Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/laola1tv.py

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from ..compat import (
   8     compat_urllib_parse_urlencode,
   9     compat_urlparse,
  10 )
  11 from ..utils import (
  12     ExtractorError,
  13     sanitized_Request,
  14     unified_strdate,
  15     urlencode_postdata,
  16     xpath_element,
  17     xpath_text,
  18 )
  19
  20
  21 class Laola1TvIE(InfoExtractor):
  22     _VALID_URL = r'https?://(?:www\.)?laola1\.tv/(?P<lang>[a-z]+)-(?P<portal>[a-z]+)/(?P<kind>[^/]+)/(?P<slug>[^/?#&]+)'
  23     _TESTS = [{
  24         'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie/227883.html',
  25         'info_dict': {
  26             'id': '227883',
  27             'display_id': 'straubing-tigers-koelner-haie',
  28             'ext': 'flv',
  29             'title': 'Straubing Tigers - Kölner Haie',
  30             'upload_date': '20140912',
  31             'is_live': False,
  32             'categories': ['Eishockey'],
  33         },
  34         'params': {
  35             'skip_download': True,
  36         },
  37     }, {
  38         'url': 'http://www.laola1.tv/de-de/video/straubing-tigers-koelner-haie',
  39         'info_dict': {
  40             'id': '464602',
  41             'display_id': 'straubing-tigers-koelner-haie',
  42             'ext': 'flv',
  43             'title': 'Straubing Tigers - Kölner Haie',
  44             'upload_date': '20160129',
  45             'is_live': False,
  46             'categories': ['Eishockey'],
  47         },
  48         'params': {
  49             'skip_download': True,
  50         },
  51     }, {
  52         'url': 'http://www.laola1.tv/de-de/livestream/2016-03-22-belogorie-belgorod-trentino-diatec-lde',
  53         'info_dict': {
  54             'id': '487850',
  55             'display_id': '2016-03-22-belogorie-belgorod-trentino-diatec-lde',
  56             'ext': 'flv',
  57             'title': 'Belogorie BELGOROD - TRENTINO Diatec',
  58             'upload_date': '20160322',
  59             'uploader': 'CEV - Europäischer Volleyball Verband',
  60             'is_live': True,
  61             'categories': ['Volleyball'],
  62         },
  63         'params': {
  64             'skip_download': True,
  65         },
  66         'skip': 'This live stream has already finished.',
  67     }]
  68
  69     def _real_extract(self, url):
  70         mobj = re.match(self._VALID_URL, url)
  71         display_id = mobj.group('slug')
  72         kind = mobj.group('kind')
  73         lang = mobj.group('lang')
  74         portal = mobj.group('portal')
  75
  76         webpage = self._download_webpage(url, display_id)
  77
  78         if 'Dieser Livestream ist bereits beendet.' in webpage:
  79             raise ExtractorError('This live stream has already finished.', expected=True)
  80
  81         iframe_url = self._search_regex(
  82             r'<iframe[^>]*?id="videoplayer"[^>]*?src="([^"]+)"',
  83             webpage, 'iframe url')
  84
  85         video_id = self._search_regex(
  86             r'videoid=(\d+)', iframe_url, 'video id')
  87
  88         iframe = self._download_webpage(compat_urlparse.urljoin(
  89             url, iframe_url), display_id, 'Downloading iframe')
  90
  91         partner_id = self._search_regex(
  92             r'partnerid\s*:\s*(["\'])(?P<partner_id>.+?)\1',
  93             iframe, 'partner id', group='partner_id')
  94
  95         hd_doc = self._download_xml(
  96             'http://www.laola1.tv/server/hd_video.php?%s'
  97             % compat_urllib_parse_urlencode({
  98                 'play': video_id,
  99                 'partner': partner_id,
 100                 'portal': portal,
 101                 'lang': lang,
 102                 'v5ident': '',
 103             }), display_id)
 104
 105         _v = lambda x, **k: xpath_text(hd_doc, './/video/' + x, **k)
 106         title = _v('title', fatal=True)
 107
 108         VS_TARGETS = {
 109             'video': '2',
 110             'livestream': '17',
 111         }
 112
 113         req = sanitized_Request(
 114             'https://club.laola1.tv/sp/laola1/api/v3/user/session/premium/player/stream-access?%s' %
 115             compat_urllib_parse_urlencode({
 116                 'videoId': video_id,
 117                 'target': VS_TARGETS.get(kind, '2'),
 118                 'label': _v('label'),
 119                 'area': _v('area'),
 120             }),
 121             urlencode_postdata(
 122                 dict((i, v) for i, v in enumerate(_v('req_liga_abos').split(',')))))
 123
 124         token_url = self._download_json(req, display_id)['data']['stream-access'][0]
 125         token_doc = self._download_xml(token_url, display_id, 'Downloading token')
 126
 127         token_attrib = xpath_element(token_doc, './/token').attrib
 128         token_auth = token_attrib['auth']
 129
 130         if token_auth in ('blocked', 'restricted', 'error'):
 131             raise ExtractorError(
 132                 'Token error: %s' % token_attrib['comment'], expected=True)
 133
 134         formats = self._extract_f4m_formats(
 135             '%s?hdnea=%s&hdcore=3.2.0' % (token_attrib['url'], token_auth),
 136             video_id, f4m_id='hds')
 137         self._sort_formats(formats)
 138
 139         categories_str = _v('meta_sports')
 140         categories = categories_str.split(',') if categories_str else []
 141
 142         return {
 143             'id': video_id,
 144             'display_id': display_id,
 145             'title': title,
 146             'upload_date': unified_strdate(_v('time_date')),
 147             'uploader': _v('meta_organisation'),
 148             'categories': categories,
 149             'is_live': _v('islive') == 'true',
 150             'formats': formats,
 151         }