process_deal.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298
  1. """
  2. @author: luojunhui
  3. """
  4. import asyncio
  5. from applications.static.config import db_article
  6. from applications.schedule import search_videos
  7. from applications.functions.log import logging
  8. class ProcessDeal(object):
  9. """
  10. 定时执行任务
  11. """
  12. def __init__(self, mysql_client):
  13. """
  14. :param mysql_client:
  15. """
  16. self.mysql_client = mysql_client
  17. async def get_task(self):
  18. """
  19. 获取任务
  20. :return:
  21. """
  22. select_sql1 = f"""
  23. SELECT DISTINCT (content_id)
  24. FROM {db_article}
  25. WHERE content_status = 0 and process_times <= 5
  26. ORDER BY request_time_stamp
  27. ASC
  28. LIMIT 4;
  29. """
  30. content_ids = await self.mysql_client.async_select(select_sql1)
  31. content_ids_tuple = tuple([i[0] for i in content_ids])
  32. select_sql = f"""
  33. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  34. FROM {db_article}
  35. WHERE content_id in {content_ids_tuple}
  36. ORDER BY request_time_stamp
  37. ASC
  38. """
  39. task_list = await self.mysql_client.async_select(sql=select_sql)
  40. task_obj_list = [
  41. {
  42. "trace_id": item[0],
  43. "content_id": item[1],
  44. "gh_id": item[2],
  45. "title": item[3],
  46. "text": item[4],
  47. "content_status": item[5],
  48. "process_times": item[6]
  49. } for item in task_list
  50. ]
  51. logging(
  52. code="9001",
  53. info="本次任务获取到 {} 条视频".format(len(task_obj_list)),
  54. data=task_obj_list
  55. )
  56. return task_obj_list
  57. async def get_history_contents(self, content_id):
  58. """
  59. check whether the content id exists
  60. :return: trace_id or None
  61. """
  62. select_sql = f"""
  63. SELECT trace_id, content_status
  64. FROM {db_article}
  65. WHERE content_id = '{content_id}'
  66. ORDER BY id DESC;
  67. """
  68. result = await self.mysql_client.async_select(select_sql)
  69. if result:
  70. for item in result:
  71. trace_id, content_status = item
  72. if content_status == 2:
  73. return trace_id
  74. else:
  75. continue
  76. return None
  77. else:
  78. return None
  79. async def judge_content_processing(self, content_id):
  80. """
  81. 判断该content_id是否在处理中
  82. :param content_id:
  83. :return:
  84. """
  85. select_sql = f"""
  86. SELECT trace_id, content_status
  87. FROM {db_article}
  88. WHERE content_id = '{content_id}'
  89. ORDER BY id DESC;
  90. """
  91. result = await self.mysql_client.async_select(select_sql)
  92. if result:
  93. for item in result:
  94. trace_id, content_status = item
  95. if content_status == 1:
  96. return False
  97. return True
  98. else:
  99. return True
  100. async def insert_history_contents_videos(self, history_trace_id, params):
  101. """
  102. 插入历史视频id
  103. :return:
  104. """
  105. select_sql = f"""
  106. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  107. FROM {db_article}
  108. WHERE trace_id = '{history_trace_id}';
  109. """
  110. info = await self.mysql_client.async_select(sql=select_sql)
  111. kimi_title, vid1, vid2, vid3 = info[0]
  112. update_sql = f"""
  113. UPDATE {db_article}
  114. SET
  115. kimi_title='{kimi_title}',
  116. recall_video_id1={vid1},
  117. recall_video_id2={"NULL" if vid2 is None else vid2},
  118. recall_video_id3={"NULL" if vid3 is None else vid3},
  119. content_status=2,
  120. process_times = {int(params['process_times']) + 1}
  121. WHERE trace_id = '{params['trace_id']}'
  122. """
  123. await self.mysql_client.async_insert(update_sql)
  124. logging(
  125. code="9002",
  126. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  127. trace_id=params['trace_id']
  128. )
  129. async def process_video_id(self, title, trace_id, process_times):
  130. """
  131. 如果video_id在标题中,则做特殊处理
  132. :return:
  133. """
  134. video_id = title.split("video_id=")[-1]
  135. update_sql = f"""
  136. UPDATE
  137. {db_article}
  138. SET
  139. recall_video_id1 = '{video_id}',
  140. content_status = 2,
  141. process_times = {int(process_times) + 1}
  142. WHERE
