process_deal.py 13 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348349350351352353354355356357358359360361362363364365366367368369370371372373374375376
  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. from applications.static.config import spider_coroutines
  9. class ProcessDeal(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_sql1 = f"""
  24. SELECT DISTINCT (content_id)
  25. FROM {db_article}
  26. WHERE content_status = 0 and process_times <= 5
  27. ORDER BY request_time_stamp
  28. ASC
  29. LIMIT {spider_coroutines};
  30. """
  31. content_ids = await self.mysql_client.async_select(select_sql1)
  32. cil = []
  33. for content_id in content_ids:
  34. cil.append(content_id[0])
  35. content_ids_tuple = str(cil).replace("[", "(").replace("]", ")")
  36. if len(content_ids_tuple) > 3:
  37. select_sql = f"""
  38. SELECT trace_id, content_id, gh_id, article_title, article_text, content_status, process_times
  39. FROM {db_article}
  40. WHERE content_id in {content_ids_tuple} and process_times <= 5
  41. ORDER BY request_time_stamp
  42. ASC;
  43. """
  44. print(select_sql)
  45. task_list = await self.mysql_client.async_select(sql=select_sql)
  46. task_obj_list = [
  47. {
  48. "trace_id": item[0],
  49. "content_id": item[1],
  50. "gh_id": item[2],
  51. "title": item[3],
  52. "text": item[4],
  53. "content_status": item[5],
  54. "process_times": item[6]
  55. } for item in task_list
  56. ]
  57. logging(
  58. code="9001",
  59. info="本次任务获取到 {} 条视频".format(len(task_obj_list)),
  60. data=task_obj_list
  61. )
  62. return task_obj_list
  63. else:
  64. return []
  65. async def get_history_contents(self, content_id):
  66. """
  67. check whether the content id exists
  68. :return: trace_id or None
  69. """
  70. select_sql = f"""
  71. SELECT trace_id, content_status
  72. FROM {db_article}
  73. WHERE content_id = '{content_id}'
  74. ORDER BY id DESC;
  75. """
  76. result = await self.mysql_client.async_select(select_sql)
  77. if result:
  78. for item in result:
  79. trace_id, content_status = item
  80. if content_status == 2:
  81. return trace_id
  82. else:
  83. continue
  84. return None
  85. else:
  86. return None
  87. async def judge_content_processing(self, content_id):
  88. """
  89. 判断该content_id是否在处理中
  90. :param content_id:
  91. :return:
  92. """
  93. select_sql = f"""
  94. SELECT trace_id, content_status
  95. FROM {db_article}
  96. WHERE content_id = '{content_id}'
  97. ORDER BY id DESC;
  98. """
  99. result = await self.mysql_client.async_select(select_sql)
  100. if result:
  101. for item in result:
  102. trace_id, content_status = item
  103. if content_status == 1:
  104. return False
  105. return True
  106. else:
  107. return True
  108. async def insert_history_contents_videos(self, history_trace_id, params):
  109. """
  110. 插入历史视频id
  111. :return:
  112. """
  113. select_sql = f"""
  114. SELECT kimi_title, recall_video_id1, recall_video_id2, recall_video_id3
  115. FROM {db_article}
  116. WHERE trace_id = '{history_trace_id}';
  117. """
  118. info = await self.mysql_client.async_select(sql=select_sql)
  119. kimi_title, vid1, vid2, vid3 = info[0]
  120. update_sql = f"""
  121. UPDATE {db_article}
  122. SET
  123. kimi_title=%s,
  124. recall_video_id1=%s,
  125. recall_video_id2=%s,
  126. recall_video_id3=%s,
  127. content_status=%s,
  128. process_times = %s
  129. WHERE trace_id = %s
  130. """
  131. await self.mysql_client.async_insert(
  132. sql=update_sql,
  133. params=(
  134. kimi_title,
  135. vid1,
  136. "NULL" if vid2 is None else vid2,
  137. "NULL" if vid3 is None else vid3,
  138. 2,
  139. int(params['process_times']) + 1,
  140. params['trace_id']
  141. )
  142. )
  143. logging(
  144. code="9002",
  145. info="已从历史文章更新,历史id: {}".format(history_trace_id),
  146. trace_id=params['trace_id']
  147. )
  148. async def process_video_id(self, title, trace_id, process_times):
  149. """
  150. 如果video_id在标题中,则做特殊处理
  151. :return:
  152. """
  153. video_id = title.split("video_id=")[-1]
  154. update_sql = f"""
  155. UPDATE
  156. {db_article}
  157. SET
  158. recall_video_id1 = %s,
  159. content_status = %s,
  160. process_times = %s
  161. WHERE
  162. trace_id = %s;"""
  163. await self.mysql_client.async_insert(
  164. sql=update_sql,
  165. params=(
  166. video_id,
  167. 2,
  168. {int(process_times) + 1},
  169. trace_id
  170. )
  171. )
  172. async def start_process(self, params):
  173. """
  174. 开始处理
  175. :param params:
  176. :return:
  177. """
  178. # 更新文章contentId为1, 说明该文章正在处理中
  179. update_sql = f"""
  180. UPDATE {db_article}
  181. SET
  182. content_status = %s
  183. WHERE
  184. trace_id = %s;
  185. """
  186. await self.mysql_client.async_insert(
  187. sql=update_sql,
  188. params=(
