X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fneteasemusic.py;h=978a05841ce68161330f9db24169dd330e51efc1;hb=HEAD;hp=a70c65ca55fa50e4ef0f964590e7c21937505b24;hpb=2da0cad6ae08e3c78f85a340260aef80a464cb20;p=youtube-dl diff --git a/youtube_dl/extractor/neteasemusic.py b/youtube_dl/extractor/neteasemusic.py index a70c65ca5..978a05841 100644 --- a/youtube_dl/extractor/neteasemusic.py +++ b/youtube_dl/extractor/neteasemusic.py @@ -2,14 +2,19 @@ from __future__ import unicode_literals from hashlib import md5 +from base64 import b64encode from datetime import datetime -import itertools import re from .common import InfoExtractor from ..compat import ( - compat_urllib_request, - compat_urllib_parse, + compat_urllib_parse_urlencode, + compat_str, + compat_itertools_count, +) +from ..utils import ( + sanitized_Request, + float_or_none, ) @@ -20,48 +25,57 @@ class NetEaseMusicBaseIE(InfoExtractor): @classmethod def _encrypt(cls, dfsid): - salt_bytes = bytearray(str(cls._NETEASE_SALT)) - string_bytes = bytearray(str(dfsid)) + salt_bytes = bytearray(cls._NETEASE_SALT.encode('utf-8')) + string_bytes = bytearray(compat_str(dfsid).encode('ascii')) salt_len = len(salt_bytes) - for i in xrange(len(string_bytes)): + for i in range(len(string_bytes)): string_bytes[i] = string_bytes[i] ^ salt_bytes[i % salt_len] m = md5() - m.update(string_bytes) - result = m.digest().encode('base64')[:-1] + m.update(bytes(string_bytes)) + result = b64encode(m.digest()).decode('ascii') return result.replace('/', '_').replace('+', '-') - @classmethod - def extract_formats(cls, info): + def extract_formats(self, info): formats = [] - for song_format in cls._FORMATS: + for song_format in self._FORMATS: details = info.get(song_format) if not details: continue - formats.append({ - 'url': 'http://m1.music.126.net/%s/%s.%s' % - (cls._encrypt(details['dfsId']), details['dfsId'], - details['extension']), - 'ext': details['extension'], - 'abr': details['bitrate'] / 1000, - 'preference': details['bitrate'], - 'format_id': song_format, - 'filesize': details['size'], - 'asr': details['sr'] - }) + song_file_path = '/%s/%s.%s' % ( + self._encrypt(details['dfsId']), details['dfsId'], details['extension']) + + # 203.130.59.9, 124.40.233.182, 115.231.74.139, etc is a reverse proxy-like feature + # from NetEase's CDN provider that can be used if m5.music.126.net does not + # work, especially for users outside of Mainland China + # via: https://github.com/JixunMoe/unblock-163/issues/3#issuecomment-163115880 + for host in ('http://m5.music.126.net', 'http://115.231.74.139/m1.music.126.net', + 'http://124.40.233.182/m1.music.126.net', 'http://203.130.59.9/m1.music.126.net'): + song_url = host + song_file_path + if self._is_valid_url(song_url, info['id'], 'song'): + formats.append({ + 'url': song_url, + 'ext': details.get('extension'), + 'abr': float_or_none(details.get('bitrate'), scale=1000), + 'format_id': song_format, + 'filesize': details.get('size'), + 'asr': details.get('sr') + }) + break return formats @classmethod def convert_milliseconds(cls, ms): - return int(round(ms/1000.0)) + return int(round(ms / 1000.0)) def query_api(self, endpoint, video_id, note): - req = compat_urllib_request.Request('%s%s' % (self._API_BASE, endpoint)) + req = sanitized_Request('%s%s' % (self._API_BASE, endpoint)) req.add_header('Referer', self._API_BASE) return self._download_json(req, video_id, note) class NetEaseMusicIE(NetEaseMusicBaseIE): IE_NAME = 'netease:song' + IE_DESC = '网易云音乐' _VALID_URL = r'https?://music\.163\.com/(#/)?song\?id=(?P[0-9]+)' _TESTS = [{ 'url': 'http://music.163.com/#/song?id=32102397', @@ -75,6 +89,7 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'timestamp': 1431878400, 'description': 'md5:a10a54589c2860300d02e1de821eb2ef', }, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'No lyrics translation.', 'url': 'http://music.163.com/#/song?id=29822014', @@ -87,6 +102,7 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'timestamp': 1419523200, 'description': 