From 3da17834a49fad2a97c308fdd89aa26781ef4d60 Mon Sep 17 00:00:00 2001 From: pukkandan Date: Tue, 28 Feb 2023 23:03:44 +0530 Subject: [PATCH] [Youtube] Construct dash formats with `range` query See yt-dlp/yt_dlp#6369 --- youtube_dl/extractor/youtube.py | 22 ++++++++++++++++------ 1 file changed, 16 insertions(+), 6 deletions(-) diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 4246d84f9..89711c84e 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -1694,8 +1694,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if n_response is None: # give up if descrambling failed break - fmt['url'] = update_url( - parsed_fmt_url, query_update={'n': [n_response]}) + for fmt_dct in traverse_obj(fmt, (None, (None, ('fragments', Ellipsis))), expected_type=dict): + fmt_dct['url'] = update_url( + fmt_dct['url'], query_update={'n': [n_response]}) # from yt-dlp, with tweaks def _extract_signature_timestamp(self, video_id, player_url, ytcfg=None, fatal=False): @@ -2047,10 +2048,19 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if no_video: dct['abr'] = tbr if no_audio or no_video: - dct['downloader_options'] = { - # Youtube throttles chunks >~10M - 'http_chunk_size': 10485760, - } + CHUNK_SIZE = 10 << 20 + # avoid Youtube throttling + dct.update({ + 'protocol': 'http_dash_segments', + 'fragments': [{ + 'url': update_url_query(dct['url'], { + 'range': '{0}-{1}'.format(range_start, min(range_start + CHUNK_SIZE - 1, dct['filesize'])) + }) + } for range_start in range(0, dct['filesize'], CHUNK_SIZE)] + } if dct['filesize'] else { + 'downloader_options': {'http_chunk_size': CHUNK_SIZE} # No longer useful? + }) + if dct.get('ext'): dct['container'] = dct['ext'] + '_dash' formats.append(dct)