From 14b76291067609e454d5c18e9ce8bc2a0d09987e Mon Sep 17 00:00:00 2001
From: Michael Orlitzky <michael@orlitzky.com>
Date: Wed, 13 Jan 2010 23:12:50 -0500
Subject: [PATCH] Added high-definition video support to the Youtube class.

---
 src/websites/youtube.rb | 85 +++++++++++++++++++++++++++++++++++++++--
 1 file changed, 82 insertions(+), 3 deletions(-)

diff --git a/src/websites/youtube.rb b/src/websites/youtube.rb
index c8a1cad..532d8ca 100644
--- a/src/websites/youtube.rb
+++ b/src/websites/youtube.rb
@@ -17,7 +17,7 @@
 #
 
 require 'src/website'
-
+require 'cgi'
 
 class Youtube < Website
 
@@ -27,6 +27,16 @@ class Youtube < Website
     return url =~ VALID_YOUTUBE_URL_REGEX
   end
 
+
+  def initialize(url)
+    super
+    
+    # The @format variable just caches the format of the video we're
+    # downloading. Storing it will prevent us from having to calculate
+    # it twice.
+    @format = 0
+  end
+
   
   def get_video_url()
     video_id = self.parse_video_id()
@@ -43,13 +53,40 @@ class Youtube < Website
     t_parameter = self.parse_t_parameter(page_data)
     
     video_url = "http://www.youtube.com/get_video?video_id=#{video_id}&t=#{t_parameter}"
-      
+
+    # Figure out which formats are available, and if any are,
+    # choose the best one.
+    available_formats = get_available_formats(page_data)
+    desired_format = get_desired_format(available_formats)
+    
+    if not desired_format.nil?
+      # First we cache the format so that when we're asked for the
+      # video filename later, we don't have to recompute the format.
+      @format = desired_format
+
+      # And then stick the format parameter on the end of the URL.
+      video_url = video_url + "&fmt=#{desired_format}"
+    end
+    
     return video_url
   end
 
 
   def get_video_filename()
-    return (self.parse_video_id() + '.flv')
+    # The format -> extension mapping is available on Wikipedia:
+    #
+    #   http://en.wikipedia.org/wiki/YouTube#Quality_and_codecs
+    #
+    # The default extension is .flv.
+    extension = '.flv'
+    
+    if [18, 22, 35, 37].include?(@format)
+      extension = '.mp4'
+    elsif (@format == 17)
+      extension = '.3gp'
+    end
+    
+    return (self.parse_video_id() + extension)
   end
 
   
@@ -93,5 +130,47 @@ class Youtube < Website
     return t_parameter
   end
   
+
+  def get_available_formats(page_data)
+    # Parse the list of available formats from the "fmt_list" Flash
+    # variable.
+    available_formats = []
+    fmt_list_regex = /\"fmt_list\"\:[[:space:]]\"([^\"]+?)\"/
+    matches = fmt_list_regex.match(page_data)
+
+    if matches.nil?
+      return nil
+    else
+      fmts_string = CGI::unescape(matches[1])
+
+      fmts_string.split(',').each do |fmt|
+        # Each "fmt" will look something like,
+        #
+        #   35/640000/9/0/115
+        #
+        # with the format identifier coming before the first slash.
+        first_slash_idx = fmt.index('/')
+        available_formats << fmt[0...first_slash_idx].to_i
+      end
+      
+    end
+    
+    return available_formats
+  end
+
+
+  def get_desired_format(available_formats)
+    # Check for the presence of formats, in order of preference
+    # (quality). That is, we check for the best formats first. As soon
+    # as a format is found to be available, we return it as the
+    # desired format, since the first format we find is going to be
+    # the best available format.
+    return 37 if available_formats.include?(37)
+    return 22 if available_formats.include?(22)
+    return 35 if available_formats.include?(35)
+    return 18 if available_formats.include?(18)
+    return 34 if available_formats.include?(34)
+    return 17 if available_formats.include?(17)
+  end
   
 end
-- 
2.43.2