PROBLEM WITH GREEK CHARACTERS
lazarus / 16-Databases

----------------------------------------------------------------------
netapproach · 2012-05-18 11:19

I just install lazarus last version(0.9.30.4) in a Windows XP SP3
system
,and I make my first connection to an mdb file.
It work just fine , but when I try to display data
inside a grid or in a TDBEdit greek chars are not display propertly.

I change the font to arial and the charset to GREEK_CHARSET but
the problem exists again.

What am I make wrong?????
Please help 

See a screenshot here:
http://www.netapproach.gr/LAZARUS_GREEK.jpg

----------------------------------------------------------------------
ludob · 2012-05-19 14:35

You're not doing anything wrong. odbconnection uses the ansi odbc api
calls instead of the unicode api calls. As a result the driver manager
is converting everything to your system code page and lazarus expects
utf8. If you are using a greek windows character set you can do the
following in the IDE:
- click on the TSQLQuery that is providing your dataset
- if FieldDefs says zero items, set active to true to fill in the
FieldDefs (the connection needs to be configured correctly)
- double click on the TSQLQuery. The Edit fields window will pop up.
Click on + and select all fields. Click on create.
- click on a field that contains text data.
- in the object inspector, double click on OnGetText to create a new
handler and complete with the following code:
Quoteprocedure TForm1.SQLQuery1yourfieldGetText(Sender: TField; var
aText: string;
  DisplayText: Boolean);
begin
  aText:=UTF8Encode(sender.AsWideString);
end;
- if there are more text fields, select in the field editor the other
fields and set their OnGetText to the same handler

When you want to save data back to the access database you'll need to
do the inverse conversion in the fields OnSetText event handler.

----------------------------------------------------------------------
Zoran · 2012-05-21 13:22

Quote from: ludob on May 19, 2012, 04:35:37 pmYou're not doing
anything wrong. odbconnection uses the ansi odbc api calls instead of
the unicode api calls.

I have the similar problem, but I cannot use the proposed solution,
because I have characters which cannot be represented in system code
page. Is there any workaround for that?

Isn't there a way to tell odbc not to do conversion... Or, at least,
can't I somehow tell odbc which ansi code page (but not what is set in
windows) to use?

----------------------------------------------------------------------
ludob · 2012-05-21 13:54

QuoteIsn't there a way to tell odbc not to do conversion... Or, at
least, can't I somehow tell odbc which ansi code page (but not what is
set in windows) to use?The problem is that the MS odbc driver manager,
the layer between the app and driver, reports all string fields as
SQL_VARCHAR if the app uses the ansi ODBC api calls. Only when using
the unicode ODBC api (functions end with 'W') the driver manager
reports fields as SQL_WVARCHAR. The sqldb odbcconnection uses only the
ansi calls. I'm updating odbcconnection now to use also the unicode
api.  I'm in the middle of testing it and so far it looks good.
Chinese and greek characters can be read and written to a MS Access
db. There is still a need for the gettext/settext workaround because
fpc will convert to the system encoding and not utf8 on windows.
I'll update this thread when I upload the patch to mantis.

----------------------------------------------------------------------
netapproach · 2012-05-21 14:53

Thanks for the solution , this works just fine!

----------------------------------------------------------------------
ludob · 2012-05-21 15:18

QuoteI'll update this thread when I upload the patch to mantis.
The patch is attached to http://bugs.freepascal.org/view.php?id=22095

To transform the unicode to utf8 use the following GetText and SetText
event handers:
Code: [Select]procedure TForm1.SQLQuery1MyUnicodeFieldGetText(Sender:
TField; var aText: string;
  DisplayText: Boolean);
begin
  aText:=UTF8Encode(sender.AsWideString);
end;

procedure TForm1.SQLQuery1MyUnicodeFieldSetText(Sender: TField; const
aText: string);
begin
  sender.AsWideString:=UTF8Decode(aText);
end;

----------------------------------------------------------------------
Zoran · 2012-05-22 09:14

Quote from: ludob on May 21, 2012, 05:18:49 pmQuoteI'll update this
thread when I upload the patch to mantis.
The patch is attached to http://bugs.freepascal.org/view.php?id=22095

To transform the unicode to utf8 use the following GetText and SetText
event handers:
Code: [Select]procedure TForm1.SQLQuery1MyUnicodeFieldGetText(Sender:
TField; var aText: string;
  DisplayText: Boolean);
begin
  aText:=UTF8Encode(sender.AsWideString);
end;

procedure TForm1.SQLQuery1MyUnicodeFieldSetText(Sender: TField; const
aText: string);
begin
  sender.AsWideString:=UTF8Decode(aText);
end;


I will install fpc from trunk and apply this patch, then try the new
functionality.
Thank you very much, Ludob!