LongArticlesJob/account_cold_start_daily.py

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

from argparse import ArgumentParser

from applications import longArticlesMySQL, bot
from coldStartTasks.crawler.weixinCategoryCrawler import weixinCategory
from coldStartTasks.publish.publishCategoryArticles import CategoryColdStartTask
from coldStartTasks.filter.title_similarity_task import ColdStartTitleSimilarityTask

DEFAULT_CATEGORY_LIST = ['1030-手动挑号', 'account_association']


class AccountColdStartDailyTask(object):
    """
    账号冷启动代码
    """

    def __init__(self):
        """
        """
        self.db_client = None

    def init_db(self):
        """
        初始化数据库
        :return:
        """
        try:
            self.db_client = longArticlesMySQL()
            return True
        except Exception as e:
            bot(
                title='账号抓取任务， 冷启动数据库连接失败',
                detail={
                    "error": str(e),
                    "error_msg": traceback.format_exc()
                }
            )
            return False

    def crawler_task(self, category_list, date_str):
        """
        :return:
        """
        # 初始化category抓取类
        try:
            weixin_category_crawler = weixinCategory(db_client=self.db_client)
            weixin_category_crawler.deal(category_list=category_list, date_str=date_str)

            # 抓取完成之后，给抓取到的标题进行相似度打分
            cold_start_title_similarity_task = ColdStartTitleSimilarityTask()
            cold_start_title_similarity_task.init_database()
            cold_start_title_similarity_task.run(meta_source='article')

            bot(
                title="账号冷启动任务，抓取完成",
                detail={
                    "finish_time": datetime.datetime.today().strftime('%Y-%m-%d %H:%M:%S'),
                    "category": category_list
                },
                mention=False
            )
        except Exception as e:
            bot(
                title="账号抓取冷启动任务，抓取失败",
                detail={
                    "error": str(e),
                    "error_msg": traceback.format_exc()
                }
            )

    def publish_task(self, category_list, article_source):
        """
        将账号文章发布到aigc抓取计划，并且绑定生成计划
        :param category_list:  文章品类
        :param article_source: 文章来源（toutiao or weixin）
        :return:
        """
        try:
            weixin_category_publisher = CategoryColdStartTask(db_client=self.db_client)
            weixin_category_publisher.do_job(
                category_list=category_list,
                article_source=article_source
            )
            bot(
                title="账号冷启任务，发布完成",
                detail={
                    "finish_time": datetime.datetime.today().strftime('%Y-%m-%d %H:%M:%S'),
                    "category": category_list
                },
                mention=False
            )
        except Exception as e:
            bot(
                title="账号发布冷启动任务，发布失败",
                detail={
                    "error": str(e),
                    "error_msg": traceback.format_exc()
                }
            )


def main(date_str, category_list=None, article_source=None):
    """
    main job, use crontab to do job daily
    :return:
    """
    if not category_list:
        category_list = DEFAULT_CATEGORY_LIST
    if not article_source:
        article_source = 'weixin'
    task = AccountColdStartDailyTask()
    if task.init_db():
        if article_source == 'weixin':
            task.crawler_task(category_list=category_list, date_str=date_str)

        task.publish_task(category_list=category_list, article_source=article_source)


if __name__ == '__main__':
    parser = ArgumentParser()
    parser.add_argument("--run_date", help="--run_date format: %Y-%m-%d")
    args = parser.parse_args()

    if args.run_date:
        run_date = args.run_date
    else:
        run_date = datetime.date.today().isoformat()

    # 执行微信抓取发布
    main(date_str=run_date)

    # 执行头条发布
    main(
        date_str=run_date,
        category_list=['history', 'tech', 'finance', 'entertainment'],
        article_source='toutiao'
    )