[fd/ffmpeg] Limit read rate for DASH livestreams (#14918)

Authored by: bashonly

yt_dlp/downloader/external.py

@@ -560,6 +560,13 @@ class FFmpegFD(ExternalFD):
                 elif isinstance(conn, str):
                     args += ['-rtmp_conn', conn]
 
+            elif protocol == 'http_dash_segments' and info_dict.get('is_live'):
+                # ffmpeg may try to read past the latest available segments for
+                # live DASH streams unless we pass `-re`. In modern ffmpeg, this
+                # is an alias of `-readrate 1`, but `-readrate` was not added
+                # until ffmpeg 5.0, so we must stick to using `-re`
+                args += ['-re']
+
             url = fmt['url']
             if self.params.get('enable_file_urls') and url.startswith('file:'):
                 # The default protocol_whitelist is 'file,crypto,data' when reading local m3u8 URLs,

yt_dlp/extractor/youtube/_tab.py

@@ -340,8 +340,9 @@ class YoutubeTabBaseInfoExtractor(YoutubeBaseInfoExtractor):
             thumbnails=self._extract_thumbnails(view_model, (
                 'contentImage', *thumb_keys, 'thumbnailViewModel', 'image'), final_key='sources'),
             duration=traverse_obj(view_model, (
-                'contentImage', 'thumbnailViewModel', 'overlays', ..., 'thumbnailOverlayBadgeViewModel',
+                'contentImage', 'thumbnailViewModel', 'overlays', ...,
-                'thumbnailBadges', ..., 'thumbnailBadgeViewModel', 'text', {parse_duration}, any)),
+                (('thumbnailBottomOverlayViewModel', 'badges'), ('thumbnailOverlayBadgeViewModel', 'thumbnailBadges')),
+                ..., 'thumbnailBadgeViewModel', 'text', {parse_duration}, any)),
             timestamp=(traverse_obj(view_model, (
                 'metadata', 'lockupMetadataViewModel', 'metadata', 'contentMetadataViewModel', 'metadataRows',
                 ..., 'metadataParts', ..., 'text', 'content', {lambda t: self._parse_time_text(t, report_failure=False)}, any))

yt_dlp/extractor/youtube/_video.py

@@ -1556,6 +1556,46 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
             'view_count': int,
         },
         'params': {'skip_download': True},
+    }, {
+        # Youtube Music Auto-generated description with dot in artist name
+        'url': 'https://music.youtube.com/watch?v=DbCvuSGfR3Y',
+        'info_dict': {
+            'id': 'DbCvuSGfR3Y',
+            'ext': 'mp4',
+            'title': 'Back Around',
+            'artists': ['half·alive'],
+            'track': 'Back Around',
+            'album': 'Conditions Of A Punk',
+            'release_date': '20221202',
+            'release_year': 2021,
+            'alt_title': 'Back Around',
+            'description': 'md5:bfc0e2b3cc903a608d8a85a13cb50f95',
+            'media_type': 'video',
+            'uploader': 'half•alive',
+            'channel': 'half•alive',
+            'channel_id': 'UCYQrYophdVI3nVDPOnXyIng',
+            'channel_url': 'https://www.youtube.com/channel/UCYQrYophdVI3nVDPOnXyIng',
+            'channel_is_verified': True,
+            'channel_follower_count': int,
+            'comment_count': int,
+            'view_count': int,
+            'like_count': int,
+            'age_limit': 0,
+            'duration': 223,
+            'thumbnail': 'https://i.ytimg.com/vi_webp/DbCvuSGfR3Y/maxresdefault.webp',
+            'heatmap': 'count:100',
+            'categories': ['Music'],
+            'tags': ['half·alive', 'Conditions Of A Punk', 'Back Around'],
+            'creators': ['half·alive'],
+            'timestamp': 1669889281,
+            'upload_date': '20221201',
+            'playable_in_embed': True,
+            'availability': 'public',
+            'live_status': 'not_live',
+        },
+        'params': {
+            'skip_download': True,
+        },
     }]
     _WEBPAGE_TESTS = [{
         # <object>
@@ -3023,8 +3063,12 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
     def _extract_formats_and_subtitles(self, video_id, player_responses, player_url, live_status, duration):
         CHUNK_SIZE = 10 << 20
-        PREFERRED_LANG_VALUE = 10
+        ORIGINAL_LANG_VALUE = 10
-        original_language = None
+        DEFAULT_LANG_VALUE = 5
+        language_map = {
+            ORIGINAL_LANG_VALUE: None,
+            DEFAULT_LANG_VALUE: None,
+        }
         itags, stream_ids = collections.defaultdict(set), []
         itag_qualities, res_qualities = {}, {0: None}
         subtitles = {}
@@ -3064,6 +3108,22 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
         # For handling potential pre-playback required waiting period
         playback_wait = int_or_none(self._configuration_arg('playback_wait', [None])[0], default=6)
 
