new video attribute 'ext'

weboob/backends/dailymotion/video.py

@@ -23,6 +23,10 @@ __all__ = ['DailymotionVideo']
 
 
 class DailymotionVideo(BaseVideo):
+    def __init__(self, *args, **kwargs):
+        BaseVideo.__init__(self, *args, **kwargs)
+        self.ext = 'flv'
+
     @classmethod
     def id2url(cls, _id):
         if _id.isdigit():

weboob/backends/youjizz/video.py

@@ -26,6 +26,7 @@ class YoujizzVideo(BaseVideo):
     def __init__(self, *args, **kwargs):
         BaseVideo.__init__(self, *args, **kwargs)
         self.nsfw = True
+        self.ext = 'flv'
 
     @classmethod
     def id2url(cls, _id):

weboob/backends/youporn/video.py

@@ -26,6 +26,7 @@ class YoupornVideo(BaseVideo):
     def __init__(self, *args, **kwargs):
         BaseVideo.__init__(self, *args, **kwargs)
         self.nsfw = True
+        self.ext = 'flv'
 
     @classmethod
     def id2url(cls, _id):

weboob/backends/youtube/backend.py

@@ -34,7 +34,32 @@ from .video import YoutubeVideo
 __all__ = ['YoutubeBackend']
 
 
-def get_video(entry):
+class YoutubeBackend(BaseBackend, ICapVideo):
+    NAME = 'youtube'
+    MAINTAINER = 'Christophe Benz'
+    EMAIL = 'christophe.benz@gmail.com'
+    VERSION = '0.7'
+    DESCRIPTION = 'Youtube videos website'
+    LICENSE = 'GPLv3'
+    BROWSER = YoutubeBrowser
+
+    URL_RE = re.compile(r'https?://.*youtube.com/watch\?v=(.*)')
+    AVAILABLE_FORMATS = [38, 37, 22, 45, 35, 34, 43, 18, 6, 5, 17, 13]
+    FORMAT_EXTENSIONS = {
+        13: '3gp',
+        17: 'mp4',
+        18: 'mp4',
+        22: 'mp4',
+        37: 'mp4',
+        38: 'video', # You actually don't know if this will be MOV, AVI or whatever
+        43: 'webm',
+        45: 'webm',
+    }
+
+    def _entry2video(self, entry):
+        """
+        Parse an entry returned by gdata and return a Video object.
+        """
         video = YoutubeVideo(to_unicode(entry.id.text.split('/')[-1].strip()),
                              title=to_unicode(entry.media.title.text.strip()),
                              duration=to_unicode(datetime.timedelta(seconds=int(entry.media.duration.seconds.strip()))),
@@ -45,16 +70,13 @@ def get_video(entry):
             video.author = to_unicode(entry.media.name.text.strip())
         return video
 
-
+    def _set_video_url(self, video, format=18):
-def get_video_url(video, format=18):
         """
-    Returns the YouTube video url for download or playback.
         In the case of a download, if the user-chosen format is not
         available, the next available format will be used.
         Much of the code for this method is borrowed from youtubeservice.py of Cutetube
         http://maemo.org/packages/view/cutetube/.
         """
-    video_url = ''
         player_url = YoutubeVideo.id2url(video.id)
         html = urllib.urlopen(player_url).read()
         html = ''.join(html.split())
@@ -70,25 +92,13 @@ def get_video_url(video, format=18):
                 value = p[:idx].replace('\\/', '/').replace('\u0026', '&').replace(',', '%2C')
                 formats[int(key)] = value
                 key = p[idx + 1:]
-    format_list = [22, 35, 34, 18, 17]
+        for format in self.AVAILABLE_FORMATS[self.AVAILABLE_FORMATS.index(format):]:
-    for format in format_list[format_list.index(format):]:
             if format in formats:
-            video_url = formats.get(format)
+                video.url = formats.get(format)
-            break
+                video.ext = self.FORMAT_EXTENSIONS.get(format, 'flv')
-        break
+                return True
-    return video_url
 
-
+        return False
-class YoutubeBackend(BaseBackend, ICapVideo):
-    NAME = 'youtube'
-    MAINTAINER = 'Christophe Benz'
-    EMAIL = 'christophe.benz@gmail.com'
-    VERSION = '0.7'
-    DESCRIPTION = 'Youtube videos website'
-    LICENSE = 'GPLv3'
-    BROWSER = YoutubeBrowser
-
-    URL_RE = re.compile(r'https?://.*youtube.com/watch\?v=(.*)')
 
     def get_video(self, _id):
         m = self.URL_RE.match(_id)
@@ -103,8 +113,8 @@ class YoutubeBackend(BaseBackend, ICapVideo):
                 return None
             raise
 
-        video = get_video(entry)
+        video = self._entry2video(entry)
-        video.url = get_video_url(video)
+        self._set_video_url(video)
         return video
 
     def iter_search_results(self, pattern=None, sortby=ICapVideo.SEARCH_RELEVANCE, nsfw=False, max_results=None):
@@ -136,7 +146,7 @@ class YoutubeBackend(BaseBackend, ICapVideo):
 
             feed = yt_service.YouTubeQuery(query)
             for entry in feed.entry:
-                yield get_video(entry)
+                yield self._entry2video(entry)
                 nb_yielded += 1
                 if nb_yielded == max_results:
                     return
@@ -145,7 +155,7 @@ class YoutubeBackend(BaseBackend, ICapVideo):
         if 'thumbnail' in fields:
             video.thumbnail.data = urllib.urlopen(video.thumbnail.url).read()
         if 'url' in fields:
-            video.url = get_video_url(video)
+            self._set_video_url(video)
         return video
 
     OBJECTS = {YoutubeVideo: fill_video}

weboob/capabilities/video.py

@@ -51,6 +51,7 @@ class BaseVideo(CapBaseObject):
 
         self.add_field('title', basestring, title)
         self.add_field('url', basestring, url)
+        self.add_field('ext', basestring)
         self.add_field('author', basestring, author)
         self.add_field('description', basestring)
         self.add_field('duration', (int,long,timedelta), duration)