View Full Version : how do video downloaders work ?

July 23rd, 2008, 09:17 AM
i googled for this, but couldnt find anything even near to the topic.
i couldnt find the source code of any downloader either, hope if you people can get me some (and only in C or java, i know nothing else)

i want to know this because, i have been through some websites (all of them were paid), which didnt allow the popular downloaders to work. when i put in the URL of the video page, the downloader says, that it couldnt find any videos on that page..
what mechanism do these downloaders follow that they afford to miss the videos feom some particular websites only ?

July 23rd, 2008, 02:41 PM
Are you talking about youtube downloaders?

July 23rd, 2008, 02:50 PM
If it isn't easy to do, then they don't want you to do it.

The easiest way to get flash videos from sites, is to just copy the cache file in /tmp.

July 23rd, 2008, 03:55 PM
Youtube DL

#!/usr/bin/env python
# -*- coding: utf-8 -*-
# Author: Ricardo Garcia Gonzalez
# License: Public domain code
import htmlentitydefs
import httplib
import math
import netrc
import os
import os.path
import re
import socket
import string
import sys
import time
import urllib
import urllib2

std_headers = {
'User-Agent': 'Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv: Gecko/2008070208 Firefox/3.0.1',
'Accept-Charset': 'ISO-8859-1,utf-8;q=0.7,*;q=0.7',
'Accept': 'text/xml,application/xml,application/xhtml+xml,text/html;q=0.9,text/plain;q=0.8,image/png,*/*;q=0.5',
'Accept-Language': 'en-us,en;q=0.5',

simple_title_chars = string.ascii_letters.decode('ascii') + string.digits.decode('ascii')

class DownloadError(Exception):
"""Download Error exception.

This exception may be thrown by FileDownloader objects if they are not
configured to continue on errors. They will contain the appropriate
error message.

class SameFileError(Exception):
"""Same File exception.

This exception will be thrown by FileDownloader objects if they detect
multiple files would have to be downloaded to the same file on disk.

class FileDownloader(object):
"""File Downloader class.

File downloader objects are the ones responsible of downloading the
actual video file and writing it to disk if the user has requested
it, among some other tasks. In most cases there should be one per
program. As, given a video URL, the downloader doesn't know how to
extract all the needed information, task that InfoExtractors do, it
has to pass the URL to one of them.

For this, file downloader objects have a method that allows
InfoExtractors to be registered in a given order. When it is passed
a URL, the file downloader handles it to the first InfoExtractor it
finds that reports being able to handle it. The InfoExtractor returns
all the information to the FileDownloader and the latter downloads the
file or does whatever it's instructed to do.

File downloaders accept a lot of parameters. In order not to saturate
the object constructor with arguments, it receives a dictionary of
options instead. These options are available through the get_params()
method for the InfoExtractors to use. The FileDownloader also registers
itself as the downloader in charge for the InfoExtractors that are
added to it, so this is a "mutual registration".

Available options:

username: Username for authentication purposes.
password: Password for authentication purposes.
usenetrc: Use netrc for authentication instead.
quiet: Do not print messages to stdout.
forceurl: Force printing final URL.
forcetitle: Force printing title.
simulate: Do not download the video files.
format: Video format code.
outtmpl: Template for output names.
ignoreerrors: Do not stop on download errors.

_params = None
_ies = []

def __init__(self, params):
"""Create a FileDownloader object with the given options."""
self._ies = []

def pmkdir(filename):
"""Create directory components in filename. Similar to Unix "mkdir -p"."""
components = filename.split(os.sep)
aggregate = [os.sep.join(components[0:x]) for x in xrange(1, len(components))]
for dir in aggregate:
if not os.path.exists(dir):

def format_bytes(bytes):
if bytes is None:
return 'N/A'
if bytes == 0:
exponent = 0
exponent = long(math.log(float(bytes), 1024.0))
suffix = 'bkMGTPEZY'[exponent]
converted = float(bytes) / float(1024**exponent)
return '%.2f%s' % (converted, suffix)

