process_deal.py 11 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304
  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. cil = []
  32. for content_id in content_ids:
  33. cil.append(content_id[0])
  34. content_ids_tuple = str(cil).replace("[", "(").replace("]", ")")
  35. if len(content_ids_tuple) > 3:
  36. select_sql = f"""
  37. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  38. FROM {db_article}
  39. WHERE content_id in {content_ids_tuple}
  40. ORDER BY request_time_stamp
  41. ASC
  42. """
  43. print(select_sql)
  44. task_list = await self.mysql_client.async_select(sql=select_sql)
  45. task_obj_list = [
  46. {
  47. "trace_id": item[0],
  48. "content_id": item[1],
  49. "gh_id": item[2],
  50. "title": item[3],
  51. "text": item[4],
  52. "content_status": item[5],
  53. "process_times": item[6]
  54. } for item in task_list
  55. ]
  56. logging(
  57. code="9001",
  58. info="本次任务获取到 {} 条视频".format(len(task_obj_list)),
  59. data=task_obj_list
  60. )
  61. return task_obj_list
  62. else:
  63. return []
  64. async def get_history_contents(self, content_id):
  65. """
  66. check whether the content id exists
  67. :return: trace_id or None
  68. """
  69. select_sql = f"""
  70. SELECT trace_id, content_status
  71. FROM {db_article}
  72. WHERE content_id = '{content_id}'
  73. ORDER BY id DESC;
  74. """
  75. result = await self.mysql_client.async_select(select_sql)
  76. if result:
  77. for item in result:
  78. trace_id, content_status = item
  79. if content_status == 2:
  80. return trace_id
  81. else:
  82. continue
  83. return None
  84. else:
  85. return None
  86. async def judge_content_processing(self, content_id):
  87. """
  88. 判断该content_id是否在处理中
  89. :param content_id:
  90. :return:
  91. """
  92. select_sql = f"""
  93. SELECT trace_id, content_status
  94. FROM {db_article}
  95. WHERE content_id = '{content_id}'
  96. ORDER BY id DESC;
  97. """
  98. result = await self.mysql_client.async_select(select_sql)
  99. if result:
  100. for item in result:
  101. trace_id, content_status = item
  102. if content_status == 1:
  103. return False
  104. return True
  105. else:
  106. return True
  107. async def insert_history_contents_videos(self, history_trace_id, params):
  108. """
  109. 插入历史视频id
  110. :return:
  111. """
  112. select_sql = f"""
  113. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  114. FROM {db_article}
  115. WHERE trace_id = '{history_trace_id}';
  116. """
  117. info = await self.mysql_client.async_select(sql=select_sql)
  118. kimi_title, vid1, vid2, vid3 = info[0]
  119. update_sql = f"""
  120. UPDATE {db_article}
  121. SET
  122. kimi_title='{kimi_title}',
  123. recall_video_id1={vid1},
  124. recall_video_id2={"NULL" if vid2 is None else vid2},
  125. recall_video_id3={"NULL" if vid3 is None else vid3},
  126. content_status=2,
  127. process_times = {int(params['process_times']) + 1}
  128. WHERE trace_id = '{params['trace_id']}'
  129. """
  130. await self.mysql_client.async_insert(update_sql)
  131. logging(
  132. code="9002",
  133. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  134. trace_id=params['trace_id']
  135. )
  136. async def process_video_id(self, title, trace_id, process_times):
  137. """
  138. 如果video_id在标题中,则做特殊处理
  139. :return:
  140. """
  141. video_id = title.split("video_id=")[-1]
  142. update_sql = f"""
  143. UPDATE
  144. {db_article}
  145. SET
  146. recall_video_id1 = '{video_id}',
  147. content_status = 2,
  148. process_times = {int(process_times) + 1}
  149. WHERE
  150. trace_id = '{trace_id}';"""
  151. await self.mysql_client.async_insert(update_sql)
  152. async def start_process(self, params):
  153. """
  154. 开始处理
  155. :param params:
  156. :return:
  157. """
  158. # 更新文章contentId为1, 说明该文章正在处理中
