UTF-8字符串转为AnsiString

最新推荐文章于 2022-02-23 15:02:25 发布

jiljil

最新推荐文章于 2022-02-23 15:02:25 发布

阅读量618

点赞数

文章标签： buffer integer byte function c string

//江雪 2001.06.11 http://asnowcn.at.china.com
//将UTF-8字符串转为代码页为CodePage的AnsiString。
function UTF8ToAnsiString(utf8str:string; CodePage:integer):AnsiString;
var
i:integer;
buffer:widestring;
ch,c1,c2:byte;

begin
result:='';
i:=1;
while i<=Length(utf8str) do begin
   ch:=byte(utf8str[i]);
   setlength(buffer,length(buffer)+1);
   if (ch and $80)=0 then //1-byte
      buffer[length(buffer)]:=widechar(ch)
   else begin
   if (ch AND $E0) = $C0 then begin // 2-byte
      inc(i);
      c1 := byte(utf8str[i]);
      buffer[length(buffer)]:=widechar((word(ch AND $1F) SHL 6) OR (c1 AND $3F));
    end
    else begin // 3-byte
      inc(i);
      c1 := byte(utf8str[i]);
      inc(i);
      c2 := byte(utf8str[i]);
      buffer[length(buffer)]:=widechar(
        (word(ch AND $0F) SHL 12) OR
        (word(c1 AND $3F) SHL 6) OR
        (c2 AND $3F));
    end;
    end;
   inc(i);
end; //while
i := WideCharToMultiByte(codePage,
           WC_COMPOSITECHECK or WC_DISCARDNS or WC_SEPCHARS or WC_DEFAULTCHAR,
           @buffer[1], -1, nil, 0, nil, nil);
if i>1 then begin
    SetLength(Result, i-1);
    WideCharToMultiByte(codePage,
        WC_COMPOSITECHECK or WC_DISCARDNS or WC_SEPCHARS or WC_DEFAULTCHAR,
        @buffer[1], -1, @Result[1], i-1, nil, nil);
end;
end;