def calc_percent(byte_counter, data_len):
if data_len is None:
return '---.-%'
return '%6s' % ('%3.1f%%' % (float(byte_counter) / float(data_len) * 100.0))

def calc_eta(start, now, total, current):
if total is None:
return '--:--'
dif = now - start
if current == 0 or dif < 0.001: # One millisecond
return '--:--'
rate = float(current) / dif
eta = long((float(total) - float(current)) / rate)
(eta_mins, eta_secs) = divmod(eta, 60)
if eta_mins > 99:
return '--:--'
return '%02d:%02d' % (eta_mins, eta_secs)

def calc_speed(start, now, bytes):
dif = now - start
if bytes == 0 or dif < 0.001: # One millisecond
return '%10s' % '---b/s'
return '%10s' % ('%s/s' % FileDownloader.format_bytes(float(bytes) / dif))

def best_block_size(elapsed_time, bytes):
new_min = max(bytes / 2.0, 1.0)
new_max = min(max(bytes * 2.0, 1.0), 4194304) # Do not surpass 4 MB
if elapsed_time < 0.001:
return int(new_max)
rate = bytes / elapsed_time
if rate > new_max:
return int(new_max)
if rate < new_min:
return int(new_min)
return int(rate)

def set_params(self, params):
"""Sets parameters."""
if type(params) != dict:
raise ValueError('params: dictionary expected')
self._params = params

def get_params(self):
"""Get parameters."""
return self._params

def add_info_extractor(self, ie):
"""Add an InfoExtractor object to the end of the list."""

def to_stdout(self, message, skip_eol=False):
"""Print message to stdout if not in quiet mode."""
if not self._params.get('quiet', False):
sys.stdout.write('%s%s' % (message, ['\n', ''][skip_eol]))

def to_stderr(self, message):
"""Print message to stderr."""
sys.stderr.write('%s\n' % message)

def fixed_template(self):
"""Checks if the output template is fixed."""
return (re.search(ur'(?u)%\(.+?\)s', self._params['outtmpl']) is None)

def trouble(self, message=None):
"""Determine action to take when a download problem appears.

Depending on if the downloader has been configured to ignore
download errors or not, this method may throw an exception or
not when errors are found, after printing the message. If it
doesn't raise, it returns an error code suitable to be returned
later as a program exit code to indicate error.
if message is not None:
if not self._params.get('ignoreerrors', False):
raise DownloadError(message)
return 1

def report_destination(self, filename):
"""Report destination filename."""
self.to_stdout('[download] Destination: %s' % filename)

def report_progress(self, percent_str, data_len_str, speed_str, eta_str):
"""Report download progress."""
self.to_stdout('\r[download] %s of %s at %s ETA %s' %
(percent_str, data_len_str, speed_str, eta_str), skip_eol=True)

def report_finish(self):
"""Report download finished."""

def download(self, url_list):
"""Download a given list of URLs."""
retcode = 0
if len(url_list) > 1 and self.fixed_template():
raise SameFileError(self._params['outtmpl'])

for url in url_list:
suitable_found = False
for ie in self._ies:
if not ie.suitable(url):
# Suitable InfoExtractor found
suitable_found = True
all_results = ie.extract(url)
results = [x for x in all_results if x is not None]
if len(results) != len(all_results):
retcode = self.trouble()

if len(results) > 1 and self.fixed_template():
raise SameFileError(self._params['outtmpl'])

for result in results:

# Forced printings
if self._params.get('forcetitle', False):
print result['title']
if self._params.get('forceurl', False):
print result['url']

