[Django]-UnicodeDecodeError : 'ascii' codec can't decode byte 0xe0 in position 0: ordinal not in range(128)

by

75👍

✅

This is a bug in mimetypes, triggered by bad data in the registry. (рєфшю/AMR is not at all a valid MIME media type.)

ctype is a registry key name returned by _winreg.EnumKey, which mimetypes is expecting to be a Unicode string, but it isn’t. Unlike _winreg.QueryValueEx, EnumKey returns a byte string (direct from the ANSI version of the Windows API; _winreg in Python 2 doesn’t use the Unicode interfaces even though it returns Unicode strings, so it’ll never read non-ANSI characters correctly).

So the attempt to .encode it fails with a Unicode​Decode​Error trying to get a Unicode string before encoding it back to ASCII!

try:
    ctype = ctype.encode(default_encoding) # omit in 3.x!
except UnicodeEncodeError:
    pass

These lines in mimetypes should simply be removed.

ETA: added to bug tracker.

👤bobince

8👍

By the way, the main culpit of the problem is QuickTime which adds non-ascii mime types to the windows registry. The easiest way to fix it is to manually find and remove from the registry the subsections of the HKCR/Mime/Database/ContentType/ starting with аудио/ and видео/.

👤newtover

5👍

There is a patch:

http://bugs.python.org/file18143/9291.patch

Works great for me.

Just replace UnicodeEncodeError to be UnicodeError

👤Wangsu

4👍

An alternative solution from python issue9291 by Alexandr Zarubkin (me21):

add file named sitecustomize.py in Lib\site-packages folder.

import sys
sys.setdefaultencoding("cp1251")
👤HaveF

1👍

its a python bug with latin MIME tipes in registry
start regedit and inspect “HKEY_CLASSES_ROOT\MIME\Database\Content Type” for non-latin names.

👤Codeboy.ru

Leave a comment