modules/arabic.js

  1. /**
  2.  * @license
  3.  * Copyright (c) 2017 Aras Abbasi
  4.  *
  5.  * Licensed under the MIT License.
  6.  * http://opensource.org/licenses/mit-license
  7.  */
  9. import { jsPDF } from "../jspdf.js";
  11. /**
  12.  * jsPDF arabic parser PlugIn
  13.  *
  14.  * @name arabic
  15.  * @module
  16.  */
  17. (function(jsPDFAPI) {
  18.   "use strict";
  20.   /**
  21.    * Arabic shape substitutions: char code => (isolated, final, initial, medial).
  22.    * Arabic Substition A
  23.    */
  24.   var arabicSubstitionA = {
  25.     0x0621: [0xfe80], // ARABIC LETTER HAMZA
  26.     0x0622: [0xfe81, 0xfe82], // ARABIC LETTER ALEF WITH MADDA ABOVE
  27.     0x0623: [0xfe83, 0xfe84], // ARABIC LETTER ALEF WITH HAMZA ABOVE
  28.     0x0624: [0xfe85, 0xfe86], // ARABIC LETTER WAW WITH HAMZA ABOVE
  29.     0x0625: [0xfe87, 0xfe88], // ARABIC LETTER ALEF WITH HAMZA BELOW
  30.     0x0626: [0xfe89, 0xfe8a, 0xfe8b, 0xfe8c], // ARABIC LETTER YEH WITH HAMZA ABOVE
  31.     0x0627: [0xfe8d, 0xfe8e], // ARABIC LETTER ALEF
  32.     0x0628: [0xfe8f, 0xfe90, 0xfe91, 0xfe92], // ARABIC LETTER BEH
  33.     0x0629: [0xfe93, 0xfe94], // ARABIC LETTER TEH MARBUTA
  34.     0x062a: [0xfe95, 0xfe96, 0xfe97, 0xfe98], // ARABIC LETTER TEH
  35.     0x062b: [0xfe99, 0xfe9a, 0xfe9b, 0xfe9c], // ARABIC LETTER THEH
  36.     0x062c: [0xfe9d, 0xfe9e, 0xfe9f, 0xfea0], // ARABIC LETTER JEEM
  37.     0x062d: [0xfea1, 0xfea2, 0xfea3, 0xfea4], // ARABIC LETTER HAH
  38.     0x062e: [0xfea5, 0xfea6, 0xfea7, 0xfea8], // ARABIC LETTER KHAH
  39.     0x062f: [0xfea9, 0xfeaa], // ARABIC LETTER DAL
  40.     0x0630: [0xfeab, 0xfeac], // ARABIC LETTER THAL
  41.     0x0631: [0xfead, 0xfeae], // ARABIC LETTER REH
  42.     0x0632: [0xfeaf, 0xfeb0], // ARABIC LETTER ZAIN
  43.     0x0633: [0xfeb1, 0xfeb2, 0xfeb3, 0xfeb4], // ARABIC LETTER SEEN
  44.     0x0634: [0xfeb5, 0xfeb6, 0xfeb7, 0xfeb8], // ARABIC LETTER SHEEN
  45.     0x0635: [0xfeb9, 0xfeba, 0xfebb, 0xfebc], // ARABIC LETTER SAD
  46.     0x0636: [0xfebd, 0xfebe, 0xfebf, 0xfec0], // ARABIC LETTER DAD
  47.     0x0637: [0xfec1, 0xfec2, 0xfec3, 0xfec4], // ARABIC LETTER TAH
  48.     0x0638: [0xfec5, 0xfec6, 0xfec7, 0xfec8], // ARABIC LETTER ZAH
  49.     0x0639: [0xfec9, 0xfeca, 0xfecb, 0xfecc], // ARABIC LETTER AIN
  50.     0x063a: [0xfecd, 0xfece, 0xfecf, 0xfed0], // ARABIC LETTER GHAIN
  51.     0x0641: [0xfed1, 0xfed2, 0xfed3, 0xfed4], // ARABIC LETTER FEH
  52.     0x0642: [0xfed5, 0xfed6, 0xfed7, 0xfed8], // ARABIC LETTER QAF
  53.     0x0643: [0xfed9, 0xfeda, 0xfedb, 0xfedc], // ARABIC LETTER KAF
  54.     0x0644: [0xfedd, 0xfede, 0xfedf, 0xfee0], // ARABIC LETTER LAM
  55.     0x0645: [0xfee1, 0xfee2, 0xfee3, 0xfee4], // ARABIC LETTER MEEM
  56.     0x0646: [0xfee5, 0xfee6, 0xfee7, 0xfee8], // ARABIC LETTER NOON
  57.     0x0647: [0xfee9, 0xfeea, 0xfeeb, 0xfeec], // ARABIC LETTER HEH
  58.     0x0648: [0xfeed, 0xfeee], // ARABIC LETTER WAW
  59.     0x0649: [0xfeef, 0xfef0, 64488, 64489], // ARABIC LETTER ALEF MAKSURA
  60.     0x064a: [0xfef1, 0xfef2, 0xfef3, 0xfef4], // ARABIC LETTER YEH
