☰

Virastar Library

Cleaning-up Persian Texts!

Tento skript by neměl být instalován přímo. Jedná se o knihovnu, kterou by měly jiné skripty využívat pomocí meta příkazu // @require https://update.greatest.deepsurf.us/scripts/527228/1538801/Virastar%20Library.js 

  1. // ==UserScript==
  2. // @name         Virastar Library
  3. // @version      0.21.0
  4. // @description  Cleaning-up Persian Texts!
  5. // @homepage     https://github.com/brothersincode/virastar/
  6. // @namespace    amm1rr.com.virastar
  7. // @name:fa      کتابخانه ویراستار
  8. // @description:fa ویراستار متنِ فارسی
  9. // @grant        none
  10. // @updateURL    https://raw.githubusercontent.com/brothersincode/virastar/master/lib/virastar.js
  11. // @downloadURL  https://raw.githubusercontent.com/brothersincode/virastar/master/lib/virastar.js
  12. // @license      MIT
  13. // ==/UserScript==
  14.  
  15. /*!
  16. * Virastar - v0.21.0 - 2020-05-14
  17. * https://github.com/brothersincode/virastar
  18. * Licensed: MIT
  19. */
  20.  
  21. (function (name, global, definition) {
  22.   if (typeof module !== 'undefined') module.exports = definition();
  23.   else if (typeof define === 'function' && typeof define.amd === 'object') define(definition);
  24.   else if (typeof window !== 'undefined') window[name] = definition();
  25.   else global[name] = definition();
  26. }('Virastar', this, function () {
  27.   function Virastar (text, options) {
  28.     if (!(this instanceof Virastar)) {
  29.       return new Virastar(text, options);
  30.     }
  31.  
  32.     text = text || {};
  33.  
  34.     if (typeof text === 'object') {
  35.       this.opts = parseOptions(text);
  36.     } else if (typeof text === 'string') {
  37.       this.opts = parseOptions(options || {});
  38.       return cleanup(text);
  39.     }
  40.  
  41.     return this;
  42.   }
  43.  
  44.   function parseOptions (options) {
  45.     // @ref: https://scotch.io/bar-talk/copying-objects-in-javascript
  46.     var parsed = Object.assign({}, defaults);
  47.  
  48.     for (var i in parsed) {
  49.       if (options.hasOwnProperty(i)) { // eslint-disable-line no-prototype-builtins
  50.         parsed[i] = options[i];
  51.       }
  52.     }
  53.  
  54.     return parsed;
  55.   }
  56.  
  57.   function charReplace (text, fromBatch, toBatch) {
  58.     var fromChars = fromBatch.split('');
  59.     var toChars = toBatch.split('');
  60.     for (var i in fromChars) {
  61.       text = text.replace(newRegExp(fromChars[i]), toChars[i]);
  62.     }
  63.     return text;
  64.   }
  65.  
  66.   function arrReplace (text, array) {
  67.     for (var i in array) {
  68.       if (array.hasOwnProperty(i)) { // eslint-disable-line no-prototype-builtins
  69.         text = text.replace(newRegExp('[' + array[i] + ']'), i);
  70.       }
  71.     }
  72.     return text;
  73.   }
  74.  
  75.   function newRegExp (pattern, flags) {
  76.     return new RegExp(pattern, flags || 'g');
  77.   }
  78.  
  79.   var charsPersian = 'ءاآأإئؤبپتثجچحخدذرزژسشصضطظعغفقکگلمنوهیةيك';
  80.  
  81.   // @REF: https://en.wikipedia.org/wiki/Persian_alphabet#Diacritics
  82.   // `\u064e\u0650\u064f\u064b\u064d\u064c\u0651\u06c0`
  83.   var charsDiacritic = 'ًٌٍَُِّْ';
  84.  
  85.   // @source: https://github.com/jhermsmeier/uri.regex
  86.   var patternURI = "([A-Za-z][A-Za-z0-9+\\-.]*):(?:(//)(?:((?:[A-Za-z0-9\\-._~!$&'()*+,;=:]|%[0-9A-Fa-f]{2})*)@)?((?:\\[(?:(?:(?:(?:[0-9A-Fa-f]{1,4}:){6}|::(?:[0-9A-Fa-f]{1,4}:){5}|(?:[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){4}|(?:(?:[0-9A-Fa-f]{1,4}:){0,1}[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){3}|(?:(?:[0-9A-Fa-f]{1,4}:){0,2}[0-9A-Fa-f]{1,4})?::(?:[0-9A-Fa-f]{1,4}:){2}|(?:(?:[0-9A-Fa-f]{1,4}:){0,3}[0-9A-Fa-f]{1,4})?::[0-9A-Fa-f]{1,4}:|(?:(?:[0-9A-Fa-f]{1,4}:){0,4}[0-9A-Fa-f]{1,4})?::)(?:[0-9A-Fa-f]{1,4}:[0-9A-Fa-f]{1,4}|(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?))|(?:(?:[0-9A-Fa-f]{1,4}:){0,5}[0-9A-Fa-f]{1,4})?::[0-9A-Fa-f]{1,4}|(?:(?:[0-9A-Fa-f]{1,4}:){0,6}[0-9A-Fa-f]{1,4})?::)|[Vv][0-9A-Fa-f]+\\.[A-Za-z0-9\\-._~!$&'()*+,;=:]+)\\]|(?:(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)\\.){3}(?:25[0-5]|2[0-4][0-9]|[01]?[0-9][0-9]?)|(?:[A-Za-z0-9\\-._~!$&'()*+,;=]|%[0-9A-Fa-f]{2})*))(?::([0-9]*))?((?:/(?:[A-Za-z0-9\\-._~!$&'()*+,;=:@]|%[0-9A-Fa-f]{2})*)*)|/((?:(?:[A-Za-z0-9\\-._~!$&'()*+,;=:@]|%[0-9A-Fa-f]{2})+(?:/(?:[A-Za-z0-9\\-._~!$&'()*+,;=:@]|%[0-9A-Fa-f]{2})*)*)?)|((?:[A-Za-z0-9\\-._~!$&'()*+,;=:@]|%[0-9A-Fa-f]{2})+(?:/(?:[A-Za-z0-9\\-._~!$&'()*+,;=:@]|%[0-9A-Fa-f]{2})*)*)|)(?:\\?((?:[A-Za-z0-9\\-._~!$&'()*+,;=:@/?]|%[0-9A-Fa-f]{2})*))?(?:\\#((?:[A-Za-z0-9\\-._~!$&'()*+,;=:@/?]|%[0-9A-Fa-f]{2})*))?";
  87.   var patternAfter = '\\s.,;،؛!؟?"\'()[\\]{}“”«»';
  88.  
