TextSerializer mistakenly calls unicode() on TEXT data

[Matt Chaput <matt@…>]

The TextSerializer.__call__() method yields the text data like this:

yield unicode(data)

I believe this is incorrect, since (a) the "data" of a TEXT event should already be Unicode, and (b) any data that IS Unicode (ie contains chars >127) will raise an error because unicode() runs its argument through the ASCII decoder.

I think it should be just yield data, or paranoid-ly:

assert isinstance(data, unicode)
yield data

[Matt Chaput <matt@…>]

Nevermind, I had accidentally "contaminated" the stream with str objects. (C'mon, Python 3000!)