Added `html_entities.js'

A JavaScript version of the `HTML::Entities' Perl module. It may be used to
decode HTML entities.

diff --git a/html_entities.js b/html_entities.js
new file mode 100644 (file)
index 0000000..13ea41c
--- /dev/null
+++ b/html_entities.js
@@ -0,0 +1,301 @@
+entityList = new Array ();
+charList   = new Array ();
+
+/* Really important entities */
+entityList.push ("amp");    charList.push ("&"); /* This MUST be the first entity! */
+entityList.push ("gt");     charList.push (">");
+entityList.push ("lt");     charList.push ("<");
+entityList.push ("quot");   charList.push ('"');
+entityList.push ("apos");   charList.push ("'");
+
+/* RFC 1866, section 14 */
+entityList.push ("nbsp");   charList.push (unescape ("%a0"));
+entityList.push ("iexcl");  charList.push (unescape ("%a1"));
+entityList.push ("cent");   charList.push (unescape ("%a2"));
+entityList.push ("pound");  charList.push (unescape ("%a3"));
+entityList.push ("curren"); charList.push (unescape ("%a4"));
+entityList.push ("yen");    charList.push (unescape ("%a5"));
+entityList.push ("brvbar"); charList.push (unescape ("%a6"));
+entityList.push ("sect");   charList.push (unescape ("%a7"));
+entityList.push ("uml");    charList.push (unescape ("%a8"));
+entityList.push ("copy");   charList.push (unescape ("%a9"));
+entityList.push ("ordf");   charList.push (unescape ("%aa"));
+entityList.push ("laquo");  charList.push (unescape ("%ab"));
+entityList.push ("not");    charList.push (unescape ("%ac"));
+entityList.push ("shy");    charList.push (unescape ("%ad"));
+entityList.push ("reg");    charList.push (unescape ("%ae"));
+entityList.push ("macr");   charList.push (unescape ("%af"));
+entityList.push ("deg");    charList.push (unescape ("%b0"));
+entityList.push ("plusmn"); charList.push (unescape ("%b1"));
+entityList.push ("sup2");   charList.push (unescape ("%b2"));
+entityList.push ("sup3");   charList.push (unescape ("%b3"));
+entityList.push ("acute");  charList.push (unescape ("%b4"));
+entityList.push ("micro");  charList.push (unescape ("%b5"));
+entityList.push ("para");   charList.push (unescape ("%b6"));
+entityList.push ("middot"); charList.push (unescape ("%b7"));
+entityList.push ("cedil");  charList.push (unescape ("%b8"));
+entityList.push ("sup1");   charList.push (unescape ("%b9"));
+entityList.push ("ordm");   charList.push (unescape ("%ba"));
+entityList.push ("raquo");  charList.push (unescape ("%bb"));
+entityList.push ("frac14"); charList.push (unescape ("%bc"));
+entityList.push ("frac12"); charList.push (unescape ("%bd"));
+entityList.push ("frac34"); charList.push (unescape ("%be"));
+entityList.push ("iquest"); charList.push (unescape ("%bf"));
+entityList.push ("Agrave"); charList.push (unescape ("%c0"));
+entityList.push ("Aacute"); charList.push (unescape ("%c1"));
+entityList.push ("Acirc");  charList.push (unescape ("%c2"));
+entityList.push ("Atilde"); charList.push (unescape ("%c3"));
+entityList.push ("Auml");   charList.push (unescape ("%c4"));
+entityList.push ("Aring");  charList.push (unescape ("%c5"));
+entityList.push ("AElig");  charList.push (unescape ("%c6"));
+entityList.push ("Ccedil"); charList.push (unescape ("%c7"));
+entityList.push ("Egrave"); charList.push (unescape ("%c8"));
+entityList.push ("Eacute"); charList.push (unescape ("%c9"));
+entityList.push ("Ecirc");  charList.push (unescape ("%ca"));
+entityList.push ("Euml");   charList.push (unescape ("%cb"));
+entityList.push ("Igrave"); charList.push (unescape ("%cc"));
+entityList.push ("Iacute"); charList.push (unescape ("%cd"));
+entityList.push ("Icirc");  charList.push (unescape ("%ce"));
+entityList.push ("Iuml");   charList.push (unescape ("%cf"));
+entityList.push ("ETH");    charList.push (unescape ("%d0"));
+entityList.push ("Ntilde"); charList.push (unescape ("%d1"));
+entityList.push ("Ograve"); charList.push (unescape ("%d2"));
+entityList.push ("Oacute"); charList.push (unescape ("%d3"));
+entityList.push ("Ocirc");  charList.push (unescape ("%d4"));
+entityList.push ("Otilde"); charList.push (unescape ("%d5"));
+entityList.push ("Ouml");   charList.push (unescape ("%d6"));
+entityList.push ("times");  charList.push (unescape ("%d7"));
+entityList.push ("Oslash"); charList.push (unescape ("%d8"));
+entityList.push ("Ugrave"); charList.push (unescape ("%d9"));
+entityList.push ("Uacute"); charList.push (unescape ("%da"));
+entityList.push ("Ucirc");  charList.push (unescape ("%db"));
+entityList.push ("Uuml");   charList.push (unescape ("%dc"));
+entityList.push ("Yacute"); charList.push (unescape ("%dd"));
+entityList.push ("THORN");  charList.push (unescape ("%de"));
+entityList.push ("szlig");  charList.push (unescape ("%df"));
+entityList.push ("agrave"); charList.push (unescape ("%e0"));
+entityList.push ("aacute"); charList.push (unescape ("%e1"));
+entityList.push ("acirc");  charList.push (unescape ("%e2"));
+entityList.push ("atilde"); charList.push (unescape ("%e3"));
+entityList.push ("auml");   charList.push (unescape ("%e4"));
+entityList.push ("aring");  charList.push (unescape ("%e5"));
+entityList.push ("aelig");  charList.push (unescape ("%e6"));
+entityList.push ("ccedil"); charList.push (unescape ("%e7"));
+entityList.push ("egrave"); charList.push (unescape ("%e8"));
+entityList.push ("eacute"); charList.push (unescape ("%e9"));
+entityList.push ("ecirc");  charList.push (unescape ("%ea"));
+entityList.push ("euml");   charList.push (unescape ("%eb"));
+entityList.push ("igrave"); charList.push (unescape ("%ec"));
+entityList.push ("iacute"); charList.push (unescape ("%ed"));
+entityList.push ("icirc");  charList.push (unescape ("%ee"));
+entityList.push ("iuml");   charList.push (unescape ("%ef"));
+entityList.push ("eth");    charList.push (unescape ("%f0"));
+entityList.push ("ntilde"); charList.push (unescape ("%f1"));
+entityList.push ("ograve"); charList.push (unescape ("%f2"));
+entityList.push ("oacute"); charList.push (unescape ("%f3"));
+entityList.push ("ocirc");  charList.push (unescape ("%f4"));
+entityList.push ("otilde"); charList.push (unescape ("%f5"));
+entityList.push ("ouml");   charList.push (unescape ("%f6"));
+entityList.push ("divide"); charList.push (unescape ("%f7"));
+entityList.push ("oslash"); charList.push (unescape ("%f8"));
+entityList.push ("ugrave"); charList.push (unescape ("%f9"));
+entityList.push ("uacute"); charList.push (unescape ("%fa"));
+entityList.push ("ucirc");  charList.push (unescape ("%fb"));
+entityList.push ("uuml");   charList.push (unescape ("%fc"));
+entityList.push ("yacute"); charList.push (unescape ("%fd"));
+entityList.push ("thorn");  charList.push (unescape ("%fe"));
+entityList.push ("yuml");   charList.push (unescape ("%ff"));
+
+/* UTF-8 characters */
+entityList.push ("OElig");  charList.push (unescape ("%u0152"));
+entityList.push ("oelig");  charList.push (unescape ("%u0153"));
+entityList.push ("Scaron"); charList.push (unescape ("%u0160"));
+entityList.push ("scaron"); charList.push (unescape ("%u0161"));
+entityList.push ("Yuml");   charList.push (unescape ("%u0178"));
+entityList.push ("fnof");   charList.push (unescape ("%u0192"));
+entityList.push ("circ");   charList.push (unescape ("%u02c6"));
+entityList.push ("tilde");  charList.push (unescape ("%u02dc"));
+entityList.push ("Alpha");  charList.push (unescape ("%u0391"));
+entityList.push ("Beta");   charList.push (unescape ("%u0392"));
+entityList.push ("Gamma");  charList.push (unescape ("%u0393"));
+entityList.push ("Delta");  charList.push (unescape ("%u0394"));
+entityList.push ("Epsilon"); charList.push (unescape ("%u0395"));
+entityList.push ("Zeta");   charList.push (unescape ("%u0396"));
+entityList.push ("Eta");    charList.push (unescape ("%u0397"));
+entityList.push ("Theta");  charList.push (unescape ("%u0398"));
+entityList.push ("Iota");   charList.push (unescape ("%u0399"));
+entityList.push ("Kappa");  charList.push (unescape ("%u039a"));
+entityList.push ("Lambda"); charList.push (unescape ("%u039b"));
+entityList.push ("Mu");     charList.push (unescape ("%u039c"));
+entityList.push ("Nu");     charList.push (unescape ("%u039d"));
+entityList.push ("Xi");     charList.push (unescape ("%u039e"));
+entityList.push ("Omicron"); charList.push (unescape ("%u039f"));
+entityList.push ("Pi");     charList.push (unescape ("%u03a0"));
+entityList.push ("Rho");    charList.push (unescape ("%u03a1"));
+entityList.push ("Sigma");  charList.push (unescape ("%u03a3"));
+entityList.push ("Tau");    charList.push (unescape ("%u03a4"));
+entityList.push ("Upsilon"); charList.push (unescape ("%u03a5"));
+entityList.push ("Phi");    charList.push (unescape ("%u03a6"));
+entityList.push ("Chi");    charList.push (unescape ("%u03a7"));
+entityList.push ("Psi");    charList.push (unescape ("%u03a8"));
+entityList.push ("Omega");  charList.push (unescape ("%u03a9"));
+entityList.push ("alpha");  charList.push (unescape ("%u03b1"));
+entityList.push ("beta");   charList.push (unescape ("%u03b2"));
+entityList.push ("gamma");  charList.push (unescape ("%u03b3"));
+entityList.push ("delta");  charList.push (unescape ("%u03b4"));
+entityList.push ("epsilon"); charList.push (unescape ("%u03b5"));
+entityList.push ("zeta");   charList.push (unescape ("%u03b6"));
+entityList.push ("eta");    charList.push (unescape ("%u03b7"));
+entityList.push ("theta");  charList.push (unescape ("%u03b8"));
+entityList.push ("iota");   charList.push (unescape ("%u03b9"));
+entityList.push ("kappa");  charList.push (unescape ("%u03ba"));
+entityList.push ("lambda"); charList.push (unescape ("%u03bb"));
+entityList.push ("mu");     charList.push (unescape ("%u03bc"));
+entityList.push ("nu");     charList.push (unescape ("%u03bd"));
+entityList.push ("xi");     charList.push (unescape ("%u03be"));
+entityList.push ("omicron"); charList.push (unescape ("%u03bf"));
+entityList.push ("pi");     charList.push (unescape ("%u03c0"));
+entityList.push ("rho");    charList.push (unescape ("%u03c1"));
+entityList.push ("sigmaf"); charList.push (unescape ("%u03c2"));
+entityList.push ("sigma");  charList.push (unescape ("%u03c3"));
+entityList.push ("tau");    charList.push (unescape ("%u03c4"));
+entityList.push ("upsilon"); charList.push (unescape ("%u03c5"));
+entityList.push ("phi");    charList.push (unescape ("%u03c6"));
+entityList.push ("chi");    charList.push (unescape ("%u03c7"));
+entityList.push ("psi");    charList.push (unescape ("%u03c8"));
+entityList.push ("omega");  charList.push (unescape ("%u03c9"));
+entityList.push ("thetasym"); charList.push (unescape ("%u03d1"));
+entityList.push ("upsih");  charList.push (unescape ("%u03d2"));
+entityList.push ("piv");    charList.push (unescape ("%u03d6"));