  89.   var defaults = {
  90.     // aggresive: true, // DEPRECATED
  91.     cleanup_begin_and_end: true,
  92.     cleanup_extra_marks: true,
  93.     cleanup_kashidas: true,
  94.     cleanup_line_breaks: true,
  95.     cleanup_rlm: true,
  96.     cleanup_spacing: true,
  97.     cleanup_zwnj: true,
  98.     decode_htmlentities: true,
  99.     fix_arabic_numbers: true,
  100.     fix_dashes: true,
  101.     fix_diacritics: true,
  102.     fix_english_numbers: true,
  103.     fix_english_quotes_pairs: true,
  104.     fix_english_quotes: true,
  105.     fix_hamzeh: true,
  106.     fix_hamzeh_arabic: false,
  107.     fix_misc_non_persian_chars: true,
  108.     fix_misc_spacing: true,
  109.     fix_numeral_symbols: true,
  110.     fix_perfix_spacing: true,
  111.     fix_persian_glyphs: true,
  112.     fix_punctuations: true,
  113.     fix_question_mark: true,
  114.     fix_spacing_for_braces_and_quotes: true,
  115.     fix_spacing_for_punctuations: true,
  116.     fix_suffix_misc: true,
  117.     fix_suffix_spacing: true,
  118.     fix_three_dots: true,
  119.     kashidas_as_parenthetic: true,
  120.     markdown_normalize_braces: true,
  121.     markdown_normalize_lists: true,
  122.     normalize_dates: true,
  123.     normalize_ellipsis: true,
  124.     normalize_eol: true,
  125.     preserve_braces: false,
  126.     preserve_brackets: false,
  127.     preserve_comments: true,
  128.     preserve_entities: true,
  129.     preserve_frontmatter: true,
  130.     preserve_HTML: true,
  131.     preserve_nbsps: true,
  132.     preserve_URIs: true,
  133.     remove_diacritics: false,
  134.     skip_markdown_ordered_lists_numbers_conversion: false
  135.   };
  136.  
  137.   var digits = '۱۲۳۴۵۶۷۸۹۰';
  138.  
  139.   /* eslint-disable */
  140.   var entities = {
  141.     'sbquo;': '\u201a',
  142.     'lsquo;': '\u2018',
  143.     'lsquor;': '\u201a',
  144.     'ldquo;': '\u201c',
  145.     'ldquor;': '\u201e',
  146.     'rdquo;': '\u201d',
  147.     'rdquor;': '\u201d',
  148.     'rsquo;': '\u2019',
  149.     'rsquor;': '\u2019',
  150.     'apos;': '\'',
  151.     'QUOT;': '"',
  152.     'QUOT': '"',
  153.     'quot;': '"',
  154.     'quot': '"',
  155.     'zwj;': '\u200d',
  156.     'ZWNJ;': '\u200c',
  157.     'zwnj;': '\u200c',
  158.     'shy;': '\u00ad' // wrongly used as zwnj
  159.   };
  160.  
  161.   // props @ebraminio/persiantools
  162.   var glyphs = {
  163.     // these two are for visually available ZWNJ #visualZwnj
  164.     '\u200cه': 'ﻫ',
  165.     'ی\u200c': 'ﻰﻲ',
  166.     'ﺃ': 'ﺄﺃ',
  167.     'ﺁ': 'ﺁﺂ',
  168.     'ﺇ': 'ﺇﺈ',
  169.     'ا': 'ﺎا',
  170.     'ب': 'ﺏﺐﺑﺒ',
  171.     'پ': 'ﭖﭗﭘﭙ',
  172.     'ت': 'ﺕﺖﺗﺘ',
  173.     'ث': 'ﺙﺚﺛﺜ',
  174.     'ج': 'ﺝﺞﺟﺠ',
  175.     'چ': 'ﭺﭻﭼﭽ',
  176.     'ح': 'ﺡﺢﺣﺤ',
  177.     'خ': 'ﺥﺦﺧﺨ',
  178.     'د': 'ﺩﺪ',
  179.     'ذ': 'ﺫﺬ',
  180.     'ر': 'ﺭﺮ',
  181.     'ز': 'ﺯﺰ',
  182.     'ژ': 'ﮊﮋ',
  183.     'س': 'ﺱﺲﺳﺴ',
  184.     'ش': 'ﺵﺶﺷﺸ',
  185.     'ص': 'ﺹﺺﺻﺼ',
  186.     'ض': 'ﺽﺾﺿﻀ',
  187.     'ط': 'ﻁﻂﻃﻄ',
  188.     'ظ': 'ﻅﻆﻇﻈ',
  189.     'ع': 'ﻉﻊﻋﻌ',
  190.     'غ': 'ﻍﻎﻏﻐ',
  191.     'ف': 'ﻑﻒﻓﻔ',
  192.     'ق': 'ﻕﻖﻗﻘ',
  193.     'ک': 'ﮎﮏﮐﮑﻙﻚﻛﻜ',
  194.     'گ': 'ﮒﮓﮔﮕ',
  195.     'ل': 'ﻝﻞﻟﻠ',
  196.     'م': 'ﻡﻢﻣﻤ',
  197.     'ن': 'ﻥﻦﻧﻨ',
  198.     'ه': 'ﻩﻪﻫﻬ',
  199.     'هٔ': 'ﮤﮥ',
  200.     'و': 'ﻭﻮ',
  201.     'ﺅ': 'ﺅﺆ',
  202.     'ی': 'ﯼﯽﯾﯿﻯﻰﻱﻲﻳﻴ',
  203.     'ئ': 'ﺉﺊﺋﺌ',
  204.     'لا': 'ﻼ',
  205.     'ﻹ': 'ﻺ',
  206.     'ﻷ': 'ﻸ',
  207.     'ﻵ': 'ﻶ'
  208.   };
  209.   /* eslint-enable */
  210.  
  211.   function cleanup (text, options) {
  212.     if (typeof text !== 'string') {
  213.       throw new TypeError('Expected a String, but received ' + typeof text);
  214.     }
  215.  
  216.     // dont bother if its empty or whitespace
  217.     if (!text.trim()) {
  218.       return text;
  219.     }
  220.  
  221.     var opts = options ? parseOptions(options) : this.opts;
  222.  
  223.     // single space paddings around the string
  224.     text = ' ' + text + ' ';
  225.  
