X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyoutube.py;h=0442906df5e02802747ae9ddc6f6bb4a909c5c85;hb=409b9324da3ead199b26081b228b042b0b494f8d;hp=677907aba39f444444a6b61a411c29f2e29ba9c2;hpb=936784b272db3f85f5ff5bdd2d5a71e0397ee7bd;p=youtube-dl diff --git a/youtube_dl/extractor/youtube.py b/youtube_dl/extractor/youtube.py index 677907aba..0442906df 100644 --- a/youtube_dl/extractor/youtube.py +++ b/youtube_dl/extractor/youtube.py @@ -64,7 +64,7 @@ class YoutubeBaseInfoExtractor(InfoExtractor): # If True it will raise an error if no login info is provided _LOGIN_REQUIRED = False - _PLAYLIST_ID_RE = r'(?:PL|LL|EC|UU|FL|RD|UL|TL)[0-9A-Za-z-_]{10,}' + _PLAYLIST_ID_RE = r'(?:PL|LL|EC|UU|FL|RD|UL|TL|OLAK5uy_)[0-9A-Za-z-_]{10,}' def _set_language(self): self._set_cookie( @@ -178,13 +178,13 @@ class YoutubeBaseInfoExtractor(InfoExtractor): warn('Unable to extract result entry') return False - tfa = try_get(res, lambda x: x[0][0], list) - if tfa: - tfa_str = try_get(tfa, lambda x: x[2], compat_str) - if tfa_str == 'TWO_STEP_VERIFICATION': + login_challenge = try_get(res, lambda x: x[0][0], list) + if login_challenge: + challenge_str = try_get(login_challenge, lambda x: x[2], compat_str) + if challenge_str == 'TWO_STEP_VERIFICATION': # SEND_SUCCESS - TFA code has been successfully sent to phone # QUOTA_EXCEEDED - reached the limit of TFA codes - status = try_get(tfa, lambda x: x[5], compat_str) + status = try_get(login_challenge, lambda x: x[5], compat_str) if status == 'QUOTA_EXCEEDED': warn('Exceeded the limit of TFA codes, try later') return False @@ -228,6 +228,17 @@ class YoutubeBaseInfoExtractor(InfoExtractor): check_cookie_url = try_get( tfa_results, lambda x: x[0][-1][2], compat_str) + else: + CHALLENGES = { + 'LOGIN_CHALLENGE': "This device isn't recognized. For your security, Google wants to make sure it's really you.", + 'USERNAME_RECOVERY': 'Please provide additional information to aid in the recovery process.', + 'REAUTH': "There is something unusual about your activity. For your security, Google wants to make sure it's really you.", + } + challenge = CHALLENGES.get( + challenge_str, + '%s returned error %s.' % (self.IE_NAME, challenge_str)) + warn('%s\nGo to https://accounts.google.com/, login and solve a challenge.' % challenge) + return False else: check_cookie_url = try_get(res, lambda x: x[2], compat_str) @@ -530,7 +541,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/justintimberlakeVEVO', 'license': 'Standard YouTube License', 'creator': 'Justin Timberlake', - 'track': 'Tunnel Vision`', + 'track': 'Tunnel Vision', 'artist': 'Justin Timberlake', 'age_limit': 18, } @@ -601,7 +612,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'id': 'IB3lcPjvWLA', 'ext': 'm4a', 'title': 'Afrojack, Spree Wilson - The Spark ft. Spree Wilson', - 'description': 'md5:12e7067fa6735a77bdcbb58cb1187d2d', + 'description': 'md5:1900ed86ee514927b9e00fbead6969a5', 'duration': 244, 'uploader': 'AfrojackVEVO', 'uploader_id': 'AfrojackVEVO', @@ -642,7 +653,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'ext': 'mp4', 'duration': 219, 'upload_date': '20100909', - 'uploader': 'The Amazing Atheist', + 'uploader': 'TJ Kirk', 'uploader_id': 'TheAmazingAtheist', 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/TheAmazingAtheist', 'license': 'Standard YouTube License', @@ -672,10 +683,10 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'url': 'https://www.youtube.com/watch?v=6kLq3WMV1nU', 'info_dict': { 'id': '6kLq3WMV1nU', - 'ext': 'mp4', + 'ext': 'webm', 'title': 'Dedication To My Ex (Miss That) (Lyric Video)', 'description': 'md5:33765bb339e1b47e7e72b5490139bb41', - 'duration': 247, + 'duration': 246, 'uploader': 'LloydVEVO', 'uploader_id': 'LloydVEVO', 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/LloydVEVO', @@ -737,7 +748,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'uploader_id': 'AllenMeow', 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/AllenMeow', 'description': 'made by Wacom from Korea | åå¹&å 油添é by TY\'s Allen | æè¬heylisa00cavey1001åå¸ç±æ æä¾æ¢åç¿»è¯', - 'uploader': 'å«è¾å«', + 'uploader': 'å«áá ', 'license': 'Standard YouTube License', 'title': '[A-made] è®æ å¦åå¹ç å¤ªå¦ æå°±æ¯é樣ç人', }, @@ -764,7 +775,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'url': 'https://www.youtube.com/watch?v=FIl7x6_3R5Y', 'info_dict': { 'id': 'FIl7x6_3R5Y', - 'ext': 