_ Git - youtube-dl/blob - youtube_dl/extractor/bilibili.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import calendar
   5 import datetime
   6 import re
   7
   8 from .common import InfoExtractor
   9 from ..compat import (
  10     compat_etree_fromstring,
  11     compat_str,
  12     compat_parse_qs,
  13     compat_xml_parse_error,
  14 )
  15 from ..utils import (
  16     ExtractorError,
  17     int_or_none,
  18     float_or_none,
  19     xpath_text,
  20 )
  21
  22
  23 class BiliBiliIE(InfoExtractor):
  24     _VALID_URL = r'https?://www\.bilibili\.(?:tv|com)/video/av(?P<id>\d+)'
  25
  26     _TESTS = [{
  27         'url': 'http://www.bilibili.tv/video/av1074402/',
  28         'md5': '9fa226fe2b8a9a4d5a69b4c6a183417e',
  29         'info_dict': {
  30             'id': '1554319',
  31             'ext': 'mp4',
  32             'title': '【金坷垃】金泡沫',
  33             'description': 'md5:ce18c2a2d2193f0df2917d270f2e5923',
  34             'duration': 308.315,
  35             'timestamp': 1398012660,
  36             'upload_date': '20140420',
  37             'thumbnail': 're:^https?://.+\.jpg',
  38             'uploader': '菊子桑',
  39             'uploader_id': '156160',
  40         },
  41     }, {
  42         'url': 'http://www.bilibili.com/video/av1041170/',
  43         'info_dict': {
  44             'id': '1507019',
  45             'ext': 'mp4',
  46             'title': '【BD1080P】刀语【诸神&异域】',
  47             'description': '这是个神奇的故事~每个人不留弹幕不给走哦~切利哦！~',
  48             'timestamp': 1396530060,
  49             'upload_date': '20140403',
  50             'uploader': '枫叶逝去',
  51             'uploader_id': '520116',
  52         },
  53     }, {
  54         'url': 'http://www.bilibili.com/video/av4808130/',
  55         'info_dict': {
  56             'id': '7802182',
  57             'ext': 'mp4',
  58             'title': '【长篇】哆啦A梦443【钉铛】',
  59             'description': '(2016.05.27)来组合客人的脸吧&amp;amp;寻母六千里锭 抱歉，又轮到周日上班现在才到家 封面www.pixiv.net/member_illust.php?mode=medium&amp;amp;illust_id=56912929',
  60             'timestamp': 1464564180,
  61             'upload_date': '20160529',
  62             'uploader': '喜欢拉面',
  63             'uploader_id': '151066',
  64         },
  65     }, {
  66         # Missing upload time
  67         'url': 'http://www.bilibili.com/video/av1867637/',
  68         'info_dict': {
  69             'id': '2880301',
  70             'ext': 'mp4',
  71             'title': '【HDTV】【喜剧】岳父岳母真难当 （2014）【法国票房冠军】',
  72             'description': '一个信奉天主教的法国旧式传统资产阶级家庭中有四个女儿。三个女儿却分别找了阿拉伯、犹太、中国丈夫，老夫老妻唯独期盼剩下未嫁的小女儿能找一个信奉天主教的法国白人，结果没想到小女儿找了一位非裔黑人……【这次应该不会跳帧了】',
  73             'uploader': '黑夜为猫',
  74             'uploader_id': '610729',
  75         },
  76         'params': {
  77             # Just to test metadata extraction
  78             'skip_download': True,
  79         },
  80         'expected_warnings': ['upload time'],
  81     }]
  82
  83     # BiliBili blocks keys from time to time. The current key is extracted from
  84     # the Android client
  85     # TODO: find the sign algorithm used in the flash player
  86     _APP_KEY = '86385cdc024c0f6c'
  87
  88     def _real_extract(self, url):
  89         mobj = re.match(self._VALID_URL, url)
  90         video_id = mobj.group('id')
  91
  92         webpage = self._download_webpage(url, video_id)
  93
  94         params = compat_parse_qs(self._search_regex(
  95             [r'EmbedPlayer\([^)]+,\s*"([^"]+)"\)',
  96              r'<iframe[^>]+src="https://secure\.bilibili\.com/secure,([^"]+)"'],
  97             webpage, 'player parameters'))
  98         cid = params['cid'][0]
  99
 100         info_xml_str = self._download_webpage(
