Restore proper support for webm formats (fixes issue #166)

1 files changed, 32 insertions, 2 deletions

diff --git a/youtube-dl b/youtube-dl
index 5fd331e79..00d9696d6 100755
--- a/youtube-dl
+++ b/youtube-dl
@@ -879,7 +879,36 @@ class YoutubeIE(InfoExtractor):
 				video_description = mobj.group(1)
 
 		# Decide which formats to download
-		if 'fmt_url_map' in video_info:
+		requested_format = self._downloader.params.get('format', None)
+
+		if requested_format in ["43", "45"]: # webm formats
+			# Join the HTML5 beta
+			html5form = { "enable_html5": "true" }
+			request = urllib2.Request('http://www.youtube.com/html5', urllib.urlencode(html5form), std_headers)
+			try:
+				self._downloader.to_stdout(u'[youtube] Joining the HTML5 Beta')
+				urllib2.urlopen(request).read()
+			except (urllib2.URLError, httplib.HTTPException, socket.error), err:
+				self._downloader.trouble(u'ERROR: unable to join the HTML5 Beta: %s' % str(err))
+				return
+
+			# Request the video webpage with webm enabled
+			request = urllib2.Request('http://www.youtube.com/watch?v=%s&webm=1' % video_id, None, std_headers)
+			try:
+				self._downloader.to_stdout(u'[youtube] Requesting HTML5 video webpage')
+				video_webpage = urllib2.urlopen(request).read()
+			except (urllib2.URLError, httplib.HTTPException, socket.error), err:
+				self._downloader.trouble(u'ERROR: unable to get the HTML5 video webpage: %s' % str(err))
+				return
+
+			# Find the URL for the requested format
+			mobj = re.search(ur'setAvailableFormat\("(.*?)".*?"%s"\);' % requested_format, video_webpage)
+			if mobj is None:
+				self._downloader.trouble(u'ERROR: format not available for video')
+				return
+			video_url_list = [(requested_format, mobj.group(1))]
+
+		elif 'fmt_url_map' in video_info:
 			url_map = dict(tuple(pair.split('|')) for pair in video_info['fmt_url_map'][0].split(','))
 			format_limit = self._downloader.params.get('format_limit', None)
 			if format_limit is not None and format_limit in self._available_formats:
@@ -890,7 +919,6 @@ class YoutubeIE(InfoExtractor):
 			if len(existing_formats) == 0:
 				self._downloader.trouble(u'ERROR: no known formats available for video')
 				return
-			requested_format = self._downloader.params.get('format', None)
 			if requested_format is None:
 				video_url_list = [(existing_formats[0], url_map[existing_formats[0]])] # Best quality
 			elif requested_format == '-1':
@@ -900,9 +928,11 @@ class YoutubeIE(InfoExtractor):
 					self._downloader.trouble(u'ERROR: format not available for video')
 					return
 				video_url_list = [(requested_format, url_map[requested_format])] # Specific format
+
 		elif 'conn' in video_info and video_info['conn'][0].startswith('rtmp'):
 			self.report_rtmp_download()
 			video_url_list = [(None, video_info['conn'][0])]
+
 		else:
 			self._downloader.trouble(u'ERROR: no fmt_url_map or conn information found in video info')
 			return