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

   1 # coding: utf-8
   2 from __future__ import unicode_literals
   3
   4 import re
   5
   6 from .common import InfoExtractor
   7 from .nexx import (
   8     NexxIE,
   9     NexxEmbedIE,
  10 )
  11 from .spiegeltv import SpiegeltvIE
  12 from ..compat import compat_urlparse
  13 from ..utils import (
  14     parse_duration,
  15     strip_or_none,
  16     unified_timestamp,
  17 )
  18
  19
  20 class SpiegelIE(InfoExtractor):
  21     _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<id>[0-9]+)(?:-embed|-iframe)?(?:\.html)?(?:#.*)?$'
  22     _TESTS = [{
  23         'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html',
  24         'md5': 'b57399839d055fccfeb9a0455c439868',
  25         'info_dict': {
  26             'id': '563747',
  27             'ext': 'mp4',
  28             'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv',
  29             'description': 'md5:8029d8310232196eb235d27575a8b9f4',
  30             'duration': 49,
  31             'upload_date': '20130311',
  32             'timestamp': 1362994320,
  33         },
  34     }, {
  35         'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html',
  36         'md5': '5b6c2f4add9d62912ed5fc78a1faed80',
  37         'info_dict': {
  38             'id': '580988',
  39             'ext': 'mp4',
  40             'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers',
  41             'description': 'md5:c2322b65e58f385a820c10fa03b2d088',
  42             'duration': 983,
  43             'upload_date': '20131115',
  44             'timestamp': 1384546642,
  45         },
  46     }, {
  47         'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-embed.html',
  48         'md5': '97b91083a672d72976faa8433430afb9',
  49         'info_dict': {
  50             'id': '601883',
  51             'ext': 'mp4',
  52             'description': 'SPIEGEL ONLINE-Nutzer durften den deutschen Astronauten Alexander Gerst über sein Leben auf der ISS-Station befragen. Hier kommen seine Antworten auf die besten sechs Fragen.',
  53             'title': 'Fragen an Astronaut Alexander Gerst: "Bekommen Sie die Tageszeiten mit?"',
  54             'upload_date': '20140904',
  55             'timestamp': 1409834160,
  56         }
  57     }, {
  58         'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-iframe.html',
  59         'only_matching': True,
  60     }, {
  61         # nexx video
  62         'url': 'http://www.spiegel.de/video/spiegel-tv-magazin-ueber-guellekrise-in-schleswig-holstein-video-99012776.html',
  63         'only_matching': True,
  64     }]
  65
  66     def _real_extract(self, url):
  67         video_id = self._match_id(url)
  68         metadata_url = 'http://www.spiegel.de/video/metadata/video-%s.json' % video_id
  69         handle = self._request_webpage(metadata_url, video_id)
  70
  71         # 302 to spiegel.tv, like http://www.spiegel.de/video/der-film-zum-wochenende-die-wahrheit-ueber-maenner-video-99003272.html
  72         if SpiegeltvIE.suitable(handle.geturl()):
  73             return self.url_result(handle.geturl(), 'Spiegeltv')
  74
  75         video_data = self._parse_json(self._webpage_read_content(
  76             handle, metadata_url, video_id), video_id)
  77         title = video_data['title']
  78         nexx_id = video_data['nexxOmniaId']
  79         domain_id = video_data.get('nexxOmniaDomain') or '748'
  80
  81         return {
  82             '_type': 'url_transparent',
  83             'id': video_id,
  84             'url': 'nexx:%s:%s' % (domain_id, nexx_id),
  85             'title': title,
  86             'description': strip_or_none(video_data.get('teaser')),
  87             'duration': parse_duration(video_data.get('duration')),
  88             'timestamp': unified_timestamp(video_data.get('datum')),
  89             'ie_key': NexxIE.ie_key(),
  90         }
  91
  92
  93 class SpiegelArticleIE(InfoExtractor):
  94     _VALID_URL = r'https?://(?:www\.)?spiegel\.de/(?!video/)[^?#]*?-(?P<id>[0-9]+)\.html'
  95     IE_NAME = 'Spiegel:Article'
  96     IE_DESC = 'Articles on spiegel.de'
  97     _TESTS = [{
  98         'url': 'http://www.spiegel.de/sport/sonst/badminton-wm-die-randsportart-soll-populaerer-werden-a-987092.html',
  99         'info_dict': {
 100             'id': '1516455',
 101             'ext': 'mp4',
 102             'title': 'Faszination Badminton: Nennt es bloß nicht Federball',
 103             'description': 're:^Patrick Kämnitz gehört.{100,}',
 104             'upload_date': '20140825',
 105         },
 106     }, {
 107         'url': 'http://www.spiegel.de/wissenschaft/weltall/astronaut-alexander-gerst-antwortet-spiegel-online-lesern-a-989876.html',
 108         'info_dict': {
 109
 110         },
 111         'playlist_count': 6,
 112     }, {
 113         # Nexx iFrame embed
 114         'url': 'http://www.spiegel.de/sptv/spiegeltv/spiegel-tv-ueber-schnellste-katapult-achterbahn-der-welt-taron-a-1137884.html',
 115         'info_dict': {
 116             'id': '161464',
 117             'ext': 'mp4',
 118             'title': 'Nervenkitzel Achterbahn',
 119             'alt_title': 'Karussellbauer in Deutschland',
 120             'description': 'md5:ffe7b1cc59a01f585e0569949aef73cc',
 121             'release_year': 2005,
 122             'creator': 'SPIEGEL TV',
 123             'thumbnail': r're:^https?://.*\.jpg$',
 124             'duration': 2761,
 125             'timestamp': 1394021479,
 126             'upload_date': '20140305',
 127         },
 128         'params': {
 129             'format': 'bestvideo',
 130             'skip_download': True,
 131         },
 132     }]
 133
 134     def _real_extract(self, url):
 135         video_id = self._match_id(url)
 136         webpage = self._download_webpage(url, video_id)
 137
 138         # Single video on top of the page
 139         video_link = self._search_regex(
 140             r'<a href="([^"]+)" onclick="return spOpenVideo\(this,', webpage,
 141             'video page URL', default=None)
 142         if video_link:
 143             video_url = compat_urlparse.urljoin(
 144                 self.http_scheme() + '//spiegel.de/', video_link)
 145             return self.url_result(video_url)
 146
 147         # Multiple embedded videos
 148         embeds = re.findall(
 149             r'<div class="vid_holder[0-9]+.*?</div>\s*.*?url\s*=\s*"([^"]+)"',
 150             webpage)
 151         entries = [
 152             self.url_result(compat_urlparse.urljoin(
 153                 self.http_scheme() + '//spiegel.de/', embed_path))
 154             for embed_path in embeds]
 155         if embeds:
 156             return self.playlist_result(entries)
 157
 158         return self.playlist_from_matches(
 159             NexxEmbedIE._extract_urls(webpage), ie=NexxEmbedIE.ie_key())