Pyhon with Oracle encoding issue

If the database content is correct, then it does not use cp1252 encoding. The character 0x8f is not defined -- see here https://en.wikipedia.org/wiki/Windows-1252 or here http://www.cp1252.com/ Firstly, I suggest to double-check, whether the database really uses that encoding.

Another possibility is that the database should contain the cp1252, but there is a bug in the data.

If you need to solve the situation, you can wrap the decoding of the character the try-except construct, replace the character by something or very visible (to be able to see the problem) or by something very invisible (to mask the problem). You can also write some warning into a log in the case.

It is difficult to say without seeing something more.

I appears that Python uses that code file, cp1252.py, by default. Can I change that?

Is the database column od the unicode type (nvarchar or so)?

Only VARCHAR2.

If that character is in the database I'm betting the database is Unicode.
select value from nls_database_parameters where parameter='NLS_CHARACTERSET';

If you cannot get Python to handle that characterset, you might look at encoding the column.  I've not done a lot with multi-byte character sets so I'm not sure the "official" Oracle way.

Off the top of my head there is utl_i18n.escape_reference:
http://docs.oracle.com/cd/E11882_01/appdev.112/e40758/u_i18n.htm#ARPLS71120

If it is VARCHAR2, then you probably have to tell the encoding explicitly somehow. The cp1252 is probably a default. However, I do not know the library. I guess it will be or the connection object parameter or the cursor parameter. The only other possibility is that the cursor.execute is also capable to return the binary type in the row (if possible, then it should be set similarly to the encoding). I do not know the answer.

The solution was suggested by a co-worker of mine and I found reference to that NLS_LANG windows key on the net.