zope security proxy doesn't handle __unicode__

[Zach Cashero]

There is no implementation of the __unicode__ method to handle objects that are wrapped in the zope.security._proxy._Proxy. This causes a UnicodeDecodeError when trying to get the unicode representation of a custom object that has a security proxy wrapped around it.

For example, this class implementation:

class ICustomClass(Interface):
    def __unicode__():
        pass

    def __str__():
        pass

@implementer(ICustomClass)
class CustomClass(object):
    def __unicode__(self):
        print "In __unicode__"
        return u'Español con caracteres especiales'

    def __str__(self):
        print "In __str__"
        return unicode(self).encode('utf-8')

With the associated zcml:
  <class class=".test.CustomClass">
    <allow
        interface=".test.ICustomClass" />
  </class>

And a simple test:
>>> from zope.security.checker import ProxyFactory
>>> obj = CustomClass()
>>> str(obj)
In __str__
In __unicode__
'Espa\xc3\xb1ol con caracteres especiales'
>>> unicode(obj)
In __unicode__
u'Espa\xf1ol con caracteres especiales'
>>> prox = ProxyFactory(obj)
>>> str(prox)
In __str__
In __unicode__
'Espa\xc3\xb1ol con caracteres especiales'
>>> unicode(prox)
In __str__
In __unicode__
*** UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 4: ordinal not in range(128)

The object, that should return a unicode string with special characters works fine on the original object, but causes an error on the proxied object because it actually calls the __str__ method which returns a 'utf-8' encoded byte string and when trying to make it into a unicode string, fails since it is trying to decode it with the system default of 'ascii.' Essentially, it is the equivalent of calling:
>>> unicode(prox.__str__())
In __str__
In __unicode__
*** UnicodeDecodeError: 'ascii' codec can't decode byte 0xc3 in position 4: ordinal not in range(128)

[Tres Seaver (tseaver)]

Now tracked on Github:

  https://github.com/zopefoundation/zope.security/issues/10