thesnarky1 · April 6, 2014 22:09 · micwallace · Mar 10, 2016 · kergalym · Mar 13, 2017
diff --git a/arabic_characterization.html b/arabic_characterization.html
 <!DOCTYPE html>
 <html>
 <head>
  <meta charset="utf-8">
  <title>JS Bin</title>
 </head>
 <body>

 </body>
 </html>
diff --git a/arabic_characterization.js b/arabic_characterization.js
 var NICE_FORM = 0;
 var INITIAL_FORM = 1;
 var MEDIAL_FORM = 2;
 var FINAL_FORM = 3;
 var ISOLATED_FORM = 4;

 var arabicTranslations = {};
 //Alif
 arabicTranslations[1575] = [1575, 65165, 65166, 65166, 65165];
 //Ba
 arabicTranslations[1576] = [1576, 65169, 65170, 65168, 65167];
 //Ta
 arabicTranslations[1578] = [1578, 65175, 65176, 65174, 65173];
 //Tha
 arabicTranslations[1579] = [1579, 65179, 65180, 65178, 65177];
 //Jim
 arabicTranslations[1580] = [1580, 65183, 65184, 65182, 65181];
 //Ha
 arabicTranslations[1581] = [1581, 65187, 65188, 65186, 65185];
 //Kha
 arabicTranslations[1582] = [1582, 65191, 65192, 65190, 65189];
 //Dal
 arabicTranslations[1583] = [1583, 65193, 65194, 65194, 65193];
 //Thal
 arabicTranslations[1584] = [1584, 65195, 65196, 65196, 65195];
 //Ra
 arabicTranslations[1585] = [1585, 65197, 65198, 65198, 65195];
 //Zain
 arabicTranslations[1586] = [1586, 65199, 65200, 65200, 65199];
 //Seen
 arabicTranslations[1587] = [1587, 65203, 65204, 65202, 65201];
 //Sheen
 arabicTranslations[1588] = [1588, 65207, 65208, 65206, 65205];
 //Sod
 arabicTranslations[1589] = [1589, 65211, 65212, 65210, 65209];
 //Dod
 arabicTranslations[1590] = [1590, 65215, 65216, 65214, 65213];
 //Tah
 arabicTranslations[1591] = [1591, 65219, 65220, 65218, 65217];
 //Thah
 arabicTranslations[1592] = [1592, 65223, 65224, 65222, 65221];
 //Ayn
 arabicTranslations[1593] = [1593, 65227, 65228, 65224, 65225];
 //Ghayn
 arabicTranslations[1594] = [1594, 65231, 65232, 65230, 65229];
 //Fah
 arabicTranslations[1601] = [1601, 65235, 65236, 65234, 65233];
 //Qaf
 arabicTranslations[1602] = [1602, 65239, 65240, 65238, 65237];
 //Kaf
 arabicTranslations[1603] = [1603, 65243, 65244, 65242, 65241];
 //Lam
 arabicTranslations[1604] = [1604, 65247, 65248, 65246, 65245];
 //Mim
 arabicTranslations[1605] = [1605, 65251, 65252, 65250, 65249];
 //Nun
 arabicTranslations[1606] = [1606, 65255, 65256, 65254, 65253];
 //Heh
 arabicTranslations[1607] = [1607, 65259, 65260, 65258, 65257];
 //Waw
 arabicTranslations[1608] = [1608, 65261, 65262, 65262, 65261];
 //Ya
 arabicTranslations[1610] = [1610, 65267, 65268, 65266, 65265];

 var nonConnectors = [1575, 1583, 1584, 1585, 1586, 1608];

 var getRealCharCodes = function(str) {
  //Can't change an empty or one-char string
  if(str.length === 0 || str.length == 1) { console.log("Empty string"); return str; } 
  
  //No arabic in here to change, let's be quick about it
  if(!/[\u0600-\u06FF]/.test(str)) { console.log("No arabic here"); return str; } 
  
  //console.log("Changing " + str);
  var toReturn = "";
  var initial = true;
  var final = false;
  for(var x = 0; x < str.length; x++) {
    var tmpCharCode = str.charCodeAt(x);
    var tmpChar = str.charAt(x);
    //console.log("Checking " + tmpChar);
    
    //We have no way to translate
    if(arabicTranslations[tmpCharCode] === undefined) { toReturn += tmpChar; console.log("Skipping unknown character: " + tmpChar + "-" + tmpCharCode); continue; }
    
    //If we're the last letter, we must be final
    if(x == str.length - 1) { 
      final = true;
    //Or if the next letter after us is not an Arabic letter we know how to deal with
    } else if(arabicTranslations[str.charCodeAt(x + 1)] === undefined) {
      final = true;
    }
    
    //Add this character
    if(initial && final) {
      //console.log("Isolated char");
      toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][ISOLATED_FORM]);
      initial = true;
      final = false;
    } else if(initial) {
      //console.log("Initial char");
      toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][INITIAL_FORM]);
      initial = false;
    } else if(final) {
      //console.log("Final char");
      toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][FINAL_FORM]);
      initial = true;
      final = false;
    } else {
      //console.log("Median char");
      toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][MEDIAL_FORM]);
    }
    
    //If this is a non-connector, the next character must be initial
    if(nonConnectors.indexOf(tmpCharCode) > -1) {
      initial = true;
    }
  }
  return toReturn;
 };

 var test = "السنجاب حيوان من القوارض، يعيش غالبا على الأشجار، وله ذيل كثيف وكبير";
 var newTest = getRealCharCodes(test);

