Unicode Console Application in Delphi 2009

Question

How can I create unicode console application with Delphi 2009?

If I do like this:

{$APPTYPE CONSOLE}
uses
  SysUtils;
begin
  writeln('öüğşç سيمانتت');
end.

I get the following:

ougsc ???????

Answer 1

You can't (At least not with the standard library). The console functions are the only non Unicode functions in Delphi 2009.

Answer 2

Windows console cannot display unicode charactes, so try to send output to file with:

my_app.exe > unicode_file.txt

and try viewing unicode_file.txt with good text editor.

Answer 3

I'm not sure that is what you're after, but you can create Unicode console applications in all 32-bit Delphi versions by using the Windows API functions. I just tried with Delphi 4:

program test;

{$APPTYPE CONSOLE}

uses
  Windows;

var
  s: WideString;
  i: integer;
  Written: Cardinal;
begin
  SetLength(s, 80);
  for i := 1 to 80 do
    s[i] := WideChar(48 + i);
  WriteConsoleW(GetStdHandle(STD_OUTPUT_HANDLE), PWideChar(s), 80, Written,
    nil);
end.

I don't have Delphi 2009 to try how entering Unicode strings in the editor works...

Answer 4

@Kabrol: With the raster font set I don't see either of the chars, but with Lucida Console I see at least widechar($11f) - "ğ".

Please see "SetConsoleOutputCP Only Effective with Unicode Fonts" and in general the description of console API functions at "Console Reference".

Answer 5

Writeln in Delphi 2009 still uses ANSI (see System TTextRec) but you can use UTF8Encode and change the console's output code page to UTF8 by calling SetConsoleOutputCP(CP_UTF8). You will also need a good font to actually display Unicode characters.

Answer 6

Actually, there is a way to do this with standard WriteLn() calls, but it involves patching a bug in the Delphi 2009 RTL. The thing is, Delphi does some compiler magic for WriteLn. For UnicodeString arguments, this results in a call to _WriteUString. That method can be seen in System.pas, although you can't call it directly. In it you'll see a call to _WriteLString, but that method receives a AnsiString argument. So when this call is taking place, your UnicodeString is being downcasted to AnsiString.

The solution is, to change this UnicodeString->AnsiString cast into a UnicodeString->UTF8String cast.

Now, when you set the console to UTF8, all your characters will go through untouched (and yes, ofcourse you'll need a font with support for the characters you want to show) :

SetConsoleOutputCP(CP_UTF8)

For this RTL fix, you'll need to do some nifty code-hooking. I've done this already, and a collegue of mine is busy writing an article about this. I'll post a link once it's available online.

Cheers!