[youtube] Add support for automatically translated subtitles (fixes #4555)

[youtube-dl] / youtube_dl / extractor / youtube.py

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 175e43272e8875e9efc6f91ae33f88d9f118e0b3..224f1b041a6a06c771a210360e1cb20a3c37dc99 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -418,6 +418,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
                 'upload_date': '20140605',
             },
         },
+        # Age-gate video with encrypted signature
+        {
+            'url': 'http://www.youtube.com/watch?v=6kLq3WMV1nU',
+            'info_dict': {
+                'id': '6kLq3WMV1nU',
+                'ext': 'mp4',
+                'title': 'Dedication To My Ex (Miss That) (Lyric Video)',
+                'description': 'md5:33765bb339e1b47e7e72b5490139bb41',
+                'uploader': 'LloydVEVO',
+                'uploader_id': 'LloydVEVO',
+                'upload_date': '20110629',
+            },
+        },
         # video_info is None (https://github.com/rg3/youtube-dl/issues/4421)
         {
             'url': '__2ABJjxzNo',
@@ -649,10 +662,11 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
             list_url = caption_url + '&' + list_params
             caption_list = self._download_xml(list_url, video_id)
             original_lang_node = caption_list.find('track')
-            if original_lang_node is None or original_lang_node.attrib.get('kind') != 'asr':
+            if original_lang_node is None:
                 self._downloader.report_warning('Video doesn\'t have automatic captions')
                 return {}
             original_lang = original_lang_node.attrib['lang_code']
+            caption_kind = original_lang_node.attrib.get('kind', '')
 
             sub_lang_list = {}
             for lang_node in caption_list.findall('target'):
@@ -662,7 +676,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor, SubtitlesInfoExtractor):
                     'tlang': sub_lang,
                     'fmt': sub_format,
                     'ts': timestamp,
-                    'kind': 'asr',
+                    'kind': caption_kind,
                 })
                 sub_lang_list[sub_lang] = caption_url + '&' + params
             return sub_lang_list