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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import itertools
   5 import re
   6
   7 from .common import InfoExtractor
   8 from ..compat import (
   9     compat_urllib_parse,
  10     compat_urllib_request,
  11 )
  12 from ..utils import (
  13     ExtractorError,
  14     parse_iso8601,
  15 )
  16
  17
  18 class TwitchIE(InfoExtractor):
  19     # TODO: One broadcast may be split into multiple videos. The key
  20     # 'broadcast_id' is the same for all parts, and 'broadcast_part'
  21     # starts at 1 and increases. Can we treat all parts as one video?
  22     _VALID_URL = r"""(?x)^(?:http://)?(?:www\.)?twitch\.tv/
  23         (?:
  24             (?P<channelid>[^/]+)|
  25             (?:(?:[^/]+)/v/(?P<vodid>[^/]+))|
  26             (?:(?:[^/]+)/b/(?P<videoid>[^/]+))|
  27             (?:(?:[^/]+)/c/(?P<chapterid>[^/]+))
  28         )
  29         /?(?:\#.*)?$
  30         """
  31     _PAGE_LIMIT = 100
  32     _API_BASE = 'https://api.twitch.tv'
  33     _LOGIN_URL = 'https://secure.twitch.tv/user/login'
  34     _TESTS = [{
  35         'url': 'http://www.twitch.tv/riotgames/b/577357806',
  36         'info_dict': {
  37             'id': 'a577357806',
  38             'title': 'Worlds Semifinals - Star Horn Royal Club vs. OMG',
  39         },
  40         'playlist_mincount': 12,
  41     }, {
  42         'url': 'http://www.twitch.tv/acracingleague/c/5285812',
  43         'info_dict': {
  44             'id': 'c5285812',
  45             'title': 'ACRL Off Season - Sports Cars @ Nordschleife',
  46         },
  47         'playlist_mincount': 3,
  48     }, {
  49         'url': 'http://www.twitch.tv/vanillatv',
  50         'info_dict': {
  51             'id': 'vanillatv',
  52             'title': 'VanillaTV',
  53         },
  54         'playlist_mincount': 412,
  55     }]
  56
  57     def _handle_error(self, response):
  58         if not isinstance(response, dict):
  59             return
  60         error = response.get('error')
  61         if error:
  62             raise ExtractorError(
  63                 '%s returned error: %s - %s' % (self.IE_NAME, error, response.get('message')),
  64                 expected=True)
  65
  66     def _download_json(self, url, video_id, note='Downloading JSON metadata'):
  67         response = super(TwitchIE, self)._download_json(url, video_id, note)
  68         self._handle_error(response)
  69         return response
  70
  71     def _extract_media(self, item, item_id):
  72         ITEMS = {
  73             'a': 'video',
  74             'v': 'vod',
  75             'c': 'chapter',
  76         }
  77         info = self._extract_info(self._download_json(
  78             '%s/kraken/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
  79             'Downloading %s info JSON' % ITEMS[item]))
  80
  81         if item == 'v':
  82             access_token = self._download_json(
  83                 '%s/api/vods/%s/access_token' % (self._API_BASE, item_id), item_id,
  84                 'Downloading %s access token' % ITEMS[item])
  85             formats = self._extract_m3u8_formats(
  86                 'http://usher.twitch.tv/vod/%s?nauth=%s&nauthsig=%s'
  87                 % (item_id, access_token['token'], access_token['sig']),
  88                 item_id, 'mp4')
  89             info['formats'] = formats
  90             return info
  91
  92         response = self._download_json(
  93             '%s/api/videos/%s%s' % (self._API_BASE, item, item_id), item_id,
  94             'Downloading %s playlist JSON' % ITEMS[item])
  95         entries = []
  96         chunks = response['chunks']
  97         qualities = list(chunks.keys())
  98         for num, fragment in enumerate(zip(*chunks.values()), start=1):
  99             formats = []
 100             for fmt_num, fragment_fmt in enumerate(fragment):
 101                 format_id = qualities[fmt_num]
 102                 fmt = {
 103                     'url': fragment_fmt['url'],
 104                     'format_id': format_id,
 105                     'quality': 1 if format_id == 'live' else 0,
 106                 }
 107                 m = re.search(r'^(?P<height>\d+)[Pp]', format_id)
 108                 if m:
 109                     fmt['height'] = int(m.group('height'))
 110                 formats.append(fmt)
 111             self._sort_formats(formats)
 112             entry = dict(info)
 113             entry['id'] = '%s_%d' % (entry['id'], num)
 114             entry['title'] = '%s part %d' % (entry['title'], num)
 115             entry['formats'] = formats
 116             entries.append(entry)
 117         return self.playlist_result(entries, info['id'], info['title'])
 118
 119     def _extract_info(self, info):
 120         return {
 121             'id': info['_id'],
 122             'title': info['title'],
 123             'description': info['description'],
 124             'duration': info['length'],
 125             'thumbnail': info['preview'],
 126             'uploader': info['channel']['display_name'],
 127             'uploader_id': info['channel']['name'],
 128             'timestamp': parse_iso8601(info['recorded_at']),
 129             'view_count': info['views'],
 130         }
 131
 132     def _real_initialize(self):
 133         self._login()
 134
 135     def _login(self):
 136         (username, password) = self._get_login_info()
