Django always returns a white page along with 400 on latin encoded URLs.

[Christian Peters]

Given a URL like /Raumh%F6he served with gunicorn and waitress (maybe others as well), django raises a UnicodeDecodeError here:

​https://github.com/django/django/blob/master/django/core/handlers/wsgi.py#L200

... which ends up in serving a white page with a 400 (sample: ​https://www.djangoproject.com/Raumh%F6he)

This does not happen with the built in runserver command, because it double encodes the query, as far as I understand:

​https://github.com/django/django/blob/master/django/core/servers/basehttp.py#L154-L157

Gunicorn on the other hand is explicitly casting to latin:

​https://github.com/benoitc/gunicorn/blob/master/gunicorn/_compat.py#L82

... which leads to the error as django is explicitly expecting utf-8.

[Christian Peters]

For what it's worth:

Pyramid raises a 500 (​http://www.pylonsproject.org/Raumh%F6he.htm) (error code: URLDecodeError: 'utf8' codec can't decode byte 0xf6 in position 11: invalid start byte (urldispatch.py, line 86)

Flask handles it correctly.

[Dheerendra Rathor]

In the part mentioned by OP,

def get_path_info(environ):
    """
    Returns the HTTP request's PATH_INFO as a unicode string.
    """
    path_info = get_bytes_from_wsgi(environ, 'PATH_INFO', '/')

    return path_info.decode(UTF_8)

I changed UTF-8 to ISO_8859_1 in my local Django installation and then it was correctly throwing 404 instead of 400.

[Jef Geskens]

Django handles this in the correct way. URI's which are encoded as latin1 are not standard, they should be UTF-8 encoded (see ​https://tools.ietf.org/html/rfc3986).

Because the url is not encoded in the standard way, Django correctly gives "400 Bad request".