UTHTML.java

package iiuf.util;

/**
   Unicode to HTML translator.

   (c) 1999, 2000, 2001, IIUF, DIUF<p>

   @author $Author: ohitz $
   @version $Revision: 1.1 $
*/
public class UTHTML 
  extends 
  UnicodeTranslator
  implements
  Unicode {
    
  final static String[] UNICODE2HTML = 
  {
    nbsp   , " ",   // non breaking space
    iexcl  , "¡",  // inverted exclamation mark
    cent   , "¢",   // cent sign
    pound  , "£",  // pund sign
    curren , "¤", // currency sign
    yen    , "¥",    // yen sign
    brvbar , "¦", // broken bar
    sect   , "§",   // section sign
    Dot    , "¨",    // diaeresis
    copy   , "©",   // copyright sign
    ordf   , "ª",   // feminine ordinal indicator
    Lt     , "≪",     // left pointing double angle quotation mark ()
    not    , "¬",    // not sign
    shy    , "­",    // soft hyphen
    reg    , "®",    // registered sign
    macr   , "¯",   // macron
    
    deg    , "°",    // degree sogn
    plusmn , "&plusm;",  // plus minus sign
    sup2   , "²",   // superscript two
    sup3   , "³",   // superscript three
    acute  , "´",  // acute accent
    mcro   , "&mcro;",   // micro sign
    para   , "¶",   // pilcrow sign
    middot , "·", // middle dot
    cedil  , "¸",  // cedilla
    sup1   , "¹",   // superscrip one
    ordm   , "º",   // masculine ordinal indicator
    Gt     , "≫",     // right pointing double angle quotation mark
    frac14 , "¼", // vulgar fraction one quarter
    frac12 , "½", // vulgar fraction one half
    frac34 , "¾", // vulagar fraction three quarter
    iquest , "¿", // inverted question mark
    
    Agrave , "À",
    Aacute , "Á",
    Acirc  , "Â",
    Atilde , "Ã",
    Auml   , "Ä",
    Aring  , "Å",
    AElig  , "Æ",
    
    Ccedil , "Ç",
    
    Egrave , "È",
    Eacute , "É",
    Ecirc  , "Ê",
    Euml   , "Ë",
    
    Igrave , "Ì",
    Iacute , "Í",
    Icirci , "Îi;",
    Iuml   , "Ï",
    
    ETH    , "Ð",    // latin capital letter ETH
    Ntilde , "Ñ",
    
    Ograve , "Ò",
    Oacute , "Ó", 
    Ocirc  , "Ô",
    Otilde , "Õ",
    Ouml   , "Ö",
    
    times  , "×",  // multiplication sign
    Ostrok , "&Ostrok;",

    Ugrave , "Ù",
    Uacute , "Ú",
    Ucircr , "Ûr;",
    Uuml   , "Ü",
    
    Yacute , "Ý",
    THORN  , "Þ", // latin capital letter thorn

    szlig  , "ß",
    
    agrave , "à",
    aacute , "á",
    acirc  , "â",
    atilde , "ã",
    auml   , "ä",
    aring  , "å",
    aelig  , "æ",
    
    ccedil , "ç",
    
    egrave , "è",
    eacute , "é",
    ecirc  , "ê",
    euml   , "ë",
    
    igrave , "ì",
    iacute , "í",
    icirc  , "î",
    iuml   , "ï",
    
    eth    , "ð",   // latin small letter eth
    ntilde , "ñ",

    ograve , "ò",
    oacute , "ó",
    ocirc  , "ô",
    otilde , "õ",
    ouml   , "ö",
    