# Do nothing else if in simulate mode
if self._params.get('simulate', False):

filename = self._params['outtmpl'] % result
except (ValueError, KeyError), err:
retcode = self.trouble('ERROR: invalid output template: %s' % str(err))
except (OSError, IOError), err:
retcode = self.trouble('ERROR: unable to create directories: %s' % str(err))
outstream = open(filename, 'wb')
except (OSError, IOError), err:
retcode = self.trouble('ERROR: unable to open for writing: %s' % str(err))
self._do_download(outstream, result['url'])
except (OSError, IOError), err:
retcode = self.trouble('ERROR: unable to write video data: %s' % str(err))
except (urllib2.URLError, httplib.HTTPException, socket.error), err:
retcode = self.trouble('ERROR: unable to download video data: %s' % str(err))
if not suitable_found:
retcode = self.trouble('ERROR: no suitable InfoExtractor: %s' % url)

return retcode

def _do_download(self, stream, url):
request = urllib2.Request(url, None, std_headers)
data = urllib2.urlopen(request)
data_len = data.info().get('Content-length', None)
data_len_str = self.format_bytes(data_len)
byte_counter = 0
block_size = 1024
start = time.time()
while True:
# Progress message
percent_str = self.calc_percent(byte_counter, data_len)
eta_str = self.calc_eta(start, time.time(), data_len, byte_counter)
speed_str = self.calc_speed(start, time.time(), byte_counter)
self.report_progress(percent_str, data_len_str, speed_str, eta_str)

# Download and write
before = time.time()
data_block = data.read(block_size)
after = time.time()
data_block_len = len(data_block)
if data_block_len == 0:
byte_counter += data_block_len
block_size = self.best_block_size(after - before, data_block_len)

if data_len is not None and str(byte_counter) != data_len:
raise ValueError('Content too short: %s/%s bytes' % (byte_counter, data_len))

class InfoExtractor(object):
"""Information Extractor class.

Information extractors are the classes that, given a URL, extract
information from the video (or videos) the URL refers to. This
information includes the real video URL, the video title and simplified
title, author and others. It is returned in a list of dictionaries when
calling its extract() method. It is a list because a URL can refer to
more than one video (think of playlists). The dictionaries must include
the following fields:

id: Video identifier.
url: Final video URL.
uploader: Nickname of the video uploader.
title: Literal title.
stitle: Simplified title.
ext: Video filename extension.

Subclasses of this one should re-define the _real_initialize() and
_real_extract() methods, as well as the suitable() static method.
Probably, they should also be instantiated and added to the main

_ready = False
_downloader = None

def __init__(self, downloader=None):
"""Constructor. Receives an optional downloader."""
self._ready = False

def suitable(url):
"""Receives a URL and returns True if suitable for this IE."""
return True

def initialize(self):
"""Initializes an instance (authentication, etc)."""
if not self._ready:
self._ready = True

def extract(self, url):
"""Extracts URL information and returns it in list of dicts."""
return self._real_extract(url)

def set_downloader(self, downloader):
"""Sets the downloader for this IE."""
self._downloader = downloader

def to_stdout(self, message):
"""Print message to stdout if downloader is not in quiet mode."""
if self._downloader is None or not self._downloader.get_params().get('quiet', False):
print message

def to_stderr(self, message):
"""Print message to stderr."""
sys.stderr.write('%s\n' % message)

def _real_initialize(self):
"""Real initialization process. Redefine in subclasses."""

def _real_extract(self, url):
"""Real extraction process. Redefine in subclasses."""

class YoutubeIE(InfoExtractor):
"""Information extractor for youtube.com."""

