[youtube] Add alternative automatic captions extraction approach (Closes #8667)

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

diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py
index 45ad881520e27e2e860c761ad635196306d3995d..ec90c211160f9f761ea25816683b9c09245fd817 100644 (file)
--- a/youtube_dl/extractor/youtube.py
+++ b/youtube_dl/extractor/youtube.py
@@ -975,40 +975,67 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             return {}
         try:
             args = player_config['args']
-            caption_url = args['ttsurl']
-            if not caption_url:
-                self._downloader.report_warning(err_msg)
-                return {}
-            timestamp = args['timestamp']
-            # We get the available subtitles
-            list_params = compat_urllib_parse.urlencode({
-                'type': 'list',
-                'tlangs': 1,
-                'asrs': 1,
-            })
-            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:
-                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', '')
+            caption_url = args.get('ttsurl')
+            if caption_url:
+                timestamp = args['timestamp']
+                # We get the available subtitles
+                list_params = compat_urllib_parse.urlencode({
+                    'type': 'list',
+                    'tlangs': 1,
+                    'asrs': 1,
+                })
+                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:
+                    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'):
+                    sub_lang = lang_node.attrib['lang_code']
+                    sub_formats = []
+                    for ext in self._SUBTITLE_FORMATS:
+                        params = compat_urllib_parse.urlencode({
+                            'lang': original_lang,
+                            'tlang': sub_lang,
+                            'fmt': ext,
+                            'ts': timestamp,
+                            'kind': caption_kind,
+                        })
+                        sub_formats.append({
+                            'url': caption_url + '&' + params,
+                            'ext': ext,
+                        })
+                    sub_lang_list[sub_lang] = sub_formats
+                return sub_lang_list
+
+            # Some videos don't provide ttsurl but rather caption_tracks and
+            # caption_translation_languages (e.g. 20LmZk1hakA)
+            caption_tracks = args['caption_tracks']
+            caption_translation_languages = args['caption_translation_languages']
+            caption_url = compat_parse_qs(caption_tracks.split(',')[0])['u'][0]
+            parsed_caption_url = compat_urlparse.urlparse(caption_url)
+            caption_qs = compat_parse_qs(parsed_caption_url.query)
 
             sub_lang_list = {}
-            for lang_node in caption_list.findall('target'):
-                sub_lang = lang_node.attrib['lang_code']
+            for lang in caption_translation_languages.split(','):
+                lang_qs = compat_parse_qs(compat_urllib_parse_unquote_plus(lang))
+                sub_lang = lang_qs.get('lc', [None])[0]
+                if not sub_lang:
+                    continue
                 sub_formats = []
                 for ext in self._SUBTITLE_FORMATS:
-                    params = compat_urllib_parse.urlencode({
-                        'lang': original_lang,
-                        'tlang': sub_lang,
-                        'fmt': ext,
-                        'ts': timestamp,
-                        'kind': caption_kind,
+                    caption_qs.update({
+                        'tlang': [sub_lang],
+                        'fmt': [ext],
                     })
+                    sub_url = compat_urlparse.urlunparse(parsed_caption_url._replace(
+                        query=compat_urllib_parse.urlencode(caption_qs, True)))
                     sub_formats.append({
-                        'url': caption_url + '&' + params,
+                        'url': sub_url,
                         'ext': ext,
                     })
                 sub_lang_list[sub_lang] = sub_formats
@@ -1701,13 +1728,7 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
 
         return self.playlist_result(self._entries(page, playlist_id), playlist_id, playlist_title)
 
-    def _real_extract(self, url):
-        # Extract playlist id
-        mobj = re.match(self._VALID_URL, url)
-        if mobj is None:
-            raise ExtractorError('Invalid URL: %s' % url)
-        playlist_id = mobj.group(1) or mobj.group(2)
-
+    def _check_download_just_video(self, url, playlist_id):
         # Check if it's a video-specific URL
         query_dict = compat_urlparse.parse_qs(compat_urlparse.urlparse(url).query)
         if 'v' in query_dict:
@@ -1718,6 +1739,17 @@ class YoutubePlaylistIE(YoutubePlaylistBaseInfoExtractor):
             else:
                 self.to_screen('Downloading playlist %s - add --no-playlist to just download video %s' % (playlist_id, video_id))
 
+    def _real_extract(self, url):
+        # Extract playlist id
+        mobj = re.match(self._VALID_URL, url)
+        if mobj is None:
+            raise ExtractorError('Invalid URL: %s' % url)
+        playlist_id = mobj.group(1) or mobj.group(2)
+
+        video = self._check_download_just_video(url, playlist_id)
+        if video:
+            return video
+
         if playlist_id.startswith('RD') or playlist_id.startswith('UL'):
             # Mixes require a custom extraction process
             return self._extract_mix(playlist_id)
@@ -2026,11 +2058,20 @@ class YoutubeFeedsInfoExtractor(YoutubeBaseInfoExtractor):
 class YoutubeWatchLaterIE(YoutubePlaylistIE):
     IE_NAME = 'youtube:watchlater'
     IE_DESC = 'Youtube watch later list, ":ytwatchlater" for short (requires authentication)'
-    _VALID_URL = r'https?://www\.youtube\.com/(?:feed/watch_later|playlist\?list=WL)|:ytwatchlater'
+    _VALID_URL = r'https?://www\.youtube\.com/(?:feed/watch_later|(?:playlist|watch)\?(?:.+&)?list=WL)|:ytwatchlater'
 
-    _TESTS = []  # override PlaylistIE tests
+    _TESTS = [{
+        'url': 'https://www.youtube.com/playlist?list=WL',
+        'only_matching': True,
+    }, {
+        'url': 'https://www.youtube.com/watch?v=bCNU9TrbiRk&index=1&list=WL',
+        'only_matching': True,
+    }]
 
     def _real_extract(self, url):
+        video = self._check_download_just_video(url, 'WL')
+        if video:
+            return video
         return self._extract_playlist('WL')