[unistra] Modernize

youtube_dl/extractor/unistra.py

@@ -1,44 +1,66 @@
+from __future__ import unicode_literals
+
 import re
 
 from .common import InfoExtractor
+from ..utils import qualities
+
 
 class UnistraIE(InfoExtractor):
-    _VALID_URL = r'http://utv\.unistra\.fr/(?:index|video)\.php\?id_video\=(\d+)'
+    _VALID_URL = r'http://utv\.unistra\.fr/(?:index|video)\.php\?id_video\=(?P<id>\d+)'
-
+
-    _TEST = {
+    _TESTS = [
-        u'url': u'http://utv.unistra.fr/video.php?id_video=154',
+        {
-        u'file': u'154.mp4',
+            'url': 'http://utv.unistra.fr/video.php?id_video=154',
-        u'md5': u'736f605cfdc96724d55bb543ab3ced24',
+            'md5': '736f605cfdc96724d55bb543ab3ced24',
-        u'info_dict': {
+            'info_dict': {
-            u'title': u'M!ss Yella',
+                'id': '154',
-            u'description': u'md5:104892c71bd48e55d70b902736b81bbf',
+                'ext': 'mp4',
+                'title': 'M!ss Yella',
+                'description': 'md5:104892c71bd48e55d70b902736b81bbf',
+            },
         },
-    }
+        {
+            'url': 'http://utv.unistra.fr/index.php?id_video=437',
+            'md5': '1ddddd6cccaae76f622ce29b8779636d',
+            'info_dict': {
+                'id': '437',
+                'ext': 'mp4',
+                'title': 'Prix Louise Weiss 2014',
+                'description': 'md5:cc3a8735f079f4fb6b0b570fc10c135a',
+            },
+        }
+    ]
 
     def _real_extract(self, url):
-        video_id = re.match(self._VALID_URL, url).group(1)
+        mobj = re.match(self._VALID_URL, url)
+        video_id = mobj.group('id')
+
         webpage = self._download_webpage(url, video_id)
-        width = re.search(r'width: "(\d*?)",', webpage).group(1)
+
-        height = re.search(r'height: "(\d*?)",', webpage).group(1)
+        files = set(re.findall(r'file\s*:\s*"([^"]+)"', webpage))
-        files = re.findall(r'file: "(.*?)"', webpage)
+
-        video_url = 'http://vod-flash.u-strasbg.fr:8080'
+        quality = qualities(['SD', 'HD'])
-        formats = [{
+        formats = []
-            'format_id': 'SD',
+        for file_path in files:
-            'url': video_url + files[0],
+            format_id = 'HD' if file_path.endswith('-HD.mp4') else 'SD'
-            'ext': 'mp4',
-            'resolution': width + 'x' + height
-            }]
-        if files[1] != files[0]:
             formats.append({
-            'format_id': 'HD',
+                'url': 'http://vod-flash.u-strasbg.fr:8080%s' % file_path,
-            'url': video_url + files[1],
+                'format_id': format_id,
-            'ext': 'mp4'
+                'quality': quality(format_id)
             })
-        title = self._html_search_regex(r'<title>UTV - (.*?)</', webpage, u'title')
+
-
+        title = self._html_search_regex(
-        return {'id': video_id,
+            r'<title>UTV - (.*?)</', webpage, 'title')
-                'title': title,
+        description = self._html_search_regex(
-                'description': self._html_search_regex(r'<meta name="Description" content="(.*?)"', webpage, u'description', flags=re.DOTALL),
+            r'<meta name="Description" content="(.*?)"', webpage, 'description', flags=re.DOTALL)
-                'thumbnail': self._search_regex(r'image: "(.*?)"', webpage, u'thumbnail'),
+        thumbnail = self._search_regex(
-                'formats': formats
+            r'image: "(.*?)"', webpage, 'thumbnail')
-                }
+
+        return {
+            'id': video_id,
+            'title': title,
+            'description': description,
+            'thumbnail': thumbnail,
+            'formats': formats
+        }