Re: [pyxser-users] Pyxser generates strange 'name' attribute values

[Max S. <ms...@gm...>]

Hi, Daniel.
I've tested version from svn trunk. It segfaults now when working with
unicode dictionaries keys.

This is a console session to reproduce an error:
>>> import pyxser
>>> import inspect
>>> selector = lambda o: dict(filter(lambda (name,value): not (name.startswith('_') or callable(value)), inspect.getmembers(o)))

# not unicode key and value
>>> class T:
...     def __init__(self, d):
...             self.d = d
...
>>> t = T({'hehe':'lala'})
>>> t.d
{'hehe': 'lala'}
>>> s = pyxser.serialize(obj=t,enc='utf-8',selector=selector)
>>> print s
<?xml version="1.0" encoding="utf-8"?>
<pyxs:obj xmlns:pyxs="http://projects.coder.cl/pyxser/model/"
version="1.0" type="T" module="pylons-admin" objid="id-33308472">
  <pyxs:col type="dict" name="d">
    <pyxs:prop type="str" name="hehe">lala</pyxs:prop>
  </pyxs:col>
</pyxs:obj>

Seems ok. Now with unicode string as a key:
>>> t = T({u'хехе': 'hehe'})   # Russian cyrillic in the dict key
>>> t.d
{u'\xd1\x85\xd0\xb5\xd1\x85\xd0\xb5': 'hehe'}
>>> s = pyxser.serialize(obj=t,enc='utf-8',selector=selector)
Segmentation Fault

Oops. But it works fine with non-unicode based dictionary keys and values:

>>> t = T({'хехе': 'ололо!'}) # Russian cyrillic in the key and value
>>> t.d
{'\xd1\x85\xd0\xb5\xd1\x85\xd0\xb5':
'\xd0\xbe\xd0\xbb\xd0\xbe\xd0\xbb\xd0\xbe!'}
>>> s = pyxser.serialize(obj=t,enc='utf-8',selector=selector)
>>> print s
<?xml version="1.0" encoding="utf-8"?>
<pyxs:obj xmlns:pyxs="http://projects.coder.cl/pyxser/model/"
version="1.0" type="T" module="pylons-admin" objid="id-33308400">
  <pyxs:col type="dict" name="d">
    <pyxs:prop type="str" name="хехе">ололо!</pyxs:prop>
  </pyxs:col>
</pyxs:obj>

It works with unicode-based dictionary values too. The problem is only
with unicode keys.
Best regards, Max Sinelnikov.