#!/usr/bin/python
# encoding: utf-8

"""
boydl – Best Of Youtube DownLoader

This script fetches the Best Of YouTube RSS feed and downloads new videos.
Via an index every video is downloaded only once.

This is a complete Python rewrite from the former Bash implementation due
to RSS feed migrations.

Usage: boydl [ --help | -h | --version ]

Options:
    --help            Display this boydl's help and exit.
    --version         Display boydl's version and exit.
"""

__author__       = "Lysander Trischler"
__copyright__    = "Copyright 2011, Lysander Trischler"
__license__      = "WTFPL"
__maintainer__   = "Lysander Trischler"
__email__        = "software@lyse.isobeef.org"
__version__      = "2.1.1"
__version_info__ = (2, 1, 1)

import os.path
import sys

me = os.path.basename(__file__)

for arg in sys.argv[1:]:
	if arg in ("--help", "-h"):
		print(__doc__.strip())
		exit(0)
	elif arg == "--version":
		print("%s Version %s" % (me, __version__))
		exit(0)
	else:
		sys.stderr.write("%s: Unknown argument `%s'!\n" % (me, arg))
		exit(2)

import xml.dom.minidom
import urllib
import re
import os

rss_url    = 'http://feeds.feedburner.com/bestofyoutubedotcom?format=xml'
ydl        = 'youtube-dl --restrict-filename -wo "' + os.path.expanduser("~") + '/Desktop/%(title)s-%(id)s.%(ext)s" -f best'
link_regex = re.compile(r'"http://img.youtube.com/vi/(.+)/[0-9]+.jpg"')
index_file = os.path.join(os.path.dirname(__file__), "boydl.index")
error_file = os.path.join(os.path.dirname(__file__), "boydl.error")

if rss_url.startswith(("http://", "https://")):
	rss_stream = urllib.urlopen(rss_url)
else:
	rss_stream = open(rss_url, "rb")

tree = xml.dom.minidom.parseString(rss_stream.read()
                                  .replace("é", "é")
                                  .replace("<br>", "&lt;br&gt;")
								  .replace("&ouml;", "ö")
                                  .replace("< here >", "here"))
rss_stream.close()
channel = tree.getElementsByTagName("rss")[0].getElementsByTagName("channel")[0]

for item in channel.getElementsByTagName("item"):

	#
	# first try to fetch the URL from the description, to bypass feed proxy
	#
	url = None
	m = link_regex.search(item.getElementsByTagName("description")[0].firstChild.data)
	if m:
		url = 'http://www.youtube.com/v/%s' % m.group(1)

	#
	# if description link cannot be parsed use feed proxy's URL
	#
	if url is None:
		print("%s: Cannot extract direct link to Best of YouTube website for '%s'."
		      " Using feed proxy's URL:" % (
				me, item.getElementsByTagName('title')[0].firstChild.data)),
		url = item.getElementsByTagName("link")[0].firstChild.data
		print(url)

	#
	# skip already downloaded videos
	# Always read the index newly so parallel running boydls may
	# not download the same video multiple times. Please note this
	# implementation is not 100% safe: It is possible two or more
	# boydl instances execute this code at the very same time and
	# therefore fetch the video twice or more often. Because this
	# scenario is considered to happen only very rarly we don't
	# care about it here. Most of the time this holds.
	#
	cont = False
	index_stream = open(index_file, "rb")
	for line in index_stream.readlines():
		if line.strip() == url:
			cont = True
			break
	index_stream.close()
	if cont: continue

	#
	# download video
	#
	if url is None:
		sys.stderr.write("%s: Cannot extract YouTube video URL!\n" % me)
	else:
		index_stream = open(index_file, "ab")
		index_stream.write("%s\n" % url)
		index_stream.close()

		cmd = ydl + ' ' + url
		print('—' * 80)
		exit_code = os.system(cmd) % 255
		if exit_code != 0:
			sys.stderr.write("'%s' failed with exit code %d\n" % (cmd, exit_code))
			error_stream = open(error_file, "ab")
			error_stream.write("%s\t%s\t%d\n" % (url, cmd, exit_code))
			error_stream.close()