added --srt-lang; updated README; extended the -g FAQ

[youtube-dl.git] / youtube-dl

diff --git a/youtube-dl b/youtube-dl
index 689427f..5224611 100755 (executable)
--- a/youtube-dl
+++ b/youtube-dl
@@ -491,6 +491,7 @@ class FileDownloader(object):
        writedescription: Write the video description to a .description file
        writeinfojson:    Write the video description to a .info.json file
        writesubtitles:   Write the video subtitles to a .srt file
+       subtitleslang:    Language of the subtitles to download
        """
 
        params = None
@@ -1443,17 +1444,24 @@ class YoutubeIE(InfoExtractor):
                        else:
                                srt_lang_list = re.findall(r'lang_code="([\w\-]+)"', srt_list)
                                if srt_lang_list:
-                                       if 'en' in srt_lang_list: srt_lang = 'en'
-                                       else: srt_lang = srt_lang_list[0] # TODO choose better and provide an override
-                                       request = urllib2.Request('http://video.google.com/timedtext?hl=en&lang=%s&v=%s' % (srt_lang, video_id))
-                                       try:
-                                               srt_xml = urllib2.urlopen(request).read()
-                                       except (urllib2.URLError, httplib.HTTPException, socket.error), err:
-                                               self._downloader.trouble(u'WARNING: unable to download video subtitles: %s' % str(err))
+                                       if self._downloader.params.get('subtitleslang', False):
+                                               srt_lang = self._downloader.params.get('subtitleslang')
+                                       elif 'en' in srt_lang_list:
+                                               srt_lang = 'en'
+                                       else:
+                                               srt_lang = srt_lang_list[0]
+                                       if not srt_lang in srt_lang_list:
+                                               self._downloader.trouble(u'WARNING: no closed captions found in the specified language')
                                        else:
-                                               video_subtitles = self._closed_captions_xml_to_srt(srt_xml.decode('utf-8'))
+                                               request = urllib2.Request('http://video.google.com/timedtext?hl=en&lang=%s&v=%s' % (srt_lang, video_id))
+                                               try:
+                                                       srt_xml = urllib2.urlopen(request).read()
+                                               except (urllib2.URLError, httplib.HTTPException, socket.error), err:
+                                                       self._downloader.trouble(u'WARNING: unable to download video subtitles: %s' % str(err))
+                                               else:
+                                                       video_subtitles = self._closed_captions_xml_to_srt(srt_xml.decode('utf-8'))
                                else:
-                                       self._downloader.trouble(u'WARNING: video has no subtitles')
+                                       self._downloader.trouble(u'WARNING: video has no closed captions')
 
                # token
                video_token = urllib.unquote_plus(video_info['token'][0])
@@ -4385,6 +4393,12 @@ def parseOpts():
                        action='store', dest='format_limit', metavar='FORMAT', help='highest quality format to download')
        video_format.add_option('-F', '--list-formats',
                        action='store_true', dest='listformats', help='list all available formats (currently youtube only)')
+       video_format.add_option('--write-srt',
+                       action='store_true', dest='writesubtitles',
+                       help='write video closed captions to a .srt file (currently youtube only)', default=False)
+       video_format.add_option('--srt-lang',
+                       action='store', dest='subtitleslang', metavar='LANG',
+                       help='language of the closed captions to download (optional) use IETF language tags like \'en\'')
 
 
        verbosity.add_option('-q', '--quiet',
@@ -4449,9 +4463,6 @@ def parseOpts():
        filesystem.add_option('--write-info-json',
                        action='store_true', dest='writeinfojson',
                        help='write video metadata to a .info.json file', default=False)
-       filesystem.add_option('--write-srt',
-                       action='store_true', dest='writesubtitles',
-                       help='write video subtitles to a .srt file', default=False)
 
 
        postproc.add_option('--extract-audio', action='store_true', dest='extractaudio', default=False,
@@ -4653,6 +4664,7 @@ def _real_main():
                'writedescription': opts.writedescription,
                'writeinfojson': opts.writeinfojson,
                'writesubtitles': opts.writesubtitles,
+               'subtitleslang': opts.subtitleslang,
                'matchtitle': opts.matchtitle,
                'rejecttitle': opts.rejecttitle,
                'max_downloads': opts.max_downloads,