法院链接

rmfy.py

#!/usr/bin/env python
# -*- coding: utf-8 -*-
import requests
from pyquery import PyQuery as pq

class Rmfy(object):
    def __init__(self):
        self.origin_url = 'http://www.rmfysszc.gov.cn/'
        self.province_url = {}
        self.city_url = {}
        self.town_url = {}

    def get_province(self):
        res1 = requests.get(self.origin_url).content
        pq_content = pq(res1)('.divbody_r1_r1_c1_r1_c1_r3 > a')
        for i in pq_content.items():
            self.province_url[i.text()] = i.attr.href
        return self.province_url

    def get_city(self):
        url = [value for key,value in self.get_province().items()]
        for i in url:
            res2 = requests.get(i).content
            for i in pq(res2)('.r2_r4_r2_r1_div > a').items():
                self.city_url[i.attr.title] = i.attr.href
        return self.city_url
    def get_town(self):
        url = [value for key,value in self.get_city().items()]
        for i in url:
            res3 = requests.get(i).content
            for i in pq(res3)('.r2_r4_r2_r1_div > a').items():
                self.town_url[i.attr.title] = i.attr.href
        return self.town_url

a = Rmfy().get_province() #省 dict
b = Rmfy().get_city()  #市 dict
c = Rmfy().get_town()  #县 dict
for x,y in a.items():
    print x,y
for x,y in b.items():
    print x,y
for x,y in c.items():
    print x,y