Überlauf bei int64

[dummzeuch]

Ich habe das mal Claude vorgelegt und der hat noch einen Fehler gefunden:

Zitat:

The most significant error is in the first multiplier: 16777246. This number appears to be incorrect. For converting IP addresses to integers, the correct multiplier should be 16777216 (which is 256³). The current value of 16777246 is off by 30, which can lead to incorrect results.

Und er gibt noch folgende Ratschläge:

Zitat:

To make the function more robust, you might also want to:

• Add range checking to ensure each octet is between 0 and 255
• Add error handling for malformed IP addresses
• Consider using Cardinal or UInt32 instead of Int64 since IPv4 addresses fit in 32 bits

Alles durchaus sinnvoll.

Das eigentliche Problem hat er allerdings in seinem "korrigierten" Code nicht gelöst:

Delphi-Quellcode:

function IP2Int(value: string): Int64;
var
  s, v1, v2, v3, v4: string;
  i1, i2, i3, i4: Integer;
begin
  Result := 0;
  s := value;
  
  if Pos('.', s) > 0 then
  begin
    v1 := Copy(s, 1, pos('.', s)-1);
    s := Copy(s, Pos('.', s)+1, Length(s));
    if Pos('.', s) > 0 then
    begin
      v2 := Copy(s, 1, pos('.', s)-1);
      s := Copy(s, Pos('.', s)+1, Length(s));
      if Pos('.', s) > 0 then
      begin
        v3 := Copy(s, 1, pos('.', s)-1);
        v4 := Copy(s, Pos('.', s)+1, Length(s));
      end;
    end;
  end;
  
  if TryStrToInt(v1, i1) then
    if TryStrToInt(v2, i2) then
      if TryStrToInt(v3, i3) then
        if TryStrToInt(v4, i4) then
        begin
          Result := i1 * 16777216 +  // Corrected from 16777246
                   i2 * 65536 +
                   i3 * 256 +
                   i4;
        end;
end;

Der Integer Overflow dürfte immer noch auftreten.