  226.     // preserves frontmatter data in the text
  227.     if (opts.preserve_frontmatter) {
  228.       var frontmatter = [];
  229.       text = text.replace(/^ ---[\S\s]*?---\n/g, function (matched) {
  230.         frontmatter.push(matched);
  231.         return ' __FRONTMATTER__PRESERVER__ ';
  232.       });
  233.     }
  234.  
  235.     // preserves all html tags in the text
  236.     // @props: @wordpress/wordcount
  237.     if (opts.preserve_HTML) {
  238.       var html = [];
  239.       text = text.replace(/<\/?[a-z][^>]*?>/gi, function (matched) {
  240.         html.push(matched);
  241.         return ' __HTML__PRESERVER__ ';
  242.       });
  243.     }
  244.  
  245.     // preserves all html comments in the text
  246.     // @props: @wordpress/wordcount
  247.     if (opts.preserve_comments) {
  248.       var comments = [];
  249.       text = text.replace(/<!--[\s\S]*?-->/g, function (matched) {
  250.         comments.push(matched);
  251.         return ' __COMMENT__PRESERVER__ ';
  252.       });
  253.     }
  254.  
  255.     // preserves strings inside square brackets (`[]`)
  256.     if (opts.preserve_brackets) {
  257.       var brackets = [];
  258.       text = text.replace(/(\[.*?\])/g, function (matched) {
  259.         brackets.push(matched);
  260.         return ' __BRACKETS__PRESERVER__ ';
  261.       });
  262.     }
  263.  
  264.     // preserves strings inside curly braces (`{}`)
  265.     if (opts.preserve_braces) {
  266.       var braces = [];
  267.       text = text.replace(/(\{.*?\})/g, function (matched) {
  268.         braces.push(matched);
  269.         return ' __BRACES__PRESERVER__ ';
  270.       });
  271.     }
  272.  
  273.     // preserves all uri strings in the text
  274.     if (opts.preserve_URIs) {
  275.       var mdlinks = [];
  276.       var uris = [];
  277.  
  278.       // stores markdown links separetly
  279.       text = text.replace(/]\((.*?)\)/g, function (matched, link) {
  280.         if (link) {
  281.           mdlinks.push(link.trim());
  282.           return '](__MD_LINK__PRESERVER__)'; // no padding!
  283.         }
  284.         return matched;
  285.       });
  286.  
  287.       text = text.replace(newRegExp(patternURI), function (matched) {
  288.         uris.push(matched);
  289.         return ' __URI__PRESERVER__ ';
  290.       });
  291.     }
  292.  
  293.     // preserves all no-break space entities in the text
  294.     if (opts.preserve_nbsps) {
  295.       var nbsps = [];
  296.       text = text.replace(/&nbsp;|&#160;/gi, function (matched) {
  297.         nbsps.push(matched);
  298.         return ' __NBSPS__PRESERVER__ ';
  299.       });
  300.     }
  301.  
  302.     if (opts.decode_htmlentities) {
  303.       text = decodeHTMLEntities(text);
  304.     }
  305.  
  306.     // preserves all html entities in the text
  307.     // @props: @substack/node-ent
  308.     if (opts.preserve_entities) {
  309.       var entities = [];
  310.       text = text.replace(/&(#?[^;\W]+;?)/g, function (matched) {
  311.         entities.push(matched);
  312.         return ' __ENTITIES__PRESERVER__ ';
  313.       });
  314.     }
  315.  
  316.     if (opts.normalize_eol) {
  317.       text = normalizeEOL(text);
  318.     }
  319.  
  320.     if (opts.fix_persian_glyphs) {
  321.       text = fixPersianGlyphs(text);
  322.     }
  323.  
  324.     if (opts.fix_dashes) {
  325.       text = fixDashes(text);
  326.     }
  327.  
  328.     if (opts.fix_three_dots) {
  329.       text = fixThreeDots(text);
  330.     }
  331.  
  332.     if (opts.normalize_ellipsis) {
  333.       text = normalizeEllipsis(text);
  334.     }
  335.  
  336.     if (opts.fix_english_quotes_pairs) {
  337.       text = fixEnglishQuotesPairs(text);
  338.     }
  339.  
  340.     if (opts.fix_english_quotes) {
  341.       text = fixEnglishQuotes(text);
  342.     }
  343.  
  344.     if (opts.fix_hamzeh) {
  345.       if (opts.fix_hamzeh_arabic) {
  346.         text = fixHamzehArabic(text);
  347.       }
  348.  
  349.       text = fixHamzeh(text);
  350.     } else if (opts.fix_suffix_spacing) {
  351.       if (opts.fix_hamzeh_arabic) {
  352.         text = fixHamzehArabicAlt(text);
  353.       }
  354.  
  355.       text = fixSuffixSpacingHamzeh(text);
  356.     }
  357.  
  358.     if (opts.cleanup_rlm) {
  359.       text = cleanupRLM(text);
  360.     }
  361.  
  362.     if (opts.cleanup_zwnj) {
  363.       text = cleanupZWNJ(text);
  364.     }
  365.  
  366.     if (opts.fix_arabic_numbers) {
  367.       text = fixArabicNumbers(text);
  368.     }
  369.  
  370.     // word tokenizer
  371.     text = text.replace(/(^|\s+)([[({"'“«]?)(\S+)([\])}"'”»]?)(?=($|\s+))/g,
  372.       function (matched, before, leadings, word, trailings, after) {
  373.         // should not replace to persian chars in english phrases
  374.         if (word.match(/[a-zA-Z\-_]{2,}/g)) {
  375.           return matched;
  376.         }
  377.  
  378.         // should not touch sprintf directives
  379.         // @source: https://stackoverflow.com/a/8915445/
  380.         if (word.match(/%(?:\d+\$)?[+-]?(?:[ 0]|'.{1})?-?\d*(?:\.\d+)?[bcdeEufFgGosxX]/g)) {
  381.           return matched;
  382.         }
  383.  
  384.         // should not touch numbers in html entities
  385.         if (word.match(/&#\d+;/g)) {
  386.           return matched;
  387.         }
  388.  
  389.         // skips converting english numbers of ordered lists in markdown
  390.         if (opts.skip_markdown_ordered_lists_numbers_conversion && (matched + trailings + after).match(/(?:(?:\r?\n)|(?:\r\n?)|(?:^|\n))\d+\.\s/)) {
  391.           return matched;
  392.         }
  393.  
  394.         if (opts.fix_english_numbers) {
  395.           matched = fixEnglishNumbers(matched);
  396.         }
  397.  
