Download videos from jeuxvideo.com
authorPierre Rudloff <pierre@rudloff.pro>
Sun, 18 Aug 2013 14:11:47 +0000 (16:11 +0200)
committerJaime Marquínez Ferrándiz <jaime.marquinez.ferrandiz@gmail.com>
Thu, 22 Aug 2013 10:12:34 +0000 (12:12 +0200)
youtube_dl/extractor/__init__.py
youtube_dl/extractor/jeuxvideo.py [new file with mode: 0644]

index 84c02c2ed99967832264e642fa084367cd4eff05..b9bd3a429132cec0c128d8b11a668ecfc960fa5c 100644 (file)
@@ -36,6 +36,7 @@ from .ign import IGNIE, OneUPIE
 from .ina import InaIE
 from .infoq import InfoQIE
 from .instagram import InstagramIE
+from .jeuxvideo import JeuxVideoIE
 from .jukebox import JukeboxIE
 from .justintv import JustinTVIE
 from .kankan import KankanIE
diff --git a/youtube_dl/extractor/jeuxvideo.py b/youtube_dl/extractor/jeuxvideo.py
new file mode 100644 (file)
index 0000000..d74a1c9
--- /dev/null
@@ -0,0 +1,33 @@
+import json
+import re
+
+from .common import InfoExtractor
+
+class JeuxVideoIE(InfoExtractor):
+    _VALID_URL = r'http://.*?\.jeuxvideo\.com/.*/(.*?)-\d+\.htm'
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        title = re.match(self._VALID_URL, url).group(1)
+        webpage = self._download_webpage(url, title)
+        m_download = re.search(r'<param name="flashvars" value="config=(.*?)" />', webpage)
+
+        xml_link = m_download.group(1)
+        
+        id = re.search(r'http://www.jeuxvideo.com/config/\w+/0011/(.*?)/\d+_player\.xml', xml_link).group(1)
+
+        xml_config = self._download_webpage(xml_link, title,
+                                                  'Downloading XML config')
+        info = re.search(r'<format\.json>(.*?)</format\.json>',
+                         xml_config, re.MULTILINE|re.DOTALL).group(1)
+        info = json.loads(info)['versions'][0]
+        
+        video_url = 'http://video720.jeuxvideo.com/' + info['file']
+
+        track_info = {'id':id,
+                      'title' : title,
+                      'ext' :   'mp4',
+                      'url' :   video_url
+                      }
+
+        return [track_info]