UnicodeConversion (Delphi)

Description

This example converts various string formats to and from Unicode. This example requires two text edits and a combo box.

Code

procedure TForm1.Button1Click(Sender: TObject);
var
  LBuffer: TBytes;
  LByteOrderMark: TBytes;
  LFileStream: TFileStream;
  LOffset: Integer;
  LEncoding, DestEncoding: TEncoding;
  ComboIndex: Integer;
  ByteString: RawByteString;
  FileText: string;
  ucs4Text: UCS4String;
  wideStringText: WideString;
  utf8Text: UTF8String;
  L: Cardinal;

begin
  LEncoding:= nil;
  ComboIndex := ComboBox1.ItemIndex;
  LFileStream := TFileStream.Create(Edit1.Text, fmOpenRead);
  try
    // Read the file into buffer.
    SetLength(LBuffer, LFileStream.Size);
    LFileStream.ReadBuffer(Pointer(LBuffer)^, Length(LBuffer));
    // Get data encoding of read data.
    LOffset := TEncoding.GetBufferEncoding(LBuffer, LEncoding);
  finally
    LFileStream.Free;
  end;

  // Process various types.
  case ComboIndex of

    // AnsiToUtf8
    0:
    begin
      // Identify the encoding and convert buffer to Ansi.
      DestEncoding := TEncoding.ASCII;
      LBuffer := LEncoding.Convert(LEncoding, DestEncoding, LBuffer,
        LOffset, Length(LBuffer) - LOffset);
      LOffset := TEncoding.GetBufferEncoding(LBuffer, DestEncoding);
      FileText := DestEncoding.GetString(LBuffer, LOffset, Length(LBuffer) - LOffset);
      ByteString := AnsiToUtf8(FileText);
      ShowMessage(UTF8ToString(ByteString));

      // Write to file.
      LFileStream := TFileStream.Create(Edit2.Text, fmCreate);
      try
        // New destination encoding
        DestEncoding := TEncoding.UTF8;
        LByteOrderMark := DestEncoding.GetPreamble;
        LFileStream.Write(LByteOrderMark[0], Length(LByteOrderMark));
        LFileStream.Write(ByteString[1], Length(ByteString));
      finally
        LFileStream.Free;
      end;
    end;

    // UCS4StringToWideString
    1:
    begin
      // Identify encoding and convert buffer to UCS4.
      DestEncoding := TEncoding.Unicode;
      LBuffer := LEncoding.Convert(LEncoding, DestEncoding, LBuffer,
        LOffset, Length(LBuffer) - LOffset);
      LOffset := TEncoding.GetBufferEncoding(LBuffer, DestEncoding);
      FileText := DestEncoding.GetString(LBuffer, LOffset, Length(LBuffer) - LOffset);
      ucs4Text := UnicodeStringToUCS4String(FileText);
      wideStringText := UCS4StringToWideString(ucs4Text);
      ShowMessage(wideStringText);

      // Write to file.
      LFileStream := TFileStream.Create(Edit2.Text, fmCreate);
      try
        // New destination encoding
//        DestEncoding := TEncoding.UTF8;
        LByteOrderMark := DestEncoding.GetPreamble;
        LFileStream.Write(LByteOrderMark[0], Length(LByteOrderMark));
        LFileStream.Write(wideStringText[1], Length(wideStringText) * SizeOf(Char));
      finally
        LFileStream.Free;
      end;
    end;

    // UnicodeToUtf8
    2:
    begin
      // Identify the encoding and convert buffer to Unicode.
      DestEncoding := TEncoding.Unicode;
      LBuffer := LEncoding.Convert(LEncoding, DestEncoding, LBuffer,
        LOffset, Length(LBuffer) - LOffset);
      LOffset := TEncoding.GetBufferEncoding(LBuffer, DestEncoding);
      FileText := DestEncoding.GetString(LBuffer, LOffset, Length(LBuffer) - LOffset);
      L := Length(FileText);
      SetLength(utf8Text, L * SizeOf(Char) + 1);
      L := UnicodeToUtf8(PAnsiChar(utf8Text), Length(utf8Text), PWideChar(FileText), L);
      ShowMessage(UTF8ToString(utf8Text));

      // Write to file.
      LFileStream := TFileStream.Create(Edit2.Text, fmCreate);
      try
        // New destination encoding
        DestEncoding := TEncoding.UTF8;
        LByteOrderMark := DestEncoding.GetPreamble;
        LFileStream.Write(LByteOrderMark[0], Length(LByteOrderMark));
        LFileStream.Write(utf8Text[1], L);
      finally
        LFileStream.Free;
      end;
    end;
  end;
end;

procedure TForm1.FormCreate(Sender: TObject);
var
  StringList: TStrings;
begin
  Edit1.Text := 'Sample.txt';
  Edit2.Text := 'Dest.txt';
  StringList := TStringList.Create;
  try
    with StringList do begin
      Add('AnsiToUtf8');
      Add('UCS4StringToWideString');
      Add('UnicodeToUtf8');
    end;
    with ComboBox1 do begin
      Items.Assign(StringList);
      ItemIndex := 0;
    end;
  finally
    StringList.free;
  end;
end;

Uses

• System.AnsiToUtf8 ( fr | de | ja )
• System.UCS4StringToWideString ( fr | de | ja )
• System.UnicodeToUtf8 ( fr | de | ja )
• SysUtils.TEncoding.GetString ( fr | de | ja )