  61.     0x0671: [0xfb50, 0xfb51], // ARABIC LETTER ALEF WASLA
  62.     0x0677: [0xfbdd], // ARABIC LETTER U WITH HAMZA ABOVE
  63.     0x0679: [0xfb66, 0xfb67, 0xfb68, 0xfb69], // ARABIC LETTER TTEH
  64.     0x067a: [0xfb5e, 0xfb5f, 0xfb60, 0xfb61], // ARABIC LETTER TTEHEH
  65.     0x067b: [0xfb52, 0xfb53, 0xfb54, 0xfb55], // ARABIC LETTER BEEH
  66.     0x067e: [0xfb56, 0xfb57, 0xfb58, 0xfb59], // ARABIC LETTER PEH
  67.     0x067f: [0xfb62, 0xfb63, 0xfb64, 0xfb65], // ARABIC LETTER TEHEH
  68.     0x0680: [0xfb5a, 0xfb5b, 0xfb5c, 0xfb5d], // ARABIC LETTER BEHEH
  69.     0x0683: [0xfb76, 0xfb77, 0xfb78, 0xfb79], // ARABIC LETTER NYEH
  70.     0x0684: [0xfb72, 0xfb73, 0xfb74, 0xfb75], // ARABIC LETTER DYEH
  71.     0x0686: [0xfb7a, 0xfb7b, 0xfb7c, 0xfb7d], // ARABIC LETTER TCHEH
  72.     0x0687: [0xfb7e, 0xfb7f, 0xfb80, 0xfb81], // ARABIC LETTER TCHEHEH
  73.     0x0688: [0xfb88, 0xfb89], // ARABIC LETTER DDAL
  74.     0x068c: [0xfb84, 0xfb85], // ARABIC LETTER DAHAL
  75.     0x068d: [0xfb82, 0xfb83], // ARABIC LETTER DDAHAL
  76.     0x068e: [0xfb86, 0xfb87], // ARABIC LETTER DUL
  77.     0x0691: [0xfb8c, 0xfb8d], // ARABIC LETTER RREH
  78.     0x0698: [0xfb8a, 0xfb8b], // ARABIC LETTER JEH
  79.     0x06a4: [0xfb6a, 0xfb6b, 0xfb6c, 0xfb6d], // ARABIC LETTER VEH
  80.     0x06a6: [0xfb6e, 0xfb6f, 0xfb70, 0xfb71], // ARABIC LETTER PEHEH
  81.     0x06a9: [0xfb8e, 0xfb8f, 0xfb90, 0xfb91], // ARABIC LETTER KEHEH
  82.     0x06ad: [0xfbd3, 0xfbd4, 0xfbd5, 0xfbd6], // ARABIC LETTER NG
  83.     0x06af: [0xfb92, 0xfb93, 0xfb94, 0xfb95], // ARABIC LETTER GAF
  84.     0x06b1: [0xfb9a, 0xfb9b, 0xfb9c, 0xfb9d], // ARABIC LETTER NGOEH
  85.     0x06b3: [0xfb96, 0xfb97, 0xfb98, 0xfb99], // ARABIC LETTER GUEH
  86.     0x06ba: [0xfb9e, 0xfb9f], // ARABIC LETTER NOON GHUNNA
  87.     0x06bb: [0xfba0, 0xfba1, 0xfba2, 0xfba3], // ARABIC LETTER RNOON
  88.     0x06be: [0xfbaa, 0xfbab, 0xfbac, 0xfbad], // ARABIC LETTER HEH DOACHASHMEE
  89.     0x06c0: [0xfba4, 0xfba5], // ARABIC LETTER HEH WITH YEH ABOVE
  90.     0x06c1: [0xfba6, 0xfba7, 0xfba8, 0xfba9], // ARABIC LETTER HEH GOAL
  91.     0x06c5: [0xfbe0, 0xfbe1], // ARABIC LETTER KIRGHIZ OE
  92.     0x06c6: [0xfbd9, 0xfbda], // ARABIC LETTER OE
  93.     0x06c7: [0xfbd7, 0xfbd8], // ARABIC LETTER U
  94.     0x06c8: [0xfbdb, 0xfbdc], // ARABIC LETTER YU
  95.     0x06c9: [0xfbe2, 0xfbe3], // ARABIC LETTER KIRGHIZ YU
  96.     0x06cb: [0xfbde, 0xfbdf], // ARABIC LETTER VE
  97.     0x06cc: [0xfbfc, 0xfbfd, 0xfbfe, 0xfbff], // ARABIC LETTER FARSI YEH
  98.     0x06d0: [0xfbe4, 0xfbe5, 0xfbe6, 0xfbe7], //ARABIC LETTER E
  99.     0x06d2: [0xfbae, 0xfbaf], // ARABIC LETTER YEH BARREE
  100.     0x06d3: [0xfbb0, 0xfbb1] // ARABIC LETTER YEH BARREE WITH HAMZA ABOVE
  101.   };
  103.   /*
  104.     var ligaturesSubstitutionA = {
  105.         0xFBEA: []// ARABIC LIGATURE YEH WITH HAMZA ABOVE WITH ALEF ISOLATED FORM
  106.     };
  107.     */
  109.   var ligatures = {
  110.     0xfedf: {
  111.       0xfe82: 0xfef5, // ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE ISOLATED FORM
