Java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv

Question

Hi all,
I am writing a servlet that connects to Oracle 8.0.6 through jdbc for jdk1.2 on NT 4.0
English version and it works fine.
But when the servlet is deployed to a solaris with Oracle 8.0.5 (not a typo, the oracle on
NT is 8.0.6 and oracle on solaris is 8.0.5) and jdbc for jdk1.2 (of course, for Solaris),
the servlet failed with the Exception:
java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv
(I am using JRun 3.0 as the application and web server for both NT and Solaris)
(The database in both the NT and solaris platform are using UTF8 charset)
My servlet looks like this: (dbConn is a Connection object proved to be connected to Oracle
in previous segment of the same method):
String strSQL = "SELECT * FROM test";
try { Statement stmt = dbConn.createStatement();
ResultSet rs = stmt.execute(strSQL);
while (rs.next()) {
out.println("id = " + rs.getInt("id"));
System.out.println("id written");
out.println("name = " + rs.getString("name")); // <-- this is the line the
exception is thrown
System.out.println("name written");
} catch (java.sql.SQLException e) {
System.out.println("SQL Exception");
System.out.println(e);
The definition of the "test" table is:
create table test(
id number(10,0),
name varchar2(30));
There are about 10 rows exists in the table "test", in which all rows contains ONLY chinese
characters in the "name" field.
And when I view the System log, the string "id written" is shown EXACTLY ONCE and then there
is:
SQL Exception
java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv
That means the resultset is fetch back from the database correctly. The problem arise only
during the getString("name") method.
Again, this problem only happens when the servlet is run on the solaris platform.
At first I would expect there are some strange code shown on the web page rather than having
an exception. I know that I should use getBytes to convert between different encodings, but
that's another story.
One more piece of information: When all the rows contains ascii characters in their "name"
field, the servlet works perfectly even in solaris.
If anyone knows why and how to tackle the problem please let me know. You can feel free to
send email to me at [email protected]
Many thanks,
Ben
null

Answer 1

Hi all,
For the problem I previously posted, I found that Oracle had had such bug filed before in Oracle 7.3.2 (something like that) and is classified to be NOT A BUG.
A further research leads me to the document of Oracle that the error message:
"java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv"
is a JDBC driver error message of error number ORA-17037.
I'm still wondering why this behaviour will happen only in Solaris platform. The servlet on an NT machine I am using (which has an Oracle 8.0.6 and jdbc for jdk 1.2 running) is working just fine. I also suspect that this may be some sort of mistakes from jdbc driver.
Nevertheless, I have found a way to work around the problem that I cannot get non-English string from Oracle in Solaris and I would like to share it with you all here.
Before I go on, I found that there are many people out there on the web that encounter the same problem. (Some of which said s/he has been working on this problem for a month). As a result, if you find this way of working around the problem does help you, please tell those who have the same problem but don't know how to tackle. Thanks very much.
Here's the way I work it out. It's kinda simple, but it does work:
Instead of using:
String abc = rs.getString("SomeColumnContainsNonEnglishCharacters");
I used this:
String abc = new String(rs.getBytes("SomeColumnContainsNonEnglishCharacters"));
This will give you a string WITH YOUR DEFAULT CHARSET (or ENCODING) from your system.
If you want to convert the string read to some other encoding type, say Big5, you can do it like this:
String abc = new String(rs.getBytes("SomeColumneContainsNonEnglishCharacters"), "BIG5");
Again, it's simple, but it works.
Finally, if anyone knows why the fail to convert problem happens, please kindly let me know by leaving a word in [email protected]
Again, thanks to those of you who had tried to help me out.
Creambun
<BLOCKQUOTE><font size="1" face="Verdana, Arial">quote:</font><HR>Originally posted by creambun creambun ([email protected]):
Hi all,
I am writing a servlet that connects to Oracle 8.0.6 through jdbc for jdk1.2 on NT 4.0
English version and it works fine.
But when the servlet is deployed to a solaris with Oracle 8.0.5 (not a typo, the oracle on
NT is 8.0.6 and oracle on solaris is 8.0.5) and jdbc for jdk1.2 (of course, for Solaris),
the servlet failed with the Exception:
java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv
(I am using JRun 3.0 as the application and web server for both NT and Solaris)
(The database in both the NT and solaris platform are using UTF8 charset)
My servlet looks like this: (dbConn is a Connection object proved to be connected to Oracle
in previous segment of the same method):
String strSQL = "SELECT * FROM test";
try { Statement stmt = dbConn.createStatement();
ResultSet rs = stmt.execute(strSQL);
while (rs.next()) {
out.println("id = " + rs.getInt("id"));
System.out.println("id written");
out.println("name = " + rs.getString("name")); // <-- this is the line the
exception is thrown
System.out.println("name written");
} catch (java.sql.SQLException e) {
System.out.println("SQL Exception");
System.out.println(e);
The definition of the "test" table is:
create table test(
id number(10,0),
name varchar2(30));
There are about 10 rows exists in the table "test", in which all rows contains ONLY chinese
characters in the "name" field.
And when I view the System log, the string "id written" is shown EXACTLY ONCE and then there
is:
SQL Exception
java.sql.SQLException: Fail to convert between UTF8 and UCS2: failUTF8Conv
That means the resultset is fetch back from the database correctly. The problem arise only
during the getString("name") method.
Again, this problem only happens when the servlet is run on the solaris platform.
At first I would expect there are some strange code shown on the web page rather than having
an exception. I know that I should use getBytes to convert between different encodings, but
that's another story.
One more piece of information: When all the rows contains ascii characters in their "name"
field, the servlet works perfectly even in solaris.
If anyone knows why and how to tackle the problem please let me know. You can feel free to
send email to me at [email protected]
Many thanks,
Ben<HR></BLOCKQUOTE>
null