Hello,

Does anyone know if there is a set of encoding conversion routines? If there isn't, how should we add one? New directory in fpc/packages?

I just needed UTF-8 to Latin 1 ISO but I can't find a cross-platform solution. Remembering that UTF8ToAnsi doesn't solve the problem because it converts to the system encoding and not Latin 1 ISO.

There do is iconv, but does it already come installed in Windows? I couldnt find any iconv.dll or something like that.

thanks,

In our previous episode, Felipe Monteiro de Carvalho said:

Does anyone know if there is a set of encoding conversion routines? If there isn't, how should we add one? New directory in fpc/packages?

Cross platform, with a library?

The lowest level is unit charset in the rtl, Unix style iconv is supported by package iconvenc.

No. Non *nix OSes have their own api. There is an abstraction, lconvencoding, but that is in the LCL afaik.

There is unit in Lazarus
lcl\LConvEncoding.pas
is that you mean?

A simpe option might be:

wides:=UTF8Decode(utf8s);
;
ansis:=wides;

Maybe set back widestringmanager.Wide2AnsiMoveProc to the original proc after that.

In our previous episode, Zaher Dirkey said:

There is unit in Lazarus lcl\LConvEncoding.pas is that you mean?

yes.

Marco van de Voort wrote

The lowest level is unit charset in the rtl, Unix style iconv is supported by package iconvenc.

The charset unit doesn't look complete or usable.

No. Non *nix OSes have their own api. There is an abstraction, lconvencoding, but that is in the LCL afaik.

In the latest lazarus svn lconvencoding has real encoding conversion tables! I don't know if they all work 100%, but I managed to strip LCL dependencies and leave only the encoding I want and the conversion from UTF-8 to Latin 1 ISO worked perfectly.

I have my own on ported from IConv in
http://www.sourceforge.net/projects/minilib

check
minilib\Unicodes\source

I used in Wince project because WinCE not support my language locale

In our previous episode, Felipe Monteiro de Carvalho said:

Afaik it is. The unit is just the main conversion, and can be plugged with units that are generated directly from unicode.org tables.

Our own cache of those tables are in ucmaps, and the tool to make units out of them is utils/creumap.pp. Examples for such units are compiler/cp*

The only problem is that it uses UTF16 internally iirc.

Maybe expanding charset with a utf16<=>utf8 conversion, and making a package with units with some common encodings would do it also.