process_deal.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301
  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. if len(content_ids_tuple) > 0:
  33. select_sql = f"""
  34. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  35. FROM {db_article}
  36. WHERE content_id in {content_ids_tuple}
  37. ORDER BY request_time_stamp
  38. ASC
  39. """
  40. task_list = await self.mysql_client.async_select(sql=select_sql)
  41. task_obj_list = [
  42. {
  43. "trace_id": item[0],
  44. "content_id": item[1],
  45. "gh_id": item[2],
  46. "title": item[3],
  47. "text": item[4],
  48. "content_status": item[5],
  49. "process_times": item[6]
  50. } for item in task_list
  51. ]
  52. logging(
  53. code="9001",
  54. info="本次任务获取到 {} 条视频".format(len(task_obj_list)),
  55. data=task_obj_list
  56. )
  57. return task_obj_list
  58. else:
  59. return []
  60. async def get_history_contents(self, content_id):
  61. """
  62. check whether the content id exists
  63. :return: trace_id or None
  64. """
  65. select_sql = f"""
  66. SELECT trace_id, content_status
  67. FROM {db_article}
  68. WHERE content_id = '{content_id}'
  69. ORDER BY id DESC;
  70. """
  71. result = await self.mysql_client.async_select(select_sql)
  72. if result:
  73. for item in result:
  74. trace_id, content_status = item
  75. if content_status == 2:
  76. return trace_id
  77. else:
  78. continue
  79. return None
  80. else:
  81. return None
  82. async def judge_content_processing(self, content_id):
  83. """
  84. 判断该content_id是否在处理中
  85. :param content_id:
  86. :return:
  87. """
  88. select_sql = f"""
  89. SELECT trace_id, content_status
  90. FROM {db_article}
  91. WHERE content_id = '{content_id}'
  92. ORDER BY id DESC;
  93. """
  94. result = await self.mysql_client.async_select(select_sql)
  95. if result:
  96. for item in result:
  97. trace_id, content_status = item
  98. if content_status == 1:
  99. return False
  100. return True
  101. else:
  102. return True
  103. async def insert_history_contents_videos(self, history_trace_id, params):
  104. """
  105. 插入历史视频id
  106. :return:
  107. """
  108. select_sql = f"""
  109. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  110. FROM {db_article}
  111. WHERE trace_id = '{history_trace_id}';
  112. """
  113. info = await self.mysql_client.async_select(sql=select_sql)
  114. kimi_title, vid1, vid2, vid3 = info[0]
  115. update_sql = f"""
  116. UPDATE {db_article}
  117. SET
  118. kimi_title='{kimi_title}',
  119. recall_video_id1={vid1},
  120. recall_video_id2={"NULL" if vid2 is None else vid2},
  121. recall_video_id3={"NULL" if vid3 is None else vid3},
  122. content_status=2,
  123. process_times = {int(params['process_times']) + 1}
  124. WHERE trace_id = '{params['trace_id']}'
  125. """
  126. await self.mysql_client.async_insert(update_sql)
  127. logging(
  128. code="9002",
  129. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  130. trace_id=params['trace_id']
  131. )
  132. async def process_video_id(self, title, trace_id, process_times):
  133. """
  134. 如果video_id在标题中,则做特殊处理
  135. :return:
  136. """
  137. video_id = title.split("video_id=")[-1]
  138. update_sql = f"""
  139. UPDATE
  140. {db_article}
  141. SET
  142. recall_video_id1 = '{video_id}',
  143. content_status = 2,
  144. process_times = {int(process_times) + 1}
  145. WHERE
  146. trace_id = '{trace_id}';"""
  147. await self.mysql_client.async_insert(update_sql)
  148. async def start_process(self, params):
  149. """
  150. 开始处理
  151. :param params:
  152. :return:
  153. """
  154. # 更新文章contentId为1, 说明该文章正在处理中
  155. update_sql = f"""
  156. UPDATE {db_article}
  157. SET
  158. content_status = 1
  159. WHERE
