dzdp获取列表

#参考资料 某评Mtgsig1.1逆向学习_轩辕丿的博客-CSDN博客 
#仅供参考学习使用  

# -*- coding: utf-8 -*-
import re,sys,os,time,datetime,requests,json
from spiderx import sx,tools
class spider():
    def __init__(self):
        pass
    def get_page(self,page):
        d = os.popen(f'node dzdp.js {page}').read()
        d=json.loads(d)
        url=d['url']
        while True:
            req=sx.post_request(url,headers=d['headers'],data=d['data'],try_num=5)
            if req.status_code==200:
                res=req.json()
                data=res ['data']['moduleInfoList'][0]['moduleData']['data']['listData']['list']
                if not data:
                    print('last')
                    return -1
                for i, d in enumerate(data):
                    #https://m.dianping.com/shop/l8jBZQcqALY28jOE?from=shoplist&shoplistqueryid=d7a2e944-3801-41ee-8bf4-c8be103ddd28
                    shopuuid=sx.json_path(d,'$..shopuuid')
                    shopurl=f'https://m.dianping.com/shop/{shopuuid}'
                    name=d['name']
                    branchName=d['branchName']
                    print(self.index,name,branchName,shopurl)
                    self.index+=1
                break
            else:
                time.sleep(1)
    def get_content(self):
        self.index=1
        page=1
        while True:
            res=self.get_page(page)
            if res==-1:
                break
    @sx.zsq_try_Exception(递归=1)
    def run(self):
        self.get_content()

if __name__ == '__main__':
    sp=spider()
    sp.run()
    os.system('pause')