Javascript Character Codes & Encodings [Script]

javascript-character-encodings.js

// Script to generate list of character codes, characters, and their respective URI/HTML encodings.


/*
 * HTML Escape Utility Function
 * Adapted from underscore library's _.escape method.
 * source: http://underscorejs.org/#escape
 */

var escape = (function() {

  var ESCAPE_MAP = {
    '&': '&',
    '<': '&lt;',
    '>': '&gt;',
    '"': '&quot;',
    "'": '&#x27;',
    '`': '&#x60;'
  };

  var escaper = function(match) {
    return ESCAPE_MAP[match];
  };

  var source = '(?:' + Object.keys(ESCAPE_MAP).join('|') + ')';
  var testRegexp = RegExp(source);
  var replaceRegexp = RegExp(source, 'g');
  return function(string) {
    string = string == null ? '' : '' + string;
    return testRegexp.test(string) ? string.replace(replaceRegexp, escaper) : string;
  };

}());


/*
 * For character codes n..0, outputs:
 * charCode, char, url_encoding, html_encoding
 */

function generateEncodings(n){
  var out = [];
  while(n--){
    var char = String.fromCharCode(n);
    out.push({
      charCode: n,
      char: char,
      uri_encoded: encodeURI(char),
      html_escaped: escape(char)
      // , html_encoded: "&#" + n + ";" // ?
    });
  }
  return out;
}

/*
 * Driver Code
 */

var result = generateEncodings(127); // first 127 characters

// console.table && console.table(result);       // tabular format (Chrome Browser)
console.log(JSON.stringify(result, true, 4));    // JSON format