  160. trace_id = '{params["trace_id"]}'
  161. """
  162. await self.mysql_client.async_insert(sql=update_sql)
  163. try:
  164. # 判断标题中是否包含video_id
  165. if "video_id=" in params['title']:
  166. logging(
  167. code="9006",
  168. info="视频生成文本测试",
  169. trace_id=params['trace_id']
  170. )
  171. await self.process_video_id(
  172. title=params['title'],
  173. trace_id=params['trace_id'],
  174. process_times=params['process_times']
  175. )
  176. else:
  177. await search_videos(
  178. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  179. trace_id=params['trace_id'],
  180. gh_id=params['gh_id'],
  181. mysql_client=self.mysql_client
  182. )
  183. # 执行完成之后,判断是否存在视频id
  184. select_sql = f"""
  185. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  186. FROM {db_article}
  187. WHERE trace_id = '{params["trace_id"]}';
  188. """
  189. result = await self.mysql_client.async_select(sql=select_sql)
  190. vid1, vid2, vid3 = result[0]
  191. if vid1:
  192. update_sql2 = f"""
  193. UPDATE {db_article}
  194. SET
  195. content_status = 2,
  196. process_times = {int(params['process_times']) + 1}
  197. WHERE trace_id = '{params["trace_id"]}';
  198. """
  199. await self.mysql_client.async_insert(sql=update_sql2)
  200. logging(
  201. code="9008",
  202. info="视频搜索成功, 状态修改为2",
  203. trace_id=params['trace_id']
  204. )
  205. else:
  206. update_sql3 = f"""
  207. UPDATE {db_article}
  208. SET
  209. content_status = 0,
  210. process_times = {int(params['process_times']) + 1}
  211. WHERE trace_id = '{params["trace_id"]}';
  212. """
  213. await self.mysql_client.async_insert(sql=update_sql3)
  214. logging(
  215. code="9018",
  216. info="视频搜索失败,回退状态为0",
  217. trace_id=params['trace_id']
  218. )
  219. except Exception as e:
  220. logging(
  221. code="9018",
  222. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  223. trace_id=params['trace_id']
  224. )
  225. update_sql4 = f"""
  226. UPDATE {db_article}
  227. SET
  228. content_status = 0,
  229. process_times = {int(params['process_times']) + 1}
  230. WHERE trace_id = '{params["trace_id"]}';
  231. """
  232. await self.mysql_client.async_insert(sql=update_sql4)
  233. async def process_task(self, params):
  234. """
  235. 异步执行
  236. :param params:
  237. :return:
  238. """
  239. content_id = params['content_id']
  240. trace_id = params['trace_id']
  241. # 判断该文章是否已经生成了
  242. history_trace_id = await self.get_history_contents(content_id)
  243. if history_trace_id:
  244. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  245. logging(
  246. code="9001",
  247. info="存在历史文章",
  248. trace_id=trace_id
  249. )
  250. await self.insert_history_contents_videos(history_trace_id, params)
  251. else:
  252. flag = await self.judge_content_processing(content_id)
  253. if flag:
  254. logging(
  255. code="9004",
  256. info="无正在处理的文章ID, 开始处理",
  257. trace_id=trace_id
  258. )
  259. await self.start_process(params=params)
  260. else:
  261. logging(
  262. code="9003",
  263. info="该文章ID正在请求--文章ID {}".format(content_id),
  264. trace_id=trace_id
  265. )
  266. async def deal(self):
  267. """
  268. 处理
  269. :return:
  270. """
  271. task_list = await self.get_task()
  272. # print(task_list)
  273. task_dict = {}
  274. for task in task_list:
  275. key = task['content_id']
  276. task_dict[key] = task
  277. process_list = []
  278. for item in task_dict:
  279. process_list.append(task_dict[item])
  280. if process_list:
  281. # for task in task_list:
  282. # await self.process_task(task)
  283. tasks = [self.process_task(params) for params in process_list]
  284. await asyncio.gather(*tasks)
  285. else:
  286. logging(
  287. code="9008",
  288. info="没有要处理的请求"
  289. )