  112.       0xfe84: 0xfef7, // ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE ISOLATED FORM
  113.       0xfe88: 0xfef9, // ARABIC LIGATURE LAM WITH ALEF WITH HAMZA BELOW ISOLATED FORM
  114.       0xfe8e: 0xfefb // ARABIC LIGATURE LAM WITH ALEF ISOLATED FORM
  115.     },
  116.     0xfee0: {
  117.       0xfe82: 0xfef6, // ARABIC LIGATURE LAM WITH ALEF WITH MADDA ABOVE FINAL FORM
  118.       0xfe84: 0xfef8, // ARABIC LIGATURE LAM WITH ALEF WITH HAMZA ABOVE FINAL FORM
  119.       0xfe88: 0xfefa, // ARABIC LIGATURE LAM WITH ALEF WITH HAMZA BELOW FINAL FORM
  120.       0xfe8e: 0xfefc // ARABIC LIGATURE LAM WITH ALEF FINAL FORM
  121.     },
  122.     0xfe8d: { 0xfedf: { 0xfee0: { 0xfeea: 0xfdf2 } } }, // ALLAH
  123.     0x0651: {
  124.       0x064c: 0xfc5e, // Shadda + Dammatan
  125.       0x064d: 0xfc5f, // Shadda + Kasratan
  126.       0x064e: 0xfc60, // Shadda + Fatha
  127.       0x064f: 0xfc61, // Shadda + Damma
  128.       0x0650: 0xfc62 // Shadda + Kasra
  129.     }
  130.   };
  132.   var arabic_diacritics = {
  133.     1612: 64606, // Shadda + Dammatan
  134.     1613: 64607, // Shadda + Kasratan
  135.     1614: 64608, // Shadda + Fatha
  136.     1615: 64609, // Shadda + Damma
  137.     1616: 64610 // Shadda + Kasra
  138.   };
  140.   var alfletter = [1570, 1571, 1573, 1575];
  142.   var noChangeInForm = -1;
  143.   var isolatedForm = 0;
  144.   var finalForm = 1;
  145.   var initialForm = 2;
  146.   var medialForm = 3;
  148.   jsPDFAPI.__arabicParser__ = {};
  150.   //private
  151.   var isInArabicSubstitutionA = (jsPDFAPI.__arabicParser__.isInArabicSubstitutionA = function(
  152.     letter
  153.   ) {
  154.     return typeof arabicSubstitionA[letter.charCodeAt(0)] !== "undefined";
  155.   });
  157.   var isArabicLetter = (jsPDFAPI.__arabicParser__.isArabicLetter = function(
  158.     letter
  159.   ) {
  160.     return (
  161.       typeof letter === "string" &&
  162.       /^[\u0600-\u06FF\u0750-\u077F\u08A0-\u08FF\uFB50-\uFDFF\uFE70-\uFEFF]+$/.test(
  163.         letter
  164.       )
  165.     );
  166.   });
  168.   var isArabicEndLetter = (jsPDFAPI.__arabicParser__.isArabicEndLetter = function(
  169.     letter
  170.   ) {
  171.     return (
  172.       isArabicLetter(letter) &&
  173.       isInArabicSubstitutionA(letter) &&
  174.       arabicSubstitionA[letter.charCodeAt(0)].length <= 2
  175.     );
  176.   });
  178.   var isArabicAlfLetter = (jsPDFAPI.__arabicParser__.isArabicAlfLetter = function(
  179.     letter
  180.   ) {
  181.     return (
  182.       isArabicLetter(letter) && alfletter.indexOf(letter.charCodeAt(0)) >= 0
  183.     );
  184.   });
  186.   jsPDFAPI.__arabicParser__.arabicLetterHasIsolatedForm = function(letter) {
  187.     return (
  188.       isArabicLetter(letter) &&
  189.       isInArabicSubstitutionA(letter) &&
  190.       arabicSubstitionA[letter.charCodeAt(0)].length >= 1
  191.     );
  192.   };
  194.   var arabicLetterHasFinalForm = (jsPDFAPI.__arabicParser__.arabicLetterHasFinalForm = function(
