Merge pull request #3684 from qstokkink/fix_tdef_name

Fixed crash on torrents with exotic names

Tribler/Core/Modules/restapi/downloads_endpoint.py

@@ -213,7 +213,7 @@ def render_GET(self, request):
 
  num_seeds, num_peers = state.get_num_seeds_peers()
 
- download_json = {"name": tdef.get_name(), "progress": state.get_progress(),
+ download_json = {"name": tdef.get_name_utf8(), "progress": state.get_progress(),
  "infohash": tdef.get_infohash().encode('hex'),
  "speed_down": state.get_current_speed(DOWNLOAD),
  "speed_up": state.get_current_speed(UPLOAD),

Tribler/Core/TorrentDef.py

@@ -502,6 +502,24 @@ def get_name(self):
  else:
  raise TorrentDefNotFinalizedException()
 
+ def get_name_utf8(self):
+ """
+ Not all names are utf-8, attempt to construct it as utf-8 anyway.
+ """
+ out = self.get_name()
+ try:
+ # Try seeing if the delivered encoding is correct and we
+ # can convert to utf8 without any issues.
+ return out.decode(self.get_encoding()).encode('utf8').decode('utf8')
+ except (LookupError, TypeError, ValueError):
+ try:
+ # The delivered encoding is incorrect, cast it to
+ # latin1 and hope for the best (minor corruption).
+ return out.decode('latin1').encode('utf8', 'ignore').decode('utf8')
+ except (TypeError, ValueError):
+ # This is a very nasty string (e.g. u'\u266b'), remove the illegal entries.
+ return out.encode('utf8', 'ignore').decode('utf8')
+
  def set_name(self, name):
  """ Set the name of this torrent
  @param name name of torrent as String

Tribler/Test/Core/test_torrent_def.py

@@ -145,6 +145,25 @@ def test_add_content_dir_and_file(self):
  reals += file['length']
  self.assertEqual(exps, reals)
 
+ def test_get_name_utf8(self):
+ """ Add a TorrentDef with non-utf8 encoding"""
+ t = TorrentDef()
+ t.set_name('\xA1\xC0')
+ t.set_encoding('euc_kr')
+ t.set_tracker(TRACKER)
+ t.finalize()
+
+ self.assertEqual(t.get_name_utf8(), u'\xf7')
+
+ def test_get_name_utf8_unknown(self):
+ """ Add a TorrentDef with non-utf8 encoding"""
+ t = TorrentDef()
+ t.set_name('\xA1\xC0')
+ t.set_tracker(TRACKER)
+ t.finalize()
+
+ self.assertEqual(t.get_name_utf8(), u'\xa1\xc0')
+
  def test_add_content_announce_list(self):
  """ Add a single file with announce-list to a TorrentDef """
  t = TorrentDef()