]>
Commit | Line | Data |
---|---|---|
55c97a03 | 1 | # encoding: utf-8 |
4baff4a4 JMF |
2 | from __future__ import unicode_literals |
3 | ||
49f5f315 | 4 | import re |
49f5f315 PH |
5 | |
6 | from .common import InfoExtractor | |
d862a4f9 | 7 | from ..compat import compat_urlparse |
2707b50f | 8 | from .spiegeltv import SpiegeltvIE |
49f5f315 PH |
9 | |
10 | ||
11 | class SpiegelIE(InfoExtractor): | |
aeb7b41d | 12 | _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<id>[0-9]+)(?:-embed|-iframe)?(?:\.html)?(?:#.*)?$' |
7150858d | 13 | _TESTS = [{ |
4baff4a4 | 14 | 'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html', |
4baff4a4 JMF |
15 | 'md5': '2c2754212136f35fb4b19767d242f66e', |
16 | 'info_dict': { | |
55c97a03 S |
17 | 'id': '1259285', |
18 | 'ext': 'mp4', | |
4baff4a4 | 19 | 'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv', |
55c97a03 S |
20 | 'description': 'md5:8029d8310232196eb235d27575a8b9f4', |
21 | 'duration': 49, | |
4baff4a4 | 22 | }, |
55c97a03 | 23 | }, { |
4baff4a4 | 24 | 'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html', |
4baff4a4 JMF |
25 | 'md5': 'f2cdf638d7aa47654e251e1aee360af1', |
26 | 'info_dict': { | |
55c97a03 S |
27 | 'id': '1309159', |
28 | 'ext': 'mp4', | |
4baff4a4 | 29 | 'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers', |
55c97a03 S |
30 | 'description': 'md5:c2322b65e58f385a820c10fa03b2d088', |
31 | 'duration': 983, | |
32 | }, | |
e4bdb37e PH |
33 | }, { |
34 | 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-embed.html', | |
35 | 'md5': 'd8eeca6bfc8f1cd6f490eb1f44695d51', | |
36 | 'info_dict': { | |
37 | 'id': '1519126', | |
38 | 'ext': 'mp4', | |
39 | '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.', | |
40 | 'title': 'Fragen an Astronaut Alexander Gerst: "Bekommen Sie die Tageszeiten mit?"', | |
41 | } | |
aeb7b41d | 42 | }, { |
43 | 'url': 'http://www.spiegel.de/video/astronaut-alexander-gerst-von-der-iss-station-beantwortet-fragen-video-1519126-iframe.html', | |
44 | 'only_matching': True, | |
7150858d | 45 | }] |
49f5f315 PH |
46 | |
47 | def _real_extract(self, url): | |
0e15e725 | 48 | video_id = self._match_id(url) |
2707b50f PH |
49 | webpage, handle = self._download_webpage_handle(url, video_id) |
50 | ||
51 | # 302 to spiegel.tv, like http://www.spiegel.de/video/der-film-zum-wochenende-die-wahrheit-ueber-maenner-video-99003272.html | |
52 | if SpiegeltvIE.suitable(handle.geturl()): | |
53 | return self.url_result(handle.geturl(), 'Spiegeltv') | |
49f5f315 | 54 | |
e4bdb37e PH |
55 | title = re.sub(r'\s+', ' ', self._html_search_regex( |
56 | r'(?s)<(?:h1|div) class="module-title"[^>]*>(.*?)</(?:h1|div)>', | |
57 | webpage, 'title')) | |
55c97a03 | 58 | description = self._html_search_meta('description', webpage, 'description') |
49f5f315 | 59 | |
8bfb6723 | 60 | base_url = self._search_regex( |
bf2c8c8f S |
61 | [r'server\s*:\s*(["\'])(?P<url>.+?)\1', r'var\s+server\s*=\s*"(?P<url>[^"]+)\"'], |
62 | webpage, 'server URL', group='url') | |
8bfb6723 EP |
63 | |
64 | xml_url = base_url + video_id + '.xml' | |
