LongArticlesJob-fork/stratrgy/strategy.py

"""
@author: luojunhui
"""
import json

from tqdm import tqdm

from applications.functions import Functions
from config import accountBaseInfo, pool_level_detail


class ArticlePoolStrategy(object):
    """
    长文策略池
    """
    Fun = Functions()

    @classmethod
    def getData(cls, article_list):
        """
        :param article_list: 每天召回的文章list
        :return: {
            "Level1": [],
            "Level2": [],
            "Level3": []
        }
        """
        detail_list = []
        print("查询文章url......")
        for i in tqdm(article_list):
            detail = cls.Fun.matchLinkById(i['article_id'])
            i['gh_id'], i['url'], i['index'] = detail
            detail_list.append(i)
        print("查询完成, 开始排序")
        return detail_list

    @classmethod
    def splitByStrategy(cls, detail_list):
        """
        账号-位置-阅读倍数
        :return:
        """
        L = []
        for line in detail_list:
            key = "{}_{}".format(line['gh_id'], line['index'])
            article_read = line['increase_read_count']
            if accountBaseInfo.get(key):
                avg_read = accountBaseInfo[key]['readAvg']
                # 计算比率
                level_rate = article_read / avg_read - 1
                obj = {
                    "key": key,
                    "avg_read": avg_read,
                    "article_read": article_read,
                    "level_rate": level_rate,
                    "url": line['url']
                }
                L.append(obj)
        L = sorted(L, key=lambda x: x["level_rate"], reverse=True)
        for index, i in enumerate(L):
            print(index,"\t", i['key'], "\t", i['level_rate'])
        result = {
            "Level1": [],
            "Level2": []
        }
        c = 0
        for line in L:
            print(json.dumps(line, ensure_ascii=False, indent=4))
            if line['level_rate'] > 0.2:
                c += 1
                gh_key = line['key']
                if pool_level_detail.get(gh_key):
                    now_level = pool_level_detail[gh_key]
                    if now_level == "3":
                        result['Level2'].append(line['url'])
                    elif now_level == "2":
                        result['Level1'].append(line['url'])
                    else:
                        continue
                else:
                    result['Level2'].append(line['url'])
        print(c)
        return result