_LOGIN_URL = 'http://www.youtube.com/login?next=/'
_AGE_URL = 'http://www.youtube.com/verify_age?next_url=/'
_NETRC_MACHINE = 'youtube'

def report_login(self):
"""Report attempt to log in."""
self.to_stdout('[youtube] Logging in')

def report_age_confirmation(self):
"""Report attempt to confirm age."""
self.to_stdout('[youtube] Confirming age')

def report_webpage_download(self, video_id):
"""Report attempt to download webpage."""
self.to_stdout('[youtube] %s: Downloading video webpage' % video_id)

def report_information_extraction(self, video_id):
"""Report attempt to extract video information."""
self.to_stdout('[youtube] %s: Extracting video information' % video_id)

def report_video_url(self, video_id, video_real_url):
"""Report extracted video URL."""
self.to_stdout('[youtube] %s: URL: %s' % (video_id, video_real_url))

def _real_initialize(self):
if self._downloader is None:

username = None
password = None
downloader_params = self._downloader.get_params()

# Attempt to use provided username and password or .netrc data
if downloader_params.get('username', None) is not None:
username = downloader_params['username']
password = downloader_params['password']
elif downloader_params.get('usenetrc', False):
info = netrc.netrc().authenticators(self._NETRC_MACHINE)
if info is not None:
username = info[0]
password = info[2]
raise netrc.NetrcParseError('No authenticators for %s' % self._NETRC_MACHINE)
except (IOError, netrc.NetrcParseError), err:
self.to_stderr('WARNING: parsing .netrc: %s' % str(err))

# No authentication to be performed
if username is None:

# Log in
login_form = {
'current_form': 'loginForm',
'next': '/',
'action_login': 'Log In',
'username': username,
'password': password,
request = urllib2.Request(self._LOGIN_URL, urllib.urlencode(login_form), std_headers)
login_results = urllib2.urlopen(request).read()
if re.search(r'(?i)<form[^>]* name="loginForm"', login_results) is not None:
self.to_stderr('WARNING: unable to log in: bad username or password')
except (urllib2.URLError, httplib.HTTPException, socket.error), err:
self.to_stderr('WARNING: unable to log in: %s' % str(err))

# Confirm age
age_form = {
'next_url': '/',
'action_confirm': 'Confirm',
request = urllib2.Request(self._AGE_URL, urllib.urlencode(age_form), std_headers)
age_results = urllib2.urlopen(request).read()
except (urllib2.URLError, httplib.HTTPException, socket.error), err:
self.to_stderr('ERROR: unable to confirm age: %s' % str(err))

def _real_extract(self, url):
# Extract video id from URL
mobj = re.match(r'^((?:http://)?(?:\w+\.)?youtube\.com/(?:(?:v/)|(?:(?:watch(?:\.php)?)?\?(?:.+&)?v=)))?([0-9A-Za-z_-]+)(?(1).+)?$', url)
if mobj is None:
self.to_stderr('ERROR: invalid URL: %s' % url)
return [None]
video_id = mobj.group(2)

# Downloader parameters
format_param = None
if self._downloader is not None:
params = self._downloader.get_params()
format_param = params.get('format', None)

# Extension
video_extension = {'18': 'mp4', '17': '3gp'}.get(format_param, 'flv')

# Normalize URL, including format
normalized_url = 'http://www.youtube.com/watch?v=%s' % video_id
if format_param is not None:
normalized_url = '%s&fmt=%s' % (normalized_url, format_param)
request = urllib2.Request(normalized_url, None, std_headers)
video_webpage = urllib2.urlopen(request).read()
except (urllib2.URLError, httplib.HTTPException, socket.error), err:
self.to_stderr('ERROR: unable to download video webpage: %s' % str(err))
return [None]

# "t" param
mobj = re.search(r', "t": "([^"]+)"', video_webpage)
if mobj is None:
self.to_stderr('ERROR: unable to extract "t" parameter')
return [None]
video_real_url = 'http://www.youtube.com/get_video?video_id=%s&t=%s' % (video_id, mobj.group(1))
if format_param is not None:
video_real_url = '%s&fmt=%s' % (video_real_url, format_param)
self.report_video_url(video_id, video_real_url)

# uploader
mobj = re.search(r'More From: ([^<]*)<', video_webpage)
if mobj is None:
self.to_stderr('ERROR: unable to extract uploader nickname')
return [None]
video_uploader = mobj.group(1)

