public static final HashMap<String, Character> chartable;
public static String decode(String s) {
String t;
Character ch;
int tmppos, i;
int maxpos = s.length();
StringBuffer sb = new StringBuffer(maxpos);
int curpos = 0;
while (curpos < maxpos) {
char c = s.charAt(curpos++);
if (c == '&') {
tmppos = curpos;
if (tmppos < maxpos) {
char d = s.charAt(tmppos++);
if (d == '#') {
if (tmppos < maxpos) {
d = s.charAt(tmppos++);
if ((d == 'x') || (d == 'x')) {
if (tmppos < maxpos) {
d = s.charAt(tmppos++);
if (ishexdigit(d)) {
while (tmppos < maxpos) {
d = s.charAt(tmppos++);
if (!ishexdigit(d)) {
if (d == ';') {
t = s.substring(curpos + 2,
tmppos - 1);
try {
i = Integer.parseInt(t,
16);
if ((i >= 0)
&& (i < 65536)) {
c = (char) i;
curpos = tmppos;
}
} catch (NumberFormatException e) {
}
}
break;
}
}
}
}
} else if (isdigit(d)) {
while (tmppos < maxpos) {
d = s.charAt(tmppos++);
if (!isdigit(d)) {
if (d == ';') {
t = s.substring(curpos + 1,
tmppos - 1);
try {
i = Integer.parseInt(t);
if ((i >= 0) && (i < 65536)) {
c = (char) i;
curpos = tmppos;
}
} catch (NumberFormatException e) {
}
}
break;
}
}
}
}
} else if (isletter(d)) {
while (tmppos < maxpos) {
d = s.charAt(tmppos++);
if (!isletterordigit(d)) {
if (d == ';') {
t = s.substring(curpos, tmppos - 1);
ch = (Character) chartable.get(t);
if (ch != null) {
c = ch.charValue();
curpos = tmppos;
}
}
break;
}
}
}
}
}
sb.append(c);
}
return sb.toString();
}
private static boolean isletterordigit(char c) {
return isletter(c) || isdigit(c);
}
private static boolean ishexdigit(char c) {
return ishexletter(c) || isdigit(c);
}
private static boolean isletter(char c) {
return ((c >= 'a') && (c <= 'z')) || ((c >= 'a') && (c <= 'z'));
}
private static boolean ishexletter(char c) {
return ((c >= 'a') && (c <= 'f')) || ((c >= 'a') && (c <= 'f'));
}
private static boolean isdigit(char c) {
return (c >= '0') && (c <= '9');
}
public static String compact(String s) {
int maxpos = s.length();
StringBuffer sb = new StringBuffer(maxpos);
int curpos = 0;
while (curpos < maxpos) {
char c = s.charAt(curpos++);
if (isWhitespace(c)) {
while ((curpos < maxpos) && isWhitespace(s.charAt(curpos))) {
curpos++;
}
c = '\u0020';
}
sb.append(c);
}
return sb.toString();
}
public static boolean isWhitespace(char ch) {
return (ch == '\u0020') || (ch == '\r') || (ch == '\n')
|| (ch == '\u0009') || (ch == '\u000c') || (ch == '\u200b');
}
static {
chartable = new HashMap<String, Character>();
chartable.put("quot", new Character((char) 34));
chartable.put("amp", new Character((char) 38));
chartable.put("apos", new Character((char) 39));
chartable.put("lt", new Character((char) 60));
chartable.put("gt", new Character((char) 62));
chartable.put("nbsp", new Character((char) 160));
chartable.put("iexcl", new Character((char) 161));
chartable.put("cent", new Character((char) 162));
chartable.put("pound", new Character((char) 163));
chartable.put("curren", new Character((char) 164));
chartable.put("yen", new Character((char) 165));
chartable.put("brvbar", new Character((char) 166));
chartable.put("sect", new Character((char) 167));
chartable.put("uml", new Character((char) 168));
chartable.put("copy", new Character((char) 169));
chartable.put("ordf", new Character((char) 170));
chartable.put("laquo", new Character((char) 171));
chartable.put("not", new Character((char) 172));
chartable.put("shy", new Character((char) 173));
chartable.put("reg", new Character((char) 174));
chartable.put("macr", new Character((char) 175));
chartable.put("deg", new Character((char) 176));
chartable.put("plusmn", new Character((char) 177));
chartable.put("sup2", new Character((char) 178));
chartable.put("sup3", new Character((char) 179));
chartable.put("acute", new Character((char) 180));
chartable.put("micro", new Character((char) 181));
chartable.put("para", new Character((char) 182));