 101             'http://interface.bilibili.com/v_cdn_play',
 102             cid, query={'appkey': self._APP_KEY, 'cid': cid},
 103             note='Downloading video info page')
 104
 105         err_msg = None
 106         durls = None
 107         info_xml = None
 108         try:
 109             info_xml = compat_etree_fromstring(info_xml_str.encode('utf-8'))
 110         except compat_xml_parse_error:
 111             info_json = self._parse_json(info_xml_str, video_id, fatal=False)
 112             err_msg = (info_json or {}).get('error_text')
 113         else:
 114             err_msg = xpath_text(info_xml, './message')
 115
 116         if info_xml is not None:
 117             durls = info_xml.findall('./durl')
 118         if not durls:
 119             if err_msg:
 120                 raise ExtractorError('%s said: %s' % (self.IE_NAME, err_msg), expected=True)
 121             else:
 122                 raise ExtractorError('No videos found!')
 123
 124         entries = []
 125
 126         for durl in durls:
 127             size = xpath_text(durl, ['./filesize', './size'])
 128             formats = [{
 129                 'url': durl.find('./url').text,
 130                 'filesize': int_or_none(size),
 131             }]
 132             for backup_url in durl.findall('./backup_url/url'):
 133                 formats.append({
 134                     'url': backup_url.text,
 135                     # backup URLs have lower priorities
 136                     'preference': -2 if 'hd.mp4' in backup_url.text else -3,
 137                 })
 138
 139             self._sort_formats(formats)
 140
 141             entries.append({
 142                 'id': '%s_part%s' % (cid, xpath_text(durl, './order')),
 143                 'duration': int_or_none(xpath_text(durl, './length'), 1000),
 144                 'formats': formats,
 145             })
 146
 147         title = self._html_search_regex('<h1[^>]+title="([^"]+)">', webpage, 'title')
 148         description = self._html_search_meta('description', webpage)
 149         datetime_str = self._html_search_regex(
 150             r'<time[^>]+datetime="([^"]+)"', webpage, 'upload time', fatal=False)
 151         timestamp = None
 152         if datetime_str:
 153             timestamp = calendar.timegm(datetime.datetime.strptime(datetime_str, '%Y-%m-%dT%H:%M').timetuple())
 154
 155         # TODO 'view_count' requires deobfuscating Javascript
 156         info = {
 157             'id': compat_str(cid),
 158             'title': title,
 159             'description': description,
 160             'timestamp': timestamp,
 161             'thumbnail': self._html_search_meta('thumbnailUrl', webpage),
 162             'duration': float_or_none(xpath_text(info_xml, './timelength'), scale=1000),
 163         }
 164
 165         uploader_mobj = re.search(
 166             r'<a[^>]+href="https?://space\.bilibili\.com/(?P<id>\d+)"[^>]+title="(?P<name>[^"]+)"',
 167             webpage)
 168         if uploader_mobj:
 169             info.update({
 170                 'uploader': uploader_mobj.group('name'),
 171                 'uploader_id': uploader_mobj.group('id'),
 172             })
 173
 174         for entry in entries:
 175             entry.update(info)
 176
 177         if len(entries) == 1:
 178             return entries[0]
 179         else:
 180             for idx, entry in enumerate(entries):
 181                 entry['id'] = '%s_part%d' % (video_id, (idx + 1))
 182
 183             return {
 184                 '_type': 'multi_video',
 185                 'id': video_id,
 186                 'title': title,
 187                 'description': description,
 188                 'entries': entries,
 189             }