# title
mobj = re.search(r'(?im)<title>YouTube - ([^<]*)</title>', video_webpage)
if mobj is None:
self.to_stderr('ERROR: unable to extract video title')
return [None]
video_title = mobj.group(1).decode('utf-8')
video_title = re.sub(ur'(?u)&(.+?);', lambda x: unichr(htmlentitydefs.name2codepoint[x.group(1)]), video_title)
video_title = video_title.replace(os.sep, u'%')

# simplified title
simple_title = re.sub(ur'(?u)([^%s]+)' % simple_title_chars, ur'_', video_title)
simple_title = simple_title.strip(ur'_')

# Return information
return [{
'id': video_id,
'url': video_real_url,
'uploader': video_uploader,
'title': video_title,
'stitle': simple_title,
'ext': video_extension,

if __name__ == '__main__':
# Modules needed only when running the main program
import getpass
import optparse

# General configuration
urllib2.install_opener(urllib2.build_opener(urllib 2.ProxyHandler()))
urllib2.install_opener(urllib2.build_opener(urllib 2.HTTPCookieProcessor()))
socket.setdefaulttimeout(300) # 5 minutes should be enough (famous last words)

# Parse command line
parser = optparse.OptionParser(
usage='Usage: %prog [options] url...',
parser.add_option('-h', '--help',
action='help', help='print this help text and exit')
parser.add_option('-v', '--version',
action='version', help='print program version and exit')
parser.add_option('-u', '--username',
dest='username', metavar='UN', help='account username')
parser.add_option('-p', '--password',
dest='password', metavar='PW', help='account password')
parser.add_option('-o', '--output',
dest='outtmpl', metavar='TPL', help='output filename template')
parser.add_option('-q', '--quiet',
action='store_true', dest='quiet', help='activates quiet mode', default=False)
parser.add_option('-s', '--simulate',
action='store_true', dest='simulate', help='do not download video', default=False)
parser.add_option('-t', '--title',
action='store_true', dest='usetitle', help='use title in file name', default=False)
parser.add_option('-l', '--literal',
action='store_true', dest='useliteral', help='use literal title in file name', default=False)
parser.add_option('-n', '--netrc',
action='store_true', dest='usenetrc', help='use .netrc authentication data', default=False)
parser.add_option('-g', '--get-url',
action='store_true', dest='geturl', help='simulate, quiet but print URL', default=False)
parser.add_option('-e', '--get-title',
action='store_true', dest='gettitle', help='simulate, quiet but print title', default=False)
parser.add_option('-f', '--format',
dest='format', metavar='FMT', help='video format code')
parser.add_option('-b', '--best-quality',
action='store_const', dest='format', help='alias for -f 18', const='18')
parser.add_option('-m', '--mobile-version',
action='store_const', dest='format', help='alias for -f 17', const='17')
parser.add_option('-i', '--ignore-errors',
action='store_true', dest='ignoreerrors', help='continue on download errors', default=False)
(opts, args) = parser.parse_args()

# Conflicting, missing and erroneous options
if len(args) < 1:
sys.exit('ERROR: you must provide at least one URL')
if opts.usenetrc and (opts.username is not None or opts.password is not None):
sys.exit('ERROR: using .netrc conflicts with giving username/password')
if opts.password is not None and opts.username is None:
sys.exit('ERROR: account username missing')
if opts.outtmpl is not None and (opts.useliteral or opts.usetitle):
sys.exit('ERROR: using output template conflicts with using title or literal title')
if opts.usetitle and opts.useliteral:
sys.exit('ERROR: using title conflicts with using literal title')
if opts.username is not None and opts.password is None:
opts.password = getpass.getpass('Type account password and press return:')

