1
0
mirror of https://codeberg.org/polarisfm/youtube-dl synced 2024-12-05 14:47:55 +01:00
youtube-dl/youtube_dl/extractor/spiegel.py

160 lines
6.2 KiB
Python
Raw Normal View History

2016-10-02 13:39:18 +02:00
# coding: utf-8
from __future__ import unicode_literals
2013-06-23 22:22:08 +02:00
import re
from .common import InfoExtractor
from .nexx import (
NexxIE,
NexxEmbedIE,
)
from .spiegeltv import SpiegeltvIE
2016-07-05 12:46:25 +02:00
from ..compat import compat_urlparse
from ..utils import (
2018-05-28 01:10:01 +02:00
parse_duration,
strip_or_none,
unified_timestamp,
2016-07-05 12:46:25 +02:00
)
2013-06-23 22:22:08 +02:00
class SpiegelIE(InfoExtractor):
_VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<id>[0-9]+)(?:-embed|-iframe)?(?:\.html)?(?:#.*)?$'
2013-11-16 01:33:12 +01:00
_TESTS = [{
'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html',
2018-05-28 01:10:01 +02:00
'md5': 'b57399839d055fccfeb9a0455c439868',
'info_dict': {
2018-05-28 01:10:01 +02:00
'id': '563747',
'ext': 'mp4',
'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv',
'description': 'md5:8029d8310232196eb235d27575a8b9f4',
'duration': 49,
2016-07-05 12:46:25 +02:00
'upload_date': '20130311',
2018-05-28 01:10:01 +02:00
'timestamp': 1362994320,
},
}, {
'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html',
2018-05-28 01:10:01 +02:00
'md5': '5b6c2f4add9d62912ed5fc78a1faed80',
'info_dict': {
2018-05-28 01:10:01 +02:00
'id': '580988',
'ext': 'mp4',
'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers',
'description': 'md5:c2322b65e58f385a820c10fa03b2d088',
'duration': 983,
2016-07-05 12:46:25 +02:00
'upload_date': '20131115',
2018-05-28 01:10:01 +02:00
'timestamp': 1384546642,
},
2014-11-13 15:02:31 +01:00
}, {
'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-embed.html',
2018-05-28 01:10:01 +02:00
'md5': '97b91083a672d72976faa8433430afb9',
2014-11-13 15:02:31 +01:00
'info_dict': {
2018-05-28 01:10:01 +02:00
'id': '601883',
2014-11-13 15:02:31 +01:00
'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?"',
2016-07-05 12:46:25 +02:00
'upload_date': '20140904',
2018-05-28 01:10:01 +02:00
'timestamp': 1409834160,
2014-11-13 15:02:31 +01:00
}
}, {
'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,
2013-11-16 01:33:12 +01:00
}]
2013-06-23 22:22:08 +02:00
def _real_extract(self, url):
2014-11-13 14:45:17 +01:00
video_id = self._match_id(url)
2018-05-28 01:10:01 +02:00
metadata_url = 'http://www.spiegel.de/video/metadata/video-%s.json' % video_id
handle = self._request_webpage(metadata_url, video_id)
# 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')
2013-06-23 22:22:08 +02:00
2018-05-28 01:10:01 +02:00
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'
2013-12-24 12:40:23 +01:00
return {
2018-05-28 01:10:01 +02:00
'_type': 'url_transparent',
2013-06-23 22:22:08 +02:00
'id': video_id,
2018-05-28 01:10:01 +02:00
'url': 'nexx:%s:%s' % (domain_id, nexx_id),
'title': title,
2018-05-28 01:10:01 +02:00
'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(),
2013-06-23 22:22:08 +02:00
}
2014-09-13 06:55:38 +02:00
class SpiegelArticleIE(InfoExtractor):
2016-09-08 13:29:05 +02:00
_VALID_URL = r'https?://(?:www\.)?spiegel\.de/(?!video/)[^?#]*?-(?P<id>[0-9]+)\.html'
2014-09-13 06:55:38 +02:00
IE_NAME = 'Spiegel:Article'
IE_DESC = 'Articles on spiegel.de'
2014-11-13 15:02:31 +01:00
_TESTS = [{
2014-09-13 06:55:38 +02:00
'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',
2014-09-13 06:55:38 +02:00
},
2014-11-13 15:02:31 +01:00
}, {
'url': 'http://www.spiegel.de/wissenschaft/weltall/astronaut-alexander-gerst-antwortet-spiegel-online-lesern-a-989876.html',
'info_dict': {
},
'playlist_count': 6,
2017-07-17 17:19:40 +02:00
}, {
# 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,
},
2014-11-13 15:02:31 +01:00
}]
2014-09-13 06:55:38 +02:00
def _real_extract(self, url):
2014-11-13 14:45:17 +01:00
video_id = self._match_id(url)
2014-09-13 06:55:38 +02:00
webpage = self._download_webpage(url, video_id)
2014-11-13 15:02:31 +01:00
# Single video on top of the page
2014-09-13 06:55:38 +02:00
video_link = self._search_regex(
r'<a href="([^"]+)" onclick="return spOpenVideo\(this,', webpage,
2014-11-13 15:02:31 +01:00
'video page URL', default=None)
if video_link:
video_url = compat_urlparse.urljoin(
self.http_scheme() + '//spiegel.de/', video_link)
return self.url_result(video_url)
# Multiple embedded videos
embeds = re.findall(
r'<div class="vid_holder[0-9]+.*?</div>\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())