Reading UTF8 strings from a file

[Nitorami]

Hi all
The attached file is a raw file generated by LTSpice, an electronic circuit simulator. The file contains a header using UTF8 characters, followed by the binary data after the word "Binary:".

I simply need to read a fews keywords and information from the header, but I do not know how to do this efficiently, because I am unexperienced with UTF strings. With old style 8-bit ASCII strings, I would simply use readln (), but how to do this with UTF8? I tried TFileStream.ReadAnsiString but it does not work because the characters seem to be stored as raw UTF8chars without length encoding. Can someone give me a bit of starting aid please?

[engkin]

The file is UTF16, not UTF8.

[wp]

Tested this to work:

Code: Pascal  [Select][+][-]

1. procedure TForm1.Button1Click(Sender: TObject);
2. var
3.   stream: TFileStream;
4.   buffer: UnicodeString = '';
5.   search: UnicodeString = 'Binary:';
6.   p: Integer;
7.   s: String = '';                
8. begin
9.   stream := TFileStream.Create('d:\Download\test_circuit.raw', fmOpenRead);
10.   try
11.     // Read the file into a single UFT-16 string. The length of the string is
12.     // half of the file size (2 bytes per UTF-16 "character").
13.     SetLength(buffer, stream.Size div SizeOf(wideChar));
14.     stream.Read(buffer[1], stream.Size);
15.     // Find the position of the unicode word "Binary:"
16.     p := pos(search, buffer);
17.     if p > 0 then
18.       // Get the string up to the found position and convert to UTF-8...
19.       s := Copy(buffer, 1, p-1)
20.     else
21.       s := '(not found)';
22.     // ... and assign to a memo
23.     Memo1.Lines.Text := s;
24.   finally
25.     stream.Free;
26.   end;
27. end;

[Nitorami]

Thanks WP, that works.

I am surprised that the old pos() function works on widestrings as well, while readln() doesn't. I had started myself to scan through the myriads of functions in unit strutils, and found a few which are similar to pos() - e.g. "containsstr","searchbuf", but if pos() works, I will use that.

I understand I cannot read the file line by line using readln, correct ? I would have to write my own readline(), e.g. in a TFileStream descendant. The function would read the file into a buffer word by word until it finds (widestring) EOL, and then copy the buffer to a string, or stringlist. I wonder why such a function does not seem to exist already.

[SymbolicFrank]

The Text property of a Delphi-compatible TStringList would do that.