  398.         if (opts.fix_numeral_symbols) {
  399.           matched = fixNumeralSymbols(matched);
  400.         }
  401.  
  402.         if (opts.fix_punctuations) {
  403.           matched = fixPunctuations(matched);
  404.         }
  405.  
  406.         if (opts.fix_misc_non_persian_chars) {
  407.           matched = fixMiscNonPersianChars(matched);
  408.         }
  409.  
  410.         if (opts.fix_question_mark) {
  411.           matched = fixQuestionMark(matched);
  412.         }
  413.  
  414.         return matched;
  415.       }
  416.     );
  417.  
  418.     if (opts.normalize_dates) {
  419.       text = normalizeDates(text);
  420.     }
  421.  
  422.     if (opts.fix_perfix_spacing) {
  423.       text = fixPerfixSpacing(text);
  424.     }
  425.  
  426.     if (opts.fix_suffix_spacing) {
  427.       text = fixSuffixSpacing(text);
  428.     }
  429.  
  430.     if (opts.fix_suffix_misc) {
  431.       text = fixSuffixMisc(text);
  432.     }
  433.  
  434.     if (opts.fix_spacing_for_braces_and_quotes) {
  435.       text = fixBracesSpacing(text);
  436.     }
  437.  
  438.     if (opts.cleanup_extra_marks) {
  439.       text = cleanupExtraMarks(text);
  440.     }
  441.  
  442.     if (opts.fix_spacing_for_punctuations) {
  443.       text = fixPunctuationSpacing(text);
  444.     }
  445.  
  446.     if (opts.kashidas_as_parenthetic) {
  447.       text = kashidasAsParenthetic(text);
  448.     }
  449.  
  450.     if (opts.cleanup_kashidas) {
  451.       text = cleanupKashidas(text);
  452.     }
  453.  
  454.     if (opts.markdown_normalize_braces) {
  455.       text = markdownNormalizeBraces(text);
  456.     }
  457.  
  458.     if (opts.markdown_normalize_lists) {
  459.       text = markdownNormalizeLists(text);
  460.     }
  461.  
  462.     // doing it again after `fixPunctuationSpacing()`
  463.     if (opts.fix_spacing_for_braces_and_quotes) {
  464.       text = fixBracesSpacingInside(text);
  465.     }
  466.  
  467.     if (opts.fix_misc_spacing) {
  468.       text = fixMiscSpacing(text);
  469.     }
  470.  
  471.     if (opts.remove_diacritics) {
  472.       text = removeDiacritics(text);
  473.     } else if (opts.fix_diacritics) {
  474.       text = fixDiacritics(text);
  475.     }
  476.  
  477.     if (opts.cleanup_spacing) {
  478.       text = cleanupSpacing(text);
  479.     }
  480.  
  481.     if (opts.cleanup_zwnj) {
  482.       text = cleanupZWNJLate(text);
  483.     }
  484.  
  485.     if (opts.cleanup_line_breaks) {
  486.       text = cleanupLineBreaks(text);
  487.     }
  488.  
  489.     // bringing back entities
  490.     if (opts.preserve_entities) {
  491.       text = text.replace(/[ ]?__ENTITIES__PRESERVER__[ ]?/g, function () {
  492.         return entities.shift();
  493.       });
  494.     }
  495.  
  496.     // bringing back nbsps
  497.     if (opts.preserve_nbsps) {
  498.       text = text.replace(/[ ]?__NBSPS__PRESERVER__[ ]?/g, function () {
  499.         return nbsps.shift();
  500.       });
  501.     }
  502.  
  503.     // bringing back URIs
  504.     if (opts.preserve_URIs) {
  505.       // no padding!
  506.       text = text.replace(/__MD_LINK__PRESERVER__/g, function () {
  507.         return mdlinks.shift();
  508.       });
  509.  
  510.       text = text.replace(/[ ]?__URI__PRESERVER__[ ]?/g, function () {
  511.         return uris.shift();
  512.       });
  513.     }
  514.  
  515.     // bringing back braces
  516.     if (opts.preserve_braces) {
  517.       text = text.replace(/[ ]?__BRACES__PRESERVER__[ ]?/g, function () {
  518.         return braces.shift();
  519.       });
  520.     }
  521.  
  522.     // bringing back brackets
  523.     if (opts.preserve_brackets) {
  524.       text = text.replace(/[ ]?__BRACKETS__PRESERVER__[ ]?/g, function () {
  525.         return brackets.shift();
  526.       });
  527.     }
  528.  
  529.     // bringing back HTML comments
  530.     if (opts.preserve_comments) {
  531.       text = text.replace(/[ ]?__COMMENT__PRESERVER__[ ]?/g, function () {
  532.         return comments.shift();
  533.       });
  534.     }
  535.  
  536.     // bringing back HTML tags
  537.     if (opts.preserve_HTML) {
  538.       text = text.replace(/[ ]?__HTML__PRESERVER__[ ]?/g, function () {
  539.         return html.shift();
  540.       });
  541.     }
  542.  
  543.     // bringing back frontmatter
  544.     if (opts.preserve_frontmatter) {
  545.       text = text.replace(/[ ]?__FRONTMATTER__PRESERVER__[ ]?/g, function () {
  546.         return frontmatter.shift();
  547.       });
  548.     }
  549.  
  550.     if (opts.cleanup_begin_and_end) {
  551.       text = cleanupBeginAndEnd(text);
  552.     } else {
  553.       // removes single space paddings around the string
  554.       text = text.replace(/^[ ]/g, '').replace(/[ ]$/g, '');
  555.     }
  556.  
  557.     return text;
  558.   }
  559.  
  560.   // props @ebraminio/persiantools
  561.   function cleanupZWNJ (text) {
  562.     return text
  563.  
  564.       // converts all soft hyphens (&shy;) into zwnj
  565.       .replace(/\u00ad/g, '\u200c')
  566.  
  567.       // removes more than one zwnj
  568.       .replace(/\u200c{2,}/g, '\u200c')
  569.  
  570.       // cleans zwnj before and after numbers, english words, spaces and punctuations
  571.       .replace(/\u200c([\w\s0-9۰-۹[\](){}«»“”.…,:;?!$%@#*=+\-/\\،؛٫٬×٪؟ـ])/g, '$1')
  572.       .replace(/([\w\s0-9۰-۹[\](){}«»“”.…,:;?!$%@#*=+\-/\\،؛٫٬×٪؟ـ])\u200c/g, '$1')
  573.  
