AccountArticleRank.py 10 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298299300301302303304305306307308309310311312313314315316317318319320321322323324325326327328
  1. """
  2. @author: luojunhui
  3. """
  4. import json
  5. import time
  6. from uuid import uuid4
  7. from applications.aliyunLog import AliyunArticleLog
  8. from applications.functions import ArticleRank, title_sim_v2_by_list
  9. from applications.pipeline import LongArticlesPipeline
  10. def deduplication(rank1, rank2, rank3):
  11. """
  12. 标题相似度去重
  13. :return:
  14. """
  15. dup_list = []
  16. final_result = []
  17. for item_list in [rank1, rank2, rank3]:
  18. result = []
  19. if item_list:
  20. for item in item_list:
  21. title = item['title']
  22. if title_sim_v2_by_list(title, dup_list):
  23. print("标题重复,已经过滤\t", title)
  24. continue
  25. else:
  26. result.append(item)
  27. dup_list.append(title)
  28. final_result.append(result)
  29. return final_result[0], final_result[1], final_result[2]
  30. class AccountArticleRank(object):
  31. """
  32. 文章排序
  33. """
  34. def __init__(self, params, mysql_client):
  35. """
  36. :param params: 请求参数
  37. :param mysql_client: 数据库链接池
  38. """
  39. self.filter_list = None
  40. self.publishArticleList = None
  41. self.publishNum = None
  42. self.strategy = None
  43. self.ghId = None
  44. self.accountName = None
  45. self.accountId = None
  46. self.params = params
  47. self.mysql_client = mysql_client
  48. self.request_id = "alg-{}-{}".format(uuid4(), int(time.time()))
  49. self.logger = AliyunArticleLog(request_id=self.request_id, alg="ArticleRank")
  50. self.pipeline = LongArticlesPipeline()
  51. def filter(self):
  52. """
  53. 过滤器
  54. """
  55. self.publishArticleList = []
  56. self.filter_list = []
  57. for item in self.params['publishArticleList']:
  58. flag = self.pipeline.deal(item)
  59. if flag:
  60. item['filterReason'] = flag['fileterReason']
  61. self.filter_list.append(item)
  62. else:
  63. self.publishArticleList.append(item)
  64. async def check_params(self):
  65. """
  66. 校验参数
  67. :return:
  68. """
  69. try:
  70. self.accountId = self.params["accountId"]
  71. self.accountName = self.params["accountName"]
  72. self.ghId = self.params["ghId"]
  73. self.strategy = self.params["strategy"]
  74. self.publishNum = self.params["publishNum"]
  75. self.filter()
  76. self.logger.log(
  77. code="1001",
  78. msg="参数校验成功",
  79. data=self.params
  80. )
  81. return None
  82. except Exception as e:
  83. response = {
  84. "msg": "params error",
  85. "info": "params check failed, params : {} is not correct".format(e),
  86. "code": 0,
  87. }
  88. self.logger.log(
  89. code="1002",
  90. msg="参数校验失败--{}".format(e),
  91. data=self.params
  92. )
  93. return response
  94. async def basic_rank(self):
  95. """
  96. 基础排序
  97. :return:
  98. """
  99. # 第一步把所有文章标题分为3组
  100. article_list1_ori = [i for i in self.publishArticleList if "【1】" in i['producePlanName']]
  101. article_list2_ori = [i for i in self.publishArticleList if "【2】" in i['producePlanName']]
  102. article_list3_ori = [i for i in self.publishArticleList if
  103. not i in article_list1_ori and not i in article_list2_ori]
  104. # # 全局去重,保留优先级由 L1 --> L2 --> L3
  105. # hash_map = {}
  106. #
  107. # article_list1 = []
  108. # for i in article_list1_ori:
  109. # title = i['title']
  110. # if hash_map.get(title):
  111. # continue
  112. # else:
  113. # article_list1.append(i)
  114. # hash_map[title] = 1
  115. #
  116. # article_list2 = []
  117. # for i in article_list2_ori:
  118. # title = i['title']
  119. # if hash_map.get(title):
  120. # continue
  121. # else:
  122. # article_list2.append(i)
  123. # hash_map[title] = 2
  124. #
  125. # article_list3 = []
  126. # for i in article_list3_ori:
  127. # title = i['title']
  128. # if hash_map.get(title):
  129. # continue
  130. # else:
  131. # article_list3.append(i)
  132. # hash_map[title] = 1
  133. # 第二步对article_list1, article_list3按照得分排序, 对article_list2按照播放量排序
  134. if article_list1_ori:
  135. rank1 = ArticleRank().rank(
  136. account_list=[self.accountName],
  137. text_list=[i['title'] for i in article_list1_ori]
  138. )
  139. score_list1 = rank1[self.accountName]['score_list']
  140. ranked_1 = []
  141. for index, value in enumerate(score_list1):
  142. obj = article_list1_ori[index]
