JdbcodbcDriver uses the wrong charSet for interpreting the result set data

Question

Hi everyone,
I?m having trouble getting results from a MS Access database which uses Greek characters. All the characters in the range 0-127 are returned good but I get ??? for all above 127.
Here is some more details:
OS: win2k
Local setting for current user: Greek (not that it makes a difference)
Default Language setting for the System: English (Astralia). This uses charSet Cp1252 or ISO 8859-1 encoding.
Environment: JBuilder 7.0
JDK: 1.4.1
DB: MS Access
DB charSet: Cp1253 (Greek encoding or ISO 8859-7)
Connection method:
     java.util.Properties prop = new java.util.Properties();
     prop.put("user" , userName);
     prop.put("password" , password);
     //driver is "sun.jdbc.odbc.JdbcOdbcDriver";     
     //none of these work
     //prop.put("charSet", "UTF-8");
     //prop.put("charSet", "Greek");
     prop.put("charSet", "Cp1253");
     connection = DriverManager.getConnection(MY_ACCESS_DB,prop);
DataSet Access method:
     // run query and get resultSet rs here
     char[] cBuff= new char[1000];
     // I have tried getString(), getBytes(), getBinaryStream also
     Reader rReader = rs.getCharacterStream(3);
     BufferedReader fIn = new BufferedReader(rReader);
     int res = fIn.read(cBuff, 0, 999);
     // contents of cBuff is incorrect here
I have also used -Dfile.encoding=Cp1253 in command line which seems to change the default charSet of the JVM. Tested
     String en = new InputStreamReader(System.in).getEncoding();
     System.out.println("Default encoding: " + en);
By enabling trace and looking at the content of the resultSet object it shows that
rs.OdbcApi.charSet = "Cp1253"
So I'm absolutely stumped. The only possible problems I can think of is either a bug in the jdbcodbcDriver or in my native ODBC driver. The latter is less likely since I connected to the same datasouce using a different application and get the right result.
One more thing that may be helpful, if I set up the default charSet under Win2K to "Greek" (that is setting default language setting in Regional Options under Control Panel)
Everything works fine.
Is there anyone out there with answer to my problem?
Thanks in advance.

Answer 1

Tried your program and it does exactly as I expected. I get a whole lot of '?'s with my OS default language setting on English but it works fine as soon as I set it to Greek. It does not make a difference if I change the "input language" on the language bar (shift+ctrl) to Greek or not. Or if I change the charSet in your program to something else. Which bring up the question what do you use the charSet setting on your program for? If it is for byte conversion, I don't think you need it since the OS default charset is used by ResultSet.GetString and that needs to be set anyway.
Anyhow your program basically behaves exactly the same way as mine does. I am almost 100% sure that the problem is with the properties used (or not used) when creating the connections. There has got to be a (provider) property like "useUnicode" or "charSet" or something that is passed through jdbcodbc bridge to ODBC to MS Access to force it to return the result set in UTF-8 format. I confirmed this by using Visual C++ data access program that i wrote (which has the same sort of problem). All other MS Office products know about this property that why they work when i imort data from them.
This means I don't have a jdbc or even a java problem but my problem is MS Access related. So I�m going to post a question on an MS Access user group (if I find one) and see what I get). Thanks for your help with this.
By the way AOKabc crashed once and froze another time. I think it had more to do with the libraries you are using rather than your code. I have a trace log if you are interested.
Soheil