task1.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347
  1. """
  2. @author: luojunhui
  3. """
  4. import asyncio
  5. from static.config import db_article, db_video
  6. from applications.schedule import search_videos
  7. from applications.functions.log import logging
  8. from static.config import spider_coroutines
  9. class MatchTask1(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 <= 3
  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 <= 3
  41. ORDER BY request_time_stamp
  42. ASC;
  43. """
  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_videos(self, content_id):
  65. """
  66. check whether the contents videos exists
  67. :param content_id:
  68. :return:
  69. """
  70. select_sql = f"""
  71. SELECT video_id
  72. FROM {db_video}
  73. where content_id = '{content_id}' and video_status = 1 order by request_time DESC;
  74. """
  75. content_videos = await self.mysql_client.async_select(select_sql)
  76. videos = [vid for vid in content_videos]
  77. if len(videos) >= 3:
  78. return videos
  79. else:
  80. return None
  81. async def judge_content_processing(self, content_id):
  82. """
  83. 判断该content_id是否在处理中
  84. :param content_id:
  85. :return:
  86. """
  87. select_sql = f"""
  88. SELECT trace_id, content_status
  89. FROM {db_article}
  90. WHERE content_id = '{content_id}'
  91. ORDER BY id DESC;
  92. """
  93. result = await self.mysql_client.async_select(select_sql)
  94. if result:
  95. for item in result:
  96. trace_id, content_status = item
  97. if content_status == 1:
  98. return False
  99. return True
  100. else:
  101. return True
  102. async def use_exists_contents_videos(self, video_id_list, params):
  103. """
  104. 使用已经存在的视频id
  105. :return:
  106. """
  107. trace_id = params['trace_id']
  108. content_id = params['content_id']
  109. select_sql = f"""
  110. SELECT kimi_title
  111. FROM {db_article}
  112. WHERE content_id = '{content_id}' and kimi_title is not null limit 1;
  113. """
  114. info = await self.mysql_client.async_select(sql=select_sql)
  115. kimi_title = info[0]
  116. update_sql = f"""
  117. UPDATE {db_article}
  118. SET
  119. kimi_title=%s,
  120. recall_video_id1=%s,
  121. recall_video_id2=%s,
  122. recall_video_id3=%s,
  123. content_status=%s,
  124. process_times = %s
  125. WHERE trace_id = %s
  126. """
  127. vid1, vid2, vid3 = video_id_list[0], video_id_list[1], video_id_list[2]
  128. await self.mysql_client.async_insert(
  129. sql=update_sql,
  130. params=(
  131. kimi_title,
  132. video_id_list[0],
  133. "NULL" if vid2 is None else vid2,
  134. "NULL" if vid3 is None else vid3,
  135. 2,
  136. int(params['process_times']) + 1,
  137. trace_id
  138. )
  139. )
  140. logging(
  141. code="9002",
  142. info="已从历史文章更新,文章id: {}".format(content_id),
  143. trace_id=trace_id
  144. )
  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 = %s
  156. WHERE
  157. trace_id = %s;
  158. """
  159. await self.mysql_client.async_insert(
  160. sql=update_sql,
  161. params=(
  162. 1, params['trace_id']
  163. )
  164. )
  165. try:
  166. video_count = await search_videos(
  167. params={
  168. "title": params['title'],
  169. "content": params['text'],
  170. "trace_id": params['trace_id'],
  171. "content_id": params['content_id']
  172. },
  173. trace_id=params['trace_id'],
  174. gh_id=params['gh_id'],
  175. mysql_client=self.mysql_client
  176. )
  177. select_sql = f"""
  178. SELECT video_id
  179. FROM {db_video}
  180. WHERE content_id = '{params['content_id']}'
  181. """
  182. result = await self.mysql_client.async_select(sql=select_sql)
  183. vid1, vid2, vid3 = result[0], result[1], result[2]
  184. if vid1 or vid2 or vid3:
  185. update_sql2 = f"""
  186. UPDATE {db_article}
  187. SET
  188. recall_video_id1 = %s,
  189. recall_video_id2 = %s,
  190. recall_video_id3 = %s,
  191. content_status = %s,
