TDbf and Encoding

[bigeno]

I fight for a long time with encoding in dbf, please help. I can't get correct Polish characters.

I use TableLevel 4. I try LanguageID as $23,  $C8, $00 (null)
strings : MyDbf.Fields.AsString:=   as UTF8, UTF8ToCP1250, UTF8ToCP852, UTF8ToANSI

I checked all combinations and nothing, (Windows 8 PL), and open files in ESRI (arsgis) or OpenOffice.

The only progress is LanguageID:=$23 and strings as UTF8ToANSI, then in OpenOffice with opened as UTF-8 I see correct chars in data, but not in FiledsDef.

[clauslack]

What charset use your dbf files?

Regards

[BigChimp]

UTF-8 or any other unicode format isn't supported in TDBF (though you might be able to do something with raw bytes or something... I vaguely remember there is an encoding none or something).

What FPC version are you using? You might try trunk/2.7.1 as it has some fixes (you could use e.g. fpcup to do a parallel independent install).

I suppose running

Code: [Select]

chcp on the command line gives

Active code page: 852

for you? Could you check?

Language ID $23 would then make sense, yes, and if your input text is in utf8, I suppose UTF8ToCP852 or UTF8ToANSI should work...

[bigeno]

You saved me, Big thanks BigChimp.
My active code page is 852
I use fpc 2.6 fixes

Now, when I set lang to $23 and use UTF8ToCP852 then I see correct chars in Excel (just open), OpenOffice (as DOS 852), ArcGis (just open).

But not in FieldDefs, maybe I put strings in wrong way ?
for FieldDefs I use

Code: [Select]

MyDbf.FieldDefs.Add(UTF8ToCP852(s),ftString,40,false)for data I use

Code: [Select]

MyDbf.Fields[i].AsString:=UTF8ToCP852(s)When I put same polish chars in Field names and data and open in hexeditor then the hex values are different.

and other question, when I use UTF8ToAnsi then I don't get correct results. Why ? If my codepage is 852 then it should be the same ?!

[howardpc]

You have to distinguish between fieldnames and field data.
Field names are normal Pascal identifiers, and so restricted to Ansi characters 'A'..'Z', 'a'..'z' '_' and (provided it is not the first character in the name) '0'..'9'. You can have a fieldname 'girl' or 'fille', or 'dziewczyna', but not a fieldname 'Mädchen'. So your problem with fieldnames may not be one of encoding but of illegal characters in the code (characters which would be perfectly OK if they were data).
Encoding is an issue then mainly for data, not for names used in Pascal programs, which are completely inflexible in this regard (at least as far as FPC is concerned, I do not know if there are other Pascal compilers that allow non-ansi characters in identifiers).
The Lazarus editor may encode Pascal identifiers and the rest of your program code as UTF8. However, this does not matter, since the lower UTF8 character range is identical to the earlier ANSI encoding.

[bigeno]

Ok, that is not big problem, I can replace national characters with ASCII equivalent for field names.
And I think I understand the difference between UTF8ToAnsi UTF8To852, 852 is for console but Windows GUI is 1250.