process_deal.py 7.4 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216
  1. """
  2. @author: luojunhui
  3. """
  4. from applications.static.config import db_article
  5. from applications.schedule import search_videos
  6. class ProcessDeal(object):
  7. """
  8. 定时执行任务
  9. """
  10. def __init__(self, mysql_client):
  11. """
  12. :param mysql_client:
  13. """
  14. self.mysql_client = mysql_client
  15. async def get_task(self):
  16. """
  17. 获取任务
  18. :return:
  19. """
  20. select_sql = f"""
  21. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  22. FROM {db_article}
  23. WHERE content_status = 0 and process_times <= 5
  24. ORDER BY request_time_stamp
  25. ASC
  26. LIMIT 10;
  27. """
  28. print(select_sql)
  29. task_list = await self.mysql_client.async_select(sql=select_sql)
  30. task_obj_list = [
  31. {
  32. "trace_id": item[0],
  33. "content_id": item[1],
  34. "gh_id": item[2],
  35. "title": item[3],
  36. "text": item[4],
  37. "content_status": item[5],
  38. "process_times": item[6]
  39. } for item in task_list
  40. ]
  41. return task_obj_list
  42. async def get_history_contents(self, content_id):
  43. """
  44. check whether the content id exists
  45. :return: trace_id or None
  46. """
  47. select_sql = f"""
  48. SELECT trace_id, content_status
  49. FROM {db_article}
  50. WHERE content_id = '{content_id}'
  51. ORDER BY id DESC;
  52. """
  53. result = await self.mysql_client.async_select(select_sql)
  54. if result:
  55. for item in result:
  56. trace_id, content_status = item
  57. if content_status == 2:
  58. return trace_id
  59. else:
  60. continue
  61. return None
  62. else:
  63. return None
  64. async def judge_content_processing(self, content_id):
  65. """
  66. 判断该content_id是否在处理中
  67. :param content_id:
  68. :return:
  69. """
  70. select_sql = f"""
  71. SELECT trace_id, content_status
  72. FROM {db_article}
  73. WHERE content_id = '{content_id}'
  74. ORDER BY id DESC;
  75. """
  76. result = await self.mysql_client.async_select(select_sql)
  77. if result:
  78. for item in result:
  79. trace_id, content_status = item
  80. if content_status == 1:
  81. return False
  82. return True
  83. else:
  84. return True
  85. async def insert_history_contents_videos(self, history_trace_id, params):
  86. """
  87. 插入历史视频id
  88. :return:
  89. """
  90. select_sql = f"""
  91. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  92. FROM {db_article}
  93. WHERE trace_id = '{history_trace_id}';
  94. """
  95. info = await self.mysql_client.async_select(sql=select_sql)
  96. kimi_title, vid1, vid2, vid3 = info[0]
  97. update_sql = f"""
  98. UPDATE {db_article}
  99. SET
  100. kimi_title='{kimi_title}',
  101. recall_video_id1={vid1},
  102. recall_video_id2={"NULL" if vid2 is None else vid2},
  103. recall_video_id3={"NULL" if vid3 is None else vid3},
  104. content_status=2,
  105. process_times = {int(params['process_times']) + 1}
  106. WHERE trace_id = '{params['trace_id']}'
  107. """
  108. await self.mysql_client.async_insert(update_sql)
  109. async def process_video_id(self, title, trace_id, process_times):
  110. """
  111. 如果video_id在标题中,则做特殊处理
  112. :return:
  113. """
  114. video_id = title.split("video_id=")[-1]
  115. update_sql = f"""
  116. UPDATE
  117. {db_article}
  118. SET
  119. recall_video_id1 = '{video_id}',
  120. content_status = 2,
  121. process_times = {int(process_times) + 1}
  122. WHERE
  123. trace_id = '{trace_id}';"""
  124. await self.mysql_client.async_insert(update_sql)
  125. async def start_process(self, params):
  126. """
  127. 开始处理
  128. :param params:
  129. :return:
  130. """
  131. # 更新文章contentId为1, 说明该文章正在处理中
  132. update_sql = f"""
  133. UPDATE {db_article}
  134. SET
  135. content_status = 1
  136. WHERE
  137. trace_id = '{params["trace_id"]}'
  138. """
  139. await self.mysql_client.async_insert(sql=update_sql)
  140. # 判断标题中是否包含video_id
  141. if "video_id=" in params['title']:
  142. await self.process_video_id(
  143. title=params['title'],
  144. trace_id=params['trace_id'],
  145. process_times=params['process_times']
  146. )
  147. else:
  148. print("开始搜索视频")
  149. await search_videos(
  150. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  151. trace_id=params['trace_id'],
  152. gh_id=params['gh_id'],
  153. mysql_client=self.mysql_client
  154. )
  155. # 执行完成之后,判断是否存在视频id
  156. select_sql = f"""
  157. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  158. FROM {db_article}
  159. WHERE trace_id = '{params["trace_id"]}';
  160. """
  161. result = await self.mysql_client.async_select(sql=select_sql)
  162. vid1, vid2, vid3 = result[0]
  163. if vid1:
  164. update_sql2 = f"""
  165. UPDATE {db_article}
  166. SET
  167. content_status = 2,
  168. process_times = {int(params['process_times']) + 1}
  169. WHERE trace_id = '{params["trace_id"]}';
  170. """
  171. await self.mysql_client.async_insert(sql=update_sql2)
  172. print("搜索视频成功")
  173. else:
  174. print("搜索视频失败")
  175. update_sql3 = f"""
  176. UPDATE {db_article}
  177. SET
  178. content_status = 0,
  179. process_times = {int(params['process_times']) + 1}
  180. WHERE trace_id = '{params["trace_id"]}';
  181. """
  182. await self.mysql_client.async_insert(sql=update_sql3)
  183. async def deal(self):
  184. """
  185. 处理
  186. :return:
  187. """
  188. task_list = await self.get_task()
  189. if task_list:
  190. for params in task_list:
  191. content_id = params['content_id']
  192. # 判断该文章是否已经生成了
  193. history_trace_id = await self.get_history_contents(content_id)
  194. if history_trace_id:
  195. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  196. print("该文章已经成功请求")
  197. await self.insert_history_contents_videos(history_trace_id, params)
  198. else:
  199. flag = await self.judge_content_processing(content_id)
  200. if flag:
  201. print("开始处理这条视频")
  202. await self.start_process(params=params)
  203. else:
  204. print("该文章id正在处理中")
  205. else:
  206. print("没有要处理的视频")