process_deal.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375
  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 6;
  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} and process_times <= 5
  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=%s,
  123. recall_video_id1=%s,
  124. recall_video_id2=%s,
  125. recall_video_id3=%s,
  126. content_status=%s,
  127. process_times = %s
  128. WHERE trace_id = %s
  129. """
  130. await self.mysql_client.async_insert(
  131. sql=update_sql,
  132. params=(
  133. kimi_title,
  134. vid1,
  135. "NULL" if vid2 is None else vid2,
  136. "NULL" if vid3 is None else vid3,
  137. 2,
  138. int(params['process_times']) + 1,
  139. params['trace_id']
  140. )
  141. )
  142. logging(
  143. code="9002",
  144. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  145. trace_id=params['trace_id']
  146. )
  147. async def process_video_id(self, title, trace_id, process_times):
  148. """
  149. 如果video_id在标题中,则做特殊处理
  150. :return:
  151. """
  152. video_id = title.split("video_id=")[-1]
  153. update_sql = f"""
  154. UPDATE
  155. {db_article}
  156. SET
  157. recall_video_id1 = %s,
  158. content_status = %s,
  159. process_times = %s
  160. WHERE
  161. trace_id = %s;"""
  162. await self.mysql_client.async_insert(
  163. sql=update_sql,
  164. params=(
  165. video_id,
  166. 2,
  167. {int(process_times) + 1},
  168. trace_id
  169. )
  170. )
  171. async def start_process(self, params):
  172. """
  173. 开始处理
  174. :param params:
  175. :return:
  176. """
  177. # 更新文章contentId为1, 说明该文章正在处理中
  178. update_sql = f"""
  179. UPDATE {db_article}
  180. SET
  181. content_status = %s
  182. WHERE
  183. trace_id = %s;
  184. """
  185. await self.mysql_client.async_insert(
  186. sql=update_sql,
  187. params=(
  188. 1, params['trace_id']
  189. )
  190. )
  191. try:
  192. # 判断标题中是否包含video_id
  193. if "video_id=" in params['title']:
  194. logging(
  195. code="9006",
  196. info="视频生成文本测试",
  197. trace_id=params['trace_id']
  198. )
  199. await self.process_video_id(
  200. title=params['title'],
  201. trace_id=params['trace_id'],
  202. process_times=params['process_times']
  203. )
  204. else:
  205. await search_videos(
  206. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  207. trace_id=params['trace_id'],
  208. gh_id=params['gh_id'],
  209. mysql_client=self.mysql_client
  210. )
  211. # 执行完成之后,判断是否存在视频id
  212. select_sql = f"""
  213. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  214. FROM {db_article}
  215. WHERE trace_id = '{params["trace_id"]}';
  216. """
  217. result = await self.mysql_client.async_select(sql=select_sql)
  218. vid1, vid2, vid3 = result[0]
  219. if vid1 or vid2 or vid3:
  220. update_sql2 = f"""
  221. UPDATE {db_article}
  222. SET
  223. content_status = %s,
  224. process_times = %s
  225. WHERE trace_id = %s;
  226. """
  227. await self.mysql_client.async_insert(
  228. sql=update_sql2,
  229. params=(
  230. 2, {int(params['process_times']) + 1}, params['trace_id']
  231. )
  232. )
  233. logging(
  234. code="9008",
  235. info="视频搜索成功, 状态修改为2",
  236. trace_id=params['trace_id']
  237. )
  238. else:
  239. if int(params['process_times']) < 5:
  240. update_sql3 = f"""
  241. UPDATE {db_article}
  242. SET
  243. content_status = %s,
  244. process_times = %s
  245. WHERE trace_id = %s;
  246. """
  247. await self.mysql_client.async_insert(
  248. sql=update_sql3,
  249. params=(0, int(params['process_times']) + 1, params['trace_id'])
  250. )
  251. logging(
  252. code="9018",
  253. info="视频搜索失败,回退状态为0",
  254. trace_id=params['trace_id']
  255. )
  256. else:
  257. update_sql3 = f"""
  258. UPDATE {db_article}
  259. SET
  260. content_status = %s,
  261. process_times = %s
  262. WHERE trace_id = %s;
  263. """
  264. await self.mysql_client.async_insert(
  265. sql=update_sql3,
  266. params=(3, int(params['process_times']) + 1, params['trace_id'])
  267. )
  268. logging(
  269. code="9019",
  270. info="视频多次搜索失败,状态修改为3",
  271. trace_id=params['trace_id']
  272. )
  273. except Exception as e:
  274. if int(params['process_times']) < 5:
  275. logging(
  276. code="9018",
  277. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  278. trace_id=params['trace_id']
  279. )
  280. update_sql4 = f"""
  281. UPDATE {db_article}
  282. SET
  283. content_status = %s,
  284. process_times = %s
  285. WHERE trace_id = %s;
  286. """
  287. await self.mysql_client.async_insert(
  288. sql=update_sql4,
  289. params=(0, int(params['process_times']) + 1, params['trace_id'])
  290. )
  291. else:
  292. logging(
  293. code="9019",
  294. info="{}异常错误:{}, 状态修改为3".format(params['trace_id'], e),
  295. trace_id=params['trace_id']
  296. )
  297. update_sql4 = f"""
  298. UPDATE {db_article}
  299. SET
  300. content_status = %s,
  301. process_times = %s
  302. WHERE trace_id = %s;
  303. """
  304. await self.mysql_client.async_insert(
  305. sql=update_sql4,
  306. params=(3, int(params['process_times']) + 1, params['trace_id'])
  307. )
  308. async def process_task(self, params):
  309. """
  310. 异步执行
  311. :param params:
  312. :return:
  313. """
  314. content_id = params['content_id']
  315. trace_id = params['trace_id']
  316. # 判断该文章是否已经生成了
  317. history_trace_id = await self.get_history_contents(content_id)
  318. if history_trace_id:
  319. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  320. logging(
  321. code="9001",
  322. info="存在历史文章",
  323. trace_id=trace_id
  324. )
  325. await self.insert_history_contents_videos(history_trace_id, params)
  326. else:
  327. flag = await self.judge_content_processing(content_id)
  328. if flag:
  329. logging(
  330. code="9004",
  331. info="无正在处理的文章ID, 开始处理",
  332. trace_id=trace_id
  333. )
  334. await self.start_process(params=params)
  335. else:
  336. logging(
  337. code="9003",
  338. info="该文章ID正在请求--文章ID {}".format(content_id),
  339. trace_id=trace_id
  340. )
  341. async def deal(self):
  342. """
  343. 处理
  344. :return:
  345. """
  346. task_list = await self.get_task()
  347. task_dict = {}
  348. for task in task_list:
  349. key = task['content_id']
  350. task_dict[key] = task
  351. process_list = []
  352. for item in task_dict:
  353. process_list.append(task_dict[item])
  354. if process_list:
  355. # for task in task_list:
  356. # await self.process_task(task)
  357. tasks = [self.process_task(params) for params in process_list]
  358. await asyncio.gather(*tasks)
  359. else:
  360. logging(
  361. code="9008",
  362. info="没有要处理的请求"
  363. )