X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fyandexmusic.py;h=d3cc1a29fa473fee2f58e91323774633be00fc4b;hb=5c2266df4b9aeb7881ed8c026a038e2a25e43734;hp=f4c0f5702e59bea80046438606dd9a28271d8a30;hpb=674fb0fcc54c72448f80a0573f7fd116f220827e;p=youtube-dl diff --git a/youtube_dl/extractor/yandexmusic.py b/youtube_dl/extractor/yandexmusic.py index f4c0f5702..d3cc1a29f 100644 --- a/youtube_dl/extractor/yandexmusic.py +++ b/youtube_dl/extractor/yandexmusic.py @@ -1,18 +1,38 @@ -# coding=utf-8 +# coding: utf-8 from __future__ import unicode_literals import re import hashlib from .common import InfoExtractor -from ..compat import compat_str +from ..compat import ( + compat_str, + compat_urllib_parse, +) from ..utils import ( int_or_none, float_or_none, + sanitized_Request, ) -class YandexMusicBaseIE(InfoExtractor): +class YandexMusicTrackIE(InfoExtractor): + IE_NAME = 'yandexmusic:track' + IE_DESC = 'Яндекс.Музыка - Трек' + _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P\d+)/track/(?P\d+)' + + _TEST = { + 'url': 'http://music.yandex.ru/album/540508/track/4878838', + 'md5': 'f496818aa2f60b6c0062980d2e00dc20', + 'info_dict': { + 'id': '4878838', + 'ext': 'mp3', + 'title': 'Carlo Ambrosio - Gypsy Eyes 1', + 'filesize': 4628061, + 'duration': 193.04, + } + } + def _get_track_url(self, storage_dir, track_id): data = self._download_json( 'http://music.yandex.ru/api/v1.5/handlers/api-jsonp.jsx?action=getTrackSrc&p=download-info/%s' @@ -26,6 +46,12 @@ class YandexMusicBaseIE(InfoExtractor): % (data['host'], key, data['ts'] + data['path'], storage[1])) def _get_track_info(self, track): + thumbnail = None + cover_uri = track.get('albums', [{}])[0].get('coverUri') + if cover_uri: + thumbnail = cover_uri.replace('%%', 'orig') + if not thumbnail.startswith('http'): + thumbnail = 'http://' + thumbnail return { 'id': track['id'], 'ext': 'mp3', @@ -33,26 +59,9 @@ class YandexMusicBaseIE(InfoExtractor): 'title': '%s - %s' % (track['artists'][0]['name'], track['title']), 'filesize': int_or_none(track.get('fileSize')), 'duration': float_or_none(track.get('durationMs'), 1000), + 'thumbnail': thumbnail, } - -class YandexMusicTrackIE(YandexMusicBaseIE): - IE_NAME = 'yandexmusic:track' - IE_DESC = 'Яндекс.Музыка - Трек' - _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P\d+)/track/(?P\d+)' - - _TEST = { - 'url': 'http://music.yandex.ru/album/540508/track/4878838', - 'md5': 'f496818aa2f60b6c0062980d2e00dc20', - 'info_dict': { - 'id': '4878838', - 'ext': 'mp3', - 'title': 'Carlo Ambrosio - Gypsy Eyes 1', - 'filesize': 4628061, - 'duration': 193.04, - } - } - def _real_extract(self, url): mobj = re.match(self._VALID_URL, url) album_id, track_id = mobj.group('album_id'), mobj.group('id') @@ -64,7 +73,15 @@ class YandexMusicTrackIE(YandexMusicBaseIE): return self._get_track_info(track) -class YandexMusicAlbumIE(YandexMusicBaseIE): +class YandexMusicPlaylistBaseIE(InfoExtractor): + def _build_playlist(self, tracks): + return [ + self.url_result( + 'http://music.yandex.ru/album/%s/track/%s' % (track['albums'][0]['id'], track['id'])) + for track in tracks if track.get('albums') and isinstance(track.get('albums'), list)] + + +class YandexMusicAlbumIE(YandexMusicPlaylistBaseIE): IE_NAME = 'yandexmusic:album' IE_DESC = 'Яндекс.