fileformatpage2charclass.js

var SYSTEM = require('system'),
    HTTP = require('http-client');

if (SYSTEM.args.length < 2)
{
    print("usage:");
    print("js " + SYSTEM.args[0] + " <fileformat.info URL>+");
    require("os").exit(1);
}

var unicodeCharacters = [],
    surrogateMappings = [],
    charRegex = /\s*<td align="center"><a href="\/info\/unicode\/char\/.+?">U+(.+?)<\/a><\/td>\s*/g;

SYSTEM.args.slice(1).forEach(function(aURL)
{
    var fileContents = HTTP.read(aURL).decodeToString('utf-8');

    while (match = charRegex.exec(fileContents))
    {
        var unicodeCharacter = parseInt(match[1], 16);

        if (unicodeCharacter > 0xFFFF)
        {
            unicodeCharacter -= 0x10000;

            var highSurrogate = ((unicodeCharacter & 0xFFC00) >> 10) + 0xD800;

            if (!surrogateMappings[highSurrogate])
                surrogateMappings[highSurrogate] = [];

            surrogateMappings[highSurrogate].push((unicodeCharacter & 0x3FF) + 0xDC00);
        }
        else
            unicodeCharacters.push(unicodeCharacter);            
    }
});

Array.prototype.toCharacterClass = function()
{
    var startChar = null,
        previousChar = null,
        string = "";

    this.forEach(function(aUnicodeCharacter)
    {
        if (startChar === null)
            startChar = aUnicodeCharacter;

        else if (aUnicodeCharacter !== previousChar + 1)
        {
            string += rangeString(startChar, previousChar);

            startChar = aUnicodeCharacter;
        }

        previousChar = aUnicodeCharacter;
    });

    if (startChar !== null)
        string += rangeString(startChar, previousChar);

    return string;

    function rangeString(fromChar, toChar)
    {
        var fromString = fromChar.toUnicodeHex();

        if (toChar === fromChar)
            return fromString;

        return fromString + (toChar - fromChar > 1 ? "-" : "") + toChar.toUnicodeHex();
    }
}

Number.prototype.toUnicodeHex = function()
{
    var string = this.toString(16).toUpperCase();

    while (string.length < 4)
        string = '0' + string;

    return '\\u' + string;
}

var classes = "[" + unicodeCharacters.toCharacterClass() + "]";

for (highSurrogate in surrogateMappings)
    if (surrogateMappings.hasOwnProperty(highSurrogate))
        classes +=  " / \"" + parseInt(highSurrogate, 10).toUnicodeHex() + 
                    "\" [" + surrogateMappings[highSurrogate].toCharacterClass() + "]";

print(classes);