JDBC-ODBC bridge pass DBCS string as 8859-1 byte string, so when
 we get string data from DB, we can't read the data correctly. To avoid this,
 application class must change character encoding. we think this function must
 be prepared by JDBC bridge or JDBC driver.


Name: krT82822 Date: 04/14/99

=20
System Environment:

- Japanese Windows NT 4.0 Service Pack 3
- Jdk 1.2.1
- ODBC driver 3.0.28.22

Problem:

I have a testing application which will updates the MS-Access97 database
 record in Japanese. The program works (no error and warning messages), how=
ever,
only part of Japanese string was stored in the database. For example, if yo=
u want
 to update 10 japanese Kanji in record 7 of database, only first 5 Kanji up=
dated.

It seems like a string byte counting problem, but how to resolve this probl=
em?

Testing database:

Database name: JDBCTEST.mdb
Table name : JDBCTEST
Field name : CODE Data Type: Text
               NAME Data Type: Text

Testing results:

--- INPUT ---
Japanese string =3D XXXXXXXXXXXXX (Japanese Kanji characters)
String length =3D 13 (16-bit Unicode chars)

--- OUTPUT ---
Read from the database:
X : [30ec]
X : [30dd]
X : [30fc]
X : [30c8]
X : [65e5]
X : [672c]
X : [8a9e]
String length =3D 7 (16-bit Unicode chars)

Only first 7 Kanji was updated correctly and other 6 was chopped off.

-------------------------------------------
Testing program:

import java.sql.*;
import java.io.*;
import java.util.*;
import java.lang.*;

class jdbcSample {
=09public static void main(String argv[]) {
        =09Properties prop =3D new Properties();
        =09prop.put("user", "guest");
        =09prop.put("password","guest");
        =09prop.put("charSet","SJIS");

        =09PreparedStatement stmtP_U;
        =09String sPrepareSQL_U =3D "UPDATE JDBCTEST SET NAME=3D? WHERE COD=
E=3D?";
                //Japanese string=20
        =09String jp =3D "=83=8C=83|=81[=83g=93=FA=96{=8C=EA=8A=BF=8E=9A=83=
=8C=83|=81[=83g";
=09
=09=09System.out.println("\n--- INPUT ---\n");
=09=09System.out.println("Japanese string =3D " + jp);
       =09=09System.out.println("String length =3D " + jp.length() + " (1=
6-bit Unicode chars)");
 =20
=09=09//update record NUM7
=09=09/////////////////////////////////////////////////////////////////////=
////
        =09try {
=09=09=09Class.forName ("sun.jdbc.odbc.JdbcOdbcDriver");
                =09String url =3D "jdbc:odbc:jdbctest;DBQ=3DC:\\JavaTest\\J=
DBCTEST.MDB";

                        Connection con =3D DriverManager.getConnection(url,=
 prop);

                        stmtP_U =3D con.prepareStatement(sPrepareSQL_U);
=09
                        stmtP_U.setString(1,jp);
=09=09=09//next code works just same as above=20
=09=09=09//stmtP_U.setString(1,convertToSJIS(jp));

=09=09=09//update NUM7=20
                        stmtP_U.setString(2, "NUM7");
                        stmtP_U.executeUpdate();

                        stmtP_U.close();
                        con.close();
                } catch (SQLException ex) {
                        System.err.println("SQL failed.");
                        ex.printStackTrace ();
                } catch (Exception ex) {
                        ex.printStackTrace ();
                }

=09=09//read updated record from JDBCTEST.mdb
=09=09/////////////////////////////////////////////////////////////////////=
////
=09=09try {
=09=09=09Class.forName ("sun.jdbc.odbc.JdbcOdbcDriver");
                =09String url =3D "jdbc:odbc:jdbctest;DBQ=3DC:\\JavaTest\\J=
DBCTEST.MDB";

=09=09=09Connection con =3D DriverManager.getConnection(url, prop);
=09=09=09Statement stmt =3D con.createStatement ();
=09=09=09String query =3D "SELECT NAME FROM JDBCTEST where code=3D'NUM7'";

=09=09=09System.out.println("\n--- OUTPUT ---\n");
            =09=09System.out.println("Read from the database:");

            =09=09ResultSet rs =3D stmt.executeQuery(query);
            =09=09rs.next();
=09=09=09String getFromDb =3D getUnicodeString(rs, 1);
            =09=09printCharCode(getFromDb);

       =09=09=09System.out.println("String length =3D " + getFromDb.lengt=
h() + " (16-bit Unicode chars)");

            =09=09rs.close();
            =09=09stmt.close();
            =09=09con.close();
        =09} catch (SQLException ex) {
            =09=09System.err.println("SQL failed.");
            =09=09ex.printStackTrace();
        =09} catch (Exception ex) {
            =09=09ex.printStackTrace();
        =09}
    =09}

    =09static void printCharCode(String s){
        =09int hex;
        =09for (int i=3D0; i<s.length(); ++i){
            =09=09hex =3D (int)s.charAt(i);
            =09=09System.out.println(s.charAt(i) + " : [" + Integer.toHexSt=
ring(hex) + "]");
        =09}
    =09}

    =09static String getUnicodeString(ResultSet rs, int index) throws SQLEx=
ception {
        =09String ustr =3D "";
        =09String s =3D rs.getString(index);
        =09if(s =3D=3D null) {
            =09=09return null;
=09=09}
        =09try {
=09=09=09byte[] b =3D s.getBytes();
            =09=09ustr =3D new String(b, "SJIS");
        =09} catch (IOException e){
            =09=09System.err.println("IO error: columnIndex=3D"+index);
        =09}
        =09return ustr;
    =09}

        public static String convertToSJIS(String s){
                try{
                        byte[] b =3D s.getBytes();
                        return new String(b, "SJIS");
                } catch(UnsupportedEncodingException ex){
                        return s;
                }
        }
}
======================================================================

Name: krT82822 Date: 07/09/99


I cannot get DBCS character data
from Access97(Japapanses) via JDBC-ODBC Bridge.

