[francetv] Add extractor for Culturebox (closes #2117)

diff --git a/youtube_dl/extractor/__init__.py b/youtube_dl/extractor/__init__.py
index e813ba976bc49a750368d8efd526b56ca07ab4ac..b887c7f101970561907c95eeba912a7656dab79f 100644 (file)
--- a/youtube_dl/extractor/__init__.py
+++ b/youtube_dl/extractor/__init__.py
@@ -65,7 +65,8 @@ from .francetv import (
     PluzzIE,
     FranceTvInfoIE,
     FranceTVIE,
-    GenerationQuoiIE
+    GenerationQuoiIE,
+    CultureboxIE,
 )
 from .freesound import FreesoundIE
 from .funnyordie import FunnyOrDieIE

diff --git a/youtube_dl/extractor/francetv.py b/youtube_dl/extractor/francetv.py
index ad85bc16d7796cfcf42331a05bb0392e773f70c5..b32ff9f867a02147d8ddc145cba685f074e6155c 100644 (file)
--- a/youtube_dl/extractor/francetv.py
+++ b/youtube_dl/extractor/francetv.py
@@ -191,3 +191,29 @@ class GenerationQuoiIE(InfoExtractor):
         info = json.loads(info_json)
         return self.url_result('http://www.dailymotion.com/video/%s' % info['id'],
             ie='Dailymotion')
+
+
+class CultureboxIE(FranceTVBaseInfoExtractor):
+    IE_NAME = u'culturebox.francetvinfo.fr'
+    _VALID_URL = r'https?://culturebox\.francetvinfo\.fr/(?P<name>.*?)(\?|$)'
+
+    _TEST = {
+        u'url': u'http://culturebox.francetvinfo.fr/einstein-on-the-beach-au-theatre-du-chatelet-146813',
+        u'info_dict': {
+            u'id': u'EV_6785',
+            u'ext': u'mp4',
+            u'title': u'Einstein on the beach au Théâtre du Châtelet',
+            u'description': u'md5:9ce2888b1efefc617b5e58b3f6200eeb',
+        },
+        u'params': {
+            # m3u8 download
+            u'skip_download': True,
+        },
+    }
+
+    def _real_extract(self, url):
+        mobj = re.match(self._VALID_URL, url)
+        name = mobj.group('name')
+        webpage = self._download_webpage(url, name)
+        video_id = self._search_regex(r'"http://videos\.francetv\.fr/video/(.*?)"', webpage, u'video id')
+        return self._extract_video(video_id)