+        def get_language_code_and_preference(fmt_stream):
+            audio_track = fmt_stream.get('audioTrack') or {}
+            display_name = audio_track.get('displayName') or ''
+            language_code = audio_track.get('id', '').split('.')[0] or None
+            if 'descriptive' in display_name.lower():
+                return join_nonempty(language_code, 'desc'), -10
+            if 'original' in display_name.lower():
+                if language_code and not language_map.get(ORIGINAL_LANG_VALUE):
+                    language_map[ORIGINAL_LANG_VALUE] = language_code
+                return language_code, ORIGINAL_LANG_VALUE
+            if audio_track.get('audioIsDefault'):
+                if language_code and not language_map.get(DEFAULT_LANG_VALUE):
+                    language_map[DEFAULT_LANG_VALUE] = language_code
+                return language_code, DEFAULT_LANG_VALUE
+            return language_code, -1
+
         for pr in player_responses:
             streaming_data = traverse_obj(pr, 'streamingData')
             if not streaming_data:
@@ -3079,7 +3139,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 return str_or_none(fmt_stream.get('itag')), traverse_obj(fmt_stream, 'audioTrack', 'id'), fmt_stream.get('isDrc')
 
             def process_format_stream(fmt_stream, proto, missing_pot):
-                nonlocal original_language
                 itag = str_or_none(fmt_stream.get('itag'))
                 audio_track = fmt_stream.get('audioTrack') or {}
                 quality = fmt_stream.get('quality')
@@ -3096,13 +3155,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     if height:
                         res_qualities[height] = quality
 
-                display_name = audio_track.get('displayName') or ''
+                language_code, language_preference = get_language_code_and_preference(fmt_stream)
-                is_original = 'original' in display_name.lower()
-                is_descriptive = 'descriptive' in display_name.lower()
-                is_default = audio_track.get('audioIsDefault')
-                language_code = audio_track.get('id', '').split('.')[0]
-                if language_code and (is_original or (is_default and not original_language)):
-                    original_language = language_code
 
                 has_drm = bool(fmt_stream.get('drmFamilies'))
 
@@ -3138,7 +3191,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     'filesize': int_or_none(fmt_stream.get('contentLength')),
                     'format_id': f'{itag}{"-drc" if fmt_stream.get("isDrc") else ""}',
                     'format_note': join_nonempty(
-                        join_nonempty(display_name, is_default and ' (default)', delim=''),
+                        join_nonempty(audio_track.get('displayName'), audio_track.get('audioIsDefault') and '(default)', delim=' '),
                         name, fmt_stream.get('isDrc') and 'DRC',
                         try_get(fmt_stream, lambda x: x['projectionType'].replace('RECTANGULAR', '').lower()),
                         try_get(fmt_stream, lambda x: x['spatialAudioType'].replace('SPATIAL_AUDIO_TYPE_', '').lower()),
@@ -3155,8 +3208,8 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     'tbr': tbr,
                     'filesize_approx': filesize_from_tbr(tbr, format_duration),
                     'width': int_or_none(fmt_stream.get('width')),
-                    'language': join_nonempty(language_code, 'desc' if is_descriptive else '') or None,
+                    'language': language_code,
-                    'language_preference': PREFERRED_LANG_VALUE if is_original else 5 if is_default else -10 if is_descriptive else -1,
+                    'language_preference': language_preference,
                     # Strictly de-prioritize damaged and 3gp formats
                     'preference': -10 if is_damaged else -2 if itag == '17' else None,
                 }
@@ -3206,6 +3259,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                     fmt_url = fmt_stream.get('url')
                     encrypted_sig, sc = None, None
                     if not fmt_url:
+                        # We still need to register original/default language information
+                        # See: https://github.com/yt-dlp/yt-dlp/issues/14883
+                        get_language_code_and_preference(fmt_stream)
                         sc = urllib.parse.parse_qs(fmt_stream.get('signatureCipher'))
                         fmt_url = url_or_none(try_get(sc, lambda x: x['url'][0]))
                         encrypted_sig = try_get(sc, lambda x: x['s'][0])
@@ -3391,9 +3447,13 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 elif itag:
                     f['format_id'] = itag
 