 137         if username is None:
 138             return
 139
 140         login_page = self._download_webpage(
 141             self._LOGIN_URL, None, 'Downloading login page')
 142
 143         authenticity_token = self._search_regex(
 144             r'<input name="authenticity_token" type="hidden" value="([^"]+)"',
 145             login_page, 'authenticity token')
 146
 147         login_form = {
 148             'utf8': '✓'.encode('utf-8'),
 149             'authenticity_token': authenticity_token,
 150             'redirect_on_login': '',
 151             'embed_form': 'false',
 152             'mp_source_action': '',
 153             'follow': '',
 154             'user[login]': username,
 155             'user[password]': password,
 156         }
 157
 158         request = compat_urllib_request.Request(
 159             self._LOGIN_URL, compat_urllib_parse.urlencode(login_form).encode('utf-8'))
 160         request.add_header('Referer', self._LOGIN_URL)
 161         response = self._download_webpage(
 162             request, None, 'Logging in as %s' % username)
 163
 164         m = re.search(
 165             r"id=([\"'])login_error_message\1[^>]*>(?P<msg>[^<]+)", response)
 166         if m:
 167             raise ExtractorError(
 168                 'Unable to login: %s' % m.group('msg').strip(), expected=True)
 169
 170     def _real_extract(self, url):
 171         mobj = re.match(self._VALID_URL, url)
 172         if mobj.group('chapterid'):
 173             return self._extract_media('c', mobj.group('chapterid'))
 174
 175             """
 176             webpage = self._download_webpage(url, chapter_id)
 177             m = re.search(r'PP\.archive_id = "([0-9]+)";', webpage)
 178             if not m:
 179                 raise ExtractorError('Cannot find archive of a chapter')
 180             archive_id = m.group(1)
 181
 182             api = api_base + '/broadcast/by_chapter/%s.xml' % chapter_id
 183             doc = self._download_xml(
 184                 api, chapter_id,
 185                 note='Downloading chapter information',
 186                 errnote='Chapter information download failed')
 187             for a in doc.findall('.//archive'):
 188                 if archive_id == a.find('./id').text:
 189                     break
 190             else:
 191                 raise ExtractorError('Could not find chapter in chapter information')
 192
 193             video_url = a.find('./video_file_url').text
 194             video_ext = video_url.rpartition('.')[2] or 'flv'
 195
 196             chapter_api_url = 'https://api.twitch.tv/kraken/videos/c' + chapter_id
 197             chapter_info = self._download_json(
 198                 chapter_api_url, 'c' + chapter_id,
 199                 note='Downloading chapter metadata',
 200                 errnote='Download of chapter metadata failed')
 201
 202             bracket_start = int(doc.find('.//bracket_start').text)
 203             bracket_end = int(doc.find('.//bracket_end').text)
 204
 205             # TODO determine start (and probably fix up file)
 206             #  youtube-dl -v http://www.twitch.tv/firmbelief/c/1757457
 207             #video_url += '?start=' + TODO:start_timestamp
 208             # bracket_start is 13290, but we want 51670615
 209             self._downloader.report_warning('Chapter detected, but we can just download the whole file. '
 210                                             'Chapter starts at %s and ends at %s' % (formatSeconds(bracket_start), formatSeconds(bracket_end)))
 211
 212             info = {
 213                 'id': 'c' + chapter_id,
 214                 'url': video_url,
 215                 'ext': video_ext,
 216                 'title': chapter_info['title'],
 217                 'thumbnail': chapter_info['preview'],
 218                 'description': chapter_info['description'],
 219                 'uploader': chapter_info['channel']['display_name'],
 220                 'uploader_id': chapter_info['channel']['name'],
 221             }
 222             return info
 223             """
 224         elif mobj.group('videoid'):
 225             return self._extract_media('a', mobj.group('videoid'))
 226         elif mobj.group('vodid'):
 227             return self._extract_media('v', mobj.group('vodid'))
 228         elif mobj.group('channelid'):
 229             channel_id = mobj.group('channelid')
 230             info = self._download_json(
 231                 '%s/kraken/channels/%s' % (self._API_BASE, channel_id),
 232                 channel_id, 'Downloading channel info JSON')
 233             channel_name = info.get('display_name') or info.get('name')
 234             entries = []
 235             offset = 0
 236             limit = self._PAGE_LIMIT
 237             for counter in itertools.count(1):
 238                 response = self._download_json(
 239                     '%s/kraken/channels/%s/videos/?offset=%d&limit=%d'
 240                     % (self._API_BASE, channel_id, offset, limit),
 241                     channel_id, 'Downloading channel videos JSON page %d' % counter)
 242                 videos = response['videos']
 243                 if not videos:
 244                     break
 245                 entries.extend([self.url_result(video['url'], 'Twitch') for video in videos])
 246                 offset += limit
 247             return self.playlist_result(entries, channel_id, channel_name)