123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286287288289290291292293294295296297298 |
- """
- @author: luojunhui
- """
- import json
- import time
- from uuid import uuid4
- from applications.aliyunLog import AliyunArticleLog
- from applications.functions import ArticleRank
- from applications.pipeline import LongArticlesPipeline
- class AccountArticleRank(object):
- """
- 文章排序
- """
- def __init__(self, params, mysql_client):
- """
- :param params: 请求参数
- :param mysql_client: 数据库链接池
- """
- self.publishArticleList = None
- self.publishNum = None
- self.strategy = None
- self.ghId = None
- self.accountName = None
- self.accountId = None
- self.params = params
- self.mysql_client = mysql_client
- self.request_id = "alg-{}-{}".format(uuid4(), int(time.time()))
- self.logger = AliyunArticleLog(request_id=self.request_id, alg="ArticleRank")
- self.pipeline = LongArticlesPipeline()
- def filter(self):
- self.publishArticleList = []
- self.filter_list = []
- for item in self.params['publishArticleList']:
- flag = self.pipeline.deal(item)
- if flag:
- item['filterReason'] = flag['fileterReason']
- self.filter_list.append(item)
- else:
- self.publishArticleList.append(item)
- async def check_params(self):
- """
- 校验参数
- :return:
- """
- try:
- self.accountId = self.params["accountId"]
- self.accountName = self.params["accountName"]
- self.ghId = self.params["ghId"]
- self.strategy = self.params["strategy"]
- self.publishNum = self.params["publishNum"]
- self.filter()
- self.logger.log(
- code="1001",
- msg="参数校验成功",
- data=self.params
- )
- return None
- except Exception as e:
- response = {
- "msg": "params error",
- "info": "params check failed, params : {} is not correct".format(e),
- "code": 0,
- }
- self.logger.log(
- code="1002",
- msg="参数校验失败--{}".format(e),
- data=self.params
- )
- return response
- async def basic_rank(self):
- # 第一步把所有文章标题分为3组
- article_list1_ori = [i for i in self.publishArticleList if "【1】" in i['producePlanName']]
- article_list2_ori = [i for i in self.publishArticleList if "【2】" in i['producePlanName']]
- article_list3_ori = [i for i in self.publishArticleList if not i in article_list1_ori and not i in article_list2_ori]
- # 全局去重,保留优先级由 L1 --> L2 --> L3
- hash_map = {}
- article_list1 = []
- for i in article_list1_ori:
- title = i['title']
- if hash_map.get(title):
- continue
- else:
- article_list1.append(i)
- hash_map[title] = 1
- article_list2 = []
- for i in article_list2_ori:
- title = i['title']
- if hash_map.get(title):
- continue
- else:
- article_list2.append(i)
- hash_map[title] = 2
- article_list3 = []
- for i in article_list3_ori:
- title = i['title']
- if hash_map.get(title):
- continue
- else:
- article_list3.append(i)
- hash_map[title] = 1
- # 第二步对article_list1, article_list3按照得分排序, 对article_list2按照播放量排序
- if article_list1:
- rank1 = ArticleRank().rank(
- account_list=[self.accountName],
- text_list=[i['title'] for i in article_list1]
- )
- score_list1 = rank1[self.accountName]['score_list']
- ranked_1 = []
- for index, value in enumerate(score_list1):
- obj = article_list1[index]
- obj['score'] = value + 1000
- ranked_1.append(obj)
- ranked_1 = sorted(ranked_1, key=lambda x:x['score'], reverse=True)
- else:
- ranked_1 = []
- # rank2
- if article_list2:
- for item in article_list2:
- item['score'] = 100
- ranked_2 = sorted(article_list2, key=lambda x:x['crawlerViewCount'], reverse=True)
- else:
- ranked_2 = []
- # rank3
- if article_list3:
- rank3 = ArticleRank().rank(
- account_list=[self.accountName],
- text_list=[i['title'] for i in article_list3]
- )
- score_list3 = rank3[self.accountName]['score_list']
- ranked_3 = []
- for index, value in enumerate(score_list3):
- obj = article_list3[index]
- obj['score'] = value
- ranked_3.append(obj)
- ranked_3 = sorted(ranked_3, key=lambda x:x['score'], reverse=True)
- else:
- ranked_3 = []
- self.logger.log(
- code="1004",
- msg="去重排序完成",
- data={
- "rank1": ranked_1,
- "rank2": ranked_2,
- "rank3": ranked_3
- }
- )
- return ranked_1, ranked_2, ranked_3
- async def rank_v1(self):
- """
- Rank Version 1
- :return:
- """
- # try:
- ranked_1, ranked_2, ranked_3 = await self.basic_rank()
- try:
- L = []
- if ranked_1:
- L.append(ranked_1[0])
- if ranked_2:
- L.append(ranked_2[0])
- else:
- if ranked_2:
- if len(ranked_2) > 1:
- for i in ranked_2[:2]:
- L.append(i)
- else:
- L.append(ranked_2[0])
- for item in ranked_3:
- L.append(item)
- result = {
- "accountId": self.accountId,
- "accountName": self.accountName,
- "ghId": self.ghId,
- "strategy": self.strategy,
- "publishNum": self.publishNum,
- "rank_list": L[:self.publishNum],
- "filter_list": self.filter_list
- }
- self.logger.log(
- code=1006,
- msg="rank successfully",
- data=result
- )
- response = {"status": "Rank Success", "data": result, "code": 1}
- except Exception as e:
- result = {
- "accountId": self.accountId,
- "accountName": self.accountName,
- "ghId": self.ghId,
- "strategy": self.strategy,
- "publishNum": self.publishNum,
- "rank_list": self.publishArticleList[: self.publishNum],
- "filter_list": self.filter_list
- }
- self.logger.log(
- code=1007,
- msg="rank failed because of {}".format(e),
- data=result
- )
- response = {"status": "Rank Fail", "data": result, "code": 1}
- return response
- # except:
- # result = {"code": 2, "info": "account is not exist"}
- # return result
- async def rank_v2(self):
- """
- Rank Version 2
- :return:
- """
- return await self.rank_v1()
- async def rank_v3(self):
- """
- Rank Version 3
- :return:
- """
- return await self.rank_v1()
- async def rank_v4(self):
- """
- Rank Version 4
- :return:
- """
- return await self.rank_v1()
- async def rank_v5(self):
- """
- Rank Version 5
- :return:
- """
- return await self.rank_v1()
- async def choose_strategy(self):
- """
- 选择排序策略
- :return:
- """
- match self.strategy:
- case "ArticleRankV1":
- self.logger.log(
- code="1003",
- msg="命中排序策略1"
- )
- return await self.rank_v1()
- case "ArticleRankV2":
- self.logger.log(
- code="1003",
- msg="命中排序策略2"
- )
- return await self.rank_v2()
- case "ArticleRankV3":
- self.logger.log(
- code="1003",
- msg="命中排序策略3"
- )
- return await self.rank_v3()
- case "ArticleRankV4":
- self.logger.log(
- code="1003",
- msg="命中排序策略4"
- )
- return await self.rank_v4()
- case "ArticleRankV5":
- self.logger.log(
- code="1003",
- msg="命中排序策略5"
- )
- return await self.rank_v5()
- async def deal(self):
- """
- Deal Function
- :return:
- """
- error_params = await self.check_params()
- if error_params:
- return error_params
- else:
- return await self.choose_strategy()
|