Error message when Model.__unicode__() returns None

[Luc Saffre]

I had already several times the situation that the unicode method of one of my models returned None in some case.
This leads to a "TypeError: coercing to Unicode: need string or buffer, NoneType found" traceback.
That's okay, but sometimes it is not easy to guess *which* model is wrong.
So I took django/db/models/base.py and added the lines marked '+' below:

    def __repr__(self):
        try:
            u = unicode(self)
        except (UnicodeEncodeError, UnicodeDecodeError):
            u = '[Bad Unicode data]'
    +   except TypeError,e:
    +       raise TypeError("%s: %s" % (self.__class__, e))
        return smart_str(u'<%s: %s>' % (self.__class__.__name__, u))

[Matthias Kestenholz]

New patch with tests and everything.

I'm not completely sure whether raising a new exception on TypeError is worth it though -- maybe we lose precious trackback information...

[Alex Gaynor]

I'm -1 on this, if you dislike the error message I highly recommended taking this up with upstream CPython (or your VM of choice I suppose), as written this will catch the issue in only some contexts (notably those that try to turn the model into a str() rather than unicode() directly).

[Jannis Leidel]

Closing as Alex requested.