Step1:Create Table;"CREATE TABLE TEST (C1 CHAR(8))"

Step2:Insert folowinf character data into table;
      0x88EA or 0x8DC5 or 0x8B49 (Note all data are Shift JIS)
     (\u4e00 or \u6700 or \u7d00 (Unicode)

Step3:Register Data Source (DataSourceName are "TEST")

Step4:Execute folowing Program;
      but i cannot get inserted character data.

======== START:JDBCTest.java ==============================
import java.sql.*;
public class JDBCTest {
    public static void main(String[] arsg) {
        Connection con = null;
        Statement stmt = null;
        ResultSet rs = null;
        try {
            Class.forName("sun.jdbc.odbc.JdbcOdbcDriver");
            con = DriverManager.getConnection("jdbc:odbc:TEST");
            stmt = con.createStatement();
            rs = stmt.executeQuery("SELECT * FROM TEST");
            while(rs.next())
                System.out.println(rs.getString(1));
            rs.close();
            rs = null;
            stmt.close();
            stmt = null;
            con.close();
            }
        catch(Exception e) {
            e.printStackTrace();
        }
        finally {
            try {
                if(rs != null) rs.close();
                if(stmt != null) stmt.close();
                if(con != null) con.close();
            }
            catch(Exception e) {
                e.printStackTrace();
            }
        }
    }
}

======== END:JDBCTest.java ==============================
(Review ID: 85399)
======================================================================

Name: krT82822 Date: 08/31/99


I am working for a java project involving database transaction with Japanese characters using JDBC-ODBC bridge under Windows NT.

I was originally using MS Access97, I get Japanese data from ResultSet using getstring() without any problem, given I set the charSet property properly when I call getConnection().

However, when I change to use MS Access2000, which stores text in UNICODE format. All characters above ascii 255 is decoded as '?' (ascii 63).

I have tried loading the same data using Visual Basic through ODBC, no such problem occurs! So, I believe that JDBC-ODBC bridge is not capable in loading UNICODE text data from database.
(Review ID: 94357)
======================================================================

Name: krT82822 Date: 09/06/99


The JDBC-ODBC bridge deriver has some problme to deal with chinese character.

I try to trace the JDBC-ODBC driver source file in JDK source src\share\classes\sun\jdbc\odbc\JdbcOdbc.c
I find that when the SQL command contains chinese characters, the GetByteArrayElements function get the error strings form JNIENV. If the original string conatins five chinese character, the GetByteArrayElements will retrive the string trim last five characters.

JNIEXPORT void JNICALL
Java_sun_jdbc_odbc_JdbcOdbc_execDirect (
    JNIEnv *env, jobject callingObject,
    jint hStmt,
    jbyteArray SQLString,
    jbyteArray errorCode)
{
    // Get the data structure portion of the error code
    UCHAR* errCode = (UCHAR*)(*env)->GetByteArrayElements(env, errorCode, 0);
    RETCODE rc;
    UCHAR* pSQL = (UCHAR*)(*env)->GetByteArrayElements(env,
                                   ^^^^^^^^^^^^^^^^^^^^
SQLString, 0);

    rc = SQLExecDirect (
         (HSTMT) hStmt, // Statement handle
  pSQL, // SQL Statement
SQL_NTS); // Length of SQL statement
    errCode[0] = (char) rc;
    (*env)->ReleaseByteArrayElements(env, errorCode, errCode, 0);
    (*env)->ReleaseByteArrayElements(env, SQLString, pSQL, 0);
}
(Review ID: 94914)
======================================================================

Name: krT82822 Date: 09/20/99


I can't update the the records to database if the reocord is chinese. But I
can update a part of it whole record. It seen a random.

Operating System: Chinese Windows NT 4.0 with Service Pacek 5
Database: Chinese Access 97
Connecting Type: JDBC-ODBC bridge

Result: I can update a part of the field randomly.

I hope that you can give a feedback or solution to me as soon as possible. I
need to make a decision of choosing a ASP or JSP (or Servlet) soon. I will
use the the Java to connect to a Sybase database server.
(Review ID: 95439)
======================================================================

Name: krT82822 Date: 10/18/99


In Bug-Id 4053824, there is reported that
we could not get these character like
  \u4e00 or \u6700 or \u7d00 (Unicode)
from resultSet.

This is sun.jdbc.odbc.JdbcOdbcObject#BytesToChars()'s bug,
and I fix this bug.

<<current source>>
|============================================
|//find the first null character
|for (ix=0; NotNull; ix++) {
| tempo=(byte)bTmp[ix];
| // When bTmp[ix] is \u4e00, tempo becomes 0 !!
| if (tempo==0) NotNull=false;
|}
|============================================

<<fixed source>>
|============================================
|//find the first null character
|for (ix=0; NotNull; ix++) {
| if (bTmp[ix]==0) NotNull=false;
| // Should compare bTmp[ix] as char-type !!
|}
|============================================

PLEASE CORRECT AS SOON AS POSSIBLE!!
(Review ID: 96603)
======================================================================