Python爬虫之——爬取全国省市对应信息

表链接

https://pan.baidu.com/s/14ywiV2aOAJwJIpPJvJ0jbA

源码：（粘贴可用）

import re
import requests


def getCity(num):
    province = ''
    city = ''
    url = 'http://www.maps7.com/china_province.php'
    page = requests.get(url)
    file = page.text
    if (num == 33):
        one = re.findall('<a name="33"(.*)</a>', file)
    else:
        one = re.findall('<a name="' + str(num) + '" href=.*?>(.*?)<a name="' + str(num + 1) + '" href=.*?>', file)
    for i in one:
        if (num == 0):
            province = re.findall('(.*?)</a></h4>', i)
        else:
            province = re.findall('<h4>(.*?)</h4>', i)
        city = re.findall('<a href="/china/dianziditu.*?>(.*?)</a>', i)
    for i in province:
        if (len(city) != 0):
            for j in city:
                print(i+' '+j)
        else:
            print(i+' '+i)

if __name__ == '__main__':
    num = 34
    for i in range(0,num):
        getCity(i)