  195.     letter
  196.   ) {
  197.     return (
  198.       isArabicLetter(letter) &&
  199.       isInArabicSubstitutionA(letter) &&
  200.       arabicSubstitionA[letter.charCodeAt(0)].length >= 2
  201.     );
  202.   });
  204.   jsPDFAPI.__arabicParser__.arabicLetterHasInitialForm = function(letter) {
  205.     return (
  206.       isArabicLetter(letter) &&
  207.       isInArabicSubstitutionA(letter) &&
  208.       arabicSubstitionA[letter.charCodeAt(0)].length >= 3
  209.     );
  210.   };
  212.   var arabicLetterHasMedialForm = (jsPDFAPI.__arabicParser__.arabicLetterHasMedialForm = function(
  213.     letter
  214.   ) {
  215.     return (
  216.       isArabicLetter(letter) &&
  217.       isInArabicSubstitutionA(letter) &&
  218.       arabicSubstitionA[letter.charCodeAt(0)].length == 4
  219.     );
  220.   });
  222.   var resolveLigatures = (jsPDFAPI.__arabicParser__.resolveLigatures = function(
  223.     letters
  224.   ) {
  225.     var i = 0;
  226.     var tmpLigatures = ligatures;
  227.     var result = "";
  228.     var effectedLetters = 0;
  230.     for (i = 0; i < letters.length; i += 1) {
  231.       if (typeof tmpLigatures[letters.charCodeAt(i)] !== "undefined") {
  232.         effectedLetters++;
  233.         tmpLigatures = tmpLigatures[letters.charCodeAt(i)];
  235.         if (typeof tmpLigatures === "number") {
  236.           result += String.fromCharCode(tmpLigatures);
  237.           tmpLigatures = ligatures;
  238.           effectedLetters = 0;
  239.         }
  240.         if (i === letters.length - 1) {
  241.           tmpLigatures = ligatures;
  242.           result += letters.charAt(i - (effectedLetters - 1));
  243.           i = i - (effectedLetters - 1);
  244.           effectedLetters = 0;
  245.         }
  246.       } else {
  247.         tmpLigatures = ligatures;
  248.         result += letters.charAt(i - effectedLetters);
  249.         i = i - effectedLetters;
  250.         effectedLetters = 0;
  251.       }
  252.     }
  254.     return result;
  255.   });
  257.   jsPDFAPI.__arabicParser__.isArabicDiacritic = function(letter) {
  258.     return (
  259.       letter !== undefined &&
  260.       arabic_diacritics[letter.charCodeAt(0)] !== undefined
  261.     );
  262.   };
  264.   var getCorrectForm = (jsPDFAPI.__arabicParser__.getCorrectForm = function(
  265.     currentChar,
  266.     beforeChar,
  267.     nextChar
  268.   ) {
  269.     if (!isArabicLetter(currentChar)) {
  270.       return -1;
  271.     }
  273.     if (isInArabicSubstitutionA(currentChar) === false) {
  274.       return noChangeInForm;
  275.     }
  276.     if (
  277.       !arabicLetterHasFinalForm(currentChar) ||
  278.       (!isArabicLetter(beforeChar) && !isArabicLetter(nextChar)) ||
  279.       (!isArabicLetter(nextChar) && isArabicEndLetter(beforeChar)) ||
  280.       (isArabicEndLetter(currentChar) && !isArabicLetter(beforeChar)) ||
  281.       (isArabicEndLetter(currentChar) && isArabicAlfLetter(beforeChar)) ||
