function UTF8Trim(const s: string; Flags: TUTF8TrimFlags): string;
var
p: PChar;
u: Cardinal;
StartP: PtrUInt;
l: Integer;
KeepAllNonASCII: boolean;
begin
Result:=s;
if Result='' then exit;
KeepAllNonASCII:=[u8tKeepControlCodes,u8tKeepNoBreakSpaces]*Flags=[u8tKeepControlCodes,u8tKeepNoBreakSpaces];
if not (u8tKeepStart in Flags) then begin
// trim start
p:=PChar(Result);
repeat
l:=1;
case p^ of
#0:
if p-PChar(Result)=length(Result) then
begin
// everything was trimmed
exit('')
end else if u8tKeepControlCodes in Flags then
break;
' ': ;
#10,#13:
if u8tKeepLineBreaks in Flags then
break;
#9:
if u8tKeepTabs in Flags then
break;
#1..#8,#11,#12,#14..#31,#127:
if u8tKeepControlCodes in Flags then
break;
#128..#255:
begin
if KeepAllNonASCII then break;
u:=UTF8CodepointToUnicode(p,l);
if (l<=1) then break; // invalid character
case u of
128..159, // C1 set of control codes
8206, 8207: // left-to-right, right-to-left mark
if u8tKeepControlCodes in Flags then break;
160, // no break space
$2007, // figure space
$2026, // narrow no-break space
$FEFF: // zero with no-break space
if u8tKeepNoBreakSpaces in Flags then break;
else
break;
end;
end;
else
break;
end;
inc(p,l);
until false;
if p>PChar(Result) then begin
Result:=copy(Result,p-PChar(Result)+1,length(Result));
if Result='' then exit;
end;
end;
if not (u8tKeepEnd in Flags) then begin
// trim end
p:=@Result[length(Result)];
repeat
case p^ of
#0:
if u8tKeepControlCodes in Flags then
break;
' ': ;
#10,#13:
if u8tKeepLineBreaks in Flags then
break;
#9:
if u8tKeepTabs in Flags then
break;
#1..#8,#11,#12,#14..#31,#127:
if u8tKeepControlCodes in Flags then
break;
#128..#255:
begin
if KeepAllNonASCII then break;
StartP:=UTF8FindNearestCharStart(PChar(Result),length(Result),p-PChar(Result));
u:=UTF8CodepointToUnicode(PChar(Result)+StartP,l);
if (l<=1) then break; // invalid character
case u of
128..159, // C1 set of control codes
8206, 8207: // left-to-right, right-to-left mark
if u8tKeepControlCodes in Flags then break;
160, // no break space
$2007, // figure space
$2026, // narrow no-break space
$FEFF: // zero with no-break space
if u8tKeepNoBreakSpaces in Flags then break;
else
break;
end;
p:=PChar(Result)+StartP;
end;
else
break;
end;
dec(p);
until p<PChar(Result);
// p is on last good byte
SetLength(Result,p+1-PChar(Result));
end;
end;