_ Git - youtube-dl/blob - youtube_dl/extractor/svt.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 SVTBaseIE(InfoExtractor):
  13     def _extract_video(self, url, video_id):
  14         info = self._download_json(url, video_id)
  15
  16         title = info['context']['title']
  17         thumbnail = info['context'].get('thumbnailImage')
  18
  19         video_info = info['video']
  20         formats = []
  21         for vr in video_info['videoReferences']:
  22             vurl = vr['url']
  23             ext = determine_ext(vurl)
  24             if ext == 'm3u8':
  25                 formats.extend(self._extract_m3u8_formats(
  26                     vurl, video_id,
  27                     ext='mp4', entry_protocol='m3u8_native',
  28                     m3u8_id=vr.get('playerType')))
  29             elif ext == 'f4m':
  30                 formats.extend(self._extract_f4m_formats(
  31                     vurl + '?hdcore=3.3.0', video_id,
  32                     f4m_id=vr.get('playerType')))
  33             else:
  34                 formats.append({
  35                     'format_id': vr.get('playerType'),
  36                     'url': vurl,
  37                 })
  38         self._sort_formats(formats)
  39
  40         # SVT does not tell us the language, so we assume swedish.
  41         subtitles = {}
  42         for sr in video_info['subtitleReferences']:
  43             if 'url' in sr:
  44                 subtitles.setdefault('sv', []).append({'url': sr['url']})
  45
  46         duration = video_info.get('materialLength')
  47         age_limit = 18 if video_info.get('inappropriateForChildren') else 0
  48
  49         return {
  50             'id': video_id,
  51             'title': title,
  52             'formats': formats,
  53             'subtitles': subtitles,
  54             'thumbnail': thumbnail,
  55             'duration': duration,
  56             'age_limit': age_limit,
  57         }
  58
  59
  60 class SVTIE(SVTBaseIE):
  61     _VALID_URL = r'https?://(?:www\.)?svt\.se/wd\?(?:.*?&)?widgetId=(?P<widget_id>\d+)&.*?\barticleId=(?P<id>\d+)'
  62     _TEST = {
  63         'url': 'http://www.svt.se/wd?widgetId=23991&sectionId=541&articleId=2900353&type=embed&contextSectionId=123&autostart=false',
  64         'md5': '9648197555fc1b49e3dc22db4af51d46',
  65         'info_dict': {
  66             'id': '2900353',
  67             'ext': 'flv',
  68             'title': 'Här trycker Jagr till Giroux (under SVT-intervjun)',
  69             'duration': 27,
  70             'age_limit': 0,
  71         },
  72     }
  73
  74     @staticmethod
  75     def _extract_url(webpage):
  76         mobj = re.search(
  77             r'(?:<iframe src|href)="(?P<url>%s[^"]*)"' % SVTIE._VALID_URL, webpage)
  78         if mobj:
  79             return mobj.group('url')
  80
  81     def _real_extract(self, url):
  82         mobj = re.match(self._VALID_URL, url)
  83         widget_id = mobj.group('widget_id')
  84         article_id = mobj.group('id')
  85         return self._extract_video(
  86             'http://www.svt.se/wd?widgetId=%s&articleId=%s&format=json&type=embed&output=json' % (widget_id, article_id),
  87             article_id)
  88
  89
  90 class SVTPlayIE(SVTBaseIE):
  91     IE_DESC = 'SVT Play and Öppet arkiv'
  92     _VALID_URL = r'https?://(?:www\.)?(?P<host>svtplay|oppetarkiv)\.se/video/(?P<id>[0-9]+)'
  93     _TESTS = [{
  94         'url': 'http://www.svtplay.se/video/2609989/sm-veckan/sm-veckan-rally-final-sasong-1-sm-veckan-rally-final',
  95         'md5': 'ade3def0643fa1c40587a422f98edfd9',
  96         'info_dict': {
  97             'id': '2609989',
  98             'ext': 'flv',
  99             'title': 'SM veckan vinter, Örebro - Rally, final',
 100             'duration': 4500,
 101             'thumbnail': 're:^https?://.*[\.-]jpg$',
 102             'age_limit': 0,
 103         },
 104     }, {
 105         'url': 'http://www.oppetarkiv.se/video/1058509/rederiet-sasong-1-avsnitt-1-av-318',
 106         'md5': 'c3101a17ce9634f4c1f9800f0746c187',
 107         'info_dict': {
 108             'id': '1058509',
 109             'ext': 'flv',
 110             'title': 'Farlig kryssning',
 111             'duration': 2566,
 112             'thumbnail': 're:^https?://.*[\.-]jpg$',
 113             'age_limit': 0,
 114         },
 115         'skip': 'Only works from Sweden',
 116     }]
 117
 118     def _real_extract(self, url):
 119         mobj = re.match(self._VALID_URL, url)
 120         video_id = mobj.group('id')
 121         host = mobj.group('host')
 122         return self._extract_video(
 123             'http://www.%s.se/video/%s?output=json' % (host, video_id),
 124             video_id)