Is this string function efficient?

[AlexTP]

Is this func (it copies utf8 string, pure ascii, to Widestring) ok in CPU code?
or need to optimize using PChar/PWChar?

Code: Pascal  [Select][+][-]

1. function SConvertUtf8ToWideForAscii(const S: string): UnicodeString;
2. var
3.   i: integer;
4. begin
5.   SetLength(Result, Length(S));
6.   for i:= 1 to Length(S) do
7.     Result[i]:= WideChar(Ord(S[i]));
8. end;
9.  

[Thaddy]

Where's the UTF8 in your "string"?  I assume you are using Lazarus?
But anyway:

Code: Pascal  [Select][+][-]

1. program untitled;
2. {$ifdef fpc}{$mode delphi}{$H+}{$I-}{$endif}
3. var
4.   s1:UTF8String = 'Whatever'; // or a Lazarus "string"...Juha...  FPC strings are either ShortString, AnsiString or UnicodeString, never UTF8: that's only the case in Lazarus. Confused???
5.   s2:Unicodestring;
6. begin
7.   s2 := S1;  // conversion is lossless and automatic and fast. Faster than you can  do by hand...
8.   writeln(s2);
9. end.

So it is a simple assignment. And your code is wrong anyway: You should have used UnicodeChar.... And the length of an UTF8 string may differ from the length of a UnicodeString... (the latter is a bug in your code)
But as I said: just a simple assignment will do.

[AlexTP]

yes, I use Lazarus. and "string" var holds raw lines of text. Usually i call WS:= UTF8Decode(s).
for pure ASCII, i want faster call, and so SConvertUtf8ToWideForAscii.
I think simple assignment is slower than SConvertUtf8ToWideForAscii?

[mse]

In my experience

Code: Pascal  [Select][+][-]

1. type
2.  card8 = byte;
3.  pcard8 = ^card8;
4.  card16 = word;
5.  pcard16 = ^card16;
6.  
7. function SConvertUtf8ToWideForAscii(const S: string): UnicodeString;
8. var
9.  ps,pe: pcard8;
10.  pd: pcard16;
11.  i1: sizeint;
12. begin
13.  i1:= Length(S);
14.  SetLength(Result,i1);
15.  ps:= pointer(s);
16.  pe:= ps + i1;
17.  pd:= pointer(result);
18.  while ps < pe do begin
19.   pd^:= ps^;
20.   inc(ps);
21.   inc(pd);
22.  end;
23. end;
24.  

is faster.

[Thaddy]

yes, I use Lazarus. and "string" var holds raw lines of text. Usually i call WS:= UTF8Decode(s).
for pure ASCII, i want faster call, and so SConvertUtf8ToWideForAscii.
I think simple assignment is slower than SConvertUtf8ToWideForAscii?

No it is not. I just timed it. A simple assignment is faster by about 20% (mse: take note)
And YOU could have timed it yourself. And solve the bug, btw.. Length(UTF8) is different from Length(Unicode) ......

[mse]

No it is not. I just timed it. A simple assignment is faster by about 20% (mse: take note)

Did you test my solution?

[Thaddy]

No need to test, you just fell in the same trap: the length(UTF8) <> Length(UnicodeString)....unless in common translations and by accident.
And you know that! that's even worse...

[mse]

For ASCII length(utf8) = length(utf16).

[Thaddy]

For ASCII length(utf8) = length(utf16).

Just for ASCII and ONLY for ASCII... You know that...

The question is about UTF8string (After I asked) to UnicodeString, which was not obvious.... If you do that in a generic function like yours or OP's you will run into trouble. It is plain wrong.
The simple assignment is fool proof and takes all into account.

[mse]

Please read the first post from Alextp.

[Thaddy]

Please read the second post from Alexp..... I am not spending any more time on this. Your answer is wrong (not complete and you use pointers) Pure ASCII? Which codepage?

[AlexTP]

I use func only for ASCII text. before in code it is checking: if Length(UTF8Decode(S))=Length(S) then item marked as ASCII.

@mse
thanks...

question was only about ASCII string.

[Thaddy]

You should not thank mse.  Even pure ASCII will run you into trouble. Use the simple assignment.

[mse]

@mse
thanks...

Any numbers on speed comparison?

[AlexTP]

@Thaddy
Simple assignment will try to decode UTF8. ie parse it. @mse's func don't parse it, it's faster?