 //console.log(test);
 for(var x = 0; x < test.length; x++) {
  //console.log(test.charCodeAt(x));
  //console.log(test.charAt(x));
 }

 console.log(newTest);
 for(var x = 0; x < newTest.length; x++) {
  console.log(newTest.charCodeAt(x));
  console.log(newTest.charAt(x));
 }
	<!DOCTYPE html>
	<html>
	<head>
	<meta charset="utf-8">
	<title>JS Bin</title>
	</head>
	<body>

	</body>
	</html>
	var NICE_FORM = 0;
	var INITIAL_FORM = 1;
	var MEDIAL_FORM = 2;
	var FINAL_FORM = 3;
	var ISOLATED_FORM = 4;

	var arabicTranslations = {};
	//Alif
	arabicTranslations[1575] = [1575, 65165, 65166, 65166, 65165];
	//Ba
	arabicTranslations[1576] = [1576, 65169, 65170, 65168, 65167];
	//Ta
	arabicTranslations[1578] = [1578, 65175, 65176, 65174, 65173];
	//Tha
	arabicTranslations[1579] = [1579, 65179, 65180, 65178, 65177];
	//Jim
	arabicTranslations[1580] = [1580, 65183, 65184, 65182, 65181];
	//Ha
	arabicTranslations[1581] = [1581, 65187, 65188, 65186, 65185];
	//Kha
	arabicTranslations[1582] = [1582, 65191, 65192, 65190, 65189];
	//Dal
	arabicTranslations[1583] = [1583, 65193, 65194, 65194, 65193];
	//Thal
	arabicTranslations[1584] = [1584, 65195, 65196, 65196, 65195];
	//Ra
	arabicTranslations[1585] = [1585, 65197, 65198, 65198, 65195];
	//Zain
	arabicTranslations[1586] = [1586, 65199, 65200, 65200, 65199];
	//Seen
	arabicTranslations[1587] = [1587, 65203, 65204, 65202, 65201];
	//Sheen
	arabicTranslations[1588] = [1588, 65207, 65208, 65206, 65205];
	//Sod
	arabicTranslations[1589] = [1589, 65211, 65212, 65210, 65209];
	//Dod
	arabicTranslations[1590] = [1590, 65215, 65216, 65214, 65213];
	//Tah
	arabicTranslations[1591] = [1591, 65219, 65220, 65218, 65217];
	//Thah
	arabicTranslations[1592] = [1592, 65223, 65224, 65222, 65221];
	//Ayn
	arabicTranslations[1593] = [1593, 65227, 65228, 65224, 65225];
	//Ghayn
	arabicTranslations[1594] = [1594, 65231, 65232, 65230, 65229];
	//Fah
	arabicTranslations[1601] = [1601, 65235, 65236, 65234, 65233];
	//Qaf
	arabicTranslations[1602] = [1602, 65239, 65240, 65238, 65237];
	//Kaf
	arabicTranslations[1603] = [1603, 65243, 65244, 65242, 65241];
	//Lam
	arabicTranslations[1604] = [1604, 65247, 65248, 65246, 65245];
	//Mim
	arabicTranslations[1605] = [1605, 65251, 65252, 65250, 65249];
	//Nun
	arabicTranslations[1606] = [1606, 65255, 65256, 65254, 65253];
	//Heh
	arabicTranslations[1607] = [1607, 65259, 65260, 65258, 65257];
	//Waw
	arabicTranslations[1608] = [1608, 65261, 65262, 65262, 65261];
	//Ya
	arabicTranslations[1610] = [1610, 65267, 65268, 65266, 65265];

	var nonConnectors = [1575, 1583, 1584, 1585, 1586, 1608];

	var getRealCharCodes = function(str) {
	//Can't change an empty or one-char string
	if(str.length === 0 \|\| str.length == 1) { console.log("Empty string"); return str; }

	//No arabic in here to change, let's be quick about it
	if(!/[\u0600-\u06FF]/.test(str)) { console.log("No arabic here"); return str; }

	//console.log("Changing " + str);
	var toReturn = "";
	var initial = true;
	var final = false;
	for(var x = 0; x < str.length; x++) {
	var tmpCharCode = str.charCodeAt(x);
	var tmpChar = str.charAt(x);
	//console.log("Checking " + tmpChar);

	//We have no way to translate
	if(arabicTranslations[tmpCharCode] === undefined) { toReturn += tmpChar; console.log("Skipping unknown character: " + tmpChar + "-" + tmpCharCode); continue; }

	//If we're the last letter, we must be final
	if(x == str.length - 1) {
	final = true;
	//Or if the next letter after us is not an Arabic letter we know how to deal with
	} else if(arabicTranslations[str.charCodeAt(x + 1)] === undefined) {
	final = true;
	}

	//Add this character
	if(initial && final) {
	//console.log("Isolated char");
	toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][ISOLATED_FORM]);
	initial = true;
	final = false;
	} else if(initial) {
	//console.log("Initial char");
	toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][INITIAL_FORM]);
	initial = false;
	} else if(final) {
	//console.log("Final char");
	toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][FINAL_FORM]);
	initial = true;
	final = false;
	} else {
	//console.log("Median char");
	toReturn += String.fromCharCode(arabicTranslations[tmpCharCode][MEDIAL_FORM]);
	}

	//If this is a non-connector, the next character must be initial
	if(nonConnectors.indexOf(tmpCharCode) > -1) {
	initial = true;
	}
	}
	return toReturn;
	};

	var test = "السنجاب حيوان من القوارض، يعيش غالبا على الأشجار، وله ذيل كثيف وكبير";
	var newTest = getRealCharCodes(test);

	//console.log(test);
	for(var x = 0; x < test.length; x++) {
	//console.log(test.charCodeAt(x));
	//console.log(test.charAt(x));
	}

	console.log(newTest);
	for(var x = 0; x < newTest.length; x++) {
	console.log(newTest.charCodeAt(x));
	console.log(newTest.charAt(x));
	}