Naraguna:Bennylin/trans-toba.js

/*!
* trans-toba.js
* https://bennylin.github.com/transliterasi/toba.html
*
* Copyright 2013-2021, Bennylin @bennylin
* Dual licensed under the MIT or GPL Version 2 licenses.
* Released under the MIT, BSD, and GPL Licenses.
*
* Versions:
* 1 April 2021 - v 1.0
*
*
* Derived from my Javanese transliteration: https://jv.wikipedia.org/wiki/Naraguna:Bennylin/trans.js
*/
var vowelPrev = false;
/***************************
Function SuperTrim, findstr
trim string, menemukan karakter di dalam string
****************************/
function SuperTrim(str) {
	str = str || '';
	return str.replace(/^\s*|\s*$/g,'').replace(/\s+/g,' ');
}
function findstr(str,tofind) {
	for (var i = 0; i < str.length; i++)
	if (str[i] == tofind)
	return true;
	return false;
}
/***************************
Function isDigit, isPunct, isVowel
cek apakah digit, tanda baca, atau huruf vokal (a, e/è/é, i, o, u, ě/ê, ô, ā/ī/ū/ō)
****************************/
function isDigit(a) {
	var str = "0123456789";
	return findstr(str,a);
}
function isPunct(a) {
	var str = ",.><?/+=-_}{[]*&^%$#@!~`\"\\|:;()";
	return findstr(str,a);
}
function isVowel(a) {
	var str = "AaEeÈèÉéIiOoUuÊêĚěXxôāīūō";
	return findstr(str,a);
}
function isLCVowel(a) {
	var str = "aeiou";
	return findstr(str,a);
}
function isConsonant(a) {
	var str = "BCDfGHJKLMNPQRSTVWXYZbcdfghjklmnpqrstvwxyz";
	//QXqx are special chars, add engma & enye
	return findstr(str,a);
}
/***************************
Function isSpecial, isHR, isLW
cek apakah karakter spesial (bikonsonan)
****************************/
function isSpecial(a) {
	var str = "GgYyDdBb";
	//untuk bikonsonan nga, nya, nda, mba
	return findstr(str,a);
}
/***************************
Function GetMatra
apabila huruf vokal, return matra (sandhangan swara)
****************************/
function GetMatra(str) {
	var i = 0;
	if (str.length < 1) {
		return ""; //virama panongonan
	}
	while (str[i] == 'h') {
		i++;
		if (i >= str.length) {
			break;
		}
	}
	if (i < str.length) {
		str = str.substring(i);
	}
	var matramap= {
		"e":'ᯧ', "i":'ᯪ', "o":'ᯬ', "u":'ᯮ', 
		"A":'ᯀ', "E":'ᯀᯧ', "I":'ᯤ', "O":'ᯀᯬ', "U":'ᯥ'
	} //TODO: talinga (ee) & tompi
    //Aksara i (ᯤ) dan u (ᯥ) hanya digunakan untuk suku kata terbuka, misal pada kata dan ina ᯤᯉ dan ulu ᯥᯞᯮ. Untuk suku kata tertutup yang diawali dengan bunyi i atau u, digunakanlah aksara a (ᯀ atau ᯁ) bersama diaktirik untuk masing-masing vokal, misal pada kata indung ᯀᯪᯉ᯲ᯑᯮᯰ dan umpama ᯀᯮᯔ᯲ᯇᯔ

	if(matramap[str]!==undefined) {
		return matramap[str];
	}
	return "";
}
/***************************
Function GetShift
apabila huruf bikonsonan, return karakter khusus
****************************/
function GetShift(str) {
	str = str.toLowerCase();
	if (str.indexOf("ng") == 0) {
		//suku kata diawali 'ng'
		return {
			"CoreSound": "ᯝ", "len": 2
		}
		;
	} 
	//nya
