python 爬虫

爬取电影天堂

返回电影名+磁力链接

import  re
import requests

url_1 = 'https://www.dytt89.com/'           # 电影天堂

re1 = re.compile(r'2021新片精品.*?<ul>(?P<ul>.*?)<ul>', re.S)   # 拿页面板块
re2 = re.compile(r"<a href='(?P<href>.*?)'", re.S)            # 拿跳转链接
re3 = re.compile(r"◎片　　名　(?P<name>.*?)<br />",re.S)        # 拿电影片名
re4 = re.compile(r"◎片　　名　(?P<name>.*?)<br />.*?magnet:\?xt=(?P<down>.*?)=",re.S)  # 片名+磁力


domain = requests.get(url_1); domain.encoding = 'gbk'
tmp1 = re.finditer(re1, domain.text)

for i in tmp1:
    # print(i.group('ul'))
    tmp2 = re.finditer(re2, i.group('ul'))
    for j in tmp2 :
        href = url_1 + j.group('href').strip("/")
        href_movie = requests.get(href)
        href_movie.encoding = 'gbk'
        res = re.findall(re4,href_movie.text)
        magnet = "magnet:?xt="+ res[0][1]
        a = res[0][0]; b = magnet
        print("{:<66s}\t\t\t{:>50}".format(a, b))