  574.       // removes unnecessary zwnj on start/end of each line
  575.       .replace(/(^\u200c|\u200c$)/gm, '')
  576.     ;
  577.   }
  578.  
  579.   // late checks for zwnjs
  580.   function cleanupZWNJLate (text) {
  581.     return text
  582.  
  583.       // cleans zwnj after characters that don't conncet to the next
  584.       .replace(/([إأةؤورزژاآدذ،؛,:«»\\/@#$٪×*()ـ\-=|])\u200c/g, '$1')
  585.     ;
  586.   }
  587.  
  588.   // converts numeral and selected html character-sets into original characters
  589.   // @props: @substack/node-ent
  590.   function decodeHTMLEntities (text) {
  591.     return text.replace(/&(#?[^;\W]+;?)/g, function (matched, match) {
  592.       var n;
  593.       if ((n = /^#(\d+);?$/.exec(match))) {
  594.         return String.fromCharCode(parseInt(n[1], 10));
  595.       } else if ((n = /^#[Xx]([A-Fa-f0-9]+);?/.exec(match))) {
  596.         return String.fromCharCode(parseInt(n[1], 16));
  597.       } else {
  598.         var hasSemi = /;$/.test(match);
  599.         var withoutSemi = hasSemi ? match.replace(/;$/, '') : match;
  600.         var target = entities[withoutSemi] || (hasSemi && entities[match]);
  601.  
  602.         if (typeof target === 'number') {
  603.           return String.fromCharCode(target);
  604.         } else if (typeof target === 'string') {
  605.           return target;
  606.         } else {
  607.           return '&' + match;
  608.         }
  609.       }
  610.     });
  611.   }
  612.  
  613.   function normalizeEOL (text) {
  614.     return text
  615.  
  616.       // replaces windows end of lines with unix eol (`\n`)
  617.       .replace(/(\r?\n)|(\r\n?)/g, '\n')
  618.     ;
  619.   }
  620.  
  621.   function fixDashes (text) {
  622.     return text
  623.  
  624.       // replaces triple dash to mdash
  625.       .replace(/-{3}/g, '—')
  626.  
  627.       // replaces double dash to ndash
  628.       .replace(/-{2}/g, '–')
  629.     ;
  630.   }
  631.  
  632.   function fixThreeDots (text) {
  633.     return text
  634.  
  635.       // removes spaces between dots
  636.       .replace(/\.([ ]+)(?=[.])/g, '.')
  637.  
  638.       // replaces three dots with ellipsis character
  639.       .replace(/[ \t]*\.{3,}/g, '…')
  640.     ;
  641.   }
  642.  
  643.   function normalizeEllipsis (text) {
  644.     return text
  645.  
  646.       // replaces more than one ellipsis with one
  647.       .replace(/(…){2,}/g, '…')
  648.  
  649.       // replaces (space|tab|zwnj) after ellipsis with one space
  650.       // NOTE: allows for space before ellipsis
  651.       .replace(/([ ]{1,})*…[ \t\u200c]*/g, '$1… ')
  652.     ;
  653.   }
  654.  
  655.   function fixEnglishQuotesPairs (text) {
  656.     return text
  657.  
  658.       // replaces english quote pairs with their persian equivalent
  659.       .replace(/(“)(.+?)(”)/g, '«$2»')
  660.     ;
  661.   }
  662.  
  663.   // replaces english quote marks with their persian equivalent
  664.   function fixEnglishQuotes (text) {
  665.     return text
  666.       .replace(/(["'`]+)(.+?)(\1)/g, '«$2»')
  667.     ;
  668.   }
  669.  
  670.   function fixHamzeh (text) {
  671.     var replacement = '$1هٔ$3';
  672.     return text
  673.  
  674.       // replaces ه followed by (space|ZWNJ|lrm) follow by ی with هٔ
  675.       .replace(/(\S)(ه[\s\u200c\u200e]+[یي])([\s\u200c\u200e])/g, replacement) // heh + ye
  676.  
  677.       // replaces ه followed by (space|ZWNJ|lrm|nothing) follow by ء with هٔ
  678.       .replace(/(\S)(ه[\s\u200c\u200e]?\u0621)([\s\u200c\u200e])/g, replacement) // heh + standalone hamza
  679.  
  680.       // replaces هٓ or single-character ۀ with the standard هٔ
  681.       // props @ebraminio/persiantools
  682.       .replace(/(ۀ|هٓ)/g, 'هٔ')
  683.     ;
  684.   }
  685.  
  686.   function fixHamzehArabic (text) {
  687.     return text
  688.  
  689.       // converts arabic hamzeh ة to هٔ
  690.       .replace(/(\S)ة([\s\u200c\u200e])/g, '$1هٔ$2')
  691.     ;
  692.   }
  693.  
  694.   function fixHamzehArabicAlt (text) {
  695.     return text
  696.       // converts arabic hamzeh ة to ه‌ی
  697.       .replace(/(\S)ة([\s\u200c\u200e])/g, '$1ه‌ی$2')
  698.     ;
  699.   }
  700.  
  701.   function cleanupRLM (text) {
  702.     return text
  703.       // converts Right-to-left marks followed by persian characters to
  704.       // zero-width non-joiners (ZWNJ)
  705.       .replace(/([^a-zA-Z\-_])(\u200F)/g, '$1\u200c')
  706.     ;
  707.   }
  708.  
  709.   // converts incorrect persian glyphs to standard characters
  710.   function fixPersianGlyphs (text) {
  711.     return arrReplace(text, glyphs);
  712.   }
  713.  
  714.   // props @ebraminio/persiantools
  715.   function fixMiscNonPersianChars (text) {
  716.     return charReplace(text, 'كڪيىۍېہە', 'ککییییههه');
  717.     // return text
  718.     //   .replace(/ك/g, 'ک') // arabic kaf
  719.     //   .replace(/ڪ/g, 'ک') // arabic letter swash kaf
  720.     //   .replace(/ي/g, 'ی') // arabic
  721.     //   .replace(/ى/g, 'ی') // urdu
  722.     //   .replace(/ۍ/g, 'ی') // pushtu
  723.     //   .replace(/ې/g, 'ی') // uyghur
  724.     //   .replace(/ہ/g, 'ه') // converts &#x06C1; to &#x0647; ہہہہ to ههه
  725.     //   .replace(/[ەھ]/g, 'ه'); // kurdish
  726.   }
  727.  
  728.   // replaces english numbers with their persian equivalent
  729.   function fixEnglishNumbers (text) {
  730.     return charReplace(text, '1234567890', digits);
  731.   }
  732.  
  733.   // replaces arabic numbers with their persian equivalent
  734.   function fixArabicNumbers (text) {
  735.     return charReplace(text, '١٢٣٤٥٦٧٨٩٠', digits);
  736.   }
  737.  
  738.   // @REF: https://github.com/shkarimpour/pholiday/pull/5/files
  739.   function convertPersianNumbers (text) {
  740.     return text.replace(/[\u0660-\u0669\u06f0-\u06f9]/g, function (char) {
  741.       return char.charCodeAt(0) & 0xf;
  742.     });
  743.   }
  744.  
  745.   function fixNumeralSymbols (text) {
  746.     return text
  747.  
  748.       // replaces english percent signs (U+066A)
  749.       // props @ebraminio/persiantools
  750.       .replace(/([۰-۹]) ?%/g, '$1٪')
  751.  
  752.       // replaces dots between numbers into decimal separator (U+066B)
  753.       // props @ebraminio/persiantools
  754.       .replace(/([۰-۹])\.(?=[۰-۹])/g, '$1٫')
  755.  
  756.       // replaces commas between numbers into thousands separator (U+066C)
  757.       // props @languagetool-org
  758.       .replace(/([۰-۹]),(?=[۰-۹])/g, '$1٬')
  759.     ;
  760.   }
  761.  
  762.   function normalizeDates (text) {
  763.     return text
  764.  
  765.       // re-orders date parts with slash as delimiter
  766.       .replace(/([0-9۰-۹]{1,2})([/-])([0-9۰-۹]{1,2})\2([0-9۰-۹]{4})/g, function (matched, day, delimiter, month, year) {
  767.         return year + '/' + month + '/' + day;
  768.       })
  769.     ;
  770.   }
  771.  
  772.   function fixPunctuations (text) {
  773.     return charReplace(text, ',;', '،؛');
  774.   }
  775.  
  776.   // replaces question marks with its persian equivalent
  777.   function fixQuestionMark (text) {
  778.     return text
  779.       .replace(/(\?)/g, '\u061F') // \u061F = ؟
  780.     ;
  781.   }
  782.  
  783.   // puts zwnj between the word and the prefix:
  784.   // - mi* nemi* bi*
  785.   // NOTE: there's a possible bug here: prefixes could be separate nouns
  786.   function fixPerfixSpacing (text) {
  787.     var replacement = '$1\u200c$3';
  788.     return text
  789.       .replace(/((\s|^)ن?می) ([^ ])/g, replacement)
  790.       .replace(/((\s|^)بی) ([^ ])/g, replacement) // props @zoghal
  791.     ;
  792.   }
  793.  
  794.   // puts zwnj between the word and the suffix
  795.   // NOTE: possible bug: suffixes could be nouns
  796.   function fixSuffixSpacing (text) {
  797.     var replacement = '$1\u200c$2';
  798.     return text
  799.  
  800.       // must done before others
  801.       // *ha *haye
  802.       .replace(newRegExp('([' + charsPersian + charsDiacritic + ']) (ها(ی)?[' + patternAfter + '])'), replacement)
  803.  
  804.       // *am *at *ash *ei *eid *eem *and *man *tan *shan
  805.       .replace(newRegExp('([' + charsPersian + charsDiacritic + ']) ((ام|ات|اش|ای|اید|ایم|اند|مان|تان|شان)[' + patternAfter + '])'), replacement)
  806.  
  807.       // *tar *tari *tarin
  808.       .replace(newRegExp('([' + charsPersian + charsDiacritic + ']) (تر((ی)|(ین))?[' + patternAfter + '])'), replacement)
  809.  
  810.       // *hayee *hayam *hayat *hayash *hayetan *hayeman *hayeshan
  811.       .replace(newRegExp('([' + charsPersian + charsDiacritic + ']) ((هایی|هایم|هایت|هایش|هایمان|هایتان|هایشان)[' + patternAfter + '])'), replacement)
  812.     ;
  813.   }
  814.  
  815.   function fixSuffixSpacingHamzeh (text) {
  816.     var replacement = '$1\u0647\u200c\u06cc$3';
  817.     return text
  818.  
  819.       // heh + ye
  820.       .replace(/(\S)(ه[\s\u200c]+[یي])([\s\u200c])/g, replacement)
  821.  
  822.       // heh + standalone hamza
  823.       .replace(/(\S)(ه[\s\u200c]?\u0621)([\s\u200c])/g, replacement)
  824.  
  825.       // heh + hamza above
  826.       .replace(/(\S)(ه[\s\u200c]?\u0654)([\s\u200c])/g, replacement)
  827.     ;
  828.   }
  829.  
  830.   function fixSuffixMisc (text) {
  831.     return text
  832.       // replaces ه followed by ئ or ی, and then by ی, with ه\u200cای,
  833.       // EXAMPLE: خانه‌ئی becomes خانه‌ای
  834.       // props @ebraminio/persiantools
  835.       .replace(/(\S)ه[\u200c\u200e][ئی]ی([\s\u200c\u200e])/g, '$1ه\u200cای$2')
  836.     ;
  837.   }
  838.  
  839.   function cleanupExtraMarks (text) {
  840.     return text
  841.  
  842.       // removes space between different/same marks (combining for cleanup)
  843.       .replace(/([؟?!])([ ]+)(?=[؟?!])/g, '$1')
  844.  
  845.       // replaces more than one exclamation mark with just one
  846.       .replace(/(!){2,}/g, '$1')
  847.       // replaces more than one english or persian question mark with just one
  848.       .replace(/(\u061F|\?){2,}/g, '$1') // \u061F = `؟`
  849.       // re-orders consecutive marks
  850.       .replace(/(!)([ \t]*)([\u061F?])/g, '$3$1') // `?!` --> `!?`
  851.     ;
  852.   }
  853.  
  854.   // replaces kashidas to ndash in parenthetic
  855.   function kashidasAsParenthetic (text) {
  856.     return text
  857.       .replace(/(\s)\u0640+/g, '$1–')
  858.       .replace(/\u0640+(\s)/g, '–$1')
  859.     ;
  860.   }
  861.  
  862.   function cleanupKashidas (text) {
  863.     return text
  864.       // converts kashida between numbers to ndash
  865.       .replace(/([0-9۰-۹]+)ـ+([0-9۰-۹]+)/g, '$1–$2')
  866.  
  867.       // removes all kashidas between non-whitespace characters
  868.       // MAYBE: more punctuations
  869.       .replace(/([^\s.])\u0640+(?![\s.])/g, '$1')
  870.     ;
  871.   }
  872.  