  192. process_times = %s
  193. WHERE trace_id = %s;
  194. """
  195. await self.mysql_client.async_insert(
  196. sql=update_sql2,
  197. params=(
  198. vid1 if vid1 else "NULL",
  199. vid2 if vid2 else "NULL",
  200. vid3 if vid3 else "NULL",
  201. 2,
  202. {int(params['process_times']) + 1},
  203. params['trace_id']
  204. )
  205. )
  206. logging(
  207. code="9008",
  208. info="视频搜索成功, 状态修改为2",
  209. trace_id=params['trace_id']
  210. )
  211. else:
  212. if int(params['process_times']) < 3:
  213. update_sql3 = f"""
  214. UPDATE {db_article}
  215. SET
  216. content_status = %s,
  217. process_times = %s
  218. WHERE trace_id = %s;
  219. """
  220. await self.mysql_client.async_insert(
  221. sql=update_sql3,
  222. params=(0, int(params['process_times']) + 1, params['trace_id'])
  223. )
  224. logging(
  225. code="9018",
  226. info="视频搜索失败,回退状态为0",
  227. trace_id=params['trace_id']
  228. )
  229. else:
  230. update_sql3 = f"""
  231. UPDATE {db_article}
  232. SET
  233. content_status = %s,
  234. process_times = %s
  235. WHERE trace_id = %s;
  236. """
  237. await self.mysql_client.async_insert(
  238. sql=update_sql3,
  239. params=(3, int(params['process_times']) + 1, params['trace_id'])
  240. )
  241. logging(
  242. code="9019",
  243. info="视频多次搜索失败,状态修改为3",
  244. trace_id=params['trace_id']
  245. )
  246. except Exception as e:
  247. if int(params['process_times']) < 3:
  248. logging(
  249. code="9018",
  250. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  251. trace_id=params['trace_id']
  252. )
  253. update_sql4 = f"""
  254. UPDATE {db_article}
  255. SET
  256. content_status = %s,
  257. process_times = %s
  258. WHERE trace_id = %s;
  259. """
  260. await self.mysql_client.async_insert(
  261. sql=update_sql4,
  262. params=(0, int(params['process_times']) + 1, params['trace_id'])
  263. )
  264. else:
  265. logging(
  266. code="9019",
  267. info="{}异常错误:{}, 状态修改为3".format(params['trace_id'], e),
  268. trace_id=params['trace_id']
  269. )
  270. update_sql4 = f"""
  271. UPDATE {db_article}
  272. SET
  273. content_status = %s,
  274. process_times = %s
  275. WHERE trace_id = %s;
  276. """
  277. await self.mysql_client.async_insert(
  278. sql=update_sql4,
  279. params=(3, int(params['process_times']) + 1, params['trace_id'])
  280. )
  281. async def process_task(self, params):
  282. """
  283. 异步执行
  284. :param params:
  285. :return:
  286. """
  287. content_id = params['content_id']
  288. trace_id = params['trace_id']
  289. # 判断该篇文章是否存在未下架的视频,且判断是否有3条, 如果没有三条,则启动新抓取任务,后续优化点
  290. video_id_list = await self.get_history_videos(content_id=content_id)
  291. if video_id_list:
  292. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  293. logging(
  294. code="9001",
  295. info="存在历史文章",
  296. trace_id=trace_id
  297. )
  298. # await self.use_exists_contents_videos(video_id_list=video_id_list, params=params)
  299. else:
  300. flag = await self.judge_content_processing(content_id)
  301. if flag:
  302. logging(
  303. code="9004",
  304. info="无正在处理的文章ID, 开始处理",
  305. trace_id=trace_id
  306. )
  307. await self.start_process(params=params)
  308. else:
  309. logging(
  310. code="9003",
  311. info="该文章ID正在请求--文章ID {}".format(content_id),
  312. trace_id=trace_id
  313. )
  314. async def deal(self):
  315. """
  316. 处理
  317. :return:
  318. """
  319. task_list = await self.get_task()
  320. task_dict = {}
  321. for task in task_list:
  322. key = task['content_id']
  323. task_dict[key] = task
  324. process_list = []
  325. for item in task_dict:
  326. process_list.append(task_dict[item])
  327. if process_list:
  328. # for task in task_list:
  329. # await self.process_task(task)
  330. tasks = [self.process_task(params) for params in process_list]
  331. await asyncio.gather(*tasks)
  332. else:
  333. logging(
  334. code="9008",
  335. info="没有要处理的请求"
  336. )