X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;ds=sidebyside;f=youtube_dl%2Fextractor%2Fpbs.py;h=81918ac6e455313c59a7b56b6c63a5bae4699e57;hb=a6571f1073eab6c9a4cc9800a0bff31cf12fe09f;hp=38cdb9975f41e559683d65f03157071a298b3040;hpb=cbc032c8b70a038a69259378c92b4ba97b42d491;p=youtube-dl

diff --git a/youtube_dl/extractor/pbs.py b/youtube_dl/extractor/pbs.py
index 38cdb9975..81918ac6e 100644
--- a/youtube_dl/extractor/pbs.py
+++ b/youtube_dl/extractor/pbs.py
@@ -512,7 +512,9 @@ class PBSIE(InfoExtractor):
         if http_url:
             for m3u8_format in m3u8_formats:
                 bitrate = self._search_regex(r'(\d+k)', m3u8_format['url'], 'bitrate', default=None)
-                if not bitrate:
+                # extract only the formats that we know that they will be available as http format.
+                # https://projects.pbs.org/confluence/display/coveapi/COVE+Video+Specifications
+                if not bitrate or bitrate not in ('400k', '800k', '1200k', '2500k'):
                     continue
                 f = m3u8_format.copy()
                 f.update({
@@ -535,6 +537,19 @@ class PBSIE(InfoExtractor):
                 'ext': 'ttml',
                 'url': closed_captions_url,
             }]
+            mobj = re.search(r'/(\d+)_Encoded\.dfxp', closed_captions_url)
+            if mobj:
+                ttml_caption_suffix, ttml_caption_id = mobj.group(0, 1)
+                ttml_caption_id = int(ttml_caption_id)
+                subtitles['en'].extend([{
+                    'url': closed_captions_url.replace(
+                        ttml_caption_suffix, '/%d_Encoded.srt' % (ttml_caption_id + 1)),
+                    'ext': 'srt',
+                }, {
+                    'url': closed_captions_url.replace(
+                        ttml_caption_suffix, '/%d_Encoded.vtt' % (ttml_caption_id + 2)),
+                    'ext': 'vtt',
+                }])
 
         # info['title'] is often incomplete (e.g. 'Full Episode', 'Episode 5', etc)
         # Try turning it to 'program - title' naming scheme if possible