  143. trace_id = '{trace_id}';"""
  144. await self.mysql_client.async_insert(update_sql)
  145. async def start_process(self, params):
  146. """
  147. 开始处理
  148. :param params:
  149. :return:
  150. """
  151. # 更新文章contentId为1, 说明该文章正在处理中
  152. update_sql = f"""
  153. UPDATE {db_article}
  154. SET
  155. content_status = 1
  156. WHERE
  157. trace_id = '{params["trace_id"]}'
  158. """
  159. await self.mysql_client.async_insert(sql=update_sql)
  160. try:
  161. # 判断标题中是否包含video_id
  162. if "video_id=" in params['title']:
  163. logging(
  164. code="9006",
  165. info="视频生成文本测试",
  166. trace_id=params['trace_id']
  167. )
  168. await self.process_video_id(
  169. title=params['title'],
  170. trace_id=params['trace_id'],
  171. process_times=params['process_times']
  172. )
  173. else:
  174. await search_videos(
  175. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  176. trace_id=params['trace_id'],
  177. gh_id=params['gh_id'],
  178. mysql_client=self.mysql_client
  179. )
  180. # 执行完成之后,判断是否存在视频id
  181. select_sql = f"""
  182. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  183. FROM {db_article}
  184. WHERE trace_id = '{params["trace_id"]}';
  185. """
  186. result = await self.mysql_client.async_select(sql=select_sql)
  187. vid1, vid2, vid3 = result[0]
  188. if vid1:
  189. update_sql2 = f"""
  190. UPDATE {db_article}
  191. SET
  192. content_status = 2,
  193. process_times = {int(params['process_times']) + 1}
  194. WHERE trace_id = '{params["trace_id"]}';
  195. """
  196. await self.mysql_client.async_insert(sql=update_sql2)
  197. logging(
  198. code="9008",
  199. info="视频搜索成功, 状态修改为2",
  200. trace_id=params['trace_id']
  201. )
  202. else:
  203. update_sql3 = f"""
  204. UPDATE {db_article}
  205. SET
  206. content_status = 0,
  207. process_times = {int(params['process_times']) + 1}
  208. WHERE trace_id = '{params["trace_id"]}';
  209. """
  210. await self.mysql_client.async_insert(sql=update_sql3)
  211. logging(
  212. code="9018",
  213. info="视频搜索失败,回退状态为0",
  214. trace_id=params['trace_id']
  215. )
  216. except Exception as e:
  217. logging(
  218. code="9018",
  219. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  220. trace_id=params['trace_id']
  221. )
  222. update_sql4 = f"""
  223. UPDATE {db_article}
  224. SET
  225. content_status = 0,
  226. process_times = {int(params['process_times']) + 1}
  227. WHERE trace_id = '{params["trace_id"]}';
  228. """
  229. await self.mysql_client.async_insert(sql=update_sql4)
  230. async def process_task(self, params):
  231. """
  232. 异步执行
  233. :param params:
  234. :return:
  235. """
  236. content_id = params['content_id']
  237. trace_id = params['trace_id']
  238. # 判断该文章是否已经生成了
  239. history_trace_id = await self.get_history_contents(content_id)
  240. if history_trace_id:
  241. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  242. logging(
  243. code="9001",
  244. info="存在历史文章",
  245. trace_id=trace_id
  246. )
  247. await self.insert_history_contents_videos(history_trace_id, params)
  248. else:
  249. flag = await self.judge_content_processing(content_id)
  250. if flag:
  251. logging(
  252. code="9004",
  253. info="无正在处理的文章ID, 开始处理",
  254. trace_id=trace_id
  255. )
  256. await self.start_process(params=params)
  257. else:
  258. logging(
  259. code="9003",
  260. info="该文章ID正在请求--文章ID {}".format(content_id),
  261. trace_id=trace_id
  262. )
  263. async def deal(self):
  264. """
  265. 处理
  266. :return:
  267. """
  268. task_list = await self.get_task()
  269. # print(task_list)
  270. task_dict = {}
  271. for task in task_list:
  272. key = task['content_id']
  273. task_dict[key] = task
  274. process_list = []
  275. for item in task_dict:
  276. process_list.append(task_dict[item])
  277. if process_list:
  278. # for task in task_list:
  279. # await self.process_task(task)
  280. tasks = [self.process_task(params) for params in process_list]
  281. await asyncio.gather(*tasks)
  282. else:
  283. logging(
  284. code="9008",
  285. info="没有要处理的请求"
  286. )