In [1]:
import requests
from bs4 import BeautifulSoup
In [8]:
class VidSource(object):
def __init__(self, url, keyclass = ['iframe','video','embed']):
self.url = url
self.soupa = ''
self.links = []
self.keys = keyclass
def __getUrlSource__(self, url):
source_object = requests.get(url)
soup = BeautifulSoup(source_object.text, 'html.parser')
return(soup)
def _getLinks_(self, termlist, soup):
for i in termlist:
self._loadIframes_(i, soup)
def _loadIframes_(self, i, soup):
soup.findAll(i)
count = count + 1
if len(k)>0:
for j in k:
if j['src'] is None:
soupb = self.__getUrlSource__(j['src'])
if count < 4:
self._getLinks_(['video', 'embed'], soupb)
if j['src'] is not None:
self._filter_(j['src'])
def _filter_(self, url):
soup = self.getUrlSource(url)
k = soup.findAll(self.query)
if len(k) > 0:
self.links.append(url)
def getResults(self):
count = 0
self.soupa = self.__getUrlSource__(self.url)
self._getLinks_(self.keys, self.soupa)
url = 'http://hindimoviesonline.watch/dangal-full-movie-online/'
source_object = requests.get(url)
soup = BeautifulSoup(source_object.text, 'html.parser')
#print(soup.prettify())
keyclass = ['iframe','video','embed']
In [3]:
videolinks = []
for i in keyclass:
k = soup.findAll(i)
#print(k)
if len(k)>0:
for j in k:
if j['src'] is not None:
videolinks.append(j['src'])
print(videolinks)
In [4]:
r = requests.head('https://redirector.googlevideo.com/videoplayback?id=7f21afcf6ec3a2de&itag=18&source=webdrive&requiressl=yes&ttl=transient&mm=30&mn=sn-2gb7sn7z&ms=nxu&mv=u&nh=IgpwcjAyLnByZzAzKgkxMjcuMC4wLjE&pl=48&mime=video/mp4&lmt=1484994549959561&mt=1486209357&ip=2001:67c:2db8:7::3e0&ipbits=128&expire=1486223899&sparams=ip%2Cipbits%2Cexpire%2Cid%2Citag%2Csource%2Crequiressl%2Cttl%2Cmm%2Cmn%2Cms%2Cmv%2Cnh%2Cpl%2Cmime%2Clmt&signature=4B892C683301E049240AF81D173768BE210CA4B6.7B4F270C05B75E369293F1F1AF92D03F8E7222D6&key=ck2&app=explorer&driveid=0B2aSpxR38G5MUVBsR3FETElZQ3c')
print (r.headers)
In [ ]: