diff options
Diffstat (limited to 'hypervideo_dl/extractor/neteasemusic.py')
| -rw-r--r-- | hypervideo_dl/extractor/neteasemusic.py | 176 |
1 files changed, 124 insertions, 52 deletions
diff --git a/hypervideo_dl/extractor/neteasemusic.py b/hypervideo_dl/extractor/neteasemusic.py index 57b4774..5957098 100644 --- a/hypervideo_dl/extractor/neteasemusic.py +++ b/hypervideo_dl/extractor/neteasemusic.py @@ -1,20 +1,25 @@ -# coding: utf-8 -from __future__ import unicode_literals - -from hashlib import md5 +import itertools +import json +import re +import time from base64 import b64encode +from binascii import hexlify from datetime import datetime -import re +from hashlib import md5 +from random import randint from .common import InfoExtractor -from ..compat import ( - compat_urllib_parse_urlencode, - compat_str, - compat_itertools_count, -) +from ..aes import aes_ecb_encrypt, pkcs7_padding +from ..compat import compat_urllib_parse_urlencode from ..utils import ( - sanitized_Request, + ExtractorError, + bytes_to_intlist, + error_to_compat_str, float_or_none, + int_or_none, + intlist_to_bytes, + sanitized_Request, + try_get, ) @@ -26,7 +31,7 @@ class NetEaseMusicBaseIE(InfoExtractor): @classmethod def _encrypt(cls, dfsid): salt_bytes = bytearray(cls._NETEASE_SALT.encode('utf-8')) - string_bytes = bytearray(compat_str(dfsid).encode('ascii')) + string_bytes = bytearray(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] @@ -35,32 +40,105 @@ class NetEaseMusicBaseIE(InfoExtractor): result = b64encode(m.digest()).decode('ascii') return result.replace('/', '_').replace('+', '-') + def make_player_api_request_data_and_headers(self, song_id, bitrate): + KEY = b'e82ckenh8dichen8' + URL = '/api/song/enhance/player/url' + now = int(time.time() * 1000) + rand = randint(0, 1000) + cookie = { + 'osver': None, + 'deviceId': None, + 'appver': '8.0.0', + 'versioncode': '140', + 'mobilename': None, + 'buildver': '1623435496', + 'resolution': '1920x1080', + '__csrf': '', + 'os': 'pc', + 'channel': None, + 'requestId': '{0}_{1:04}'.format(now, rand), + } + request_text = json.dumps( + {'ids': '[{0}]'.format(song_id), 'br': bitrate, 'header': cookie}, + separators=(',', ':')) + message = 'nobody{0}use{1}md5forencrypt'.format( + URL, request_text).encode('latin1') + msg_digest = md5(message).hexdigest() + + data = '{0}-36cd479b6b5-{1}-36cd479b6b5-{2}'.format( + URL, request_text, msg_digest) + data = pkcs7_padding(bytes_to_intlist(data)) + encrypted = intlist_to_bytes(aes_ecb_encrypt(data, bytes_to_intlist(KEY))) + encrypted_params = hexlify(encrypted).decode('ascii').upper() + + cookie = '; '.join( + ['{0}={1}'.format(k, v if v is not None else 'undefined') + for [k, v] in cookie.items()]) + + headers = { + 'User-Agent': self.extractor.get_param('http_headers')['User-Agent'], + 'Content-Type': 'application/x-www-form-urlencoded', + 'Referer': 'https://music.163.com', + 'Cookie': cookie, + } + return ('params={0}'.format(encrypted_params), headers) + + def _call_player_api(self, song_id, bitrate): + url = 'https://interface3.music.163.com/eapi/song/enhance/player/url' + data, headers = self.make_player_api_request_data_and_headers(song_id, bitrate) + try: + msg = 'empty result' + result = self._download_json( + url, song_id, data=data.encode('ascii'), headers=headers) + if result: + return result + except ExtractorError as e: + if type(e.cause) in (ValueError, TypeError): + # JSON load failure + raise + except Exception as e: + msg = error_to_compat_str(e) + self.report_warning('%s API call (%s) failed: %s' % ( + song_id, bitrate, msg)) + return {} + def extract_formats(self, info): + err = 0 formats = [] + song_id = info['id'] for song_format in self._FORMATS: details = info.get(song_format) if not details: continue - 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 + + bitrate = int_or_none(details.get('bitrate')) or 999000 + data = self._call_player_api(song_id, bitrate) + for song in try_get(data, lambda x: x['data'], list) or []: + song_url = try_get(song, lambda x: x['url']) + if not song_url: + continue 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), + 'abr': float_or_none(song.get('br'), scale=1000), 'format_id': song_format, - 'filesize': details.get('size'), - 'asr': details.get('sr') + 'filesize': int_or_none(song.get('size')), + 'asr': int_or_none(details.get('sr')), }) - break + elif err == 0: + err = try_get(song, lambda x: x['code'], int) + + if not formats: + msg = 'No media links found' + if err != 0 and (err < 200 or err >= 400): + raise ExtractorError( + '%s (site code %d)' % (msg, err, ), expected=True) + else: + self.raise_geo_restricted( + msg + ': probably this video is not available from your location due to geo restriction.', + countries=['CN']) + return formats @classmethod @@ -76,33 +154,19 @@ class NetEaseMusicBaseIE(InfoExtractor): class NetEaseMusicIE(NetEaseMusicBaseIE): IE_NAME = 'netease:song' IE_DESC = '网易云音乐' - _VALID_URL = r'https?://music\.163\.com/(#/)?song\?id=(?P<id>[0-9]+)' + _VALID_URL = r'https?://(y\.)?music\.163\.com/(?:[#m]/)?song\?.*?\bid=(?P<id>[0-9]+)' _TESTS = [{ 'url': 'http://music.163.com/#/song?id=32102397', - 'md5': 'f2e97280e6345c74ba9d5677dd5dcb45', + 'md5': '3e909614ce09b1ccef4a3eb205441190', 'info_dict': { 'id': '32102397', 'ext': 'mp3', - 'title': 'Bad Blood (feat. Kendrick Lamar)', + 'title': 'Bad Blood', 'creator': 'Taylor Swift / Kendrick Lamar', - 'upload_date': '20150517', - 'timestamp': 1431878400, - 'description': 'md5:a10a54589c2860300d02e1de821eb2ef', + 'upload_date': '20150516', + 'timestamp': 1431792000, + 'description': 'md5:25fc5f27e47aad975aa6d36382c7833c', }, - 'skip': 'Blocked outside Mainland China', - }, { - 'note': 'No lyrics translation.', - 'url': 'http://music.163.com/#/song?id=29822014', - 'info_dict': { - 'id': '29822014', - 'ext': 'mp3', - 'title': '听见下雨的声音', - 'creator': '周杰伦', - 'upload_date': '20141225', - 'timestamp': 1419523200, - 'description': 'md5:a4d8d89f44656af206b7b2555c0bce6c', - }, - 'skip': 'Blocked outside Mainland China', }, { 'note': 'No lyrics.', 'url': 'http://music.163.com/song?id=17241424', @@ -112,9 +176,9 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'title': 'Opus 28', 'creator': 'Dustin O\'Halloran', 'upload_date': '20080211', + 'description': 'md5:f12945b0f6e0365e3b73c5032e1b0ff4', 'timestamp': 1202745600, }, - 'skip': 'Blocked outside Mainland China', }, { 'note': 'Has translated name.', 'url': 'http://music.163.com/#/song?id=22735043', @@ -128,7 +192,18 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 'timestamp': 1264608000, 'alt_title': '说出愿望吧(Genie)', }, - 'skip': 'Blocked outside Mainland China', + }, { + 'url': 'https://y.music.163.com/m/song?app_version=8.8.45&id=95670&uct2=sKnvS4+0YStsWkqsPhFijw%3D%3D&dlt=0846', + 'md5': '95826c73ea50b1c288b22180ec9e754d', + 'info_dict': { + 'id': '95670', + 'ext': 'mp3', + 'title': '国际歌', + 'creator': '马备', + 'upload_date': '19911130', + 'timestamp': 691516800, + 'description': 'md5:1ba2f911a2b0aa398479f595224f2141', + }, }] def _process_lyrics(self, lyrics_info): @@ -161,7 +236,6 @@ class NetEaseMusicIE(NetEaseMusicBaseIE): 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, @@ -337,7 +411,6 @@ class NetEaseMusicMvIE(NetEaseMusicBaseIE): {'url': mv_url, 'ext': 'mp4', 'format_id': '%sp' % brs, 'height': int(brs)} for brs, mv_url in info['brs'].items() ] - self._sort_formats(formats) return { 'id': mv_id, @@ -407,7 +480,6 @@ class NetEaseMusicProgramIE(NetEaseMusicBaseIE): if not self._yes_playlist(info['songs'] and program_id, info['mainSong']['id']): formats = self.extract_formats(info['mainSong']) - self._sort_formats(formats) return { 'id': info['mainSong']['id'], @@ -452,7 +524,7 @@ class NetEaseMusicDjRadioIE(NetEaseMusicBaseIE): name = None desc = None entries = [] - for offset in compat_itertools_count(start=0, step=self._PAGE_SIZE): + for offset in 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), |