task2.py 9.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284
  1. """
  2. @author: luojunhui
  3. """
  4. import asyncio
  5. from static.config import db_article
  6. from applications.schedule import search_videos
  7. from applications.functions.log import logging
  8. from static.config import mysql_coroutines
  9. class MatchTask2(object):
  10. """
  11. 定时执行任务
  12. """
  13. def __init__(self, mysql_client):
  14. """
  15. :param mysql_client:
  16. """
  17. self.mysql_client = mysql_client
  18. async def get_task(self):
  19. """
  20. 获取任务
  21. :return:
  22. """
  23. select_sql = f"""
  24. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  25. FROM {db_article}
  26. WHERE content_status = 0 and process_times <= 5
  27. ORDER BY request_time_stamp
  28. ASC
  29. LIMIT {mysql_coroutines};
  30. """
  31. task_list = await self.mysql_client.async_select(sql=select_sql)
  32. task_obj_list = [
  33. {
  34. "trace_id": item[0],
  35. "content_id": item[1],
  36. "gh_id": item[2],
  37. "title": item[3],
  38. "text": item[4],
  39. "content_status": item[5],
  40. "process_times": item[6]
  41. } for item in task_list
  42. ]
  43. logging(
  44. code="9001",
  45. info="本次任务获取到 {} 条视频".format(len(task_obj_list)),
  46. data=task_obj_list
  47. )
  48. return task_obj_list
  49. async def get_history_contents(self, content_id):
  50. """
  51. check whether the content id exists
  52. :return: trace_id or None
  53. """
  54. select_sql = f"""
  55. SELECT trace_id, content_status
  56. FROM {db_article}
  57. WHERE content_id = '{content_id}'
  58. ORDER BY id DESC;
  59. """
  60. result = await self.mysql_client.async_select(select_sql)
  61. if result:
  62. for item in result:
  63. trace_id, content_status = item
  64. if content_status == 2:
  65. return trace_id
  66. elif content_status == 3:
  67. update_sql = f"""
  68. UPDATE {db_article}
  69. SET content_status = 3
  70. WHERE trace_id = %s;
  71. """
  72. await self.mysql_client.async_insert(update_sql, trace_id)
  73. else:
  74. continue
  75. return None
  76. else:
  77. return None
  78. async def insert_history_contents_videos(self, history_trace_id, params):
  79. """
  80. 插入历史视频id
  81. :return:
  82. """
  83. select_sql = f"""
  84. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  85. FROM {db_article}
  86. WHERE trace_id = '{history_trace_id}';
  87. """
  88. info = await self.mysql_client.async_select(sql=select_sql)
  89. kimi_title, vid1, vid2, vid3 = info[0]
  90. update_sql = f"""
  91. UPDATE {db_article}
  92. SET
  93. kimi_title=%s,
  94. recall_video_id1=%s,
  95. recall_video_id2=%s,
  96. recall_video_id3=%s,
  97. content_status=%s,
  98. process_times = %s
  99. WHERE trace_id = %s;
  100. """
  101. await self.mysql_client.async_insert(
  102. update_sql,
  103. params=(
  104. kimi_title,
  105. vid1,
  106. {"NULL" if vid2 is None else vid2},
  107. {"NULL" if vid2 is None else vid3},
  108. 2,
  109. int(params['process_times']) + 1,
  110. params['trace_id']
  111. )
  112. )
  113. logging(
  114. code="9002",
  115. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  116. trace_id=params['trace_id']
  117. )
  118. async def process_video_id(self, title, trace_id, process_times):
  119. """
  120. 如果video_id在标题中,则做特殊处理
  121. :return:
  122. """
  123. video_id = title.split("video_id=")[-1]
  124. update_sql = f"""
  125. UPDATE
  126. {db_article}
  127. SET
  128. recall_video_id1 = %s,
  129. content_status = %s,
  130. process_times = %s
  131. WHERE
  132. trace_id = %s;"""
  133. await self.mysql_client.async_insert(
  134. sql=update_sql,
  135. params=(video_id, 2, int(process_times) + 1, trace_id)
