youtube_dl/extractor/eagleplatform.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 compat_HTTPError
   8 from ..utils import (
   9     ExtractorError,
  10     int_or_none,
  11 )
  12
  13
  14 class EaglePlatformIE(InfoExtractor):
  15     _VALID_URL = r'''(?x)
  16                     (?:
  17                         eagleplatform:(?P<custom_host>[^/]+):|
  18                         https?://(?P<host>.+?\.media\.eagleplatform\.com)/index/player\?.*\brecord_id=
  19                     )
  20                     (?P<id>\d+)
  21                 '''
  22     _TESTS = [{
  23         # http://lenta.ru/news/2015/03/06/navalny/
  24         'url': 'http://lentaru.media.eagleplatform.com/index/player?player=new&record_id=227304&player_template_id=5201',
  25         'md5': '70f5187fb620f2c1d503b3b22fd4efe3',
  26         'info_dict': {
  27             'id': '227304',
  28             'ext': 'mp4',
  29             'title': 'Навальный вышел на свободу',
  30             'description': 'md5:d97861ac9ae77377f3f20eaf9d04b4f5',
  31             'thumbnail': 're:^https?://.*\.jpg$',
  32             'duration': 87,
  33             'view_count': int,
  34             'age_limit': 0,
  35         },
  36     }, {
  37         # http://muz-tv.ru/play/7129/
  38         # http://media.clipyou.ru/index/player?record_id=12820&width=730&height=415&autoplay=true
  39         'url': 'eagleplatform:media.clipyou.ru:12820',
  40         'md5': '90b26344ba442c8e44aa4cf8f301164a',
  41         'info_dict': {
  42             'id': '12820',
  43             'ext': 'mp4',
  44             'title': "'O Sole Mio",
  45             'thumbnail': 're:^https?://.*\.jpg$',
  46             'duration': 216,
  47             'view_count': int,
  48         },
  49         'skip': 'Georestricted',
  50     }]
  51
  52     @staticmethod
  53     def _handle_error(response):
  54         status = int_or_none(response.get('status', 200))
  55         if status != 200:
  56             raise ExtractorError(' '.join(response['errors']), expected=True)
  57
  58     def _download_json(self, url_or_request, video_id, note='Downloading JSON metadata'):
  59         try:
  60             response = super(EaglePlatformIE, self)._download_json(url_or_request, video_id, note)
  61         except ExtractorError as ee:
  62             if isinstance(ee.cause, compat_HTTPError):
  63                 response = self._parse_json(ee.cause.read().decode('utf-8'), video_id)
  64                 self._handle_error(response)
  65             raise
  66         return response
  67
  68     def _get_video_url(self, url_or_request, video_id, note='Downloading JSON metadata'):
  69         return self._download_json(url_or_request, video_id, note)['data'][0]
  70
  71     def _real_extract(self, url):
  72         mobj = re.match(self._VALID_URL, url)
  73         host, video_id = mobj.group('custom_host') or mobj.group('host'), mobj.group('id')
  74
  75         player_data = self._download_json(
  76             'http://%s/api/player_data?id=%s' % (host, video_id), video_id)
  77
  78         media = player_data['data']['playlist']['viewports'][0]['medialist'][0]
  79
  80         title = media['title']
  81         description = media.get('description')
  82         thumbnail = self._proto_relative_url(media.get('snapshot'), 'http:')
  83         duration = int_or_none(media.get('duration'))
  84         view_count = int_or_none(media.get('views'))
  85
  86         age_restriction = media.get('age_restriction')
  87         age_limit = None
  88         if age_restriction:
  89             age_limit = 0 if age_restriction == 'allow_all' else 18
  90
  91         secure_m3u8 = self._proto_relative_url(media['sources']['secure_m3u8']['auto'], 'http:')
  92
  93         m3u8_url = self._get_video_url(secure_m3u8, video_id, 'Downloading m3u8 JSON')
  94         formats = self._extract_m3u8_formats(
  95             m3u8_url, video_id,
  96             'mp4', entry_protocol='m3u8_native', m3u8_id='hls')
  97
  98         mp4_url = self._get_video_url(
  99             # Secure mp4 URL is constructed according to Player.prototype.mp4 from
 100             # http://lentaru.media.eagleplatform.com/player/player.js
 101             re.sub(r'm3u8|hlsvod|hls|f4m', 'mp4', secure_m3u8),
 102             video_id, 'Downloading mp4 JSON')
 103         formats.append({'url': mp4_url, 'format_id': 'mp4'})
 104
 105         self._sort_formats(formats)
 106
 107         return {
 108             'id': video_id,
 109             'title': title,
 110             'description': description,
 111             'thumbnail': thumbnail,
 112             'duration': duration,
 113             'view_count': view_count,
 114             'age_limit': age_limit,
 115             'formats': formats,
 116         }