Don't encode and decode when copying nearly-unicode objects.

Why? Because BeautifulSoup's string class has a broken encode
implementation... and it appears that the unicode class is smart enough
to not reuse the object when it's a subclass anyway.

diff --git a/rawdoglib/rawdog.py b/rawdoglib/rawdog.py
index 0880cf56c458fc62cb2a50209e21c3cf68ffd7c1..988aacdef404998e07e4da1e1e70d39d0628a7f7 100644 (file)
--- a/rawdoglib/rawdog.py
+++ b/rawdoglib/rawdog.py
@@ -290,7 +290,7 @@ def ensure_unicode(value, encoding):
                # This is a subclass of unicode (e.g.  BeautifulSoup's
                # NavigableString, which is unpickleable in some versions of
                # the library), so force it to be a real unicode object.
-               return value.encode("UTF-8").decode("UTF-8")
+               return unicode(value)
        elif isinstance(value, dict):
                d = {}
                for (k, v) in value.items():