Localize file www.TrSek.com/zadania/winutf.pas

{ WINUTF.PAS                Copyright (c) TrSek alias Zdeno Sekerak }
{ Program konvertuje textovy subor kodovany v znakovej sade Win1250 }
{ (Latin2), alebo UTF do suboru v UNICODE.                          }
{                                                                   }
{ Datum:06.06.2005                             http://www.trsek.com }

program Win1250_UTF_UNICODE;
uses crt,dos;

const WIN_1250 = 'WIN1250.txt';    { prevodna tabulka z Win1250, Latin2 }
      WIN_UTF8 = 'UTF8.txt';       { prevodna tabulka z UTF-8  }
      VST_TXT1 = 'v_W1250.txt';    { kodovany text vo Win-1250 }
      VST_TXT2 = 'v_UTF8.txt';     { kodovany text v  UTF-8    }
      VYS_TXT  = 'Unicode.txt';    { kodovany text v  Unicode  }

var  fin:file of byte;
    fout:file of word;
     win:array[0..255] of word;
     utf:array[0..255] of word;
   vstup:string;
  choice:byte;
      ch:byte;
       w:word;

{ prekonvertuje Hex do Word tak ze zavola val }
function HexToWord(s:string):word;
var hex:word;
    err:integer;
begin
  Val('$'+s,hex,err);
  HexToWord := hex;
end;

{ precitame definicny subor pre kodovanie win1250, UTF-8 -> Unicode }
procedure LoadCodePage(sour:string; var win:array of word);
var i:byte;
   fw:text;
  pom:string;
begin
  Assign(fw, sour);
  ReSet(fw);

  for i:=0 to 255 do begin
     readln(fw,pom);
     win[i]:=HexToWord(pom);
  end;

  Close(fw);
end;

{ najde v UTF-8 sade }
function NajdiUTF(w:word):byte;
var i:byte;
begin
  NajdiUTF:=0;

  for i:=0 to 255 do begin
    if( utf[i]=w )then
        NajdiUTF:=i;
  end;
end;

begin
  Writeln('Program konvertuje textovy subor zo sady Win1250 (Latin2), UTF do Unicode.');
  WriteLn('Pozadovane znakove sady su definovane v suboroch ', WIN_1250, ' a ', WIN_UTF8, '.');

  WriteLn;
  WriteLn('Vyber:');
  WriteLn('1. Win1250 -> Unicode  (', VST_TXT1,')');
  WriteLn('2. UTF-8   -> Unicode  (', VST_TXT2,')');
  ReadLn(choice);

  { meno suboru }
  Write('Zadaj meno suboru:');
  ReadLn(vstup);

  { Nacita znakove sady }
  LoadCodePage(WIN_1250, win);
  LoadCodePage(WIN_UTF8, utf);

  { zacneme citat kodovany text }
  Assign(fin, vstup);
  ReSet(fin);

  { do tohoto suboru budeme text ukladat }
  Assign(fout, VYS_TXT);
  ReWrite(fout);

  { priznak UNICODE }
  w:=$FEFF;
  Write(fout, w);
  w:=0;

  { prekodujem na unicode }
  while( not(eof(fin))) do begin
    Read(fin,ch);

    if( choice=1 )then          { win1250 }
      Write(fout, win[ch])
    else
    begin                       { UTF-8 }
      { znak je kodovany na 2 alebo 1 bajte }
      if( ch in [$C2, $C3, $C4, $C5 ])then
          w:= word(ch) * $100
      else
      begin
          Write( fout, win[ NajdiUTF(w+ch)]);
          w:=0;
      end;

    end;  { if( choice }
  end;

  { zavriem subory }
  Close(fin);
  Close(fout);

  WriteLn('Hotovo. Vysledok je v subore ', VYS_TXT);
  ReadLn;
end.

Copyrigth by Zdeno Sekerak 2007, http://www.trsek.com