'md5:a4d8d89f44656af206b7b2555c0bce6c', }, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'No lyrics.', 'url': 'http://music.163.com/song?id=17241424', @@ -98,6 +114,21 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'upload_date': '20080211', 'timestamp': 1202745600, }, + 'skip': 'Blocked outside Mainland China', + }, { + 'note': 'Has translated name.', + 'url': 'http://music.163.com/#/song?id=22735043', + 'info_dict': { + 'id': '22735043', + 'ext': 'mp3', + 'title': '소원을 말해봐 (Genie)', + 'creator': '少女时代', + 'description': 'md5:79d99cc560e4ca97e0c4d86800ee4184', + 'upload_date': '20100127', + 'timestamp': 1264608000, + 'alt_title': '说出愿望吧(Genie)', + }, + 'skip': 'Blocked outside Mainland China', }] def _process_lyrics(self, lyrics_info): @@ -109,10 +140,9 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): lyrics_expr = r'(\[[0-9]{2}:[0-9]{2}\.[0-9]{2,}\])([^\n]+)' original_ts_texts = re.findall(lyrics_expr, original) - translation_ts_dict = { - time_stamp: text for time_stamp, text in re.findall(lyrics_expr, translated) - } - + translation_ts_dict = dict( + (time_stamp, text) for time_stamp, text in re.findall(lyrics_expr, translated) + ) lyrics = '\n'.join([ '%s%s / %s' % (time_stamp, text, translation_ts_dict.get(time_stamp, '')) for time_stamp, text in original_ts_texts @@ -127,20 +157,20 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'ids': '[%s]' % song_id } info = self.query_api( - 'song/detail?' + compat_urllib_parse.urlencode(params), + 'song/detail?' + compat_urllib_parse_urlencode(params), song_id, 'Downloading song info')['songs'][0] formats = self.extract_formats(info) self._sort_formats(formats) lyrics_info = self.query_api( - 'song/lyric?id=%s&lv=-1&tv=-1' % song_id, + 'song/lyric?id=%s&lv=-1&tv=-1' % song_id, song_id, 'Downloading lyrics data') lyrics = self._process_lyrics(lyrics_info) alt_title = None - if info.get('alias'): - alt_title = '/'.join(info.get('alias')) + if info.get('transNames'): + alt_title = '/'.join(info.get('transNames')) return { 'id': song_id, @@ -157,6 +187,7 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): class NetEaseMusicAlbumIE(NetEaseMusicBaseIE): IE_NAME = 'netease:album' + IE_DESC = '网易云音乐 - 专辑' _VALID_URL = r'https?://music\.163\.com/(#/)?album\?id=(?P[0-9]+)' _TEST = { 'url': 'http://music.163.com/#/album?id=220780', @@ -165,13 +196,14 @@ class NetEaseMusicAlbumIE(NetEaseMusicBaseIE): 'title': 'B\'day', }, 'playlist_count': 23, + 'skip': 'Blocked outside Mainland China', } def _real_extract(self, url): album_id = self._match_id(url) info = self.query_api( - 'album/%s?id=%s' % (album_id, album_id), + 'album/%s?id=%s' % (album_id, album_id), album_id, 'Downloading album data')['album'] name = info['name'] @@ -186,6 +218,7 @@ class NetEaseMusicAlbumIE(NetEaseMusicBaseIE): class NetEaseMusicSingerIE(NetEaseMusicBaseIE): IE_NAME = 'netease:singer' + IE_DESC = '网易云音乐 - 歌手' _VALID_URL = r'https?://music\.163\.com/(#/)?artist\?id=(?P[0-9]+)' _TESTS = [{ 'note': 'Singer has aliases.', @@ -195,6 +228,7 @@ class NetEaseMusicSingerIE(NetEaseMusicBaseIE): 'title': '张惠妹 - aMEI;阿密特', }, 'playlist_count': 50, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'Singer has translated name.', 'url': 'http://music.163.com/#/artist?id=124098', @@ -203,20 +237,21 @@ class NetEaseMusicSingerIE(NetEaseMusicBaseIE): 'title': '李昇基 - 이승기', }, 'playlist_count': 50, + 'skip': 'Blocked outside Mainland China', }] def _real_extract(self, url): singer_id = self._match_id(url) info = self.query_api( - 'artist/%s?id=%s' % (singer_id, singer_id), + 'artist/%s?id=%s' % (singer_id, singer_id), singer_id, 'Downloading singer data') name = info['artist']['name'] if info['artist']['trans']: name = '%s - %s' % (name, info['artist']['trans']) if info['artist']['alias']: - name = '%s - %s' % (name, ";".join(info['artist']['alias'])) + name = '%s - %s' % (name, ';'.join(info['artist']['alias'])) entries = [ self.url_result('http://music.163.com/#/song?id=%s' % song['id'], @@ -228,6 +263,7 @@ class NetEaseMusicSingerIE(NetEaseMusicBaseIE): class NetEaseMusicListIE(NetEaseMusicBaseIE): IE_NAME = 'netease:playlist' + IE_DESC = '网易云音乐 - 歌单' _VALID_URL = r'https?://music\.163\.com/(#/)?