task1.py 12 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328329330331332333334335336337338339340341342343344345346347348
  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 <= 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_videos(self, content_id):
  66. """
  67. check whether the contents videos exists
  68. :param content_id:
  69. :return:
  70. """
  71. select_sql = f"""
  72. SELECT video_id
  73. FROM {db_video}
  74. where content_id = '{content_id}' and video_status = 1 order by request_time DESC;
  75. """
  76. content_videos = await self.mysql_client.async_select(select_sql)
  77. videos = [vid for vid in content_videos]
  78. if len(videos) >= 3:
  79. return videos
  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 use_exists_contents_videos(self, video_id_list, params):
  104. """
  105. 使用已经存在的视频id
  106. :return:
  107. """
  108. trace_id = params['trace_id']
  109. content_id = params['content_id']
  110. select_sql = f"""
  111. SELECT kimi_title
  112. FROM {db_article}
  113. WHERE content_id = '{content_id}' and kimi_title is not null limit 1;
  114. """
  115. info = await self.mysql_client.async_select(sql=select_sql)
  116. kimi_title = info[0]
  117. update_sql = f"""
  118. UPDATE {db_article}
  119. SET
  120. kimi_title=%s,
  121. recall_video_id1=%s,
  122. recall_video_id2=%s,
  123. recall_video_id3=%s,
  124. content_status=%s,
  125. process_times = %s
  126. WHERE trace_id = %s
  127. """
  128. vid1, vid2, vid3 = video_id_list[0], video_id_list[1], video_id_list[2]
  129. await self.mysql_client.async_insert(
  130. sql=update_sql,
  131. params=(
  132. kimi_title,
  133. video_id_list[0],
  134. "NULL" if vid2 is None else vid2,
  135. "NULL" if vid3 is None else vid3,
  136. 2,
  137. int(params['process_times']) + 1,
  138. trace_id
  139. )
  140. )
  141. logging(
  142. code="9002",
  143. info="已从历史文章更新,文章id: {}".format(content_id),
  144. trace_id=trace_id
  145. )
  146. async def start_process(self, params):
  147. """
  148. 开始处理
  149. :param params:
  150. :return:
  151. """
  152. # 更新文章contentId为1, 说明该文章正在处理中
  153. update_sql = f"""
  154. UPDATE {db_article}
  155. SET
  156. content_status = %s
  157. WHERE
  158. trace_id = %s;
  159. """
  160. await self.mysql_client.async_insert(
  161. sql=update_sql,
  162. params=(
  163. 1, params['trace_id']
  164. )
  165. )
  166. try:
  167. video_count = await search_videos(
  168. params={
  169. "title": params['title'],
  170. "content": params['text'],
  171. "trace_id": params['trace_id'],
  172. "content_id": params['content_id']
  173. },
  174. trace_id=params['trace_id'],
  175. gh_id=params['gh_id'],
  176. mysql_client=self.mysql_client
  177. )
  178. select_sql = f"""
  179. SELECT video_id
  180. FROM {db_video}
  181. WHERE content_id = '{params['content_id']}'
  182. """
  183. result = await self.mysql_client.async_select(sql=select_sql)
  184. vid1, vid2, vid3 = result[0], result[1], result[2]
  185. if vid1 or vid2 or vid3:
  186. update_sql2 = f"""
  187. UPDATE {db_article}
  188. SET
  189. recall_video_id1 = %s,
  190. recall_video_id2 = %s,
  191. recall_video_id3 = %s,
  192. content_status = %s,
  193. process_times = %s
  194. WHERE trace_id = %s;
  195. """
  196. await self.mysql_client.async_insert(
  197. sql=update_sql2,
  198. params=(
  199. vid1 if vid1 else "NULL",
  200. vid2 if vid2 else "NULL",
  201. vid3 if vid3 else "NULL",
  202. 2,
  203. {int(params['process_times']) + 1},
  204. params['trace_id']
  205. )
  206. )
