src/websites/youtube.rb

   1 #
   2 # Copyright Michael Orlitzky
   3 #
   4 # http://michael.orlitzky.com/
   5 #
   6 # This program is free software: you can redistribute it and/or modify
   7 # it under the terms of the GNU General Public License as published by
   8 # the Free Software Foundation, either version 3 of the License, or
   9 # (at your option) any later version.
  10 #
  11 # This program is distributed in the hope that it will be useful,
  12 # but WITHOUT ANY WARRANTY; without even the implied warranty of
  13 # MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
  14 # GNU General Public License for more details.
  15 #
  16 # http://www.fsf.org/licensing/licenses/gpl.html
  17 #
  18
  19 require 'src/website'
  20 require 'cgi'
  21
  22 class Youtube < Website
  23
  24   VALID_YOUTUBE_URL_REGEX = /^(http:\/\/)?([a-z0-9]+\.)?youtube\.com\/((watch\?v=)|(v\/)|([a-z]+\#[a-z]\/[a-z]\/[0-9]\/))[a-z0-9_\-]+(\&.*)?\#?$/i
  25
  26   def self.owns_url?(url)
  27     return url =~ VALID_YOUTUBE_URL_REGEX
  28   end
  29
  30
  31   def initialize(url)
  32     super
  33
  34     # The @format variable just caches the format of the video we're
  35     # downloading. Storing it will prevent us from having to calculate
  36     # it twice.
  37     @format = 0
  38   end
  39
  40
  41   def get_video_url()
  42     video_id = self.parse_video_id()
  43
  44     # The video's URL (the "page data" URL) may be different from the
  45     # URL that was passed to the program. We support the /v/video_id
  46     # URL format, but that is *not* the main video page where we can
  47     # retrieve the "t" parameter. We can only get that from the
  48     # /watch?v=video_id form.
  49     page_data_url = "http://www.youtube.com/watch?v=#{video_id}"
  50     page_data = self.get_page_data(page_data_url)
  51
  52     begin
  53       # Get the URL map from the page.
  54       fmt_url_map = get_format_url_map(page_data)
  55
  56       # Figure out which formats are available, and if any are,
  57       # choose the best one.
  58       available_formats = fmt_url_map.keys()
  59       desired_format = get_desired_format(available_formats)
  60
  61       # First we cache the format so that when we're asked for the
  62       # video filename later, we don't have to recompute the format.
  63       @format = desired_format
  64
  65       # And then use whatever URL is available for the desired format.
  66       # We assume that all available formats will have an entry in the
  67       # fmt_url_map hash.
  68       video_url = fmt_url_map[desired_format]
  69
  70       return video_url
  71     rescue StandardError => e
  72       # If at first you do not succeed, maybe someone decided to
  73       # change some shit. This alternate method parses
  74       # url_encoded_fmt_stream_map.
  75       fmt_streams = get_fmt_stream_list(page_data)
  76       video_url = self.unicode_unescape(fmt_streams[0])
  77       video_url = CGI::unescape(video_url)
  78
  79       # Strip off everything after the first space in the URL.
  80       # I don't know why this works, but if we leave the space
  81       # in (encoded, even), Youtube throws us 403 errors.
  82       video_url.gsub!(/ .+$/, '')
  83     end
  84
  85     return video_url
  86   end
  87
  88
  89   def get_video_filename()
  90     # The format -> extension mapping is available on Wikipedia:
  91     #
  92     #   http://en.wikipedia.org/wiki/YouTube#Quality_and_codecs
  93     #
  94     # The default extension is .flv.
  95     extension = '.flv'
  96
  97     if [18, 22, 35, 37].include?(@format)
  98       extension = '.mp4'
  99     elsif (@format == 17)
 100       extension = '.3gp'
 101     end
 102
 103     return (self.parse_video_id() + extension)
 104   end
 105
 106
 107   protected;
 108
 109   def unicode_unescape(string)
 110     # Unescape sequences like '\u0026'.
 111     # Ok, only '\u0026' for now.
 112     return string.gsub('\u0026', '&')
 113   end
 114
 115   def get_fmt_stream_list(page_data)
 116     # This is another (new?) method of embedding the video URLs.