	if (str.indexOf("ny") == 0) {
		//suku kata diawali 'ny'
		return {
			"CoreSound": "ᯠ", "len": 2
		}
		;
	} 
	if (str.indexOf("nd") == 0) {
		//nd-
		return {
			"CoreSound": "ᯢ", "len": 2
		}
		;
	} 
	if (str.indexOf("mb") == 0) {
		//mb-
		return {
			"CoreSound": "ᯣ", "len": 2
		}
		;
	} 
	return {
		"CoreSound": null, "len": 1
	}
	;
}
/***************************
Function GetCoreSound
return aksara nglegana
****************************/
function GetCoreSound(str) {
	var consonantMap1 = {
		"A":"ᯀ", //A
		"B":"ᯅ", //B
		"C":"ᯡ", //C
		"D":"ᯑ", //D
		"E":"ᯀᯧ", //E
		"F":"ᯇ", //P
		"G":"ᯎ", //G
		"H":"ᯂ", //H
		"I":"ᯤ", //I
		"J":"ᯐ", //J
		"K":"ᯂ", //K
		"L":"ᯞ", //L
		"M":"ᯔ", //M
		"N":"ᯉ", //N
		"O":"ᯀᯬ", //O
		"P":"ᯇ", //P
		"Q":"ᯂ", //K
		"R":"ᯒ", //R
		"S":"ᯘ", //S
		"T":"ᯖ", //T
		"U":"ᯥ", //U
		"V":"ᯋ", //W
		"W":"ᯋ", //W
		"X":"ᯂ", //K
		"Y":"ᯛ", //Y
		"Z":"ᯐ", //J
		"a":"ᯀ", //A
		"b":"ᯅ", //B
		"c":"ᯡ", //C
		"d":"ᯑ", //D
		"e":"ᯀᯧ", //E
		"f":"ᯇ", //P
		"g":"ᯎ", //G
		"h":"ᯂ", //H
		"i":"ᯀᯪ", //I
		"j":"ᯐ", //J
		"k":"ᯂ", //K
		"l":"ᯞ", //L
		"m":"ᯔ", //M
		"n":"ᯉ", //N
		"o":"ᯀᯬ", //O
		"p":"ᯇ", //P
		"q":"ᯂ", //K
		"r":"ᯒ", //R
		"s":"ᯘ", //S
		"t":"ᯖ", //T
		"u":"ᯀᯮ", //U
		"v":"ᯋ", //W
		"w":"ᯋ", //W
		"x":"ᯂ", //K
		"y":"ᯛ", //Y
		"z":"ᯐ" //J
	}
	var consonantMap, murda;
	var modeMurda = document.getElementsByName("murda");
	for (var rad in modeMurda) {
		if(modeMurda[rad].checked)
		murda = modeMurda[rad].value;
	}
	if(murda == "pakai")
	consonantMap = consonantMap2; else //if(murda == "tidak")
	consonantMap = consonantMap1;
	var h_shift = GetShift(str);
	var core = str;
	if (h_shift["CoreSound"] == null) {
		if (consonantMap[str.charAt(0)]) core = consonantMap[str.charAt(0)];
		return {
			"CoreSound": core,
			"len": 1
		}
		;
	} else {
		return h_shift;
	}
}
/***************************
Function ResolveCharacterSound
return tanda baca, digit, vokal, maupun nglegana+pangkon
****************************/
function ResolveCharacterSound(c) {
	var str = "" + c;
	var len = 0;
	if (isDigit(c)) {
		return "" + ('' + (c - '0'));
	} else if (isConsonant(str[0])) {
		return "" + GetCoreSound(str).CoreSound + "᯳"; //virama panongonan
	} else {
		//if (isVowel(str[0])) {
		return "" + GetCoreSound(str).CoreSound;
	}
	/**/
}
/***************************
Function GetSound
fungsi yang mentransliterasi masing-masing suku kata
****************************/
function GetSound(str) {
	var len = 0;
	str = SuperTrim(str);
	if (str == null || str == "") {
		return "";
	}
	if (str.length == 1) {
		return ResolveCharacterSound(str[0]);
	} else {
		var core_sound = GetCoreSound(str);
		//return "1"+core_sound.CoreSound+"2";
		var matra = "";
		var konsonan = "";
		if (core_sound.len >= 1) {
			matra = GetMatra(str.substring(core_sound.len));
			//aeiou (suku, wulu, pepet, taling, taling tarung, dll.)