  207. logging(
  208. code="9008",
  209. info="视频搜索成功, 状态修改为2",
  210. trace_id=params['trace_id']
  211. )
  212. else:
  213. if int(params['process_times']) < 5:
  214. update_sql3 = f"""
  215. UPDATE {db_article}
  216. SET
  217. content_status = %s,
  218. process_times = %s
  219. WHERE trace_id = %s;
  220. """
  221. await self.mysql_client.async_insert(
  222. sql=update_sql3,
  223. params=(0, int(params['process_times']) + 1, params['trace_id'])
  224. )
  225. logging(
  226. code="9018",
  227. info="视频搜索失败,回退状态为0",
  228. trace_id=params['trace_id']
  229. )
  230. else:
  231. update_sql3 = f"""
  232. UPDATE {db_article}
  233. SET
  234. content_status = %s,
  235. process_times = %s
  236. WHERE trace_id = %s;
  237. """
  238. await self.mysql_client.async_insert(
  239. sql=update_sql3,
  240. params=(3, int(params['process_times']) + 1, params['trace_id'])
  241. )
  242. logging(
  243. code="9019",
  244. info="视频多次搜索失败,状态修改为3",
  245. trace_id=params['trace_id']
  246. )
  247. except Exception as e:
  248. if int(params['process_times']) < 5:
  249. logging(
  250. code="9018",
  251. info="{}异常错误:{}, 回退状态为0".format(params['trace_id'], e),
  252. trace_id=params['trace_id']
  253. )
  254. update_sql4 = f"""
  255. UPDATE {db_article}
  256. SET
  257. content_status = %s,
  258. process_times = %s
  259. WHERE trace_id = %s;
  260. """
  261. await self.mysql_client.async_insert(
  262. sql=update_sql4,
  263. params=(0, int(params['process_times']) + 1, params['trace_id'])
  264. )
  265. else:
  266. logging(
  267. code="9019",
  268. info="{}异常错误:{}, 状态修改为3".format(params['trace_id'], e),
  269. trace_id=params['trace_id']
  270. )
  271. update_sql4 = f"""
  272. UPDATE {db_article}
  273. SET
  274. content_status = %s,
  275. process_times = %s
  276. WHERE trace_id = %s;
  277. """
  278. await self.mysql_client.async_insert(
  279. sql=update_sql4,
  280. params=(3, int(params['process_times']) + 1, params['trace_id'])
  281. )
  282. async def process_task(self, params):
  283. """
  284. 异步执行
  285. :param params:
  286. :return:
  287. """
  288. content_id = params['content_id']
  289. trace_id = params['trace_id']
  290. # 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
  291. video_id_list = await self.get_history_videos(content_id=content_id)
  292. if video_id_list:
  293. # 说明已经存在了结果, 将该条记录下的video_id拿出来
  294. logging(
  295. code="9001",
  296. info="存在历史文章",
  297. trace_id=trace_id
  298. )
  299. await self.use_exists_contents_videos(video_id_list=video_id_list, params=params)
  300. else:
  301. flag = await self.judge_content_processing(content_id)
  302. if flag:
  303. logging(
  304. code="9004",
  305. info="无正在处理的文章ID, 开始处理",
  306. trace_id=trace_id
  307. )
  308. await self.start_process(params=params)
  309. else:
  310. logging(
  311. code="9003",
  312. info="该文章ID正在请求--文章ID {}".format(content_id),
  313. trace_id=trace_id
  314. )
  315. async def deal(self):
  316. """
  317. 处理
  318. :return:
  319. """
  320. task_list = await self.get_task()
  321. task_dict = {}
  322. for task in task_list:
  323. key = task['content_id']
  324. task_dict[key] = task
  325. process_list = []
  326. for item in task_dict:
  327. process_list.append(task_dict[item])
  328. if process_list:
  329. # for task in task_list:
  330. # await self.process_task(task)
  331. tasks = [self.process_task(params) for params in process_list]
  332. await asyncio.gather(*tasks)
  333. else:
  334. logging(
  335. code="9008",
  336. info="没有要处理的请求"
  337. )