 117     # The url_encoded_fmt_stream_map variable contains a list of URLs
 118     # in the form url=foo1,url=foo2...
 119     #
 120     # It looks like the first one in the list is the highest
 121     # quality?  Let's just take that one for now.
 122     fmt_stream_regex = /\"url_encoded_fmt_stream_map\": \"(.+?)\"/
 123
 124     matches = fmt_stream_regex.match(page_data)
 125
 126     if (matches.nil? || matches.length < 2)
 127       raise StandardError.new("Could not parse the url_encoded_fmt_stream_map Flash variable.")
 128     end
 129
 130     urlstring = matches[1]
 131     urlstring.gsub!('url=', '')
 132     urls = urlstring.split(',')
 133     return urls
 134   end
 135
 136
 137   # Get the video id from the URL. Should be relatively easy,
 138   # unless Youtube supports some URL formats of which I'm unaware.
 139   def parse_video_id()
 140     # Both URLs are fairly easy to parse if you handle
 141     # them one at a time. The only tricky situation is when
 142     # parameters like "&hl=en" are tacked on to the end.
 143     # We'll call /watch?v=video_id the "first form."
 144     first_form_video_id_regex = /v=([0-9a-z_\-]+)/i
 145     first_form_matches = first_form_video_id_regex.match(@url)
 146     return first_form_matches[1] if not (first_form_matches.nil? ||
 147                                          first_form_matches.length < 2)
 148
 149     # First form didn't work? Try the second.
 150     second_form_video_id_regex = /\/v\/([0-9a-z_\-]+)/i
 151     second_form_matches = second_form_video_id_regex.match(@url)
 152     return second_form_matches[1] if not (second_form_matches.nil? ||
 153                                           second_form_matches.length < 2)
 154
 155     # ...and the third.
 156     third_form_video_id_regex = /\/([[:alnum:]]+)$/i
 157     third_form_matches = third_form_video_id_regex.match(@url)
 158     return third_form_matches[1] if not (third_form_matches.nil? ||
 159                                          third_form_matches.length < 2)
 160
 161     # If we made it here, we couldn't figure out the video id. Yes,
 162     # this is fatal, since we don't know where the video file is
 163     # located.
 164     raise StandardError.new("Could not parse the video id.")
 165   end
 166
 167
 168
 169   def get_format_url_map(page_data)
 170     # Youtube has implemented a new fmt_url_map that (perhaps
 171     # unsurprisingly) maps formats to video URLs. This makes it
 172     # easyish to parse the video URLs.
 173     url_map = {}
 174     url_map_regex = /fmt_url_map=([^&\"]+)/
 175
 176     matches = url_map_regex.match(page_data)
 177
 178     if (matches.nil? || matches.length < 1)
 179       raise StandardError.new("Could not parse the fmt_url_map Flash variable.")
 180     end
 181
 182     # The map is stored entirely in one Flash variable. The format is
 183     # key|value,key|value,...
 184     maptext = CGI::unescape(matches[1])
 185     entries = maptext.split(',')
 186     entries.each do |entry|
 187       key = entry.split('|')[0].to_i
 188       value = entry.split('|')[1]
 189       url_map[key] = value
 190     end
 191
 192     if (url_map.length < 1)
 193       raise StandardError.new("Could not find any valid format URLs.")
 194     end
 195
 196     return url_map
 197   end
 198
 199
 200   def get_desired_format(available_formats)
 201     # Check for the presence of formats, in order of preference
 202     # (quality). That is, we check for the best formats first. As soon
 203     # as a format is found to be available, we return it as the
 204     # desired format, since the first format we find is going to be
 205     # the best available format.
 206     return 37 if available_formats.include?(37)
 207     return 22 if available_formats.include?(22)
 208     return 35 if available_formats.include?(35)
 209     return 18 if available_formats.include?(18)
 210     return 34 if available_formats.include?(34)
 211     return 17 if available_formats.include?(17)
 212
 213     # Available formats can't be empty (we would have raised an error
 214     # in get_available_formats), so if there's some unknown format
 215     # here we might as well return it as a last resort.
 216     return available_formats[0]
 217   end
 218
 219 end