X-Git-Url: https://git.rapsys.eu/youtubedl/blobdiff_plain/9815bb0a551468e4939cacfffbc2d5cb8dd12431..64669d882d7a41b7f3c25a7191f4cc7db0684ef2:/youtube_dl/extractor/spiegel.py?ds=sidebyside diff --git a/youtube_dl/extractor/spiegel.py b/youtube_dl/extractor/spiegel.py index 98a65b7..4df7f4d 100644 --- a/youtube_dl/extractor/spiegel.py +++ b/youtube_dl/extractor/spiegel.py @@ -1,37 +1,159 @@ +# coding: utf-8 +from __future__ import unicode_literals + import re -import xml.etree.ElementTree from .common import InfoExtractor +from .nexx import ( + NexxIE, + NexxEmbedIE, +) +from .spiegeltv import SpiegeltvIE +from ..compat import compat_urlparse +from ..utils import ( + parse_duration, + strip_or_none, + unified_timestamp, +) class SpiegelIE(InfoExtractor): - _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P[0-9]+)(?:\.html)?(?:#.*)?$' + _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P[0-9]+)(?:-embed|-iframe)?(?:\.html)?(?:#.*)?$' + _TESTS = [{ + 'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html', + 'md5': 'b57399839d055fccfeb9a0455c439868', + 'info_dict': { + 'id': '563747', + 'ext': 'mp4', + 'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv', + 'description': 'md5:8029d8310232196eb235d27575a8b9f4', + 'duration': 49, + 'upload_date': '20130311', + 'timestamp': 1362994320, + }, + }, { + 'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html', + 'md5': '5b6c2f4add9d62912ed5fc78a1faed80', + 'info_dict': { + 'id': '580988', + 'ext': 'mp4', + 'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers', + 'description': 'md5:c2322b65e58f385a820c10fa03b2d088', + 'duration': 983, + 'upload_date': '20131115', + 'timestamp': 1384546642, + }, + }, { + 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-embed.html', + 'md5': '97b91083a672d72976faa8433430afb9', + 'info_dict': { + 'id': '601883', + 'ext': 'mp4', + '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.', + 'title': 'Fragen an Astronaut Alexander Gerst: "Bekommen Sie die Tageszeiten mit?"', + 'upload_date': '20140904', + 'timestamp': 1409834160, + } + }, { + 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-iframe.html', + 'only_matching': True, + }, { + # nexx video + 'url': 'http://www.spiegel.de/video/spiegel-tv-magazin-ueber-guellekrise-in-schleswig-holstein-video-99012776.html', + 'only_matching': True, + }] def _real_extract(self, url): - m = re.match(self._VALID_URL, url) - video_id = m.group('videoID') - - webpage = self._download_webpage(url, video_id) + video_id = self._match_id(url) + metadata_url = 'http://www.spiegel.de/video/metadata/video-%s.json' % video_id + handle = self._request_webpage(metadata_url, video_id) - video_title = self._html_search_regex(r'
(.*?)
', - webpage, u'title') + # 302 to spiegel.tv, like http://www.spiegel.de/video/der-film-zum-wochenende-die-wahrheit-ueber-maenner-video-99003272.html + if SpiegeltvIE.suitable(handle.geturl()): + return self.url_result(handle.geturl(), 'Spiegeltv') - xml_url = u'http://video2.spiegel.de/flash/' + video_id + u'.xml' - xml_code = self._download_webpage(xml_url, video_id, - note=u'Downloading XML', errnote=u'Failed to download XML') + video_data = self._parse_json(self._webpage_read_content( + handle, metadata_url, video_id), video_id) + title = video_data['title'] + nexx_id = video_data['nexxOmniaId'] + domain_id = video_data.get('nexxOmniaDomain') or '748' - idoc = xml.etree.ElementTree.fromstring(xml_code) - last_type = idoc[-1] - filename = last_type.findall('./filename')[0].text - duration = float(last_type.findall('./duration')[0].text) - - video_url = 'http://video2.spiegel.de/flash/' + filename - video_ext = filename.rpartition('.')[2] - info = { + return { + '_type': 'url_transparent', 'id': video_id, - 'url': video_url, - 'ext': video_ext, - 'title': video_title, - 'duration': duration, + 'url': 'nexx:%s:%s' % (domain_id, nexx_id), + 'title': title, + 'description': strip_or_none(video_data.get('teaser')), + 'duration': parse_duration(video_data.get('duration')), + 'timestamp': unified_timestamp(video_data.get('datum')), + 'ie_key': NexxIE.ie_key(), } - return [info] + + +class SpiegelArticleIE(InfoExtractor): + _VALID_URL = r'https?://(?:www\.)?spiegel\.de/(?!video/)[^?#]*?-(?P[0-9]+)\.html' + IE_NAME = 'Spiegel:Article' + IE_DESC = 'Articles on spiegel.de' + _TESTS = [{ + 'url': 'http://www.spiegel.de/sport/sonst/badminton-wm-die-randsportart-soll-populaerer-werden-a-987092.html', + 'info_dict': { + 'id': '1516455', + 'ext': 'mp4', + 'title': 'Faszination Badminton: Nennt es bloß nicht Federball', + 'description': 're:^Patrick Kämnitz gehört.{100,}', + 'upload_date': '20140825', + }, + }, { + 'url': 'http://www.spiegel.de/wissenschaft/weltall/astronaut-alexander-gerst-antwortet-spiegel-online-lesern-a-989876.html', + 'info_dict': { + + }, + 'playlist_count': 6, + }, { + # Nexx iFrame embed + 'url': 'http://www.spiegel.de/sptv/spiegeltv/spiegel-tv-ueber-schnellste-katapult-achterbahn-der-welt-taron-a-1137884.html', + 'info_dict': { + 'id': '161464', + 'ext': 'mp4', + 'title': 'Nervenkitzel Achterbahn', + 'alt_title': 'Karussellbauer in Deutschland', + 'description': 'md5:ffe7b1cc59a01f585e0569949aef73cc', + 'release_year': 2005, + 'creator': 'SPIEGEL TV', + 'thumbnail': r're:^https?://.*\.jpg$', + 'duration': 2761, + 'timestamp': 1394021479, + 'upload_date': '20140305', + }, + 'params': { + 'format': 'bestvideo', + 'skip_download': True, + }, + }] + + def _real_extract(self, url): + video_id = self._match_id(url) + webpage = self._download_webpage(url, video_id) + + # Single video on top of the page + video_link = self._search_regex( + r'\s*.*?url\s*=\s*"([^"]+)"', + webpage) + entries = [ + self.url_result(compat_urlparse.urljoin( + self.http_scheme() + '//spiegel.de/', embed_path)) + for embed_path in embeds] + if embeds: + return self.playlist_result(entries) + + return self.playlist_from_matches( + NexxEmbedIE._extract_urls(webpage), ie=NexxEmbedIE.ie_key())