LongArticlesJob-fork/tasks/task3.py

"""
@author: luojunhui
"""
import datetime

from tqdm import tqdm

from applications import AIDTApi, DeNetMysql, PQMySQL, Functions, ODPSApi
from config import poolTagMap
from stratrgy import ArticlePoolStrategy


class SendToMultiLevels(object):
    """
    冷启分配任务
    """
    AidApi = AIDTApi()
    DeMysql = DeNetMysql()
    PqMysql = PQMySQL()
    Fun = Functions()
    OA = ODPSApi()

    @classmethod
    def getYesterdayData(cls):
        """
        获取前一天数据表现
        :return:
        """
        odps_sql = "select * from loghubods.changwen_article_datastat where dt = '20240724';"
        result = cls.OA.select(sql=odps_sql)
        response_list = [
            {
                "article_id": record["article_id"],
                "increase_read_count": record["increase_read_count"],
                "read_count": record["read_count"],
                "increase_income": record["increase_income"],
                "income": record["income"],
                "increase_share_count": record["increase_share_count"],
                "share_count": record["share_count"],
                "update_timestamp": record["update_timestamp"]
            } for record in result if record['increase_read_count'] >= 1000
        ]
        return response_list

    @classmethod
    def splitToDifferentPools(cls, yesterday_data):
        """
        分类至Pools
        :return:
        """
        pool_level_1 = [cls.Fun.matchLinkById(i['article_id']) for i in tqdm(yesterday_data) if
                        i['increase_read_count'] >= 9000]
        pool_level_2 = [cls.Fun.matchLinkById(i['article_id']) for i in tqdm(yesterday_data) if
                        3500 <= i['increase_read_count'] < 9000]
        pool_level_3 = [cls.Fun.matchLinkById(i['article_id']) for i in tqdm(yesterday_data) if
                        1000 <= i['increase_read_count'] < 3500]
        L = {
            "Level1": pool_level_1,
            "Level2": pool_level_2,
            "Level3": pool_level_3
        }
        return L

    @classmethod
    def sendToEachCrawlerPlan(cls, key, url_list):
        """

        :param key:
        :param url_list:
        :return:
        """
        # daily自动创建新抓取计划
        # cls.AidApi.updateArticleIntoCrawlerPlan(
        #     plan_id=None,
        #     plan_name="{}--{}".format(datetime.datetime.today().__str__().split(" ")[0], key),
        #     plan_tag=poolTagMap[key],
        #     url_list=url_list
        # )

    @classmethod
    def sendToDifferentPools(cls, pool_info):
        """
        获取文章url
        :return:
        """
        for key in pool_info:
            cls.sendToEachCrawlerPlan(key, pool_info[key])

    @classmethod
    def deal(cls):
        """
        Dealing function
        :return:
        """
        yesterday_data = cls.getYesterdayData()
        level_url_list_map = cls.splitToDifferentPools(yesterday_data)
        cls.sendToDifferentPools(pool_info=level_url_list_map)


STML = SendToMultiLevels()
yesterday_data = STML.getYesterdayData()
S = ArticlePoolStrategy()
detail_list = S.getData(article_list=yesterday_data)
S.splitByStrategy(detail_list=detail_list)