Get a hex substring & convert to binary Delphi XE2

Question

My program reads from a device via a serial port and returns this string. 'IC'#$0088#$0080'Ô'#$0080#$0080 I need to get the 5 hex values and convert to binary. #$0088 = 10001000, #$0080 = 10000000, Ô = 11010100.

I can convert the 80 & 88, but am having difficulty extracting them from the whole string. The Ô(xD4) I can neither extract or convert. An extended character like the Ô could be at any or all locations.

The read methods in my serial component are:

function Read(var Buffer; Count: Integer): Integer;
function ReadStr(var Str: string; Count: Integer): Integer;
function ReadAsync(var Buffer; Count: Integer;   var AsyncPtr: PAsync): Integer;
function ReadStrAsync(var Str: Ansistring; Count: Integer;  var AsyncPtr: PAsync): Integer;

Can you give me an example of reading binary?

Answer 1

It looks like the real problem is that you are treating binary data as though it were UTF-16 encoded text.

Whatever is feeding you this data, is not feeding you UTF-16 encoded text. What the device is really feeding you is a byte array. Treat it as such rather than as text. Then you can pick out the five values you are interested in by index.

So, declare an array of bytes:

var
  Data: TArray<Byte>; // dynamic array

or

var
  Data: TBytes; // shorthand for the same

or

var
  Data: array [0..N-1] of Byte; // fixed length array

And then read into those arrays. To pick out values, use Data[i].

Note that I am using a significant amount of guesswork here, based on the question and your comments. Don't take my word for it. My guessing could be wrong. Consult the specification of the communication protocol for the device. And learn carefully the difference between text and binary.

Answer 2

As I wrote earlier in the comments, the problem with the message in your question is that it consists partly of non-ASCII characters. The ASCII range is from $00 to $7F and have the same characters as Unicode U+0000 to U+007F. Therefore no conversion (except for the leading 0). AnsiCharacters ($80 to $FF) on the other hand are subject to conversion according to the code page in use, in order to keep the same glyph for both. F.Ex. AnsiChar $80 (Euro sign in CP1252) is therefore converted to Unicode U+02C6. Bit patten for the lower byte doesn't match anymore.

Ref: https://msdn.microsoft.com/en-us/library/cc195054.aspx

Following code shows the result of two tests, Using Char vs. AnsiChar

procedure TMainForm.Button2Click(Sender: TObject);
const
  Buffer: array[0..7] of AnsiChar = ('I','C', #$88, #$80, #$D4, #$80, #$80, ';');
//  Buffer: array[0..7] of Char = ('I','C', #$88, #$80, #$D4, #$80, #$80, ';');
  BinChars: array[0..1] of Char = ('0','1');
var
  i, k: integer;
  c: AnsiChar;
//  c: Char;
  s: string;
begin
  for k := 2 to 6 do
  begin
    c := Buffer[k];
    SetLength(s, 8);
    for i := 0 to 7 do
      s[8-i] := BinChars[(ord(c) shr i) and 1];
    Memo1.Lines.Add(format('Character %d in binary format: %s',[k, s]));
  end;
end;

Using Char (UTF-16 WideChar)

AnsiChar #$88 is converted to U+02C6 
AnsiChar #$80 is converted to U+20AC 
AnsiChar #$D4 is converted to U+00D4 !

Lower byte gives

Character 2 in binary format: 11000110 
Character 3 in binary format: 10101100 
Character 4 in binary format: 11010100
Character 5 in binary format: 10101100 
Character 6 in binary format: 10101100

Using AnsiChar

Character 2 in binary format: 10001000
Character 3 in binary format: 10000000
Character 4 in binary format: 11010100
Character 5 in binary format: 10000000
Character 6 in binary format: 10000000

Unfortunately a conversion from Unicode to Ansi (even if originally converted from Ansi to Unicode) is lossy and will fail.

I really don't see any easy solution with the information available.