-                if original_language and f.get('language') == original_language:
+                lang_code = f.get('language')
+                if lang_code and lang_code == language_map[ORIGINAL_LANG_VALUE]:
+                    f['format_note'] = join_nonempty(f.get('format_note'), '(original)', delim=' ')
+                    f['language_preference'] = ORIGINAL_LANG_VALUE
+                elif lang_code and lang_code == language_map[DEFAULT_LANG_VALUE]:
                     f['format_note'] = join_nonempty(f.get('format_note'), '(default)', delim=' ')
-                    f['language_preference'] = PREFERRED_LANG_VALUE
+                    f['language_preference'] = DEFAULT_LANG_VALUE
 
                 if itag in ('616', '235'):
                     f['format_note'] = join_nonempty(f.get('format_note'), 'Premium', delim=' ')
@@ -3988,20 +4048,14 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
 
         # Youtube Music Auto-generated description
         if (video_description or '').strip().endswith('\nAuto-generated by YouTube.'):
-            # XXX: Causes catastrophic backtracking if description has "·"
-            # E.g. https://www.youtube.com/watch?v=DoPaAxMQoiI
-            # Simulating atomic groups:  (?P<a>[^xy]+)x  =>  (?=(?P<a>[^xy]+))(?P=a)x
-            # reduces it, but does not fully fix it. https://regex101.com/r/8Ssf2h/2
             mobj = re.search(
                 r'''(?xs)
-                    (?=(?P<track>[^\n·]+))(?P=track)·
+                    (?:\n|^)(?P<track>[^\n·]+)\ ·\ (?P<artist>[^\n]+)\n+
-                    (?=(?P<artist>[^\n]+))(?P=artist)\n+
+                    (?P<album>[^\n]+)\n+
-                    (?=(?P<album>[^\n]+))(?P=album)\n
+                    (?:℗\s*(?P<release_year>\d{4}))?
-                    (?:.+?℗\s*(?P<release_year>\d{4})(?!\d))?
+                    (?:.+?\nReleased\ on\s*:\s*(?P<release_date>\d{4}-\d{2}-\d{2}))?
-                    (?:.+?Released\ on\s*:\s*(?P<release_date>\d{4}-\d{2}-\d{2}))?
+                    (?:.+?\nArtist\s*:\s*(?P<clean_artist>[^\n]+)\n)?
-                    (.+?\nArtist\s*:\s*
+                    .+\nAuto-generated\ by\ YouTube\.\s*$
-                        (?=(?P<clean_artist>[^\n]+))(?P=clean_artist)\n
-                    )?.+\nAuto-generated\ by\ YouTube\.\s*$
                 ''', video_description)
             if mobj:
                 release_year = mobj.group('release_year')
@@ -4013,7 +4067,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor):
                 info.update({
                     'album': mobj.group('album'.strip()),
                     'artists': ([a] if (a := mobj.group('clean_artist'))
-                                else [a.strip() for a in mobj.group('artist').split('·')]),
+                                else [a.strip() for a in mobj.group('artist').split(' · ')]),
                     'track': mobj.group('track').strip(),
                     'release_date': release_date,
                     'release_year': int_or_none(release_year),