[Handbook]Python一键某狐首页新闻

Code

"""
@author: MR.N
@created: 2022/9/20 Mon. 23:08
@file: se_sohu_1
@version: 1.0
"""

from httpkit import *
from scrapy import Selector
# import re
from filtertags import *
import time
import os


def test(url='https://www.sohu.com'):
    remote_task = RemoteTask(url=url)
    ret = []
    code = get_res_objects2(remote_task=remote_task, ret=ret, dtype=0)
    if code != 'success':
        return None
    content = ret[0]
    sel = Selector(text=content)
    # 新闻标题包含于具有特定特征码的A标签
    links = sel.xpath('//a').getall()
    sources = []
    for link in links:
    	# 包含新闻标题的A标签具有title属性
        if 'title="' in link:
            sources.append(filter_html_tags(link).replace('  ', '').replace('\n', ''))
    if len(sources) < 1:
        print('err', content)
    return sources


if __name__ == '__main__':
    items = test()
    # print('\n\n'.join(items))
    for item in items:
        print('\n\n\t' + item)
        time.sleep(.6)

Example