youtube-dl

commit 3d60bb96e138ce8221f35b7f9d1e1b28f235083e
parent 38d025b3f0f6f349c36a4531f3b36d7e7553f417
Author: Jaime Marquínez Ferrándiz <jaime.marquinez.ferrandiz@gmail.com>
Date:   Fri, 20 Sep 2013 16:55:50 +0200

Add an extractor for ebaumsworld.com (closes #1462)

Diffstat:
M
youtube_dl/extractor/__init__.py
 | 
1
+
A
youtube_dl/extractor/ebaumsworld.py
 | 
37
+++++++++++++++++++++++++++++++++++++

2 files changed, 38 insertions(+), 0 deletions(-)
diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
@@ -24,6 +24,7 @@ from .depositfiles import DepositFilesIE
 from .dotsub import DotsubIE
 from .dreisat import DreiSatIE
 from .defense import DefenseGouvFrIE
+from .ebaumsworld import EbaumsWorldIE
 from .ehow import EHowIE
 from .eighttracks import EightTracksIE
 from .escapist import EscapistIE
diff --git a/youtube_dl/extractor/ebaumsworld.py b/youtube_dl/extractor/ebaumsworld.py
@@ -0,0 +1,37 @@
+import re
+import xml.etree.ElementTree
+
+from .common import InfoExtractor
+from ..utils import determine_ext
+
+
+class EbaumsWorldIE(InfoExtractor):
+    _VALID_URL = r'https?://www\.ebaumsworld\.com/video/watch/(?P<id>\d+)'
+
+    _TEST = {
+        u'url': u'http://www.ebaumsworld.com/video/watch/83367677/',
+        u'file': u'83367677.mp4',
+        u'info_dict': {
+            u'title': u'A Giant Python Opens The Door',
+            u'description': u'This is how nightmares start...',
+            u'uploader': u'jihadpizza',
+        },
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+        config_xml = self._download_webpage(
+            'http://www.ebaumsworld.com/video/player/%s' % video_id, video_id)
+        config = xml.etree.ElementTree.fromstring(config_xml.encode('utf-8'))
+        video_url = config.find('file').text
+
+        return {
+            'id': video_id,
+            'title': config.find('title').text,
+            'url': video_url,
+            'ext': determine_ext(video_url),
+            'description': config.find('description').text,
+            'thumbnail': config.find('image').text,
+            'uploader': config.find('username').text,
+        }