  282.       (isArabicEndLetter(currentChar) && isArabicEndLetter(beforeChar))
  283.     ) {
  284.       return isolatedForm;
  285.     }
  287.     if (
  288.       arabicLetterHasMedialForm(currentChar) &&
  289.       isArabicLetter(beforeChar) &&
  290.       !isArabicEndLetter(beforeChar) &&
  291.       isArabicLetter(nextChar) &&
  292.       arabicLetterHasFinalForm(nextChar)
  293.     ) {
  294.       return medialForm;
  295.     }
  297.     if (isArabicEndLetter(currentChar) || !isArabicLetter(nextChar)) {
  298.       return finalForm;
  299.     }
  300.     return initialForm;
  301.   });
  303.   /**
  304.    * @name processArabic
  305.    * @function
  306.    * @param {string} text
  307.    * @returns {string}
  308.    */
  309.   var parseArabic = function(text) {
  310.     text = text || "";
  312.     var result = "";
  313.     var i = 0;
  314.     var j = 0;
  315.     var position = 0;
  316.     var currentLetter = "";
  317.     var prevLetter = "";
  318.     var nextLetter = "";
  320.     var words = text.split("\\s+");
  321.     var newWords = [];
  322.     for (i = 0; i < words.length; i += 1) {
  323.       newWords.push("");
  324.       for (j = 0; j < words[i].length; j += 1) {
  325.         currentLetter = words[i][j];
  326.         prevLetter = words[i][j - 1];
  327.         nextLetter = words[i][j + 1];
  328.         if (isArabicLetter(currentLetter)) {
  329.           position = getCorrectForm(currentLetter, prevLetter, nextLetter);
  330.           if (position !== -1) {
  331.             newWords[i] += String.fromCharCode(
  332.               arabicSubstitionA[currentLetter.charCodeAt(0)][position]
  333.             );
  334.           } else {
  335.             newWords[i] += currentLetter;
  336.           }
  337.         } else {
  338.           newWords[i] += currentLetter;
  339.         }
  340.       }
  342.       newWords[i] = resolveLigatures(newWords[i]);
  343.     }
  344.     result = newWords.join(" ");
  346.     return result;
  347.   };
  349.   var processArabic = (jsPDFAPI.__arabicParser__.processArabic = jsPDFAPI.processArabic = function() {
  350.     var text =
  351.       typeof arguments[0] === "string" ? arguments[0] : arguments[0].text;
  352.     var tmpText = [];
  353.     var result;
  355.     if (Array.isArray(text)) {
  356.       var i = 0;
  357.       tmpText = [];
  358.       for (i = 0; i < text.length; i += 1) {
  359.         if (Array.isArray(text[i])) {
  360.           tmpText.push([parseArabic(text[i][0]), text[i][1], text[i][2]]);
  361.         } else {
  362.           tmpText.push([parseArabic(text[i])]);
  363.         }
  364.       }
  365.       result = tmpText;
  366.     } else {
  367.       result = parseArabic(text);
  368.     }
  369.     if (typeof arguments[0] === "string") {
  370.       return result;
  371.     } else {
  372.       arguments[0].text = result;
  373.       return arguments[0];
  374.     }
  375.   });
  377.   jsPDFAPI.events.push(["preProcessText", processArabic]);
  378. })(jsPDF.API);