55c97a03 | 65 | idoc = self._download_xml(xml_url, video_id) |
49f5f315 | 66 | |
e92d4a11 S |
67 | formats = [] |
68 | for n in list(idoc): | |
69 | if n.tag.startswith('type') and n.tag != 'type6': | |
70 | format_id = n.tag.rpartition('type')[2] | |
71 | video_url = base_url + n.find('./filename').text | |
e92d4a11 S |
72 | formats.append({ |
73 | 'format_id': format_id, | |
74 | 'url': video_url, | |
75 | 'width': int(n.find('./width').text), | |
76 | 'height': int(n.find('./height').text), | |
77 | 'abr': int(n.find('./audiobitrate').text), | |
78 | 'vbr': int(n.find('./videobitrate').text), | |
79 | 'vcodec': n.find('./codec').text, | |
80 | 'acodec': 'MP4A', | |
81 | }) | |
7150858d PH |
82 | duration = float(idoc[0].findall('./duration')[0].text) |
83 | ||
d862a4f9 | 84 | self._check_formats(formats, video_id) |
e6812ac9 PH |
85 | self._sort_formats(formats) |
86 | ||
4baff4a4 | 87 | return { |
49f5f315 | 88 | 'id': video_id, |
55c97a03 S |
89 | 'title': title, |
90 | 'description': description, | |
49f5f315 | 91 | 'duration': duration, |
7150858d | 92 | 'formats': formats, |
49f5f315 | 93 | } |
89fb6a97 PH |
94 | |
95 | ||
96 | class SpiegelArticleIE(InfoExtractor): | |
97 | _VALID_URL = 'https?://www\.spiegel\.de/(?!video/)[^?#]*?-(?P<id>[0-9]+)\.html' | |
98 | IE_NAME = 'Spiegel:Article' | |
99 | IE_DESC = 'Articles on spiegel.de' | |
e4bdb37e | 100 | _TESTS = [{ |
89fb6a97 PH |
101 | 'url': 'http://www.spiegel.de/sport/sonst/badminton-wm-die-randsportart-soll-populaerer-werden-a-987092.html', |
102 | 'info_dict': { | |
103 | 'id': '1516455', | |
104 | 'ext': 'mp4', | |
105 | 'title': 'Faszination Badminton: Nennt es bloß nicht Federball', | |
106 | 'description': 're:^Patrick Kämnitz gehört.{100,}', | |
107 | }, | |
e4bdb37e PH |
108 | }, { |
109 | 'url': 'http://www.spiegel.de/wissenschaft/weltall/astronaut-alexander-gerst-antwortet-spiegel-online-lesern-a-989876.html', | |
110 | 'info_dict': { | |
111 | ||
112 | }, | |
113 | 'playlist_count': 6, | |
114 | }] | |
89fb6a97 PH |
115 | |
116 | def _real_extract(self, url): | |
0e15e725 | 117 | video_id = self._match_id(url) |
89fb6a97 | 118 | webpage = self._download_webpage(url, video_id) |
e4bdb37e PH |
119 | |
120 | # Single video on top of the page | |
89fb6a97 PH |
121 | video_link = self._search_regex( |
122 | r'<a href="([^"]+)" onclick="return spOpenVideo\(this,', webpage, | |
e4bdb37e PH |
123 | 'video page URL', default=None) |
124 | if video_link: | |
125 | video_url = compat_urlparse.urljoin( | |
126 | self.http_scheme() + '//spiegel.de/', video_link) | |
127 | return self.url_result(video_url) | |
128 | ||
129 | # Multiple embedded videos | |
130 | embeds = re.findall( | |
131 | r'<div class="vid_holder[0-9]+.*?</div>\s*.*?url\s*=\s*"([^"]+)"', | |
132 | webpage) | |
133 | entries = [ | |
134 | self.url_result(compat_urlparse.urljoin( | |
135 | self.http_scheme() + '//spiegel.de/', embed_path)) | |
136 | for embed_path in embeds | |
137 | ] | |
138 | return self.playlist_result(entries) |