X-Git-Url: http://git.bitcoin.ninja/index.cgi?a=blobdiff_plain;f=youtube_dl%2Fextractor%2Fneteasemusic.py;h=0d36474fa069b793ff32e90e6de1804de09058ac;hb=ffa2cecf7291a7b9aa40dba84a76c71fba20cc57;hp=e73c77f89b47f5ac3e9b5b460cb173be3247c86a;hpb=15830339eff9076b1c4ca2cb6322fc1ec55233a8;p=youtube-dl diff --git a/youtube_dl/extractor/neteasemusic.py b/youtube_dl/extractor/neteasemusic.py index e73c77f89..0d36474fa 100644 --- a/youtube_dl/extractor/neteasemusic.py +++ b/youtube_dl/extractor/neteasemusic.py @@ -4,13 +4,17 @@ 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, ) @@ -21,33 +25,42 @@ class NetEaseMusicBaseIE(InfoExtractor): @classmethod def _encrypt(cls, dfsid): - salt_bytes = bytearray(cls._NETEASE_SALT, 'utf-8') - 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 range(len(string_bytes)): string_bytes[i] = string_bytes[i] ^ salt_bytes[i % salt_len] m = md5() - m.update(string_bytes) - result = b64encode(m.digest()) + 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.get('extension'), - 'abr': details.get('bitrate', 0) / 1000, - 'format_id': song_format, - 'filesize': details.get('size'), - 'asr': details.get('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 @@ -55,13 +68,14 @@ class NetEaseMusicBaseIE(InfoExtractor): 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', @@ -139,7 +153,7 @@ 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) @@ -169,6 +183,7 @@ 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', @@ -198,6 +213,7 @@ 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.', @@ -228,7 +244,7 @@ class NetEaseMusicSingerIE(NetEaseMusicBaseIE): 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'], @@ -240,6 +256,7 @@ 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', @@ -285,6 +302,7 @@ 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', @@ -325,6 +343,7 @@ 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', @@ -409,6 +428,7 @@ 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', @@ -427,7 +447,7 @@ 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' % (self._PAGE_SIZE, dj_id, offset),