Raphaël G. Git Repositories - youtubedl/blob - youtube_dl/extractor/svtplay.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     determine_ext,
   9 )
  10
  11
  12 class SVTPlayIE(InfoExtractor):
  13     IE_DESC = 'SVT Play and Öppet arkiv'
  14     _VALID_URL = r'https?://(?:www\.)?(?P<host>svtplay|oppetarkiv)\.se/video/(?P<id>[0-9]+)'
  15     _TESTS = [{
  16         'url': 'http://www.svtplay.se/video/2609989/sm-veckan/sm-veckan-rally-final-sasong-1-sm-veckan-rally-final',
  17         'md5': 'ade3def0643fa1c40587a422f98edfd9',
  18         'info_dict': {
  19             'id': '2609989',
  20             'ext': 'flv',
  21             'title': 'SM veckan vinter, Örebro - Rally, final',
  22             'duration': 4500,
  23             'thumbnail': 're:^https?://.*[\.-]jpg$',
  24             'age_limit': 0,
  25         },
  26     }, {
  27         'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318',
  28         'md5': 'c3101a17ce9634f4c1f9800f0746c187',
  29         'info_dict': {
  30             'id': '1058509',
  31             'ext': 'flv',
  32             'title': 'Farlig kryssning',
  33             'duration': 2566,
  34             'thumbnail': 're:^https?://.*[\.-]jpg$',
  35             'age_limit': 0,
  36         },
  37         'skip': 'Only works from Sweden',
  38     }]
  39
  40     def _real_extract(self, url):
  41         mobj = re.match(self._VALID_URL, url)
  42         video_id = mobj.group('id')
  43         host = mobj.group('host')
  44
  45         info = self._download_json(
  46             'http://www.%s.se/video/%s?output=json' % (host, video_id), video_id)
  47
  48         title = info['context']['title']
  49         thumbnail = info['context'].get('thumbnailImage')
  50
  51         video_info = info['video']
  52         formats = []
  53         for vr in video_info['videoReferences']:
  54             vurl = vr['url']
  55             ext = determine_ext(vurl)
  56             if ext == 'm3u8':
  57                 formats.extend(self._extract_m3u8_formats(
  58                     vurl, video_id,
  59                     ext='mp4', entry_protocol='m3u8_native',
  60                     m3u8_id=vr.get('playerType')))
  61             elif ext == 'f4m':
  62                 formats.extend(self._extract_f4m_formats(
  63                     vurl + '?hdcore=3.3.0', video_id,
  64                     f4m_id=vr.get('playerType')))
  65             else:
  66                 formats.append({
  67                     'format_id': vr.get('playerType'),
  68                     'url': vurl,
  69                 })
  70         self._sort_formats(formats)
  71
  72         duration = video_info.get('materialLength')
  73         age_limit = 18 if video_info.get('inappropriateForChildren') else 0
  74
  75         return {
  76             'id': video_id,
  77             'title': title,
  78             'formats': formats,
  79             'thumbnail': thumbnail,
  80             'duration': duration,
  81             'age_limit': age_limit,
  82         }