Lazarus 字符集转换 Utf8ToAnsi，UTF8ToWinCP，UTF8ToSys，UTF8ToConsole

Posted 2021-12-01 hieroly

由于Lazarus从1.2版开始默认字符集就是UTF8，如果要转到系统正常显示或文本保存，就必须对字符集进行转换。Lazarus提供了很多函数。如题。

那么这里面有什么关系呢？

UTF8ToSys 需要 启用编译参数 –dDisableUTF8RTL,否则还是UTF8，如果变码还是Utf8ToAnsi，但做了基本的处理。

对应代码

function UTF8ToSys(const s: string): string;
begin
  {$IFDEF UTF8_RTL}
  Result:=s;
  {$ELSE}
  if NeedRTLAnsi and (not IsASCII(s)) then
    Result:=UTF8ToAnsi(s)
  else
    Result:=s;
  {$ENDIF}
end;

Utf8ToAnsi是freePascal 自带的，变码道Ansi ，在ustringh.inc文件里面

function Utf8ToAnsi(const s : RawByteString) : RawByteString;{$ifdef SYSTEMINLINE}inline;{$endif}
  begin
    Result:=RawByteString(Utf8Decode(s));
  end;

{$ifdef FPC_HAS_CPSTRING}
  RawByteString       = type AnsiString(CP_NONE);
{$else FPC_HAS_CPSTRING}
  RawByteString       = ansistring;
{$endif FPC_HAS_CPSTRING}

UTF8ToWinCP 是Lazarus 针对Windows 功能封包，不变化代码页，仅仅变化字符编码

{$ifdef WinCe}
function UTF8ToWinCP(const s: string): string; inline;
begin
  Result := Utf8ToSys(s);
end;
{$else}
function UTF8ToWinCP(const s: string): string;
// result has codepage CP_ACP
var
  src: UnicodeString;
  len: LongInt;
begin
  Result:=s;
  if IsASCII(Result) then begin
    {$ifdef FPC_HAS_CPSTRING}
    // prevent codepage conversion magic
    SetCodePage(RawByteString(Result), CP_ACP, False);
    {$endif}
    exit;
  end;
  src:=UTF8Decode(s);
  if src=‘‘ then
    exit;
  len:=WideCharToMultiByte(CP_ACP,0,PUnicodeChar(src),length(src),nil,0,nil,nil);
  SetLength(Result,len);
  if len>0 then begin
    WideCharToMultiByte(CP_ACP,0,PUnicodeChar(src),length(src),@Result[1],length(Result),nil,nil);
    {$ifdef FPC_HAS_CPSTRING}
    // prevent codepage conversion magic
    SetCodePage(RawByteString(Result), CP_ACP, False);
    {$endif}
  end;
end;
{$endif not wince}

UTF8ToConsole 是Lazarus 针对Windows 功能封包 ，变码后，同时会配置成默认代码页文本。

{$ifdef WinCe}
function UTF8ToConsole(const s: string): string; // converts UTF8 to console string (used by Write, WriteLn)
begin
  Result := UTF8ToSys(s);
end;
{$else}
function UTF8ToConsole(const s: string): string; // converts UTF8 to console string (used by Write, WriteLn)
var
  Dst: PChar;
begin
  {$ifndef NO_CP_RTL}
  Result := UTF8ToWinCP(s);
  {$else NO_CP_RTL}
  Result := UTF8ToSys(s); // Kept for compatibility
  {$endif NO_CP_RTL}
  Dst := AllocMem((Length(Result) + 1) * SizeOf(Char));
  if CharToOEM(PChar(Result), Dst) then
    Result := StrPas(Dst);
  FreeMem(Dst);
  {$ifndef NO_CP_RTL}
  SetCodePage(RawByteString(Result), CP_OEMCP, False);
  {$endif NO_CP_RTL}
end;
{$endif not WinCE}