  136. )
  137. async def start_process(self, params):
  138. """
  139. 开始处理
  140. :param params:
  141. :return:
  142. """
  143. # 更新文章contentId为1, 说明该文章正在处理中
  144. update_sql = f"""
  145. UPDATE {db_article}
  146. SET
  147. content_status = %s
  148. WHERE
  149. trace_id = %s;
  150. """
  151. await self.mysql_client.async_insert(
  152. sql=update_sql,
  153. params=(1, params['trace_id'])
  154. )
  155. try:
  156. # 判断标题中是否包含video_id
  157. if "video_id=" in params['title']:
  158. logging(
  159. code="9006",
  160. info="视频生成文本测试",
  161. trace_id=params['trace_id']
  162. )
  163. await self.process_video_id(
  164. title=params['title'],
  165. trace_id=params['trace_id'],
  166. process_times=params['process_times']
  167. )
  168. else:
  169. await search_videos(
  170. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  171. trace_id=params['trace_id'],
  172. gh_id=params['gh_id'],
  173. mysql_client=self.mysql_client
  174. )
  175. # 执行完成之后,判断是否存在视频id
  176. select_sql = f"""
  177. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  178. FROM {db_article}
  179. WHERE trace_id = '{params["trace_id"]}';
  180. """
  181. result = await self.mysql_client.async_select(sql=select_sql)
  182. vid1, vid2, vid3 = result[0]
  183. if vid1:
  184. update_sql2 = f"""
  185. UPDATE {db_article}
  186. SET
  187. content_status = %s,
  188. process_times = %s
  189. WHERE trace_id = %s;
  190. """
  191. await self.mysql_client.async_insert(
  192. sql=update_sql2,
  193. params=(2, int(params['process_times']) + 1, params['trace_id'])
  194. )
  195. logging(
  196. code="9008",
  197. info="视频搜索成功, 状态修改为2",
  198. trace_id=params['trace_id']
  199. )
  200. else:
  201. update_sql3 = f"""
  202. UPDATE {db_article}
  203. SET
  204. content_status = %s,
  205. process_times = %s
  206. WHERE trace_id = %s;
  207. """
  208. await self.mysql_client.async_insert(
  209. sql=update_sql3,
  210. params=(0, int(params['process_times']) + 1, params["trace_id"])
  211. )
  212. logging(
  213. code="9018",
  214. info="视频搜索失败,回退状态为0",
  215. trace_id=params['trace_id']
  216. )
  217. except Exception as e:
  218. logging(
  219. code="9018",
  220. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  221. trace_id=params['trace_id']
  222. )
  223. update_sql4 = f"""
  224. UPDATE {db_article}
  225. SET
  226. content_status = %s,
  227. process_times = %s
  228. WHERE trace_id = %s;
  229. """
  230. await self.mysql_client.async_insert(
  231. sql=update_sql4,
  232. params=(0, int(params['process_times']) + 1, params["trace_id"])
  233. )
  234. async def process_task(self, params):
  235. """
  236. 异步执行
  237. :param params:
  238. :return:
  239. """
  240. content_id = params['content_id']
  241. trace_id = params['trace_id']
  242. # 判断该文章是否已经生成了
  243. history_trace_id = await self.get_history_contents(content_id)
  244. if history_trace_id:
  245. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  246. logging(
  247. code="9001",
  248. info="存在历史文章",
  249. trace_id=trace_id,
  250. function="find_history_article"
  251. )
  252. await self.insert_history_contents_videos(history_trace_id, params)
  253. else:
  254. logging(
  255. code="9003",
  256. info="未找到历史文章",
  257. trace_id=trace_id,
  258. function="find_history_article"
  259. )
  260. async def deal(self):
  261. """
  262. 处理
  263. :return:
  264. """
  265. task_list = await self.get_task()
  266. if task_list:
  267. tasks = [self.process_task(params) for params in task_list]
  268. await asyncio.gather(*tasks)
  269. else:
  270. logging(
  271. code="9008",
  272. info="没有要处理的请求"
  273. )