# Information extractors
youtube_ie = YoutubeIE()

# File downloader
fd = FileDownloader({
'usenetrc': opts.usenetrc,
'username': opts.username,
'password': opts.password,
'quiet': (opts.quiet or opts.geturl or opts.gettitle),
'forceurl': opts.geturl,
'forcetitle': opts.gettitle,
'simulate': (opts.simulate or opts.geturl or opts.gettitle),
'format': opts.format,
'outtmpl': ((opts.outtmpl is not None and opts.outtmpl)
or (opts.usetitle and '%(stitle)s-%(id)s.%(ext)s')
or (opts.useliteral and '%(title)s-%(id)s.%(ext)s')
or '%(id)s.%(ext)s'),
'ignoreerrors': opts.ignoreerrors,
retcode = fd.download(args)

except DownloadError:
except SameFileError:
sys.exit('ERROR: fixed output name but more than one file to download')
except KeyboardInterrupt:
sys.exit('\nERROR: Interrupted by user')

ORIGINAL LINK (http://www.arrakis.es/~rggi3/youtube-dl/youtube-dl)

July 23rd, 2008, 04:39 PM
You will have to do a *lot* of digging to build your own downloader. Remember: every website uses a different mechanism to protect their videos.

An easy way out is that you remove the official Adobe Flash Player and install the free and open source player swfdec instead. It lets you view and save the embedded media inside SWF files. I think Gnash offers this functionality, too. The only problem is that swfdec and Gnash do not support the full Adobe Flash spec yet, and some SWFs might not work correctly with them.

July 23rd, 2008, 04:43 PM
Or you can just copy the cache file like I do :-)

July 23rd, 2008, 04:46 PM
Well I think open-source programs help you write your own programs as long as you don't copy the same source code and change the trade marks as it helps alot in educations purpose,and I think Youtube-DL will help you out here,and if you know Python well and another language for example C,you might write your program with C based on the python code.And I suggest searching around in websites like SourceForge and open-source websites,that will help you find what you are lookig for.Good luck anyway :-)

July 23rd, 2008, 05:22 PM
I use Miro to download files:)

July 23rd, 2008, 06:32 PM
thanks a lot.
@saratchandra, i am trying to find a more generic approach, not the one specific to youtube

@LaRoza, even i do the same ..its the safest bet around :)
but i wanted to have an automation of this copy-paste process...have no design yet, but i will try to make a bash-script for this.

@Kadrus, thanks a lot for the code, but unfortunately i dont know python yet, i am pretty limited to C , C++ and java.
but i will try some reverse-engineering on this code.

regarding the original context, i exactly wanted to have the security mechanism of those "paid-sites" on my website too ( i am using tomcat6 on apache)..if someone can help me achieve this goal, i would be delighted !

July 23rd, 2008, 07:21 PM
i am pretty limited to C , C++ and java.
I found this at Source Forge that might be useful to you,it's also in synaptic if I am not mistkaen.
YOUGRABBER (http://sourceforge.net/project/showfiles.php?group_id=178664)
Written in C
Project Home Page (http://yougrabber.sourceforge.net/)

July 25th, 2008, 08:40 AM
Hey... I don't know if you want this, but this thread made me think of a bash script that was posted in the Tutorials / How Tos section a while ago...

See here (http://ubuntuforums.org/showpost.php?p=3359861&postcount=4) (I want to give credit where credit is due).


July 25th, 2008, 12:43 PM

I've written a GUI tool that allows users to download youtube videos.

this is for people who prefer GUI's rather than command line scripts.

It is here:

Please e-mail feedback/comments/problems you have with it !


July 25th, 2008, 07:14 PM
One thing which will almost always work is to run a network analyzer like Wireshark to capture your traffic, then open the page with the video and start watching it, as soon as the video starts to load you can go into wireshark, filter out all requests except HTTP GET requests, then you'll have a list of files that you've downloaded through HTTP, it will almost always be a ".flv" file you're looking for, or one with no extension.
If you do it a lot for one site you can note down the page URL and the URL of the flash video you're downloading, if you can spot the connection between them you can then just set up a nice script to download for you in future..

Although I've never thought of copying the file from the cache, that's probably much easier.