Codierung/Decodierung von Umlauten

[Sir Rufo]

Oder man klimpert sich das eben fix zusammen

Delphi-Quellcode:

unit JavaString;

interface

function JavaString_Decode( const AStr: string ): string;

implementation

uses
  System.SysUtils;

function OctalStrToInt( AOctal: string ): Integer;
var
  LChar: Char;
begin
  Result := 0;
  for LChar in AOctal do
    begin
      Result := Result * 8 + Ord( LChar ) - Ord( '0' );
    end;
end;

function JavaString_Decode( const AStr: string ): string;
var
  LTempBuffer: string;
  LChar: Char;
  LState: Integer;
  procedure Normal;
  begin
    case LChar of
      '\':
        LState := 1;
    else
      Result := Result + LChar;
    end;
  end;
  procedure Escape;
  begin
    case LChar of
      '\', '/', '"', '''':
        ;
      'b': // Backspace
        LChar := #$0008;
      't': // Horizontal Tab
        LChar := #$0009;
      'n': // New Line
        LChar := #$000A;
      'v': // Vertical Tab
        LChar := #$000B;
      'f': // Form Feed
        LChar := #$000C;
      'r': // Carriage Return
        LChar := #$000D;
      '0' .. '3': // Latin1 Octal
        begin
          LTempBuffer := LChar;
          LState := 2;
          Exit;
        end;
      'x': // Latin1 Hex
        begin
          LTempBuffer := '';
          LState := 3;
          Exit;
        end;
      'u': // Unicode Hex
        begin
          LTempBuffer := '';
          LState := 4;
          Exit;
        end;
    else
      raise EArgumentException.Create( 'Parse Error' );
    end;
    Result := Result + LChar;
    LState := 0;
  end;
  procedure Latin1Octal;
  var
    LByte: Byte;
  begin
    case LChar of
      '0' .. '7':
        LTempBuffer := LTempBuffer + LChar;
    else
      raise EArgumentException.Create( 'Parse Error' );
    end;
    if Length( LTempBuffer ) = 3
    then
      begin
        LByte := OctalStrToInt( LTempBuffer );
        Result := Result + TEncoding.Default.GetString( TArray<Byte>.Create( LByte ) );
        LState := 0;
      end;
  end;
  procedure Latin1Hex;
  var
    LByte: Byte;
  begin
    case LChar of
      '0' .. '9', 'A' .. 'F', 'a' .. 'f':
        LTempBuffer := LTempBuffer + LChar;
    else
      raise EArgumentException.Create( 'Parse Error' );
    end;
    if Length( LTempBuffer ) = 2
    then
      begin
        LByte := StrToInt( '$' + LTempBuffer );
        Result := Result + TEncoding.Default.GetString( TArray<Byte>.Create( LByte ) );
        LState := 0;
      end;
  end;
  procedure UnicodeHex;
  var
    LWord: Word;
  begin
    case LChar of
      '0' .. '9', 'A' .. 'F', 'a' .. 'f':
        LTempBuffer := LTempBuffer + LChar;
    else
      raise EArgumentException.Create( 'Parse Error' );
    end;
    if Length( LTempBuffer ) = 4
    then
      begin
        LWord := StrToInt( '$' + LTempBuffer );
        Result := Result + TEncoding.Unicode.GetString( TArray<Byte>.Create( Lo( LWord ), Hi( LWord ) ) );
        LState := 0;
      end;
  end;

begin
  Result := '';
  LState := 0;
  for LChar in AStr do
    case LState of
      0:
        Normal;
      1:
        Escape;
      2:
        Latin1Octal;
      3:
        Latin1Hex;
      4:
        UnicodeHex;
    end;
  if LState <> 0
  then
    raise EArgumentException.Create( 'Parse Error' );
end;

end.

Delphi-Quellcode:

procedure Main;
begin
  Writeln( JavaString_Decode( 'Copyright \251 \xA9 \u00A9' ) );
end;

ergibt dann

Code:

Copyright © © ©