chartable.put("middot", new Character((char) 183));
chartable.put("cedil", new Character((char) 184));
chartable.put("sup1", new Character((char) 185));
chartable.put("ordm", new Character((char) 186));
chartable.put("raquo", new Character((char) 187));
chartable.put("frac14", new Character((char) 188));
chartable.put("frac12", new Character((char) 189));
chartable.put("frac34", new Character((char) 190));
chartable.put("iquest", new Character((char) 191));
chartable.put("Agrave", new Character((char) 192));
chartable.put("Aacute", new Character((char) 193));
chartable.put("Acirc", new Character((char) 194));
chartable.put("Atilde", new Character((char) 195));
chartable.put("Auml", new Character((char) 196));
chartable.put("Aring", new Character((char) 197));
chartable.put("AElig", new Character((char) 198));
chartable.put("Ccedil", new Character((char) 199));
chartable.put("Egrave", new Character((char) 200));
chartable.put("Eacute", new Character((char) 201));
chartable.put("Ecirc", new Character((char) 202));
chartable.put("Euml", new Character((char) 203));
chartable.put("Igrave", new Character((char) 204));
chartable.put("Iacute", new Character((char) 205));
chartable.put("Icirc", new Character((char) 206));
chartable.put("Iuml", new Character((char) 207));
chartable.put("ETH", new Character((char) 208));
chartable.put("Ntilde", new Character((char) 209));
chartable.put("Ograve", new Character((char) 210));
chartable.put("Oacute", new Character((char) 211));
chartable.put("Ocirc", new Character((char) 212));
chartable.put("Otilde", new Character((char) 213));
chartable.put("Ouml", new Character((char) 214));
chartable.put("times", new Character((char) 215));
chartable.put("Oslash", new Character((char) 216));
chartable.put("Ugrave", new Character((char) 217));
chartable.put("Uacute", new Character((char) 218));
chartable.put("Ucirc", new Character((char) 219));
chartable.put("Uuml", new Character((char) 220));
chartable.put("Yacute", new Character((char) 221));
chartable.put("THORN", new Character((char) 222));
chartable.put("szlig", new Character((char) 223));
chartable.put("agrave", new Character((char) 224));
chartable.put("aacute", new Character((char) 225));
chartable.put("acirc", new Character((char) 226));
chartable.put("atilde", new Character((char) 227));
chartable.put("auml", new Character((char) 228));
chartable.put("aring", new Character((char) 229));
chartable.put("aelig", new Character((char) 230));
chartable.put("ccedil", new Character((char) 231));
chartable.put("egrave", new Character((char) 232));
chartable.put("eacute", new Character((char) 233));
chartable.put("ecirc", new Character((char) 234));
chartable.put("euml", new Character((char) 235));
chartable.put("igrave", new Character((char) 236));
chartable.put("iacute", new Character((char) 237));
chartable.put("icirc", new Character((char) 238));
chartable.put("iuml", new Character((char) 239));
chartable.put("eth", new Character((char) 240));
chartable.put("ntilde", new Character((char) 241));
chartable.put("ograve", new Character((char) 242));
chartable.put("oacute", new Character((char) 243));
chartable.put("ocirc", new Character((char) 244));
chartable.put("otilde", new Character((char) 245));
chartable.put("ouml", new Character((char) 246));
chartable.put("divide", new Character((char) 247));
chartable.put("oslash", new Character((char) 248));
chartable.put("ugrave", new Character((char) 249));
chartable.put("uacute", new Character((char) 250));
chartable.put("ucirc", new Character((char) 251));
chartable.put("uuml", new Character((char) 252));
chartable.put("yacute", new Character((char) 253));
chartable.put("thorn", new Character((char) 254));
chartable.put("yuml", new Character((char) 255));
chartable.put("OElig", new Character((char) 338));
