Keep video IDs verbatim if possible (Closes #571)

test/test_utils.py

@@ -79,6 +79,11 @@ class TestUtil(unittest.TestCase):
         self.assertTrue(sanitize_filename('-', restricted=True) != '')
         self.assertTrue(sanitize_filename(':', restricted=True) != '')
 
+    def test_sanitize_ids(self):
+        self.assertEquals(sanitize_filename('_n_cd26wFpw', is_id=True), '_n_cd26wFpw')
+        self.assertEquals(sanitize_filename('_BD_eEpuzXw', is_id=True), '_BD_eEpuzXw')
+        self.assertEquals(sanitize_filename('N0Y__7-UOdI', is_id=True), 'N0Y__7-UOdI')
+
     def test_ordered_set(self):
         self.assertEqual(orderedSet([1, 1, 2, 3, 4, 4, 5, 6, 7, 3, 5]), [1, 2, 3, 4, 5, 6, 7])
         self.assertEqual(orderedSet([]), [])

youtube_dl/FileDownloader.py

@@ -334,8 +334,11 @@ class FileDownloader(object):
             template_dict['epoch'] = int(time.time())
             template_dict['autonumber'] = u'%05d' % self._num_downloads
 
-            template_dict = dict((key, u'NA' if val is None else val) for key, val in template_dict.items())
-            template_dict = dict((k, sanitize_filename(compat_str(v), self.params.get('restrictfilenames'))) for k,v in template_dict.items())
+            sanitize = lambda k,v: sanitize_filename(
+                u'NA' if v is None else compat_str(v),
+                restricted=self.params.get('restrictfilenames'),
+                is_id=(k==u'id'))
+            template_dict = dict((k, sanitize(k, v)) for k,v in template_dict.items())
 
             filename = self.params['outtmpl'] % template_dict
             return filename

youtube_dl/utils.py

@@ -317,9 +317,10 @@ def timeconvert(timestr):
         timestamp = email.utils.mktime_tz(timetuple)
     return timestamp
 
-def sanitize_filename(s, restricted=False):
+def sanitize_filename(s, restricted=False, is_id=False):
     """Sanitizes a string so it could be used as part of a filename.
     If restricted is set, use a stricter subset of allowed characters.
+    Set is_id if this is not an arbitrary string, but an ID that should be kept if possible
     """
     def replace_insane(char):
         if char == '?' or ord(char) < 32 or ord(char) == 127:
@@ -337,6 +338,7 @@ def sanitize_filename(s, restricted=False):
         return char
 
     result = u''.join(map(replace_insane, s))
+    if not is_id:
         while '__' in result:
             result = result.replace('__', '_')
         result = result.strip('_')