From 2a9c6dcd22b2d93632e8bb78686df40547c1fb8b Mon Sep 17 00:00:00 2001 From: pukkandan Date: Mon, 26 Jul 2021 03:33:42 +0530 Subject: [PATCH] [youtube] Fix format sorting when using alternate clients --- yt_dlp/extractor/youtube.py | 60 +++++++++++++++++++++---------------- 1 file changed, 35 insertions(+), 25 deletions(-) diff --git a/yt_dlp/extractor/youtube.py b/yt_dlp/extractor/youtube.py index 6e34dc25d5..153cb29978 100644 --- a/yt_dlp/extractor/youtube.py +++ b/yt_dlp/extractor/youtube.py @@ -2499,11 +2499,12 @@ def _extract_player_responses(self, clients, video_id, webpage, master_ytcfg, pl def _extract_formats(self, streaming_data, video_id, player_url, is_live): itags, stream_ids = [], [] - itag_qualities = {} + itag_qualities, res_qualities = {}, {} q = qualities([ - # "tiny" is the smallest video-only format. But some audio-only formats - # was also labeled "tiny". It is not clear if such formats still exist - 'tiny', 'audio_quality_low', 'audio_quality_medium', 'audio_quality_high', # Audio only formats + # Normally tiny is the smallest video-only formats. But + # audio-only formats with unknown quality may get tagged as tiny + 'tiny', + 'audio_quality_ultralow', 'audio_quality_low', 'audio_quality_medium', 'audio_quality_high', # Audio only formats 'small', 'medium', 'large', 'hd720', 'hd1080', 'hd1440', 'hd2160', 'hd2880', 'highres' ]) streaming_formats = traverse_obj(streaming_data, (..., ('formats', 'adaptiveFormats'), ...), default=[]) @@ -2519,10 +2520,18 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live): continue quality = fmt.get('quality') + height = int_or_none(fmt.get('height')) if quality == 'tiny' or not quality: quality = fmt.get('audioQuality', '').lower() or quality - if itag and quality: - itag_qualities[itag] = quality + # The 3gp format (17) in android client has a quality of "small", + # but is actually worse than other formats + if itag == '17': + quality = 'tiny' + if quality: + if itag: + itag_qualities[itag] = quality + if height: + res_qualities[height] = quality # FORMAT_STREAM_TYPE_OTF(otf=1) requires downloading the init fragment # (adding `&sq=0` to the URL) and parsing emsg box to determine the # number of fragment that would subsequently requested with (`&sq=N`) @@ -2553,13 +2562,14 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live): 'filesize': int_or_none(fmt.get('contentLength')), 'format_id': itag, 'format_note': ', '.join(filter(None, ( - audio_track.get('displayName'), fmt.get('qualityLabel') or quality))), + audio_track.get('displayName'), + fmt.get('qualityLabel') or quality.replace('audio_quality_', '')))), 'fps': int_or_none(fmt.get('fps')), - 'height': int_or_none(fmt.get('height')), + 'height': height, 'quality': q(quality), 'tbr': tbr, 'url': fmt_url, - 'width': fmt.get('width'), + 'width': int_or_none(fmt.get('width')), 'language': audio_track.get('id', '').split('.')[0], } mime_mobj = re.match( @@ -2567,11 +2577,6 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live): if mime_mobj: dct['ext'] = mimetype2ext(mime_mobj.group(1)) dct.update(parse_codecs(mime_mobj.group(2))) - # The 3gp format in android client has a quality of "small", - # but is actually worse than all other formats - if dct['ext'] == '3gp': - dct['quality'] = q('tiny') - dct['preference'] = -10 no_audio = dct.get('acodec') == 'none' no_video = dct.get('vcodec') == 'none' if no_audio: @@ -2591,11 +2596,16 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live): get_dash = not is_live and 'dash' not in skip_manifests and self.get_param('youtube_include_dash_manifest', True) get_hls = 'hls' not in skip_manifests and self.get_param('youtube_include_hls_manifest', True) + def guess_quality(f): + for val, qdict in ((f.get('format_id'), itag_qualities), (f.get('height'), res_qualities)): + if val in qdict: + return q(qdict[val]) + return -1 + for sd in streaming_data: hls_manifest_url = get_hls and sd.get('hlsManifestUrl') if hls_manifest_url: - for f in self._extract_m3u8_formats( - hls_manifest_url, video_id, 'mp4', fatal=False): + for f in self._extract_m3u8_formats(hls_manifest_url, video_id, 'mp4', fatal=False): itag = self._search_regex( r'/itag/(\d+)', f['url'], 'itag', default=None) if itag in itags: @@ -2603,19 +2613,18 @@ def _extract_formats(self, streaming_data, video_id, player_url, is_live): if itag: f['format_id'] = itag itags.append(itag) + f['quality'] = guess_quality(f) yield f dash_manifest_url = get_dash and sd.get('dashManifestUrl') if dash_manifest_url: - for f in self._extract_mpd_formats( - dash_manifest_url, video_id, fatal=False): + for f in self._extract_mpd_formats(dash_manifest_url, video_id, fatal=False): itag = f['format_id'] if itag in itags: continue if itag: itags.append(itag) - if itag in itag_qualities: - f['quality'] = q(itag_qualities[itag]) + f['quality'] = guess_quality(f) filesize = int_or_none(self._search_regex( r'/clen/(\d+)', f.get('fragment_base_url') or f['url'], 'file size', default=None)) @@ -2740,13 +2749,14 @@ def feed_entry(name): self.raise_no_formats(reason, expected=True) for f in formats: - # TODO: detect if throttled - if '&n=' in f['url']: # possibly throttled + if '&c=WEB&' in f['url'] and '&ratebypass=yes&' not in f['url']: # throttled f['source_preference'] = -10 - # note = f.get('format_note') - # f['format_note'] = f'{note} (throttled)' if note else '(throttled)' + note = f.get('format_note') + f['format_note'] = f'{note} (throttled)' if note else '(throttled)' - self._sort_formats(formats) + # Source is given priority since formats that throttle are given lower source_preference + # When throttling issue is fully fixed, remove this + self._sort_formats(formats, ('quality', 'height', 'fps', 'source')) keywords = get_first(video_details, 'keywords', expected_type=list) or [] if not keywords and webpage: