/* Unicode chart - http://www.unicode.org/charts/PDF/U0B80.pdf */

function ent(u, s, t)
{
	this.u = u; /* unicode number */
	this.s = s; /* english string */
	/*
	 * Type: special action
	 *	For uyir-mei letters, should the
	 *	vowel be 1 - appended, 2 - prepended or 3 - both
	 */
	this.t = t;
}

function rv(s, n)
{
	this.s = s; /* unicode number */
	this.n = n; /* no. of letters consumed */
}

/* Longest pattern should come first */
var uyir = [	/* vowels */
	new ent("ஆ",	"A",	1),
	new ent("ஐ",	"ai",	1),
	new ent("ஔ",	"au",	1),
	new ent("ஆ",	"aa",	1),
	new ent("அ",	"a",	1),
	new ent("ஈ",	"I",	1),
	new ent("ஈ",	"ii",	1),
	new ent("இ",	"i",	1),
	new ent("ஊ",	"U",	1),
	new ent("ஊ",	"uu",	1),
	new ent("உ",	"u",	1),
	new ent("ஏ",	"E",	1),
	new ent("ஏ",	"ee",	1),
	new ent("எ",	"e",	1),
	new ent("ஓ",	"O",	1),
	new ent("ஊ",	"oo",	1),
	new ent("ஒ",	"o",	1),
	// special letters
	new ent("ஃ",	"@",	0),	// ayutha ezhuthu
];

var mei = [	/* consonants */
	new ent("க",	"k",	0),
	new ent("க",	"g",	0),
	new ent("ங",	"ng",	0),
	new ent("ங",	"nG",	0),
	new ent("ச",	"ch",	0),
	new ent("ச",	"c",	0),
	new ent("ஜ",	"j",	0),
	new ent("ஞ",	"ny",	0),
	new ent("ஞ",	"nY",	0),
	new ent("த",	"th",	0),
	new ent("த",	"dh",	0),
	new ent("ட",	"d",	0),
	new ent("ண",	"nN",	0),	// 3 suzhi
	new ent("ட",	"t",	0),
	new ent("ந",	"N",	0),	// 'tha' type
	new ent("ன",	"n",	0),	// 2 suzhi
	new ent("ப",	"p",	0),
	new ent("ப",	"b",	0),
	new ent("ம",	"m",	0),
	new ent("ய",	"y",	0),
	new ent("ற",	"R",	0),
	new ent("ர",	"r",	0),
	new ent("ள",	"L",	0),
	new ent("ல",	"l",	0),
	new ent("ழ",	"zh",	0),
	new ent("ழ",	"z",	0),
	new ent("வ",	"v",	0),
	new ent("வ",	"w",	0),
	new ent("ஶ",	"ss",	0),	// 'w' type sha
	new ent("ஷ",	"sh",	0),	//
	new ent("ஸ",	"s",	0),
	new ent("ஹ",	"h",	0),
];

var uyirmei = [
	new ent("ா",	"A",	0),
	new ent("ா",	"aa",	0),
	new ent("ௌ",	"au",	0),
	new ent("ை",	"ai",	0),
	// Consume the 'kuril' 'a'. If a mei ezhuthu joins 'a',
	// it just looses its puLLi, unlike other uyir ezhuthus
	// that add a 'kaal' or 'kombu'
	new ent("",		"a",	1),
	new ent("ீ",	"I",	0),
	new ent("ீ",	"ii",	0),
	new ent("ி",	"i",	0),
	new ent("ு",	"u",	0),
	new ent("ூ",	"UU",	0),
	new ent("ூ",	"U",	0),
	new ent("ே",	"E",	0),
	new ent("ீ",	"ee",	0),
	new ent("ெ",	"e",	0),
	new ent("ோ",	"O",	0),
	new ent("ூ",	"oo",	0),
	new ent("ொ",	"o",	0),
];

var puLLi = "்";

function get_char(s, n)
{
	if (s.length > n) {
		return s.charAt(n);
	}
	return "";
}