'mp4', + 'ext': 'webm', 'title': 'md5:7b81415841e02ecd4313668cde88737a', 'description': 'md5:116377fd2963b81ec4ce64b542173306', 'duration': 220, @@ -773,8 +784,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/dorappi2000', 'uploader': 'dorappi2000', 'license': 'Standard YouTube License', - 'formats': 'mincount:32', + 'formats': 'mincount:31', }, + 'skip': 'not actual anymore', }, # DASH manifest with segment_list { @@ -889,7 +901,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'id': 'lsguqyKfVQg', 'ext': 'mp4', 'title': '{dark walk}; Loki/AC/Dishonored; collab w/Elflover21', - 'alt_title': 'Dark Walk', + 'alt_title': 'Dark Walk - Position Music', 'description': 'md5:8085699c11dc3f597ce0410b0dcbb34a', 'duration': 133, 'upload_date': '20151119', @@ -897,7 +909,9 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'uploader_url': r're:https?://(?:www\.)?youtube\.com/user/IronSoulElf', 'uploader': 'IronSoulElf', 'license': 'Standard YouTube License', - 'creator': 'Todd Haberman, Daniel Law Heath & Aaron Kaplan', + 'creator': 'Todd Haberman, Daniel Law Heath and Aaron Kaplan', + 'track': 'Dark Walk - Position Music', + 'artist': 'Todd Haberman, Daniel Law Heath and Aaron Kaplan', }, 'params': { 'skip_download': True, @@ -954,7 +968,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'description': 'md5:dda0d780d5a6e120758d1711d062a867', 'duration': 4060, 'upload_date': '20151119', - 'uploader': 'Bernie 2016', + 'uploader': 'Bernie Sanders', 'uploader_id': 'UCH1dpzjCEiGAt8CXkryhkZg', 'uploader_url': r're:https?://(?:www\.)?youtube\.com/channel/UCH1dpzjCEiGAt8CXkryhkZg', 'license': 'Creative Commons Attribution license (reuse allowed)', @@ -989,6 +1003,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'params': { 'skip_download': True, }, + 'skip': 'This video is not available.', }, { # YouTube Red video with episode data @@ -997,7 +1012,7 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'id': 'iqKdEhx-dD4', 'ext': 'mp4', 'title': 'Isolation - Mind Field (Ep 1)', - 'description': 'md5:8013b7ddea787342608f63a13ddc9492', + 'description': 'md5:25b78d2f64ae81719f5c96319889b736', 'duration': 2085, 'upload_date': '20170118', 'uploader': 'Vsauce', @@ -1030,7 +1045,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor): 'uploader_id': 'UCEJYpZGqgUob0zVVEaLhvVg', 'uploader_url': r're:https?://(?:www\.)?youtube\.com/channel/UCEJYpZGqgUob0zVVEaLhvVg', 'license': 'Standard YouTube License', - 'view_count': int, }, 'params': { 'skip_download': True, @@ -1698,136 +1712,6 @@ class YoutubeIE(YoutubeBaseInfoExtractor): if 'ypc_video_rental_bar_text' in video_info and 'author' not in video_info: raise ExtractorError('"rental" videos not supported. See https://github.com/rg3/youtube-dl/issues/359 for more information.', expected=True) - # Start extracting information - self.report_information_extraction(video_id) - - # uploader - video_uploader = try_get(video_info, lambda x: x['author'][0], compat_str) - if video_uploader: - video_uploader = compat_urllib_parse_unquote_plus(video_uploader) - else: - self._downloader.report_warning('unable to extract uploader name') - - # uploader_id - video_uploader_id = None - video_uploader_url = None - mobj = re.search( - r'', - video_webpage) - if mobj is not None: - video_uploader_id = mobj.group('uploader_id') - video_uploader_url = mobj.group('uploader_url') - else: - self._downloader.report_warning('unable to extract uploader nickname') - - # thumbnail image - # We try first to get a high quality image: - m_thumb = re.search(r'', - video_webpage, re.DOTALL) - if m_thumb is not None: - video_thumbnail = m_thumb.group(1) - elif 'thumbnail_url' not in video_info: - self._downloader.report_warning('unable to extract video thumbnail') - video_thumbnail = None - else: # don't panic if we can't find it - video_thumbnail = compat_urllib_parse_unquote_plus(video_info['thumbnail_url'][0]) - - # upload date - upload_date = self._html_search_meta( - 'datePublished', video_webpage, 'upload date', default=None) - if not upload_date: - upload_date = self._search_regex( - [r'(?s)id="eow-date.*?>(.*?)', - r'(?:id="watch-uploader-info".*?>.*?|["\']simpleText["\']\s*:\s*["\'])(?:Published|Uploaded|Streamed live|Started) on (.+?)[<"\']'], - video_webpage, 'upload date', default=None) - upload_date = unified_strdate(upload_date) - - video_license = self._html_search_regex( - r'