163 Open Course downloader

163.py

#!/usr/bin/env python3
import requests
from bs4 import BeautifulSoup as bs
import re
import xmlrpc.client
import sys
import xml.etree.ElementTree as et

"""
Usage: python 163.py <url of the content>
Dependency:python-requests BeautifulSoup and a running daemon-mode Aria2 in localhost
"""

a = xmlrpc.client.Server('http://127.0.0.1:6800/rpc')

def getClasses(url):
	soup = bs(requests.get(url).text)
	name = soup.find(class_='pos').text
	return name,[(''.join(i.text.split()),i.a['href']) for i in soup.find(id='list2').findAll('td',class_='u-ctitle')]

def parseVideo(url):
	cid,mid = re.findall(r'[0-9A-Z]{9}',url)
	url = 'http://live.ws.126.net/movie/{mid[7]}/{mid[8]}/2_{cid}_{mid}.xml'.format(cid=cid,mid=mid)
	# print(url)
	xml = requests.get(url).content.decode("GBK",'IGNORE')
	# FIXME: here is a workround for the GBK encoding of the xml file
	soup = bs(xml)
	
	video_url = soup.find('flvurlorigin').text if soup.find('flvurlorigin').text else soup.find('flvurl').text
	
	if soup.find('subs'):
		sub_url = [(i.find('name').text,i.url.text)for i in soup.findAll('sub')]
	else:
		sub_url = None
	return video_url,sub_url

def downloadFile(url,filename):
	option = {'max-connection-per-server':20,'split':10,'continue':True,'pause':True}
	option['out'] = filename
	a.aria2.addUri([url],option)

def main():
	url = sys.argv[1]
	name,clist = getClasses(url)
	for mname,mlink in clist:
		vlink,sub_urls = parseVideo(mlink)
		if sub_urls:
			for i,j in sub_urls:
				subname = "{}/{}-{}.srt".format(name,mname,i)
				downloadFile(j,subname)
		mname = '{}/{}.flv'.format(name,mname)
		downloadFile(vlink,mname)
		print(mname,vlink)
if __name__ == '__main__':
	main()