Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/vk.py

   1 # encoding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5 import json
   6
   7 from .common import InfoExtractor
   8 from ..utils import (
   9     ExtractorError,
  10     compat_urllib_request,
  11     compat_urllib_parse,
  12     compat_str,
  13     unescapeHTML,
  14 )
  15
  16
  17 class VKIE(InfoExtractor):
  18     IE_NAME = 'vk.com'
  19     _VALID_URL = r'https?://vk\.com/(?:video_ext\.php\?.*?\boid=(?P<oid>-?\d+).*?\bid=(?P<id>\d+)|(?:videos.*?\?.*?z=)?video(?P<videoid>.*?)(?:\?|%2F|$))'
  20     _NETRC_MACHINE = 'vk'
  21
  22     _TESTS = [
  23         {
  24             'url': 'http://vk.com/videos-77521?z=video-77521_162222515%2Fclub77521',
  25             'md5': '0deae91935c54e00003c2a00646315f0',
  26             'info_dict': {
  27                 'id': '162222515',
  28                 'ext': 'flv',
  29                 'title': 'ProtivoGunz - Хуёвая песня',
  30                 'uploader': 'Noize MC',
  31                 'duration': 195,
  32             },
  33         },
  34         {
  35             'url': 'http://vk.com/video4643923_163339118',
  36             'md5': 'f79bccb5cd182b1f43502ca5685b2b36',
  37             'info_dict': {
  38                 'id': '163339118',
  39                 'ext': 'mp4',
  40                 'uploader': 'Elya Iskhakova',
  41                 'title': 'Dream Theater - Hollow Years Live at Budokan 720*',
  42                 'duration': 558,
  43             }
  44         },
  45         {
  46             'note': 'Embedded video',
  47             'url': 'http://vk.com/video_ext.php?oid=32194266&id=162925554&hash=7d8c2e0d5e05aeaa&hd=1',
  48             'md5': 'c7ce8f1f87bec05b3de07fdeafe21a0a',
  49             'info_dict': {
  50                 'id': '162925554',
  51                 'ext': 'mp4',
  52                 'uploader': 'Vladimir Gavrin',
  53                 'title': 'Lin Dan',
  54                 'duration': 101,
  55             }
  56         },
  57         {
  58             'url': 'http://vk.com/video-8871596_164049491',
  59             'md5': 'a590bcaf3d543576c9bd162812387666',
  60             'note': 'Only available for registered users',
  61             'info_dict': {
  62                 'id': '164049491',
  63                 'ext': 'mp4',
  64                 'uploader': 'Триллеры',
  65                 'title': '► Бойцовский клуб / Fight Club 1999 [HD 720]\u00a0',
  66                 'duration': 8352,
  67             },
  68             'skip': 'Requires vk account credentials',
  69         },
  70     ]
  71
  72     def _login(self):
  73         (username, password) = self._get_login_info()
  74         if username is None:
  75             return
  76
  77         login_form = {
  78             'act': 'login',
  79             'role': 'al_frame',
  80             'expire': '1',
  81             'email': username,
  82             'pass': password,
  83         }
  84
  85         request = compat_urllib_request.Request('https://login.vk.com/?act=login',
  86             compat_urllib_parse.urlencode(login_form).encode('utf-8'))
  87         login_page = self._download_webpage(request, None, note='Logging in as %s' % username)
  88
  89         if re.search(r'onLoginFailed', login_page):
  90             raise ExtractorError('Unable to login, incorrect username and/or password', expected=True)
  91
  92     def _real_initialize(self):
  93         self._login()
  94
  95     def _real_extract(self, url):
  96         mobj = re.match(self._VALID_URL, url)
  97         video_id = mobj.group('videoid')
  98
  99         if not video_id:
 100             video_id = '%s_%s' % (mobj.group('oid'), mobj.group('id'))
 101
 102         info_url = 'http://vk.com/al_video.php?act=show&al=1&video=%s' % video_id
 103         info_page = self._download_webpage(info_url, video_id)
 104
 105         if re.search(r'<!>Please log in or <', info_page):
 106             raise ExtractorError('This video is only available for registered users, '
 107                 'use --username and --password options to provide account credentials.', expected=True)
 108
 109         m_yt = re.search(r'src="(http://www.youtube.com/.*?)"', info_page)
 110         if m_yt is not None:
 111             self.to_screen('Youtube video detected')
 112             return self.url_result(m_yt.group(1), 'Youtube')
 113         data_json = self._search_regex(r'var vars = ({.*?});', info_page, 'vars')
 114         data = json.loads(data_json)
 115
 116         formats = [{
 117             'format_id': k,
 118             'url': v,
 119             'width': int(k[len('url'):]),
 120         } for k, v in data.items()
 121             if k.startswith('url')]
 122         self._sort_formats(formats)
 123
 124         return {
 125             'id': compat_str(data['vid']),
 126             'formats': formats,
 127             'title': unescapeHTML(data['md_title']),
 128             'thumbnail': data.get('jpg'),
 129             'uploader': data.get('md_author'),
 130             'duration': data.get('duration')
 131         }