]> Raphaël G. Git Repositories - youtubedl/blobdiff - youtube_dl/extractor/dropbox.py
debian/changelog: Wrap long line.
[youtubedl] / youtube_dl / extractor / dropbox.py
index 41208c97691aafc1c2c96ed06d4a326bca8886a6..14b6c00b0bd1c4d3a306b5513477e2bb6c4cd52d 100644 (file)
@@ -5,25 +5,33 @@ import os.path
 import re
 
 from .common import InfoExtractor
 import re
 
 from .common import InfoExtractor
+from ..compat import compat_urllib_parse_unquote
+from ..utils import url_basename
 
 
 class DropboxIE(InfoExtractor):
 
 
 class DropboxIE(InfoExtractor):
-    _VALID_URL = r'https?://(?:www\.)?dropbox[.]com/s/(?P<id>[a-zA-Z0-9]{15})/(?P<title>[^?#]*)'
-    _TEST = {
-        'url': 'https://www.dropbox.com/s/0qr9sai2veej4f8/THE_DOCTOR_GAMES.mp4',
-        'md5': '8ae17c51172fb7f93bdd6a214cc8c896',
-        'info_dict': {
-            'id': '0qr9sai2veej4f8',
-            'ext': 'mp4',
-            'title': 'THE_DOCTOR_GAMES'
-        }
-    }
+    _VALID_URL = r'https?://(?:www\.)?dropbox[.]com/sh?/(?P<id>[a-zA-Z0-9]{15})/.*'
+    _TESTS = [
+        {
+            'url': 'https://www.dropbox.com/s/nelirfsxnmcfbfh/youtube-dl%20test%20video%20%27%C3%A4%22BaW_jenozKc.mp4?dl=0',
+            'info_dict': {
+                'id': 'nelirfsxnmcfbfh',
+                'ext': 'mp4',
+                'title': 'youtube-dl test video \'ä"BaW_jenozKc'
+            }
+        }, {
+            'url': 'https://www.dropbox.com/sh/662glsejgzoj9sr/AAByil3FGH9KFNZ13e08eSa1a/Pregame%20Ceremony%20Program%20PA%2020140518.m4v',
+            'only_matching': True,
+        },
+    ]
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
 
     def _real_extract(self, url):
         mobj = re.match(self._VALID_URL, url)
         video_id = mobj.group('id')
-        title = os.path.splitext(mobj.group('title'))[0]
-        video_url = url + '?dl=1'
+        fn = compat_urllib_parse_unquote(url_basename(url))
+        title = os.path.splitext(fn)[0]
+        video_url = re.sub(r'[?&]dl=0', '', url)
+        video_url += ('?' if '?' not in video_url else '&') + 'dl=1'
 
         return {
             'id': video_id,
 
         return {
             'id': video_id,