(playlist|discover/toplist)\?id=(?P[0-9]+)' _TESTS = [{ 'url': 'http://music.163.com/#/playlist?id=79177352', @@ -237,6 +273,7 @@ class NetEaseMusicListIE(NetEaseMusicBaseIE): 'description': 'md5:12fd0819cab2965b9583ace0f8b7b022' }, 'playlist_count': 99, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'Toplist/Charts sample', 'url': 'http://music.163.com/#/discover/toplist?id=3733003', @@ -246,13 +283,14 @@ class NetEaseMusicListIE(NetEaseMusicBaseIE): 'description': 'md5:73ec782a612711cadc7872d9c1e134fc', }, 'playlist_count': 50, + 'skip': 'Blocked outside Mainland China', }] def _real_extract(self, url): list_id = self._match_id(url) info = self.query_api( - 'playlist/detail?id=%s&lv=-1&tv=-1' % list_id, + 'playlist/detail?id=%s&lv=-1&tv=-1' % list_id, list_id, 'Downloading playlist data')['result'] name = info['name'] @@ -273,6 +311,7 @@ class NetEaseMusicListIE(NetEaseMusicBaseIE): class NetEaseMusicMvIE(NetEaseMusicBaseIE): IE_NAME = 'netease:mv' + IE_DESC = '网易云音乐 - MV' _VALID_URL = r'https?://music\.163\.com/(#/)?mv\?id=(?P[0-9]+)' _TEST = { 'url': 'http://music.163.com/#/mv?id=415350', @@ -284,6 +323,7 @@ class NetEaseMusicMvIE(NetEaseMusicBaseIE): 'creator': '白雅言', 'upload_date': '20150520', }, + 'skip': 'Blocked outside Mainland China', } def _real_extract(self, url): @@ -294,7 +334,7 @@ class NetEaseMusicMvIE(NetEaseMusicBaseIE): mv_id, 'Downloading mv info')['data'] formats = [ - {'url': mv_url, 'ext': 'mp4', 'format_id': '%sp' % brs, 'preference': int(brs)} + {'url': mv_url, 'ext': 'mp4', 'format_id': '%sp' % brs, 'height': int(brs)} for brs, mv_url in info['brs'].items() ] self._sort_formats(formats) @@ -313,6 +353,7 @@ class NetEaseMusicMvIE(NetEaseMusicBaseIE): class NetEaseMusicProgramIE(NetEaseMusicBaseIE): IE_NAME = 'netease:program' + IE_DESC = '网易云音乐 - 电台节目' _VALID_URL = r'https?://music\.163\.com/(#/?)program\?id=(?P[0-9]+)' _TESTS = [{ 'url': 'http://music.163.com/#/program?id=10109055', @@ -326,6 +367,7 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): 'upload_date': '20150613', 'duration': 900, }, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'This program has accompanying songs.', 'url': 'http://music.163.com/#/program?id=10141022', @@ -335,6 +377,7 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): 'description': 'md5:8d594db46cc3e6509107ede70a4aaa3b', }, 'playlist_count': 4, + 'skip': 'Blocked outside Mainland China', }, { 'note': 'This program has accompanying songs.', 'url': 'http://music.163.com/#/program?id=10141022', @@ -348,7 +391,8 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): }, 'params': { 'noplaylist': True - } + }, + 'skip': 'Blocked outside Mainland China', }] def _real_extract(self, url): @@ -366,7 +410,7 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): self.to_screen( 'Downloading just the main audio %s because of --no-playlist' % info['mainSong']['id']) - + formats = self.extract_formats(info['mainSong']) self._sort_formats(formats) @@ -397,6 +441,7 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE): IE_NAME = 'netease:djradio' + IE_DESC = '网易云音乐 - 电台' _VALID_URL = r'https?://music\.163\.com/(#/)?djradio\?id=(?P[0-9]+)' _TEST = { 'url': 'http://music.163.com/#/djradio?id=42', @@ -406,6 +451,7 @@ class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE): 'description': 'md5:766220985cbd16fdd552f64c578a6b15' }, 'playlist_mincount': 40, + 'skip': 'Blocked outside Mainland China', } _PAGE_SIZE = 1000 @@ -415,9 +461,9 @@ class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE): name = None desc = None entries = [] - for offset in itertools.count(start=0, step=self._PAGE_SIZE): + for offset in compat_itertools_count(start=0, step=self._PAGE_SIZE): info = self.query_api( - 'dj/program/byradio?asc=false&limit=%d&radioId=%s&offset=%d' + 'dj/program/byradio?asc=false&limit=%d&radioId=%s&offset=%d' % (self._PAGE_SIZE, dj_id, offset), dj_id, 'Downloading dj programs - %d' % offset)