Double byte character strings from database

[nolasco]

I am trying to read double byte character strings (DBCS) from a table in an Oracle database (via Oracle ODBC driver). I can't seem to get instances of DBString from the DBCS fields, only instances of String. Below is the sample code I am using:

| connectionSpec conn querySpec resultTable row |

connectionSpec := AbtDatabaseConnectionSpec new
dbmClass: AbtOdbcDatabaseManager;
dsn: 'FINANCE';
promptEnabled: true;
yourself.

conn := connectionSpec connectUsingAlias: 'alias'.
querySpec := AbtQuerySpec new statement: 'select * from ps_vendor where setid = ''USA'' and vendor_id = ''0000046820'''.
resultTable := conn resultTableFromQuerySpec: querySpec.
row := resultTable first.
row at: #NAME2


Inspecting the code gives me '¿¿¿¿¿¿ ¿¿¿¿¿ ¿¿¿¿¿¿¿¿¿¿' for field #NAME2 instead of the expected Cyrillic result "Клевец Лилия Николаевна".

Is there something I need to do/set to get the right field value?

[tc]

Hello,

After the string is read from the DB, can you send the message 'asByteArray' to the string and post the byte array here?

Thanks.

--tc

[nolasco]

Oops! I forgot that I have left a post on this site about a problem that still has me stumped ...

***********
tc,

The byte array is [191 191 191 191 191 191 32 191 191 191 191 191 32 191 191 191 191 191 191 191 191 191 191]. As you can see, all the Cyrillic characters are being converted to "¿" (i.e. character 191).

Thanks.

[tc]

Hello,

It seems the first issue is you will have to determine what language you are dealing with and add that code page to your abt.ini file.

I am also wondering if there is a DB issue. Even if a string does not display correctly, the bytes should be the same. The bytes would not be changed to 191 by ST for any reason I can think of. You need to make sure the DB connection has the proper NLS support. Read a string back using sql*plus to check this and/or get it working.

--tc

[nolasco]

WinSQL (using same ODBC driver as VS) returns the correct Cyrillic characters. As such, I am guessing the issue is on the VS side. Time to find the correct NLS setting.

[tc]

Hello,

From Oracle.com:

Code: Select all

select * from nls_database_parameters;

PARAMETER                      VALUE
------------------------------ ----------------------------------------
NLS_LANGUAGE                   AMERICAN
NLS_TERRITORY                  AMERICA
NLS_CURRENCY                   $
NLS_ISO_CURRENCY               AMERICA
NLS_NUMERIC_CHARACTERS         .,
NLS_CHARACTERSET               US7ASCII
NLS_CALENDAR                   GREGORIAN
NLS_DATE_FORMAT                DD-MON-RR
NLS_DATE_LANGUAGE              AMERICAN
NLS_SORT                       BINARY
NLS_TIME_FORMAT                HH.MI.SSXFF AM
NLS_TIMESTAMP_FORMAT           DD-MON-RR HH.MI.SSXFF AM
NLS_TIME_TZ_FORMAT             HH.MI.SSXFF AM TZH:TZM
NLS_TIMESTAMP_TZ_FORMAT        DD-MON-RR HH.MI.SSXFF AM TZH:TZM
NLS_DUAL_CURRENCY              $
NLS_COMP                       BINARY
NLS_NCHAR_CHARACTERSET         US7ASCII
NLS_RDBMS_VERSION              8.1.7.0.0

. . . the NCHAR characterset in the database should be something that supports your other
characterset.

--tc

[nolasco]

We have it set as AL16UTF16, which should cover Cyrillic.

WinSQL returns the correct characters. But, Oracle's own SQLPlus, whether in Windows or Solaris, seems to return the same '¿¿¿¿¿¿ ¿¿¿¿¿ ¿¿¿¿¿¿¿¿¿¿' result as VA Smalltalk. As such, I am now even more confused.

The search for a solution continues ...

[benvandijk]

Hi,

I guess you are running sqlplus on a client machine, not on the database server.
In that case you should check the NLS_LANG settings on the ciient in the windows registry.

For more information on charactersets you can look at the oracle dosumentation: "Oracle9i Database Globalization Support Guide"

I am not familiar with solaris, but i assume there will be some client settings too.

Greetz, Ben.