Tuesday, 26 February 2008

Allow Chinese characters on a Client-Server application

I recently had to provide a working solution for Traditional Chinese for Hongkong.

It involved both ODBC on the Linux server to read correct data from a Progress database, and process that data in .net on Mono, but also on the client side to be able to read data through ODBC.

Here is the solution and also some warnings about what can go wrong:

For Mono, you need to have installed the RPM package mono-locale-extras, which installs the dll I18N.CJK.dll (CJK stands for Chinese Japanese Korean).

Make sure the Progress database is in codepage big-5.
proutil mydb -C convchar convert undefined -G 15
proutil mydb -C convchar convert big-5 -G 15
_progres -pf batch.pf -1 -b -rx -p upgrade.p -param $DLC/prolang/tch/big-5/_tran.df | cat
proutil mydb -C idxbuild all

Make sure that your pf files (or even the Progress startup.pf) contain these lines:
-cpinternal big-5
-cpstream big-5
-cprcodein undefined

For the Progress server, you need to set this variable before the server is started (note: if you have both sql and 4gl server, the first one needs all settings already):
export SQL_CLIENT_CHARSET=BIG-5

Mono would rather expect BIG5 (without the dash), so we need to set another variable so that the default encoding is correct.
Before you start the server in the Mono environment, make sure that you export this variable:
export LANG=zh_TW.big5

The following command shows what languages are available on your Linux:
locale -a
Unfortunately there is no big5 for Hongkong, and zh_HK.big5hkscs does not seem to work for me. zh_TW.big5 works fine.

This code converts the data from the codepage coming through ODBC from the Progress database into Unicode:

FDBEncoding := System.Text.Encoding.Default;
try
sqlClientCodePage := Convert.ToInt32(System.Environment.GetEnvironmentVariable('SQL_CLIENT_CHARSET'));
FDBEncoding := System.Text.Encoding.GetEncoding(sqlClientCodePage);
except on exp: Exception do
end;

function ConvertToUnicode(s: System.String): System.String;
var
bytes: array of byte;
begin
bytes := Encoding.Default.GetBytes(s);
result := FDBEncoding.GetChars(bytes);
end;

function ConvertFromUnicode(s: System.String): System.String;
var
bytes: array of byte;
begin
bytes := FDBEncoding.GetBytes(s);
result := Encoding.Default.GetChars(bytes);
end;

For the client to be able to access through ODBC, we also need to set the environment variable in Windows XP SQL_CLIENT_CHARSET=BIG-5
(Start/Settings/Control Panel/System/Advanced/Environment Variables/System or User Variables)

No comments: