From c9d0f685a43d95d653db56a00efe520e3a04d0d2 Mon Sep 17 00:00:00 2001 From: James Taylor Date: Mon, 19 Oct 2020 13:53:57 -0700 Subject: Use get_video_info to get video urls if player response missing Fixes failure mode 1 in #22 --- youtube/watch.py | 12 ++++++++---- youtube/yt_data_extract/watch_extraction.py | 10 ++++++++-- 2 files changed, 16 insertions(+), 6 deletions(-) (limited to 'youtube') diff --git a/youtube/watch.py b/youtube/watch.py index 1a9e6c4..b1d4665 100644 --- a/youtube/watch.py +++ b/youtube/watch.py @@ -226,15 +226,19 @@ def extract_info(video_id, use_invidious, playlist_id=None, index=None): return {'error': 'Failed to parse json response'} info = yt_data_extract.extract_watch_info(polymer_json) - # age restriction bypass - if info['age_restricted']: - print('Fetching age restriction bypass page') + # request player if it's missing + # see https://github.com/user234683/youtube-local/issues/22#issuecomment-706395160 + if info['age_restricted'] or info['player_response_missing']: + if info['age_restricted']: + print('Age restricted video. Fetching get_video_info page') + else: + print('Missing player. Fetching get_video_info page') data = { 'video_id': video_id, 'eurl': 'https://youtube.googleapis.com/v/' + video_id, } url = 'https://www.youtube.com/get_video_info?' + urllib.parse.urlencode(data) - video_info_page = util.fetch_url(url, debug_name='get_video_info', report_text='Fetched age restriction bypass page').decode('utf-8') + video_info_page = util.fetch_url(url, debug_name='get_video_info', report_text='Fetched get_video_info page').decode('utf-8') yt_data_extract.update_with_age_restricted_info(info, video_info_page) # signature decryption diff --git a/youtube/yt_data_extract/watch_extraction.py b/youtube/yt_data_extract/watch_extraction.py index 340a367..f89cec1 100644 --- a/youtube/yt_data_extract/watch_extraction.py +++ b/youtube/yt_data_extract/watch_extraction.py @@ -447,7 +447,8 @@ def _extract_playability_error(info, player_response, error_prefix=''): SUBTITLE_FORMATS = ('srv1', 'srv2', 'srv3', 'ttml', 'vtt') def extract_watch_info(polymer_json): - info = {'playability_error': None, 'error': None} + info = {'playability_error': None, 'error': None, + 'player_response_missing': None} if isinstance(polymer_json, dict): top_level = polymer_json @@ -477,6 +478,10 @@ def extract_watch_info(polymer_json): else: embedded_player_response = {} + # see https://github.com/user234683/youtube-local/issues/22#issuecomment-706395160 + info['player_response_missing'] = not ( + player_response or embedded_player_response) + # captions info['automatic_caption_languages'] = [] info['manual_caption_languages'] = [] @@ -580,7 +585,8 @@ def get_caption_url(info, language, format, automatic=False, translation_languag return url def update_with_age_restricted_info(info, video_info_page): - ERROR_PREFIX = 'Error bypassing age-restriction: ' + '''Inserts urls from 'player_response' in get_video_info page''' + ERROR_PREFIX = 'Error getting missing player or bypassing age-restriction: ' video_info = urllib.parse.parse_qs(video_info_page) player_response = deep_get(video_info, 'player_response', 0) -- cgit v1.2.3