#?-*-?coding:?utf-8?-*-
“““
@author:?Administrator
“““

import?urllib
import?requests
import?re
import?pandas?as?pd?
import?numpy?as?np
import?ssl
import?warnings
warnings.filterwarnings（“ignore“）

base_url?=?‘https://umbraco.tv‘





????
if?__name__==‘__main__‘:????
????#第一步：完成登錄
????url?=?‘https://umbraco.tv/login/‘
????resp?=?requests.get（url?verify=False?allow_redirects=False）
????headers?=?{
????		‘Content-type‘:‘application/x-www-form-urlencoded‘
??????????‘Cookie‘:resp.headers[‘set-cookie‘]
????	}
????formdata=resp.content
????#從formdata中提取兩個變量的值__RequestVerificationTokenufprt并把值放到postdata里
????reqToken?=?re.findall（r‘‘formdatare.S|re.M）[0]
????ufprt?=?re.findall（r‘‘formdatare.S|re.M）[0]
????
????data?=?{?‘__RequestVerificationToken‘:reqToken
?????????????‘Username‘:‘haierol@qq.com‘
?????????????‘Password‘:‘EEjnMYL3‘
?????????????‘ReturnUrl‘:‘‘
?????????????‘ufprt‘:ufprt?????????????
?????????????}
????	
????#這里才是真正的登陸過程，data里面是抓包獲取的賬號密碼及其他信息。
????resp?=?requests.post（url?urllib.urlencode（data）
????		headers=headers
????		verify=False
????		allow_redirects=False）??	
????	
????	
????headers?=?{‘Cookie‘:resp.headers[‘set-cookie‘]}
????	
????url1=‘https://umbraco.tv/videos/umbraco-v7/developer/fundamentals/api-controllers/introduction/‘
????
????resp1?=?requests.get（url1headers=headersverify=Falseallow_redirects=False）
????t=resp1.content?#這時應該可以取到這個需要登錄的頁面里的mp4信息了?
????
????
????#第二步：爬所有鏈接的子網頁????
????mp4_list?=?[]????
????access_list?=?[]
????
????#爬5層
????index_url?=?‘https://umbraco.tv/videos/‘????????
????i_linklist?=?[re.findall（r‘?href=“（.{1100}?）“‘urllib.urlopen（index_url）.read（）re.S|re.M）]
????i?=?0
????
????all_links?=?[]
????
????while?i<10:
????????l?=?i_linklist[i]????#第i層所有link
????????new_l?=?[]
????????for?link?in?l:?
????????????if?link.startswith（‘href=‘）:
????????????????link?=?link[6:-1]	
????????????#過濾.png?.css?...
????????????if?not?（（‘http://‘?in?link）?or?（‘https://‘?in?link）?or?（‘.png‘?in?link）?or?（‘.css‘?in?link）?or?（‘.js‘?in?link）?or?（‘.ico‘??in?link））:
????????????????print?（link）????????????????
????????????????if?base_url?+?link?