Prepare for upload.

[youtubedl] / youtube_dl / extractor / spiegel.py

diff --git a/youtube_dl/extractor/spiegel.py b/youtube_dl/extractor/spiegel.py
index 051a34d5b8b048db9112ac2df5f3f04115447a87..9ed7d3b39e227806971fe98f43e1c1018b84ad3c 100644 (file)
--- a/youtube_dl/extractor/spiegel.py
+++ b/youtube_dl/extractor/spiegel.py
@@ -1,25 +1,34 @@
+# encoding: utf-8
+from __future__ import unicode_literals
+
 import re
 
 from .common import InfoExtractor
+from ..utils import compat_urlparse
 
 
 class SpiegelIE(InfoExtractor):
     _VALID_URL = r'https?://(?:www\.)?spiegel\.de/video/[^/]*-(?P<videoID>[0-9]+)(?:\.html)?(?:#.*)?$'
     _TESTS = [{
-        u'url': u'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html',
-        u'file': u'1259285.mp4',
-        u'md5': u'2c2754212136f35fb4b19767d242f66e',
-        u'info_dict': {
-            u"title": u"Vulkanausbruch in Ecuador: Der \"Feuerschlund\" ist wieder aktiv"
-        }
-    },
-    {
-        u'url': u'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html',
-        u'file': u'1309159.mp4',
-        u'md5': u'f2cdf638d7aa47654e251e1aee360af1',
-        u'info_dict': {
-            u'title': u'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers'
-        }
+        'url': 'http://www.spiegel.de/video/vulkan-tungurahua-in-ecuador-ist-wieder-aktiv-video-1259285.html',
+        'md5': '2c2754212136f35fb4b19767d242f66e',
+        'info_dict': {
+            'id': '1259285',
+            'ext': 'mp4',
+            'title': 'Vulkanausbruch in Ecuador: Der "Feuerschlund" ist wieder aktiv',
+            'description': 'md5:8029d8310232196eb235d27575a8b9f4',
+            'duration': 49,
+        },
+    }, {
+        'url': 'http://www.spiegel.de/video/schach-wm-videoanalyse-des-fuenften-spiels-video-1309159.html',
+        'md5': 'f2cdf638d7aa47654e251e1aee360af1',
+        'info_dict': {
+            'id': '1309159',
+            'ext': 'mp4',
+            'title': 'Schach-WM in der Videoanalyse: Carlsen nutzt die Fehlgriffe des Titelverteidigers',
+            'description': 'md5:c2322b65e58f385a820c10fa03b2d088',
+            'duration': 983,
+        },
     }]
 
     def _real_extract(self, url):
@@ -28,18 +37,20 @@ class SpiegelIE(InfoExtractor):
 
         webpage = self._download_webpage(url, video_id)
 
-        video_title = self._html_search_regex(
-            r'<div class="module-title">(.*?)</div>', webpage, u'title')
+        title = self._html_search_regex(
+            r'<div class="module-title">(.*?)</div>', webpage, 'title')
+        description = self._html_search_meta('description', webpage, 'description')
+
+        base_url = self._search_regex(
+            r'var\s+server\s*=\s*"([^"]+)\"', webpage, 'server URL')
 
-        xml_url = u'http://video2.spiegel.de/flash/' + video_id + u'.xml'
-        idoc = self._download_xml(
-            xml_url, video_id,
-            note=u'Downloading XML', errnote=u'Failed to download XML')
+        xml_url = base_url + video_id + '.xml'
+        idoc = self._download_xml(xml_url, video_id)
 
         formats = [
             {
                 'format_id': n.tag.rpartition('type')[2],
-                'url': u'http://video2.spiegel.de/flash/' + n.find('./filename').text,
+                'url': base_url + n.find('./filename').text,
                 'width': int(n.find('./width').text),
                 'height': int(n.find('./height').text),
                 'abr': int(n.find('./audiobitrate').text),
@@ -55,10 +66,41 @@ class SpiegelIE(InfoExtractor):
 
         self._sort_formats(formats)
 
-        info = {
+        return {
             'id': video_id,
-            'title': video_title,
+            'title': title,
+            'description': description,
             'duration': duration,
             'formats': formats,
         }
-        return info
+
+
+class SpiegelArticleIE(InfoExtractor):
+    _VALID_URL = 'https?://www\.spiegel\.de/(?!video/)[^?#]*?-(?P<id>[0-9]+)\.html'
+    IE_NAME = 'Spiegel:Article'
+    IE_DESC = 'Articles on spiegel.de'
+    _TEST = {
+        '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,}',
+        },
+    }
+
+    def _real_extract(self, url):
+        m = re.match(self._VALID_URL, url)
+        video_id = m.group('id')
+
+        webpage = self._download_webpage(url, video_id)
+        video_link = self._search_regex(
+            r'<a href="([^"]+)" onclick="return spOpenVideo\(this,', webpage,
+            'video page URL')
+        video_url = compat_urlparse.urljoin(
+            self.http_scheme() + '//spiegel.de/', video_link)
+
+        return {
+            '_type': 'url',
+            'url': video_url,
+        }