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 ..utils import (
   8     ExtractorError,
   9     int_or_none,
  10 )
  11
  12
  13 class EaglePlatformIE(InfoExtractor):
  14     _VALID_URL = r'''(?x)
  15                     (?:
  16                         eagleplatform:(?P<custom_host>[^/]+):|
  17                         https?://(?P<host>.+?\.media\.eagleplatform\.com)/index/player\?.*\brecord_id=
  18                     )
  19                     (?P<id>\d+)
  20                 '''
  21     _TESTS = [{
  22         # http://lenta.ru/news/2015/03/06/navalny/
  23         'url': 'http://lentaru.media.eagleplatform.com/index/player?player=new&record_id=227304&player_template_id=5201',
  24         'md5': '0b7994faa2bd5c0f69a3db6db28d078d',
  25         'info_dict': {
  26             'id': '227304',
  27             'ext': 'mp4',
  28             'title': 'Навальный вышел на свободу',
  29             'description': 'md5:d97861ac9ae77377f3f20eaf9d04b4f5',
  30             'thumbnail': 're:^https?://.*\.jpg$',
  31             'duration': 87,
  32             'view_count': int,
  33             'age_limit': 0,
  34         },
  35     }, {
  36         # http://muz-tv.ru/play/7129/
  37         # http://media.clipyou.ru/index/player?record_id=12820&width=730&height=415&autoplay=true
  38         'url': 'eagleplatform:media.clipyou.ru:12820',
  39         'md5': '6c2ebeab03b739597ce8d86339d5a905',
  40         'info_dict': {
  41             'id': '12820',
  42             'ext': 'mp4',
  43             'title': "'O Sole Mio",
  44             'thumbnail': 're:^https?://.*\.jpg$',
  45             'duration': 216,
  46             'view_count': int,
  47         },
  48     }]
  49
  50     def _handle_error(self, response):
  51         status = int_or_none(response.get('status', 200))
  52         if status != 200:
  53             raise ExtractorError(' '.join(response['errors']), expected=True)
  54
  55     def _download_json(self, url_or_request, video_id, note='Downloading JSON metadata'):
  56         response = super(EaglePlatformIE, self)._download_json(url_or_request, video_id, note)
  57         self._handle_error(response)
  58         return response
  59
  60     def _real_extract(self, url):
  61         mobj = re.match(self._VALID_URL, url)
  62         host, video_id = mobj.group('custom_host') or mobj.group('host'), mobj.group('id')
  63
  64         player_data = self._download_json(
  65             'http://%s/api/player_data?id=%s' % (host, video_id), video_id)
  66
  67         media = player_data['data']['playlist']['viewports'][0]['medialist'][0]
  68
  69         title = media['title']
  70         description = media.get('description')
  71         thumbnail = media.get('snapshot')
  72         duration = int_or_none(media.get('duration'))
  73         view_count = int_or_none(media.get('views'))
  74
  75         age_restriction = media.get('age_restriction')
  76         age_limit = None
  77         if age_restriction:
  78             age_limit = 0 if age_restriction == 'allow_all' else 18
  79
  80         m3u8_data = self._download_json(
  81             media['sources']['secure_m3u8']['auto'],
  82             video_id, 'Downloading m3u8 JSON')
  83
  84         formats = self._extract_m3u8_formats(
  85             m3u8_data['data'][0], video_id,
  86             'mp4', entry_protocol='m3u8_native')
  87         self._sort_formats(formats)
  88
  89         return {
  90             'id': video_id,
  91             'title': title,
  92             'description': description,
  93             'thumbnail': thumbnail,
  94             'duration': duration,
  95             'view_count': view_count,
  96             'age_limit': age_limit,
  97             'formats': formats,
  98         }