/*
 * s - string
 * n - index in the string
 * m - was the previous character a 'mei' ezhuthu
 */
function is_uyir(s, n, m)
{
	var	i, arr;

	if (m != 0) {
		arr = uyirmei;
	} else {
		arr = uyir;
	}

	for (i = 0; arr[i]; i++) {

		if (get_char(s, n) == get_char(arr[i].s, 0)) {
			if (arr[i].s.length > 1) {
				if (get_char(s, n + 1) == get_char(arr[i].s, 1)) {
					return new rv(arr[i].u, 2);
				}
			} else {
				return new rv(arr[i].u, 1);
			}
		}
	}
	return new rv("", 0);
}

function is_mei(s, n)
{
	var	i;

	for (i = 0; mei[i]; i++) {
		if (get_char(s, n) == get_char(mei[i].s, 0)) {
			if (mei[i].s.length > 1) {
				if (get_char(s, n + 1) == get_char(mei[i].s, 1)) {
					return new rv(mei[i].u, 2);
				}
			} else {
				return new rv(mei[i].u, 1);
			}
		}
	}
	return new rv("", 0);
}

function e2u(s)
{
	var	m, o, i, ret;
	o = "";

	for (i = 0; i < s.length;) {
		m = 0;
		ret = is_mei(s, i);
		if (ret.n) {
			o += ret.s;
			i += ret.n;
			m = 1;
		}
		ret = is_uyir(s, i, m);
		if (ret.n) {
			o += ret.s;
			i += ret.n;
			m = 1;
		} else if (m) {
			// 'puLLi' for the 'mei' ezhuthu
			o += puLLi;
			m = 1;
		}

		if (!m) {
			if (s.charAt(i) == "<") {
				if (s.charAt(i + 1) == "~") {
					// To avoid parsing enclose it in <~...>
					for (i += 2;i < s.length && s.charAt(i) != ">"; i++) {
						o += s.charAt(i);
					}
					// skip ">"
					i++;
					continue;
				} else {
					for (;i < s.length && s.charAt(i) != ">"; i++) {
						// skip html tags
						o += s.charAt(i);
					}
				}
			}
			// Non tamil character. Skip and proceed
			o += s.charAt(i);
			i++;
		}
	}

	if (i != s.length) {
		// alert("parse length doesnt match: " + s.length + " ~ " + i);
	}


	return o;
}

function is_mei_u(u)
{
	var	i;
	for (i = 0; mei[i]; i++) {
		if (mei[i].u == u) {
			return new rv(mei[i].s, mei[i].s.length);
		}
	}
	return new rv("", 0);
}

function is_uyir_u(u)
{
	var	i;
	for (i = 0; uyir[i]; i++) {
		if (uyir[i].u == u) {
			return new rv(uyir[i].s, uyir[i].s.length);
		}
	}
	for (i = 0; uyirmei[i]; i++) {
		if (uyirmei[i].u == u) {
			return new rv(uyirmei[i].s, uyirmei[i].s.length);
		}
	}
	return new rv("", 0);
}

function u2e(s)
{
	var	e, i, t, ret, a;

	if (s.length < 7) { return s; }

	a = e = "";

	for (i = 0; i < s.length;) {
		t = s.substr(i, 7);
		if ((s.charAt(i) != "&") ||
		    (t.search(/&#[0-9][0-9][0-9][0-9];/) == -1)) {
			// Not tamil unicode
			e += a + s.charAt(i);
			a = "";
			i ++;
			continue;
		}

		ret = is_mei_u(t);
		if (ret.n) {
			e += a + ret.s;
			i += 7;
			if (s.charAt(i) == "&") {
				t = s.substr(i, 7);
				if (t == puLLi) {
					i += 7;
					a = "";
				} else {
					// kuril 'a'
					a = 'a';
				}
			} else {
					// kuril 'a'
					a = 'a';
			}
		} else {
			ret = is_uyir_u(t);
			if (ret.n) {
				e += ret.s;
				i += 7;
			} else {
				e += a;
			}
			a = "";
		}
	}
	return e + a;
}