Codeschool Downloader

Readme.md

Nothing

codeschool.py

import os
import json
import asyncio
import aiohttp
import pyquery
from pyquery import PyQuery as pq

base = 'https://www.codeschool.com'
url = 'https://www.codeschool.com/paths'
paths = ['git', 'javascript']
headers = {
    'cookie': ''  # give your cookie
}

sem = asyncio.Semaphore(5)

async def fetch_video(session, url):
    path = './downloads/' + url.split('/')[-2]
    if not os.path.exists(path):
        os.makedirs(path)
    async with session.get(url) as response:
        assert response.status == 200
        body = await response.text()
        text = 'new CS.Classes.VideoManager('
        if text not in body:
            return
        idx = body.rindex(text)
        videos = body[idx + len(text):body.index('});', idx) + 1]
        videos = json.loads(videos)
        print(path, ' ', len(videos['media']))
        for video in videos['media']:
            async with session.get(video['media']) as response:
                body = await response.json()
                with open(path + '/videos.txt', 'a') as f:
                    f.write(body['media'][0]['download'] + '\n')


async def fetch_course(session, url):
    with (await sem):
        async with session.get(url) as response:
            assert response.status == 200
            body = await response.text()
            d = pq(body)
            for course in d('.course-title-link'):
                await fetch_video(session, base + pq(course).attr('href') + '/videos')


loop = asyncio.get_event_loop()
with aiohttp.ClientSession(headers=headers, loop=loop) as session:
    f = asyncio.wait([fetch_course(session, url + '/' + path) for path in paths])
    loop.run_until_complete(f)

download.sh

#!/bin/bash
trap "exit" 2

for path in `ls downloads/`; do
  echo "downloading $path"
  aria2c -d "downloads/$path" -i "downloads/$path/videos.txt"
done