    divide , "÷", // division sign
    ostrok , "&ostrok;",

    ugrave , "ù",
    ucute  , "&ucute;",
    ucircr , "ûr;",
    uuml   , "ü",

    yacute , "ý",
    thorn  , "þ", // latin small letter thorn
    yuml   , "ÿ",
    
    inodot      , "ı", // LATIN SMALL LETTER DOTLESS I
    OElig       , "Œ", // LATIN CAPITAL LIGATURE OE
    oelig       , "œ", // LATIN SMALL LIGATURE OE
    Yuml        , "Ÿ", // LATIN CAPITAL LETTER Y WITH DIAERESIS
    fnof        , "ƒ", // LATIN SMALL LETTER F WITH HOOK
    circ        , "ˆ", // MODIFIER LETTER CIRCUMFLEX ACCENT
    caron       , "ˇ", // CARON
    breve       , "˘", // BREVE
    dot         , "˙", // DOT ABOVE
    ring        , "˚", // RING ABOVE
    ogon        , "˛", // OGONEK
    tilde       , "˜", // SMALL TILDE
    dblac       , "˝", // DOUBLE ACUTE ACCENT
    OHgr        , "&OHgr;", // GREEK CAPITAL LETTER OMEGA
    b_pi        , "&b_pi;", // GREEK SMALL LETTER PI
    ndash       , "–", // EN DASH
    mdash       , "—", // EM DASH
    rsquor      , "’", // LEFT SINGLE QUOTATION MARK
    rsquo       , "’", // RIGHT SINGLE QUOTATION MARK
    lsquor      , "‚", // SINGLE LOW-9 QUOTATION MARK
    ldquo       , "“", // LEFT DOUBLE QUOTATION MARK
    rdquo       , "”", // RIGHT DOUBLE QUOTATION MARK
    bdquo       , "„", // DOUBLE LOW-9 QUOTATION MARK
    dagger      , "†", // DAGGER
    Dagger      , "‡", // DOUBLE DAGGER
    bull        , "•", // BULLET
    hellip      , "…", // HORIZONTAL ELLIPSIS
    permil      , "‰", // PER MILLE SIGN
    lsaquo      , "‹", // SINGLE LEFT-POINTING ANGLE QUOTATION MARK
    rsaquo      , "›", // SINGLE RIGHT-POINTING ANGLE QUOTATION MARK
    frasl       , "⁄", // FRACTION SLASH
    euro        , "€", // EURO SIGN
    trade       , "™", // TRADE MARK SIGN
    part        , "∂", // PARTIAL DIFFERENTIAL
    Delta       , "Δ", // INCREMENT
    prod        , "∏", // N-ARY PRODUCT
    sum         , "∑", // N-ARY SUMMATION
    radic       , "√", // SQUARE ROOT
    infin       , "∞", // INFINITY
    int_        , "∫", // INTEGRAL
    thkap       , "≈", // ALMOST EQUAL TO
    ne          , "&nr;", // NOT EQUAL TO
    le          , "≤", // LESS-THAN OR EQUAL TO
    ge          , "≥", // GREATER-THAN OR EQUAL TO
    loz         , "◊", // LOZENGE
    _APPLE_LOGO , "_APPLE_LOGO", // Apple logo
    filig       , "fi", // LATIN SMALL LIGATURE FI
    fllig       , "fl", // LATIN SMALL LIGATURE FL
  };
  static String[] HTML2UNICODE = invTable(UNICODE2HTML);
  
  /** The default translator instance. */  
  public static UnicodeTranslator trans = new UTHTML();

  private UTHTML() {
    super(UNICODE2HTML);
  }
  
  public String getUnicode(String nstr) {
    return UnicodeTrans.trans(nstr, HTML2UNICODE);
  }
  
  public String getNative(String unicode) {
    return UnicodeTrans.trans(unicode, UNICODE2HTML);
  }

  public static void main(String[] argv) {
    System.out.println("getNative("  + argv[0] + ") = " + UTHTML.trans.getNative (argv[0]));
    System.out.println("getUnicode(" + argv[0] + ") = " + UTHTML.trans.getUnicode(argv[0]));
  }
}
/*
  $Log: UTHTML.java,v $
  Revision 1.1  2002/07/11 12:00:11  ohitz
  Initial checkin

  Revision 1.3  2001/01/04 16:28:42  schubige
  Header update for 2001 and DIUF

  Revision 1.2  2000/05/01 12:41:42  schubige
  intermediate checkin after UT update

  Revision 1.1  2000/04/25 12:11:15  schubige
  pre bibtex restart commit

*/