  189. 1, params['trace_id']
  190. )
  191. )
  192. try:
  193. # 判断标题中是否包含video_id
  194. if "video_id=" in params['title']:
  195. logging(
  196. code="9006",
  197. info="视频生成文本测试",
  198. trace_id=params['trace_id']
  199. )
  200. await self.process_video_id(
  201. title=params['title'],
  202. trace_id=params['trace_id'],
  203. process_times=params['process_times']
  204. )
  205. else:
  206. await search_videos(
  207. params={"title": params['title'], "content": params['text'], "trace_id": params['trace_id']},
  208. trace_id=params['trace_id'],
  209. gh_id=params['gh_id'],
  210. mysql_client=self.mysql_client
  211. )
  212. # 执行完成之后,判断是否存在视频id
  213. select_sql = f"""
  214. SELECT recall_video_id1, recall_video_id2, recall_video_id3
  215. FROM {db_article}
  216. WHERE trace_id = '{params["trace_id"]}';
  217. """
  218. result = await self.mysql_client.async_select(sql=select_sql)
  219. vid1, vid2, vid3 = result[0]
  220. if vid1 or vid2 or vid3:
  221. update_sql2 = f"""
  222. UPDATE {db_article}
  223. SET
  224. content_status = %s,
  225. process_times = %s
  226. WHERE trace_id = %s;
  227. """
  228. await self.mysql_client.async_insert(
  229. sql=update_sql2,
  230. params=(
  231. 2, {int(params['process_times']) + 1}, params['trace_id']
  232. )
  233. )
  234. logging(
  235. code="9008",
  236. info="视频搜索成功, 状态修改为2",
  237. trace_id=params['trace_id']
  238. )
  239. else:
  240. if int(params['process_times']) < 5:
  241. update_sql3 = f"""
  242. UPDATE {db_article}
  243. SET
  244. content_status = %s,
  245. process_times = %s
  246. WHERE trace_id = %s;
  247. """
  248. await self.mysql_client.async_insert(
  249. sql=update_sql3,
  250. params=(0, int(params['process_times']) + 1, params['trace_id'])
  251. )
  252. logging(
  253. code="9018",
  254. info="视频搜索失败,回退状态为0",
  255. trace_id=params['trace_id']
  256. )
  257. else:
  258. update_sql3 = f"""
  259. UPDATE {db_article}
  260. SET
  261. content_status = %s,
  262. process_times = %s
  263. WHERE trace_id = %s;
  264. """
  265. await self.mysql_client.async_insert(
  266. sql=update_sql3,
  267. params=(3, int(params['process_times']) + 1, params['trace_id'])
  268. )
  269. logging(
  270. code="9019",
  271. info="视频多次搜索失败,状态修改为3",
  272. trace_id=params['trace_id']
  273. )
  274. except Exception as e:
  275. if int(params['process_times']) < 5:
  276. logging(
  277. code="9018",
  278. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  279. trace_id=params['trace_id']
  280. )
  281. update_sql4 = f"""
  282. UPDATE {db_article}
  283. SET
  284. content_status = %s,
  285. process_times = %s
  286. WHERE trace_id = %s;
  287. """
  288. await self.mysql_client.async_insert(
  289. sql=update_sql4,
  290. params=(0, int(params['process_times']) + 1, params['trace_id'])
  291. )
  292. else:
  293. logging(
  294. code="9019",
  295. info="{}异常错误:{}, 状态修改为3".format(params['trace_id'], e),
  296. trace_id=params['trace_id']
  297. )
  298. update_sql4 = f"""
  299. UPDATE {db_article}
  300. SET
  301. content_status = %s,
  302. process_times = %s
  303. WHERE trace_id = %s;
  304. """
  305. await self.mysql_client.async_insert(
  306. sql=update_sql4,
  307. params=(3, int(params['process_times']) + 1, params['trace_id'])
  308. )
  309. async def process_task(self, params):
  310. """
  311. 异步执行
  312. :param params:
  313. :return:
  314. """
  315. content_id = params['content_id']
  316. trace_id = params['trace_id']
  317. # 判断该文章是否已经生成了
  318. history_trace_id = await self.get_history_contents(content_id)
  319. if history_trace_id:
  320. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  321. logging(
  322. code="9001",
  323. info="存在历史文章",
  324. trace_id=trace_id
  325. )
  326. await self.insert_history_contents_videos(history_trace_id, params)
  327. else:
  328. flag = await self.judge_content_processing(content_id)
  329. if flag:
  330. logging(
  331. code="9004",
  332. info="无正在处理的文章ID, 开始处理",
  333. trace_id=trace_id
  334. )
  335. await self.start_process(params=params)
  336. else:
  337. logging(
  338. code="9003",
  339. info="该文章ID正在请求--文章ID {}".format(content_id),
  340. trace_id=trace_id
  341. )
  342. async def deal(self):
  343. """
  344. 处理
  345. :return:
  346. """
  347. task_list = await self.get_task()
  348. task_dict = {}
  349. for task in task_list:
  350. key = task['content_id']
  351. task_dict[key] = task
  352. process_list = []
  353. for item in task_dict:
  354. process_list.append(task_dict[item])
  355. if process_list:
  356. # for task in task_list:
  357. # await self.process_task(task)
  358. tasks = [self.process_task(params) for params in process_list]
  359. await asyncio.gather(*tasks)
  360. else:
  361. logging(
  362. code="9008",
  363. info="没有要处理的请求"
  364. )