  143. obj['score'] = value + 1000
  144. ranked_1.append(obj)
  145. ranked_1 = sorted(ranked_1, key=lambda x: x['score'], reverse=True)
  146. else:
  147. ranked_1 = []
  148. # rank2
  149. if article_list2_ori:
  150. for item in article_list2_ori:
  151. item['score'] = 100
  152. ranked_2 = sorted(article_list2_ori, key=lambda x: x['crawlerViewCount'], reverse=True)
  153. else:
  154. ranked_2 = []
  155. # rank3
  156. if article_list3_ori:
  157. rank3 = ArticleRank().rank(
  158. account_list=[self.accountName],
  159. text_list=[i['title'] for i in article_list3_ori]
  160. )
  161. score_list3 = rank3[self.accountName]['score_list']
  162. ranked_3 = []
  163. for index, value in enumerate(score_list3):
  164. obj = article_list3_ori[index]
  165. obj['score'] = value
  166. ranked_3.append(obj)
  167. ranked_3 = sorted(ranked_3, key=lambda x: x['score'], reverse=True)
  168. else:
  169. ranked_3 = []
  170. self.logger.log(
  171. code="1004",
  172. msg="排序完成",
  173. data={
  174. "rank1": ranked_1,
  175. "rank2": ranked_2,
  176. "rank3": ranked_3
  177. }
  178. )
  179. return ranked_1, ranked_2, ranked_3
  180. async def rank_v1(self):
  181. """
  182. Rank Version 1
  183. :return:
  184. """
  185. try:
  186. ranked_1_d, ranked_2_d, ranked_3_d = await self.basic_rank()
  187. ranked_1, ranked_2, ranked_3 = deduplication(ranked_1_d, ranked_2_d, ranked_3_d)
  188. try:
  189. L = []
  190. if ranked_1:
  191. L.append(ranked_1[0])
  192. if ranked_2:
  193. L.append(ranked_2[0])
  194. else:
  195. if ranked_2:
  196. if len(ranked_2) > 1:
  197. for i in ranked_2[:2]:
  198. L.append(i)
  199. else:
  200. L.append(ranked_2[0])
  201. for item in ranked_3:
  202. L.append(item)
  203. result = {
  204. "accountId": self.accountId,
  205. "accountName": self.accountName,
  206. "ghId": self.ghId,
  207. "strategy": self.strategy,
  208. "publishNum": self.publishNum,
  209. "rank_list": L[:self.publishNum],
  210. "filter_list": self.filter_list
  211. }
  212. self.logger.log(
  213. code=1006,
  214. msg="rank successfully",
  215. data=result
  216. )
  217. response = {"status": "Rank Success", "data": result, "code": 1}
  218. except Exception as e:
  219. result = {
  220. "accountId": self.accountId,
  221. "accountName": self.accountName,
  222. "ghId": self.ghId,
  223. "strategy": self.strategy,
  224. "publishNum": self.publishNum,
  225. "rank_list": self.publishArticleList[: self.publishNum],
  226. "filter_list": self.filter_list
  227. }
  228. self.logger.log(
  229. code=1007,
  230. msg="rank failed because of {}".format(e),
  231. data=result
  232. )
  233. response = {"status": "Rank Fail", "data": result, "code": 1}
  234. return response
  235. except:
  236. result = {"code": 2, "info": "account is not exist"}
  237. return result
  238. async def rank_v2(self):
  239. """
  240. Rank Version 2
  241. :return:
  242. """
  243. return await self.rank_v1()
  244. async def rank_v3(self):
  245. """
  246. Rank Version 3
  247. :return:
  248. """
  249. return await self.rank_v1()
  250. async def rank_v4(self):
  251. """
  252. Rank Version 4
  253. :return:
  254. """
  255. return await self.rank_v1()
  256. async def rank_v5(self):
  257. """
  258. Rank Version 5
  259. :return:
  260. """
  261. return await self.rank_v1()
  262. async def choose_strategy(self):
  263. """
  264. 选择排序策略
  265. :return:
  266. """
  267. match self.strategy:
  268. case "ArticleRankV1":
  269. self.logger.log(
  270. code="1003",
  271. msg="命中排序策略1"
  272. )
  273. return await self.rank_v1()
  274. case "ArticleRankV2":
  275. self.logger.log(
  276. code="1003",
  277. msg="命中排序策略2"
  278. )
  279. return await self.rank_v2()
  280. case "ArticleRankV3":
  281. self.logger.log(
  282. code="1003",
  283. msg="命中排序策略3"
  284. )
  285. return await self.rank_v3()
  286. case "ArticleRankV4":
  287. self.logger.log(
  288. code="1003",
  289. msg="命中排序策略4"
  290. )
  291. return await self.rank_v4()
  292. case "ArticleRankV5":
  293. self.logger.log(
  294. code="1003",
  295. msg="命中排序策略5"
  296. )
  297. return await self.rank_v5()
  298. async def deal(self):
  299. """
  300. Deal Function
  301. :return:
  302. """
  303. error_params = await self.check_params()
  304. if error_params:
  305. return error_params
  306. else:
  307. return await self.choose_strategy()