I'm not positive that this is a bug.  The buit-in
compile function acepts unicode with non-ascii text in
literals:

>>> text = u"print u'''\u0442\u0435\u0441\u0442'''"
>>> exec compile(text, 's', 'exec')
тест
>>> import compiler
>>> exec compiler.compile(text, 's', 'exec')
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File
"/usr/local/python/2.3.4/lib/python2.3/compiler/pycodegen.py",
line 64, in compile
    gen.compile()
  File
"/usr/local/python/2.3.4/lib/python2.3/compiler/pycodegen.py",
line 111, in compile
    tree = self._get_tree()
  File
"/usr/local/python/2.3.4/lib/python2.3/compiler/pycodegen.py",
line 77, in _get_tree
    tree = parse(self.source, self.mode)
  File
"/usr/local/python/2.3.4/lib/python2.3/compiler/transformer.py",
line 50, in parse
    return Transformer().parsesuite(buf)
  File
"/usr/local/python/2.3.4/lib/python2.3/compiler/transformer.py",
line 120, in parsesuite
    return self.transform(parser.suite(text))
UnicodeEncodeError: 'ascii' codec can't encode
characters in position 10-13: ordinal not in range(128)
>>>