chartable.put("oelig", new Character((char) 339));
chartable.put("Scaron", new Character((char) 352));
chartable.put("scaron", new Character((char) 353));
chartable.put("fnof", new Character((char) 402));
chartable.put("circ", new Character((char) 710));
chartable.put("tilde", new Character((char) 732));
chartable.put("Alpha", new Character((char) 913));
chartable.put("Beta", new Character((char) 914));
chartable.put("Gamma", new Character((char) 915));
chartable.put("Delta", new Character((char) 916));
chartable.put("Epsilon", new Character((char) 917));
chartable.put("Zeta", new Character((char) 918));
chartable.put("Eta", new Character((char) 919));
chartable.put("Theta", new Character((char) 920));
chartable.put("Iota", new Character((char) 921));
chartable.put("Kappa", new Character((char) 922));
chartable.put("Lambda", new Character((char) 923));
chartable.put("Mu", new Character((char) 924));
chartable.put("Nu", new Character((char) 925));
chartable.put("Xi", new Character((char) 926));
chartable.put("Omicron", new Character((char) 927));
chartable.put("Pi", new Character((char) 928));
chartable.put("Rho", new Character((char) 929));
chartable.put("Sigma", new Character((char) 931));
chartable.put("Tau", new Character((char) 932));
chartable.put("Upsilon", new Character((char) 933));
chartable.put("Phi", new Character((char) 934));
chartable.put("Chi", new Character((char) 935));
chartable.put("Psi", new Character((char) 936));
chartable.put("Omega", new Character((char) 937));
chartable.put("alpha", new Character((char) 945));
chartable.put("beta", new Character((char) 946));
chartable.put("gamma", new Character((char) 947));
chartable.put("delta", new Character((char) 948));
chartable.put("epsilon", new Character((char) 949));
chartable.put("zeta", new Character((char) 950));
chartable.put("eta", new Character((char) 951));
chartable.put("theta", new Character((char) 952));
chartable.put("iota", new Character((char) 953));
chartable.put("kappa", new Character((char) 954));
chartable.put("lambda", new Character((char) 955));
chartable.put("mu", new Character((char) 956));
chartable.put("nu", new Character((char) 957));
chartable.put("xi", new Character((char) 958));
chartable.put("omicron", new Character((char) 959));
chartable.put("pi", new Character((char) 960));
chartable.put("rho", new Character((char) 961));
chartable.put("sigmaf", new Character((char) 962));
chartable.put("sigma", new Character((char) 963));
chartable.put("tau", new Character((char) 964));
chartable.put("upsilon", new Character((char) 965));
chartable.put("phi", new Character((char) 966));
chartable.put("chi", new Character((char) 967));
chartable.put("psi", new Character((char) 968));
chartable.put("omega", new Character((char) 969));
chartable.put("thetasym", new Character((char) 977));
chartable.put("upsih", new Character((char) 978));
chartable.put("piv", new Character((char) 982));
chartable.put("ensp", new Character((char) 8194));
chartable.put("emsp", new Character((char) 8195));
chartable.put("thinsp", new Character((char) 8201));
chartable.put("zwnj", new Character((char) 8204));
chartable.put("zwj", new Character((char) 8205));
chartable.put("lrm", new Character((char) 8206));
chartable.put("rlm", new Character((char) 8207));
chartable.put("ndash", new Character((char) 8211));
chartable.put("mdash", new Character((char) 8212));
chartable.put("lsquo", new Character((char) 8216));
chartable.put("rsquo", new Character((char) 8217));
chartable.put("sbquo", new Character((char) 8218));
chartable.put("ldquo", new Character((char) 8220));
chartable.put("rdquo", new Character((char) 8221));
chartable.put("bdquo", new Character((char) 8222));
chartable.put("dagger", new Character((char) 8224));
chartable.put("Dagger", new Character((char) 8225));
chartable.put("bull", new Character((char) 8226));