  873.   function fixPunctuationSpacing (text) {
  874.     return text
  875.       // removes space before punctuations
  876.       .replace(/[ \t\u200c]*([:;,؛،.؟?!]{1})/g, '$1')
  877.  
  878.       // removes more than one space after punctuations
  879.       // except followed by new-lines (or preservers)
  880.       .replace(/([:;,؛،.؟?!]{1})[ \t\u200c]*(?!\n|_{2})/g, '$1 ')
  881.  
  882.       // removes space after colon that separates time parts
  883.       .replace(/([0-9۰-۹]+):\s+([0-9۰-۹]+)/g, '$1:$2')
  884.  
  885.       // removes space after dots in numbers
  886.       .replace(/([0-9۰-۹]+)\. ([0-9۰-۹]+)/g, '$1.$2')
  887.  
  888.       // removes space before common domain tlds
  889.       .replace(/([\w\-_]+)\. (ir|com|org|net|info|edu|me)([\s/\\\])»:;.])/g, '$1.$2$3')
  890.  
  891.       // removes space between different/same marks (double-check)
  892.       .replace(/([؟?!])([ ]+)(?=[؟?!])/g, '$1')
  893.     ;
  894.   }
  895.  
  896.   function fixBracesSpacing (text) {
  897.     var replacement = ' $1$2$3 ';
  898.     return text
  899.       // removes inside spaces and more than one outside
  900.       // for `()`, `[]`, `{}`, `“”` and `«»`
  901.       .replace(/[ \t\u200c]*(\()\s*([^)]+?)\s*?(\))[ \t\u200c]*/g, replacement)
  902.       .replace(/[ \t\u200c]*(\[)\s*([^\]]+?)\s*?(\])[ \t\u200c]*/g, replacement)
  903.       .replace(/[ \t\u200c]*(\{)\s*([^}]+?)\s*?(\})[ \t\u200c]*/g, replacement)
  904.       .replace(/[ \t\u200c]*(“)\s*([^”]+?)\s*?(”)[ \t\u200c]*/g, replacement)
  905.       .replace(/[ \t\u200c]*(«)\s*([^»]+?)\s*?(»)[ \t\u200c]*/g, replacement)
  906.     ;
  907.   }
  908.  
  909.   function fixBracesSpacingInside (text) {
  910.     var replacement = '$1$2$3';
  911.     return text
  912.       // removes inside spaces for `()`, `[]`, `{}`, `“”` and `«»`
  913.       .replace(/(\()\s*([^)]+?)\s*?(\))/g, replacement)
  914.       .replace(/(\[)\s*([^\]]+?)\s*?(\])/g, replacement)
  915.       .replace(/(\{)\s*([^}]+?)\s*?(\})/g, replacement)
  916.       .replace(/(“)\s*([^”]+?)\s*?(”)/g, replacement)
  917.       .replace(/(«)\s*([^»]+?)\s*?(»)/g, replacement)
  918.  
  919.       // NOTE: must be here, wierd not working if on `markdownNormalizeBraces()`
  920.       // removes markdown link spaces inside normal ()
  921.       .replace(/(\(\[.*?\]\(.*?\))\s+(\))/g, '$1$2')
  922.     ;
  923.   }
  924.  
  925.   function markdownNormalizeBraces (text) {
  926.     return text
  927.       // removes space between ! and opening brace on markdown images
  928.       // EXAMPLE: `! [alt] (src)` --> `![alt](src)`
  929.       .replace(/! (\[.*?\])[ ]?(\(.*?\))[ ]?/g, '!$1$2')
  930.  
  931.       // removes spaces between [] and ()
  932.       // EXAMPLE: `[text] (link)` --> `[text](link)`
  933.       .replace(/(\[.*?\])[ \t]+(\(.*?\))/g, '$1$2')
  934.  
  935.       // removes spaces inside double () [] {}
  936.       // EXAMPLE: `[[ text ]]` --> `[[text]]`
  937.       .replace(/\(\([ \t]*(.*?)[ \t]*\)\)/g, '(($1))')
  938.       .replace(/\[\[[ \t]*(.*?)[ \t]*\]\]/g, '[[$1]]')
  939.       .replace(/\{\{[ \t]*(.*?)[ \t]*\}\}/g, '{{$1}}')
  940.       .replace(/\{\{\{[ \t]*(.*?)[ \t]*\}\}\}/g, '{{{$1}}}') // mustache escape
  941.  
  942.       // removes spaces between double () [] {}
  943.       // EXAMPLE: `[[text] ]` --> `[[text]]`
  944.       .replace(/(\(\(.*\))[ \t]+(\))/g, '$1$2')
  945.       .replace(/(\[\[.*\])[ \t]+(\])/g, '$1$2')
  946.       .replace(/(\{\{.*\})[ \t]+(\})/g, '$1$2')
  947.     ;
  948.   }
  949.  
  950.   function markdownNormalizeLists (text) {
  951.     return text
  952.       // removes extra line between two items list
  953.       .replace(/((\n|^)\*.*?)\n+(?=\n\*)/g, '$1')
  954.       .replace(/((\n|^)-.*?)\n+(?=\n-)/g, '$1')
  955.       .replace(/((\n|^)#.*?)\n+(?=\n#)/g, '$1')
  956.     ;
  957.   }
  958.  
  959.   function fixMiscSpacing (text) {
  960.     return text
  961.  
  962.       // removes space before parentheses on misc cases
  963.       .replace(/ \((ص|عج|س|ع|ره)\)/g, '($1)')
  964.  
  965.       // removes space before braces containing numbers
  966.       .replace(/ \[([0-9۰-۹]+)\]/g, '[$1]')
  967.     ;
  968.   }
  969.  
  970.   function fixDiacritics (text) {
  971.     return text
  972.       // cleans zwnj before diacritic characters
  973.       .replace(newRegExp('\u200c([' + charsDiacritic + '])'), '$1')
  974.  
  975.       // cleans more than one diacritic characters
  976.       // props @languagetool-org
  977.       .replace(newRegExp('(.*)([' + charsDiacritic + ']){2,}(.*)'), '$1$2$3')
  978.  
  979.       // cleans spaces before diacritic characters
  980.       .replace(newRegExp('(\\S)[ ]+([' + charsDiacritic + '])'), '$1$2')
  981.     ;
  982.   }
  983.  
  984.   function removeDiacritics (text) {
  985.     return text
  986.  
  987.       // removes all diacritic characters
  988.       .replace(newRegExp('[' + charsDiacritic + ']+'), '')
  989.     ;
  990.   }
  991.  