  159. update_sql = f"""
  160. UPDATE {db_article}
  161. SET
  162. content_status = 1
  163. WHERE
  164. trace_id = '{params["trace_id"]}'
  165. """
  166. await self.mysql_client.async_insert(sql=update_sql)
  167. try:
  168. # 判断标题中是否包含video_id
  169. if "video_id=" in params['title']:
  170. logging(
  171. code="9006",
  172. info="视频生成文本测试",
  173. trace_id=params['trace_id']
  174. )
  175. await self.process_video_id(
  176. title=params['title'],
  177. trace_id=params['trace_id'],
  178. process_times=params['process_times']
  179. )
  180. else:
  181. await search_videos(
  182. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  183. trace_id=params['trace_id'],
  184. gh_id=params['gh_id'],
  185. mysql_client=self.mysql_client
  186. )
  187. # 执行完成之后,判断是否存在视频id
  188. select_sql = f"""
  189. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  190. FROM {db_article}
  191. WHERE trace_id = '{params["trace_id"]}';
  192. """
  193. result = await self.mysql_client.async_select(sql=select_sql)
  194. vid1, vid2, vid3 = result[0]
  195. if vid1:
  196. update_sql2 = f"""
  197. UPDATE {db_article}
  198. SET
  199. content_status = 2,
  200. process_times = {int(params['process_times']) + 1}
  201. WHERE trace_id = '{params["trace_id"]}';
  202. """
  203. await self.mysql_client.async_insert(sql=update_sql2)
  204. logging(
  205. code="9008",
  206. info="视频搜索成功, 状态修改为2",
  207. trace_id=params['trace_id']
  208. )
  209. else:
  210. update_sql3 = f"""
  211. UPDATE {db_article}
  212. SET
  213. content_status = 0,
  214. process_times = {int(params['process_times']) + 1}
  215. WHERE trace_id = '{params["trace_id"]}';
  216. """
  217. await self.mysql_client.async_insert(sql=update_sql3)
  218. logging(
  219. code="9018",
  220. info="视频搜索失败,回退状态为0",
  221. trace_id=params['trace_id']
  222. )
  223. except Exception as e:
  224. logging(
  225. code="9018",
  226. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  227. trace_id=params['trace_id']
  228. )
  229. update_sql4 = f"""
  230. UPDATE {db_article}
  231. SET
  232. content_status = 0,
  233. process_times = {int(params['process_times']) + 1}
  234. WHERE trace_id = '{params["trace_id"]}';
  235. """
  236. await self.mysql_client.async_insert(sql=update_sql4)
  237. async def process_task(self, params):
  238. """
  239. 异步执行
  240. :param params:
  241. :return:
  242. """
  243. content_id = params['content_id']
  244. trace_id = params['trace_id']
  245. # 判断该文章是否已经生成了
  246. history_trace_id = await self.get_history_contents(content_id)
  247. if history_trace_id:
  248. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  249. logging(
  250. code="9001",
  251. info="存在历史文章",
  252. trace_id=trace_id
  253. )
  254. await self.insert_history_contents_videos(history_trace_id, params)
  255. else:
  256. flag = await self.judge_content_processing(content_id)
  257. if flag:
  258. logging(
  259. code="9004",
  260. info="无正在处理的文章ID, 开始处理",
  261. trace_id=trace_id
  262. )
  263. await self.start_process(params=params)
  264. else:
  265. logging(
  266. code="9003",
  267. info="该文章ID正在请求--文章ID {}".format(content_id),
  268. trace_id=trace_id
  269. )
  270. async def deal(self):
  271. """
  272. 处理
  273. :return:
  274. """
  275. task_list = await self.get_task()
  276. task_dict = {}
  277. for task in task_list:
  278. key = task['content_id']
  279. task_dict[key] = task
  280. process_list = []
  281. for item in task_dict:
  282. process_list.append(task_dict[item])
  283. if process_list:
  284. # for task in task_list:
  285. # await self.process_task(task)
  286. tasks = [self.process_task(params) for params in process_list]
  287. await asyncio.gather(*tasks)
  288. else:
  289. logging(
  290. code="9008",
  291. info="没有要处理的请求"
  292. )