chartable.put("hellip", new Character((char) 8230));
chartable.put("permil", new Character((char) 8240));
chartable.put("prime", new Character((char) 8242));
chartable.put("Prime", new Character((char) 8243));
chartable.put("lsaquo", new Character((char) 8249));
chartable.put("rsaquo", new Character((char) 8250));
chartable.put("oline", new Character((char) 8254));
chartable.put("frasl", new Character((char) 8260));
chartable.put("euro", new Character((char) 8364));
chartable.put("image", new Character((char) 8465));
chartable.put("weierp", new Character((char) 8472));
chartable.put("real", new Character((char) 8476));
chartable.put("trade", new Character((char) 8482));
chartable.put("alefsym", new Character((char) 8501));
chartable.put("larr", new Character((char) 8592));
chartable.put("uarr", new Character((char) 8593));
chartable.put("rarr", new Character((char) 8594));
chartable.put("darr", new Character((char) 8595));
chartable.put("harr", new Character((char) 8596));
chartable.put("crarr", new Character((char) 8629));
chartable.put("lArr", new Character((char) 8656));
chartable.put("uArr", new Character((char) 8657));
chartable.put("rArr", new Character((char) 8658));
chartable.put("dArr", new Character((char) 8659));
chartable.put("hArr", new Character((char) 8660));
chartable.put("forall", new Character((char) 8704));
chartable.put("part", new Character((char) 8706));
chartable.put("exist", new Character((char) 8707));
chartable.put("empty", new Character((char) 8709));
chartable.put("nabla", new Character((char) 8711));
chartable.put("isin", new Character((char) 8712));
chartable.put("notin", new Character((char) 8713));
chartable.put("ni", new Character((char) 8715));
chartable.put("prod", new Character((char) 8719));
chartable.put("sum", new Character((char) 8721));
chartable.put("minus", new Character((char) 8722));
chartable.put("lowast", new Character((char) 8727));
chartable.put("radic", new Character((char) 8730));
chartable.put("prop", new Character((char) 8733));
chartable.put("infin", new Character((char) 8734));
chartable.put("ang", new Character((char) 8736));
chartable.put("and", new Character((char) 8743));
chartable.put("or", new Character((char) 8744));
chartable.put("cap", new Character((char) 8745));
chartable.put("cup", new Character((char) 8746));
chartable.put("int", new Character((char) 8747));
chartable.put("there4", new Character((char) 8756));
chartable.put("sim", new Character((char) 8764));
chartable.put("cong", new Character((char) 8773));
chartable.put("asymp", new Character((char) 8776));
chartable.put("ne", new Character((char) 8800));
chartable.put("equiv", new Character((char) 8801));
chartable.put("le", new Character((char) 8804));
chartable.put("ge", new Character((char) 8805));
chartable.put("sub", new Character((char) 8834));
chartable.put("sup", new Character((char) 8835));
chartable.put("nsub", new Character((char) 8836));
chartable.put("sube", new Character((char) 8838));
chartable.put("supe", new Character((char) 8839));
chartable.put("oplus", new Character((char) 8853));
chartable.put("otimes", new Character((char) 8855));
chartable.put("perp", new Character((char) 8869));
chartable.put("sdot", new Character((char) 8901));
chartable.put("lceil", new Character((char) 8968));
chartable.put("rceil", new Character((char) 8969));
chartable.put("lfloor", new Character((char) 8970));
chartable.put("rfloor", new Character((char) 8971));
chartable.put("lang", new Character((char) 9001));
chartable.put("rang", new Character((char) 9002));
chartable.put("loz", new Character((char) 9674));
chartable.put("spades", new Character((char) 9824));
chartable.put("clubs", new Character((char) 9827));
chartable.put("hearts", new Character((char) 9829));
chartable.put("diams", new Character((char) 9830));
}