python 2.3.4, cx_Oracle 4.1 and utf-8 - trouble

[Serge.Orlov at gmail.com]

Maxim Kuleshov wrote:
> Hello!
>
> Trying to fetch long varchar2 column and get the following error:
>
> cx_Oracle.DatabaseError: column at array pos 0 fetched with error:
1406
>
> i.e. string buffer is not much enough to fetch the string.
>
> # fragment of code...
> myCon = cx_Oracle.connect(user, psw, dsn)
> myCur = myCon.cursor()
> myCur.execute("""
> select COLUMN from TABLE where ID=1
> """)
> for record in myCur.fetchall():
> # ...
>
> Error is reproduced only if actual string value longer than half of
> declared column size. For short strings all is ok.
>
> If I print myCur.description, I get:
>
> [('COLUMN', <type 'cx_Oracle.STRING'>, 250, 250, 0, 0, 1)]
>
> 250 - declared column max size, but I guess cx_Oracle allocates only
250
> bytes(!), so if my string longer than 125 chars (utf-8 national char
> occupies > 1 byte) - I get the error.
>
> Is it bug or what? Any suggestions?

Googling for "varchar2 utf-8" suggests that it's an Oracle's
feature (or issue). varchar2 is allocated in bytes. Finding
out the largest number of bytes per character is left as
an exercise :)

  Sergey.