Музыка - Альбом' _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/album/(?P\d+)/?(\?|$)' @@ -85,7 +102,7 @@ class YandexMusicAlbumIE(YandexMusicBaseIE): 'http://music.yandex.ru/handlers/album.jsx?album=%s' % album_id, album_id, 'Downloading album JSON') - entries = [self._get_track_info(track) for track in album['volumes'][0]] + entries = self._build_playlist(album['volumes'][0]) title = '%s - %s' % (album['artists'][0]['name'], album['title']) year = album.get('year') @@ -95,12 +112,12 @@ class YandexMusicAlbumIE(YandexMusicBaseIE): return self.playlist_result(entries, compat_str(album['id']), title) -class YandexMusicPlaylistIE(YandexMusicBaseIE): +class YandexMusicPlaylistIE(YandexMusicPlaylistBaseIE): IE_NAME = 'yandexmusic:playlist' IE_DESC = 'Яндекс.Музыка - Плейлист' _VALID_URL = r'https?://music\.yandex\.(?:ru|kz|ua|by)/users/[^/]+/playlists/(?P\d+)' - _TEST = { + _TESTS = [{ 'url': 'http://music.yandex.ru/users/music.partners/playlists/1245', 'info_dict': { 'id': '1245', @@ -108,20 +125,54 @@ class YandexMusicPlaylistIE(YandexMusicBaseIE): 'description': 'md5:3b9f27b0efbe53f2ee1e844d07155cc9', }, 'playlist_count': 6, - } + }, { + # playlist exceeding the limit of 150 tracks shipped with webpage (see + # https://github.com/rg3/youtube-dl/issues/6666) + 'url': 'https://music.yandex.ru/users/ya.playlist/playlists/1036', + 'info_dict': { + 'id': '1036', + 'title': 'Музыка 90-х', + }, + 'playlist_count': 310, + }] def _real_extract(self, url): playlist_id = self._match_id(url) webpage = self._download_webpage(url, playlist_id) - playlist = self._parse_json( + mu = self._parse_json( self._search_regex( r'var\s+Mu\s*=\s*({.+?});\s*', webpage, 'player'), - playlist_id)['pageData']['playlist'] - - entries = [self._get_track_info(track) for track in playlist['tracks']] + playlist_id) + + playlist = mu['pageData']['playlist'] + tracks, track_ids = playlist['tracks'], playlist['trackIds'] + + # tracks dictionary shipped with webpage is limited to 150 tracks, + # missing tracks should be retrieved manually. + if len(tracks) < len(track_ids): + present_track_ids = set([compat_str(track['id']) for track in tracks if track.get('id')]) + missing_track_ids = set(map(compat_str, track_ids)) - set(present_track_ids) + request = sanitized_Request( + 'https://music.yandex.ru/handlers/track-entries.jsx', + compat_urllib_parse.urlencode({ + 'entries': ','.join(missing_track_ids), + 'lang': mu.get('settings', {}).get('lang', 'en'), + 'external-domain': 'music.yandex.ru', + 'overembed': 'false', + 'sign': mu.get('authData', {}).get('user', {}).get('sign'), + 'strict': 'true', + }).encode('utf-8')) + request.add_header('Referer', url) + request.add_header('X-Requested-With', 'XMLHttpRequest') + + missing_tracks = self._download_json( + request, playlist_id, 'Downloading missing tracks JSON', fatal=False) + if missing_tracks: + tracks.extend(missing_tracks) return self.playlist_result( - entries, compat_str(playlist_id), + self._build_playlist(tracks), + compat_str(playlist_id), playlist['title'], playlist.get('description'))