql=update_sql3)
  185. except Exception as e:
  186. print("{}异常错误:{}".format(params['trace_id'], e))
  187. update_sql4 = f"""
  188. UPDATE {db_article}
  189. SET
  190. content_status = 0,
  191. process_times = {int(params['process_times']) + 1}
  192. WHERE trace_id = '{params["trace_id"]}';
  193. """
  194. await self.mysql_client.async_insert(sql=update_sql4)
  195. async def process_task(self, params):
  196. """
  197. 异步执行
  198. :param params:
  199. :return:
  200. """
  201. content_id = params['content_id']
  202. # 判断该文章是否已经生成了
  203. history_trace_id = await self.get_history_contents(content_id)
  204. if history_trace_id:
  205. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  206. print("该文章已经成功请求")
  207. await self.insert_history_contents_videos(history_trace_id, params)
  208. else:
  209. flag = await self.judge_content_processing(content_id)
  210. if flag:
  211. print("开始处理这条视频")
  212. await self.start_process(params=params)
  213. else:
  214. print("该文章id正在处理中")
  215. async def deal(self):
  216. """
  217. 处理
  218. :return:
  219. """
  220. task_list = await self.get_task()
  221. if task_list:
  222. tasks = [self.process_task(params) for params in task_list]
  223. await asyncio.gather(*tasks)
  224. else:
  225. print("没有要处理的视频")