用request.get(url)获取js代码

执行函数eval()语句修改为return语句返回cookie值

调用execjs执行js代码获得cookie值

将cookie值转化为字典格式，用request.get(url, headers=headers)方法获取得到正确的网页信息

代码如下：

def getResponse():

"""

获取response

:return:

"""

response = requests.get(startUrl, headers=headers)

return response

def getJslid(response):

"""

:param response:

:return:

"""

cook = response.cookies

return '; '.join(['='.join(item) for item in cook.items()])

def getClearance(response):

"""

:return:

"""

txt = ''.join(re.findall('', response.text))

func_return = txt.replace('eval', 'return')

print(func_return)

content = execjs.compile(func_return)

eval_func = content.call('x')

name = re.findall(r'var (.*?)=function.*', eval_func)[0]

mode_func = eval_func.replace('while(window._phantom||window.__phantomas){};', ''). \

replace('document.cookie=', 'return').replace('if((function(){try{return !!window.addEventListener;}', ''). \

replace("catch(e){return false;}})()){document.addEventListener('DOMContentLoaded',%s,false)}" % name, ''). \

replace("else{document.attachEvent('onreadystatechange',%s)}" % name, '').replace(

r"setTimeout('location.href=location.pathname+location.search.replace(/[\?|&]captcha-challenge/,\'\')',1500);",

'')

content = execjs.compile(mode_func)

cookies = content.call(name)

# print(cookies)

clearance = cookies.split(';')[0]

return clearance

def structureHeaders(cook, clearance):

"""

构造新的headers

:return:

"""

cookie = {

'cookie': cook + ';' + clearance

}

return dict(headers, **cookie)