			/*if () {
} else {
}*/
		} else {
			matra = "";
		}
		if (str.indexOf("nggr") == 0) {
			//nggr-
			if (vowelPrev) konsonan = "ᬂᬕ᭄ᬭ";
			//<vowel>nggr-, e.g. panggrahita else konsonan = "ᬗ᭄ᬕ᭄ᬭ";
			//<nonvowel>nggr-, i.e. nggronjal
		} else if (str.indexOf("ngg") == 0) {
			//ngg-
			if (vowelPrev) konsonan = "ᬂᬕ";
			//<vowel>ngg-, e.g. tunggal else konsonan = "ᬗ᭄ᬕ";
			//<nonvowel>ngg-, i.e. nggambar
		} else if (str.indexOf("njl") == 0) {
			//njl-
			konsonan = "ᬜ᭄ᬚ᭄ᬮ";
		} else if (str.indexOf("njr") == 0) {
			//njr-
			konsonan = "ᬜ᭄ᬚ᭄ᬭ";
		} else if (str.indexOf("ngg") == 0) {
			//njr-
			if (vowelPrev) konsonan = "ᬂᬕ";
			//<vowel>ngg-, e.g. tunggal else konsonan = "ᬗ᭄ᬕ";
			//<nonvowel>ngg-, i.e. nggambar
		} else if (core_sound.CoreSound == "ᬦᬜ᭄ᬘ") {
			// -nc-
			konsonan = "ᬜ᭄ᬘ";
			//-nyc-/*
		} else if (core_sound.CoreSound == "ᬦᬜ᭄ᬚ᭄ᬮ") {
			// -njl-
			konsonan = "ᬜ᭄ᬚ᭄ᬮ";
			//-njl-
		} else if (core_sound.CoreSound == "ᬦᬜ᭄ᬚ") {
			// -njr-
			konsonan = "ᬜ᭄ᬚ᭄ᬭ";
			//-njr-*/
		} else if (core_sound.CoreSound == "ᬦᬜ᭄ᬚ") {
			// -nj-
			konsonan = "ᬜ᭄ᬚ";
			//-nyj-
		} else if (core_sound.CoreSound == "ᬤᬟ᭄ᬯ") {
			// -dhw-
			konsonan = "ᬟ᭄ᬯ";
			//-dhw-
		} else if (core_sound.CoreSound == "ᬤᬟ") {
			// -dhy-
			konsonan = "ᬟ";
			//-dhy-
		} else if (core_sound.CoreSound == "ᬢᬝ᭄ᬯ") {
			// -thw-
			konsonan = "ᬝ᭄ᬯ";
			//-dhw-
		} else if (core_sound.CoreSound == "ᬢᬝ") {
			// -thy-
			konsonan = "ᬝ";
			//-dhy-
		} else if (findstr(core_sound.CoreSound,'') && matra == "᭄") {
			// pengkal
			konsonan = core_sound.CoreSound;
			matra = "";
			//-y-
		} else if (findstr(core_sound.CoreSound,'') && matra == "᭄") {
			// cakra
			konsonan = core_sound.CoreSound;
			matra = "᭄ᬭ";
			//-r-
		} else if (findstr(core_sound.CoreSound,'᭄ᬭ') && matra == "ᭂ") {
			// cakra keret
			if ((str[0] == "n" && str[1] == "y") || ((str[0] == "t" || str[0] == "d") && str[1] == "h")) {
				konsonan = GetCoreSound(str[0]+str[1]).CoreSound + "";
				matra = "";
				//nyrê-, thrê-, dhrê-
			} else if (str[0] == "n" && str[1] == "g") {
				if (str[2] == "g") konsonan = "ᬗ᭄ᬕ"; else konsonan = "ᬗ";
				matra = "";
				//nggrê-/ngrê-
			} else {
				konsonan = GetCoreSound(str[0]).CoreSound + "";
				matra = "";
				//-rê-
			}
		} else if (findstr(core_sound.CoreSound, 'ᬮ') && matra == "ᭂ") {
			// nga lelet
			if ((str[0] == "n" && str[1] == "y") || ((str[0] == "t" || str[0] == "d") && str[1] == "h")) {
				konsonan = GetCoreSound(str[0]+str[1]).CoreSound + "᭄ᬮᭂ";
				matra = "";
				//nylê-, thlê-, dhlê-
			} else if (str[0] == "n" && str[1] == "g") {
				if (str[2] == "g") konsonan = "ᬗ᭄ᬕ᭄ᬮᭂ"; else konsonan = "ᬗ᭄ᬮᭂ";
				matra = "";
				//ngglê-/nglê-
			} else if (str[0] == "l") {
				konsonan = "ᬍ";
				matra = "";
				//-lê-
			} else {
				konsonan = GetCoreSound(str[0]).CoreSound + "᭄ᬮᭂ";
				matra = "";
				//-lê-
			}
		} else if (core_sound.CoreSound == 'ᬝ᭄ᬭ' || core_sound.CoreSound == 'ᬟ᭄ᬭ' || core_sound.CoreSound == 'ᬗ᭄ᬭ' || core_sound.CoreSound == 'ᬜ᭄ᬭ') {
			// i.e. nyruput
			konsonan = core_sound.CoreSound;
			if (matra == "᭄") matra = "";
		} else if (core_sound.CoreSound == "ᬮᬮ᭄ᬮ") {
			// -ll-
			konsonan = "ᬮ᭄ᬮ";
			//double -l-
		} else if (core_sound.CoreSound == "ᬃᬃᬭ") {
			// -rr-
			konsonan = "ᬃᬭ";
			//double -r-
		} else if (core_sound.CoreSound == "ᬃᬃᬳ") {
			// -rh-
			konsonan = "ᬃᬳ";
			//-rh-
		} else if (core_sound.CoreSound == "ᬃᬃᬮ") {
			// -rl-
			konsonan = "ᬃᬮ";
			//-rl-
		} else if (core_sound.CoreSound == "ᬃᬃᬯ") {
			// -rw-
			if (vowelPrev) konsonan = "ᬃᬯ";
			//-rw- -- arwana else konsonan = "ᬭ᭄ᬯ";
			//rw- -- rwa/rwi/rwab
		} else if (core_sound.CoreSound == "ᬃᬃᬘ") {
			// -rc-
			konsonan = "ᬃᬘ";
			//-rc-
		} else if (core_sound.CoreSound == "ᬄᬄᬳ") {
			// -hh-
			konsonan = "ᬄᬳ";
			//double -h-
		} else if (core_sound.CoreSound == "ᬄᬄᬮ") {
			// -hl-
			if (vowelPrev) konsonan = "ᬄᬮ";
			//-hl- else konsonan = "ᬳ᭄ᬮ";
			//hlam
		} else if (core_sound.CoreSound == "ᬄᬄᬯ") {
			// -hw-
			if (vowelPrev) konsonan = "ᬄᬯ";
			//-hw- else konsonan = "ᬳ᭄ᬯ";
			//hwab,hwan
		} else if (core_sound.CoreSound == "ᬄᬳ") {
			// -hy-
			if (vowelPrev) konsonan = "ᬄᬬ";
			//sembahyang else konsonan = "ᬳ";
			//hyang/*
		} else if (core_sound.CoreSound == "ᬄᬄ") {
			// hrx-
			konsonan = "ᬳ᭄ᬭ";
			//hrx-
		} else if (core_sound.CoreSound == "ᬄᬄ᭄ᬭ") {
			// hr-
			if (matra == "ᭂ") konsonan = "ᬳ";
			//hr- else konsonan = "ᬳ᭄ᬭ";
			//hr-
		} else if (core_sound.CoreSound == "ᬄᬳ᭄ᬭ") {
			// hr-
			if (matra == "ᭂ") konsonan = "ᬳ";
			//hr- else konsonan = "ᬳ";
			//hr-
		} else if (core_sound.CoreSound == 'ᬄ' && matra == "᭄") {
			// wignyan - 12 April
			konsonan = "ᬳ";
			//ha
		} else if (core_sound.CoreSound == 'ᬄ' && matra != "᭄") {
			// wignyan
			konsonan = "ᬳ";
			//ha
		} else if (core_sound.CoreSound == 'ᬃ' && matra == "ᭂ") {
			// pa cerek
			konsonan = "ᬋ";
			matra = "";
			//rê
		} else if (core_sound.CoreSound == 'ᬃ' && matra != "᭄") {
			// layar
			konsonan = "ᬭ";
			//ra
		} else if (core_sound.CoreSound == 'ᬂ' && matra != "᭄") {
			// cecak
			konsonan = "ᬗ";
			//nga
		} else if (core_sound.CoreSound == 'ᬂ' && matra == "᭄") {
			// cecak
			konsonan = "ᬂ";
			matra = "";
			//cecak
		} else {
			konsonan = core_sound.CoreSound;
		}
		return "" + konsonan + matra;
	}
}
/***************************
Function DoTransliterate
fungsi utama yang dipanggil (main function)
****************************/
function DoTransliterate(str) {
	var i = 0;
	var ret = "";
	var pi = 0;
	var vowelFlag = false;
	startVowel = false;
	str = SuperTrim(str);
	while (i < str.length) {

        if (i > 0 && isVowel(str[i]) && isVowel(str[i-1])) { //deal with words that start with multiple vocals
               str = str.substring(0, i)+'​'+str.substring(i, str.length);
        }
		if ((isSpecial(str[i])) && !vowelFlag) {
			//i++;
		} else if ((str[i] == 'h' && vowelFlag) || (!isVowel(str[i]) && i > 0) || (str[i] == ' ') || isPunct(str[i]) || isDigit(str[i]) || ((i - pi) > 5)) {
			ret += GetSound(str.substring(pi, i));
            if (str[i] == ' ') { ret += ' '; }
			if (isPunct(str[i])) {
					ret += str[i];
					pi = i + 1;
			} else {
				pi = i;
			}
			vowelFlag = false;
		} else if (isVowel(str[i]) && str[i] != 'h') {
			//ret += GetSound(str.substring(pi, i));
			vowelFlag = true;
		}
		if (pi > 0 && isVowel(str[pi-1])) {
			//<vowel>ngg
			vowelPrev = true;
		} else vowelPrev = false;
		i++;
	}
	//endwhile
	if (pi < i) {
		ret += GetSound(str.substring(pi, i));
	}
	return SuperTrim(ret);
}