+entityList.push ("ensp");   charList.push (unescape ("%u2002"));
+entityList.push ("emsp");   charList.push (unescape ("%u2003"));
+entityList.push ("thinsp"); charList.push (unescape ("%u2009"));
+entityList.push ("zwnj");   charList.push (unescape ("%u200c"));
+entityList.push ("zwj");    charList.push (unescape ("%u200d"));
+entityList.push ("lrm");    charList.push (unescape ("%u200e"));
+entityList.push ("rlm");    charList.push (unescape ("%u200f"));
+entityList.push ("ndash");  charList.push (unescape ("%u2013"));
+entityList.push ("mdash");  charList.push (unescape ("%u2014"));
+entityList.push ("lsquo");  charList.push (unescape ("%u2018"));
+entityList.push ("rsquo");  charList.push (unescape ("%u2019"));
+entityList.push ("sbquo");  charList.push (unescape ("%u201a"));
+entityList.push ("ldquo");  charList.push (unescape ("%u201c"));
+entityList.push ("rdquo");  charList.push (unescape ("%u201d"));
+entityList.push ("bdquo");  charList.push (unescape ("%u201e"));
+entityList.push ("dagger"); charList.push (unescape ("%u2020"));
+entityList.push ("Dagger"); charList.push (unescape ("%u2021"));
+entityList.push ("bull");   charList.push (unescape ("%u2022"));
+entityList.push ("hellip"); charList.push (unescape ("%u2026"));
+entityList.push ("permil"); charList.push (unescape ("%u2030"));
+entityList.push ("prime");  charList.push (unescape ("%u2032"));
+entityList.push ("Prime");  charList.push (unescape ("%u2033"));
+entityList.push ("lsaquo"); charList.push (unescape ("%u2039"));
+entityList.push ("rsaquo"); charList.push (unescape ("%u203a"));
+entityList.push ("oline");  charList.push (unescape ("%u203e"));
+entityList.push ("frasl");  charList.push (unescape ("%u2044"));
+entityList.push ("euro");   charList.push (unescape ("%u20ac"));
+entityList.push ("image");  charList.push (unescape ("%u2111"));
+entityList.push ("weierp"); charList.push (unescape ("%u2118"));
+entityList.push ("real");   charList.push (unescape ("%u211c"));
+entityList.push ("trade");  charList.push (unescape ("%u2122"));
+entityList.push ("alefsym"); charList.push (unescape ("%u2135"));
+entityList.push ("larr");   charList.push (unescape ("%u2190"));
+entityList.push ("uarr");   charList.push (unescape ("%u2191"));
+entityList.push ("rarr");   charList.push (unescape ("%u2192"));
+entityList.push ("darr");   charList.push (unescape ("%u2193"));
+entityList.push ("harr");   charList.push (unescape ("%u2194"));
+entityList.push ("crarr");  charList.push (unescape ("%u21b5"));
+entityList.push ("lArr");   charList.push (unescape ("%u21d0"));
+entityList.push ("uArr");   charList.push (unescape ("%u21d1"));
+entityList.push ("rArr");   charList.push (unescape ("%u21d2"));
+entityList.push ("dArr");   charList.push (unescape ("%u21d3"));
+entityList.push ("hArr");   charList.push (unescape ("%u21d4"));
+entityList.push ("forall"); charList.push (unescape ("%u2200"));
+entityList.push ("part");   charList.push (unescape ("%u2202"));
+entityList.push ("exist");  charList.push (unescape ("%u2203"));
+entityList.push ("empty");  charList.push (unescape ("%u2205"));
+entityList.push ("nabla");  charList.push (unescape ("%u2207"));
+entityList.push ("isin");   charList.push (unescape ("%u2208"));
+entityList.push ("notin");  charList.push (unescape ("%u2209"));
+entityList.push ("ni");     charList.push (unescape ("%u220b"));
+entityList.push ("prod");   charList.push (unescape ("%u220f"));
+entityList.push ("sum");    charList.push (unescape ("%u2211"));
+entityList.push ("minus");  charList.push (unescape ("%u2212"));
