Author: William Gerbert

I have an application that reads data from a server via winsock. The data sent are 
in Unicode format and I need to parse out the constituent strings and display in a 
ListView. They are sent as C strings so the data looks like this: array of 
chars#0array of chars#0array of chars#0#0. Since the 'array of chars' is actually 
an array of widechars it also contains #0 bytes in the msb of the character. I 
tried StringReplace(Intext, #0, '', [rfReplaceAll]); but it does not convert, maybe 
it cannot go past the first #0 in the input string?

Answer:

Yes. What you need to do here is work with PWideChars. It would have helped, of 
course, to post a bit more specific information, e.g. what the type of Intext is. 
Anyway, all you need is a way to get the address of the first widechar in the data. 
Assuming intext is a String (even though it contains widechars) the process would 
look like this:

1   procedure SplitServerWidecharList(const intext: string; list: TStrings);
2   var
3     p: PWideChar;
4   begin
5     Assert(Assigned(list));
6     list.Clear;
7     if intext <> '' then
8     begin
9       p := PWideChar(@intext[1]); {points to first widechar}
10      while p^ <> #0000 do
11      begin
12        {Convert this widestring to Ansi and store it}
13        list.add(WidecharToString(p));
14        {Find end of this widestring}
15        while p^ <> #0000 do
16          Inc(p);
17        {Hop to start of the next one  }
18        Inc(p);
19      end;
20    end;
21  end;


Can you be sure of the byte order of the received Unicode characters? The code above assumes little-endian byte order, if the data comes in in big-endian byte order you would have to swap the bytes in every widechar before you could process it as above.