  992.   function cleanupSpacing (text) {
  993.     return text
  994.  
  995.       // replaces more than one space with just a single one
  996.       // except before/after preservers and before new-lines
  997.       // .replace(/(?<![_]{2})([ ]{2,})(?![_]{2}|\n)/g, ' ') // WORKS: using lookbehind
  998.       .replace(/([^_])([ ]{2,})(?![_]{2}|\n)/g, '$1 ')
  999.  
  1000.       // cleans tab/space/zwnj/zwj/nbsp between two new-lines(\n)
  1001.       // @REF: https://stackoverflow.com/a/10965543/
  1002.       .replace(/^\n([\t\u0020\u200c\u200d\u00a0]*)\n$/gm, '\n\n')
  1003.     
  1004.   }
  1005.  
  1006.   function cleanupLineBreaks (text) {
  1007.     return text
  1008.  
  1009.       // cleans more than two contiguous line-breaks
  1010.       .replace(/\n{2,}/g, '\n\n')
  1011.     ;
  1012.   }
  1013.  
  1014.   function cleanupBeginAndEnd (text) {
  1015.     return text
  1016.  
  1017.       // removes space/tab/zwnj/nbsp from the beginning of the new-lines
  1018.       .replace(/([\n]+)[ \t\u200c\u00a0]*/g, '$1')
  1019.  
  1020.       // removes spaces, tabs, zwnj, direction marks and new lines from
  1021.       // the beginning and end of text
  1022.       // @REF: http://stackoverflow.com/a/38490203
  1023.       .replace(/^[\s\u200c\u200e\u200f]+|[\s\u200c\u200e\u200f]+$/g, '')
  1024.     ;
  1025.   }
  1026.  
  1027.   function flipPunctuations (text) {
  1028.     var end = ['-'];
  1029.     var start = ['!', '.', '،', '…', '"'];
  1030.     var before = [];
  1031.     var after = [];
  1032.  
  1033.     text = fixThreeDots(text);
  1034.  
  1035.     for (var iStart = 0; iStart < start.length; iStart++) {
  1036.       var sElement = start[iStart];
  1037.       var sReg = newRegExp('^\\' + sElement, 'i');
  1038.       if (sReg.test(text)) {
  1039.         text = text.replace(sReg, '').trim();
  1040.         after.push(sElement);
  1041.       }
  1042.     }
  1043.  
  1044.     for (var iEnd = 0; iEnd < end.length; iEnd++) {
  1045.       var eElement = end[iEnd];
  1046.       var eReg = newRegExp('\\' + eElement + '$', 'i');
  1047.       if (eReg.test(text)) {
  1048.         text = text.replace(eReg, '').trim();
  1049.         before.push(eElement);
  1050.       }
  1051.     }
  1052.  
  1053.     for (var iBefore = 0; iBefore < before.length; iBefore++) {
  1054.       text = before[iBefore] + ' ' + text;
  1055.     }
  1056.  
  1057.     for (var iAfter = 0; iAfter < after.length; iAfter++) {
  1058.       text += after[iAfter];
  1059.     }
  1060.  
  1061.     return normalizeEllipsis(text);
  1062.   }
  1063.  
  1064.   // swap incorrect quotes pairs `»«` to `«»` and `”“` to `“”`
  1065.   function swapQuotes (text) {
  1066.     return text
  1067.       .replace(/(»)(.+?)(«)/g, '«$2»')
  1068.       .replace(/(”)(.+?)(“)/g, '“$2”')
  1069.     ;
  1070.   }
  1071.  
  1072.   Virastar.prototype = {
  1073.  
  1074.     // public methods
  1075.     defaults: defaults,
  1076.     cleanup: cleanup,
  1077.  
  1078.     // internal methods
  1079.     // cleanupZWNJ: cleanupZWNJ,
  1080.     // cleanupZWNJLate: cleanupZWNJLate,
  1081.     // decodeHTMLEntities: decodeHTMLEntities,
  1082.     // normalizeEOL: normalizeEOL,
  1083.     // fixDashes: fixDashes,
  1084.     // fixThreeDots: fixThreeDots,
  1085.     // normalizeEllipsis: normalizeEllipsis,
  1086.     // fixEnglishQuotesPairs: fixEnglishQuotesPairs,
  1087.     // fixEnglishQuotes: fixEnglishQuotes,
  1088.     // fixHamzeh: fixHamzeh,
  1089.     // fixHamzehArabic: fixHamzehArabic,
  1090.     // fixHamzehArabicAlt: fixHamzehArabicAlt,
  1091.     // cleanupRLM: cleanupRLM,
  1092.     // fixPersianGlyphs: fixPersianGlyphs,
  1093.     // fixMiscNonPersianChars: fixMiscNonPersianChars,
  1094.     // fixEnglishNumbers: fixEnglishNumbers,
  1095.     // fixArabicNumbers: fixArabicNumbers,
  1096.     // fixNumeralSymbols: fixNumeralSymbols,
  1097.     // fixPunctuations: fixPunctuations,
  1098.     // fixQuestionMark: fixQuestionMark,
  1099.     // fixPerfixSpacing: fixPerfixSpacing,
  1100.     // fixSuffixSpacing: fixSuffixSpacing,
  1101.     // fixSuffixSpacingHamzeh: fixSuffixSpacingHamzeh,
  1102.     // fixSuffixMisc: fixSuffixMisc,
  1103.     // cleanupExtraMarks: cleanupExtraMarks,
  1104.     // kashidasAsParenthetic: kashidasAsParenthetic,
  1105.     // cleanupKashidas: cleanupKashidas,
  1106.     // fixPunctuationSpacing: fixPunctuationSpacing,
  1107.     // fixBracesSpacing: fixBracesSpacing,
  1108.     // fixBracesSpacingInside: fixBracesSpacingInside,
  1109.     // markdownNormalizeBraces: markdownNormalizeBraces,
  1110.     // markdownNormalizeLists: markdownNormalizeLists,
  1111.     // fixDiacritics: fixDiacritics,
  1112.     // cleanupSpacing: cleanupSpacing,
  1113.     // cleanupLineBreaks: cleanupLineBreaks,
  1114.     // cleanupBeginAndEnd: cleanupBeginAndEnd,
  1115.  
  1116.     // extra methods
  1117.     convertPersianNumbers: convertPersianNumbers,
  1118.     flipPunctuations: flipPunctuations,
  1119.     swapQuotes: swapQuotes
  1120.   };
  1121.  
  1122.   return Virastar;
  1123. }));