+entityList.push ("lowast"); charList.push (unescape ("%u2217"));
+entityList.push ("radic");  charList.push (unescape ("%u221a"));
+entityList.push ("prop");   charList.push (unescape ("%u221d"));
+entityList.push ("infin");  charList.push (unescape ("%u221e"));
+entityList.push ("ang");    charList.push (unescape ("%u2220"));
+entityList.push ("and");    charList.push (unescape ("%u2227"));
+entityList.push ("or");     charList.push (unescape ("%u2228"));
+entityList.push ("cap");    charList.push (unescape ("%u2229"));
+entityList.push ("cup");    charList.push (unescape ("%u222a"));
+entityList.push ("int");    charList.push (unescape ("%u222b"));
+entityList.push ("there4"); charList.push (unescape ("%u2234"));
+entityList.push ("sim");    charList.push (unescape ("%u223c"));
+entityList.push ("cong");   charList.push (unescape ("%u2245"));
+entityList.push ("asymp");  charList.push (unescape ("%u2248"));
+entityList.push ("ne");     charList.push (unescape ("%u2260"));
+entityList.push ("equiv");  charList.push (unescape ("%u2261"));
+entityList.push ("le");     charList.push (unescape ("%u2264"));
+entityList.push ("ge");     charList.push (unescape ("%u2265"));
+entityList.push ("sub");    charList.push (unescape ("%u2282"));
+entityList.push ("sup");    charList.push (unescape ("%u2283"));
+entityList.push ("nsub");   charList.push (unescape ("%u2284"));
+entityList.push ("sube");   charList.push (unescape ("%u2286"));
+entityList.push ("supe");   charList.push (unescape ("%u2287"));
+entityList.push ("oplus");  charList.push (unescape ("%u2295"));
+entityList.push ("otimes"); charList.push (unescape ("%u2297"));
+entityList.push ("perp");   charList.push (unescape ("%u22a5"));
+entityList.push ("sdot");   charList.push (unescape ("%u22c5"));
+entityList.push ("lceil");  charList.push (unescape ("%u2308"));
+entityList.push ("rceil");  charList.push (unescape ("%u2309"));
+entityList.push ("lfloor"); charList.push (unescape ("%u230a"));
+entityList.push ("rfloor"); charList.push (unescape ("%u230b"));
+entityList.push ("lang");   charList.push (unescape ("%u2329"));
+entityList.push ("rang");   charList.push (unescape ("%u232a"));
+entityList.push ("loz");    charList.push (unescape ("%u25ca"));
+entityList.push ("spades"); charList.push (unescape ("%u2660"));
+entityList.push ("clubs");  charList.push (unescape ("%u2663"));
+entityList.push ("hearts"); charList.push (unescape ("%u2665"));
+entityList.push ("diams");  charList.push (unescape ("%u2666"));
+
+function encode_entities (str)
+{
+       for (var i = 0; (i < entityList.length) && (i < charList.length); i++)
+       {
+               var entity = entityList[i];
+               var mchar  = charList[i];
+
+               var re = new RegExp (mchar);
+               while (str.match (re))
+               {
+                       str = str.replace (re, "&" + entity + ";");
+               }
+       }
+
+       return (str);
+}
+
+function decode_entities (str)
+{
+       if (!str.match (/&[A-Za-z0-9]+;/))
+               return (str);
+
+       for (var i = 0; (i < entityList.length) && (i < charList.length); i++)
+       {
+               var entity = entityList[i];
+               var mchar  = charList[i];
+
+               var re = new RegExp ("&" + entity + ";");
+               while (str.match (re))
+               {
+                       str = str.replace (re, mchar);
+               }
+
+               if (!str.match (/&[A-Za-z0-9]+;/))
+                       break;
+       }
+
+       return (str);
+}