youtube_dl/extractor/bilibili.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..utils import (
   8     compat_parse_qs,
   9     ExtractorError,
  10     int_or_none,
  11     unified_strdate,
  12 )
  13
  14
  15 class BiliBiliIE(InfoExtractor):
  16     _VALID_URL = r'http://www\.bilibili\.(?:tv|com)/video/av(?P<id>[0-9]+)/'
  17
  18     _TEST = {
  19         'url': 'http://www.bilibili.tv/video/av1074402/',
  20         'md5': '2c301e4dab317596e837c3e7633e7d86',
  21         'info_dict': {
  22             'id': '1074402',
  23             'ext': 'flv',
  24             'title': '【金坷垃】金泡沫',
  25             'duration': 308,
  26             'upload_date': '20140420',
  27             'thumbnail': 're:^https?://.+\.jpg',
  28         },
  29     }
  30
  31     def _real_extract(self, url):
  32         mobj = re.match(self._VALID_URL, url)
  33         video_id = mobj.group('id')
  34
  35         webpage = self._download_webpage(url, video_id)
  36         video_code = self._search_regex(
  37             r'(?s)<div itemprop="video".*?>(.*?)</div>', webpage, 'video code')
  38
  39         title = self._html_search_meta(
  40             'media:title', video_code, 'title', fatal=True)
  41         duration_str = self._html_search_meta(
  42             'duration', video_code, 'duration')
  43         if duration_str is None:
  44             duration = None
  45         else:
  46             duration_mobj = re.match(
  47                 r'^T(?:(?P<hours>[0-9]+)H)?(?P<minutes>[0-9]+)M(?P<seconds>[0-9]+)S$',
  48                 duration_str)
  49             duration = (
  50                 int_or_none(duration_mobj.group('hours'), default=0) * 3600 +
  51                 int(duration_mobj.group('minutes')) * 60 +
  52                 int(duration_mobj.group('seconds')))
  53         upload_date = unified_strdate(self._html_search_meta(
  54             'uploadDate', video_code, fatal=False))
  55         thumbnail = self._html_search_meta(
  56             'thumbnailUrl', video_code, 'thumbnail', fatal=False)
  57
  58         player_params = compat_parse_qs(self._html_search_regex(
  59             r'<iframe .*?class="player" src="https://secure\.bilibili\.(?:tv|com)/secure,([^"]+)"',
  60             webpage, 'player params'))
  61
  62         if 'cid' in player_params:
  63             cid = player_params['cid'][0]
  64
  65             lq_doc = self._download_xml(
  66                 'http://interface.bilibili.cn/v_cdn_play?cid=%s' % cid,
  67                 video_id,
  68                 note='Downloading LQ video info'
  69             )
  70             lq_durl = lq_doc.find('.//durl')
  71             formats = [{
  72                 'format_id': 'lq',
  73                 'quality': 1,
  74                 'url': lq_durl.find('./url').text,
  75                 'filesize': int_or_none(
  76                     lq_durl.find('./size'), get_attr='text'),
  77             }]
  78
  79             hq_doc = self._download_xml(
  80                 'http://interface.bilibili.cn/playurl?cid=%s' % cid,
  81                 video_id,
  82                 note='Downloading HQ video info',
  83                 fatal=False,
  84             )
  85             if hq_doc is not False:
  86                 hq_durl = hq_doc.find('.//durl')
  87                 formats.append({
  88                     'format_id': 'hq',
  89                     'quality': 2,
  90                     'ext': 'flv',
  91                     'url': hq_durl.find('./url').text,
  92                     'filesize': int_or_none(
  93                         hq_durl.find('./size'), get_attr='text'),
  94                 })
  95         else:
  96             raise ExtractorError('Unsupported player parameters: %r' % (player_params,))
  97
  98         self._sort_formats(formats)
  99         return {
 100             'id': video_id,
 101             'title': title,
 102             'formats': formats,
 103             'duration': duration,
 104             'upload_date': upload_date,
 105             'thumbnail': thumbnail,
 106         }