fix unicode encoding errors in sparql processor #446

[joernhees]

attempt to fix #446.

The problem consist of two problems: in parserutils a SPARQLError is returned instead of raised, making it end up in operators.py, in which it ends up in another SPARQLError('... %s' % here) causing str() to be called on it... if the original expression contained a unicode char it is thus causing an encoding error:

In [7]: Exception('name was: %s' % u'jörn')
Out[7]: Exception(u'name was: j\xf6rn')

In [8]: Exception('something went wrong: %s' % Exception('name was: %s' % u'jörn'))
---------------------------------------------------------------------------
UnicodeEncodeError                        Traceback (most recent call last)
<ipython-input-8-b264e31546f5> in <module>()
----> 1 Exception('something went wrong: %s' % Exception('name was: %s' % u'jörn'))

UnicodeEncodeError: 'ascii' codec can't encode character u'\xf6' in position 11: ordinal not in range(128)

this fixes the problem by using %r instead of %s, which might be a good idea in general.

[gromgull]

a SPARQLError is returned instead of raised,

this is intentional - it's a marker for a certain type of error that occured, that isn't really "exceptional"

[gromgull]

we could argue that we should use a different object as error marker though, and not one that is ALSO used as an exception

[joernhees]

yupp, it seems weird to me to return exceptions, but i already tried raising it or returning object() or None... both cause a lot of failed tests, which is why i opted to just make the exception wrapping more robust...

i think we can live with this for now and move the "don't return exceptions" part to a more boring time ;)