simple-sql-lexer.sql

var quoted = [
    '"(?:[^"\\\\]|\\\\.)*"',
    "'(?:[^'\\\\]|\\\\.)*'",
    "`(?:[^`\\\\]|\\\\.)*`"

],
keywords = ['SELECT', 'FROM', 'DELETE FROM', 'INSERT INTO', 'UPDATE', 'JOIN',
    'LEFT JOIN', 'INNER JOIN', 'ORDER BY', 'GROUP BY', 'HAVING', 'WHERE',
    'LIMIT', 'VALUES', 'SET'],
values = [
    '\\*',
    '\\.',
    '\\w+'
],
space = [
    '\\s+'
],
operators = [
    ',',
    '\\(',
    '\\)',
    '\\+',
    '-',
    '!=',
    '<>',
    '=',
    '/',
    '<',
    '<=',
    '>',
    '>=',
    ';'
];

var tokenize = (function(){
    _regex = function(){
        var list = [quoted, keywords, space, values, operators, ['(.+)']]
        var l2 = [];
        for (var i=0; i<list.length; i++){
            l2.push(
                (list[i]).join('|')
            )
        }
        console.debug(RegExp(l2.join('|'), 'gmi'))
        return RegExp(l2.join('|'), 'gmi')
    }();


    return function(sql_string){
        _regex.lastIndex = 0;
        var tokens = [], match;
        while(match = _regex.exec(sql_string)){
            if(match[1]){

                console.error("syntax error, invalid token ", match[1]);
                return;
            } else{
                tokens.push(match[0]);
            }
        }
        return tokens;
            // sql_string.match(_regex)
        // return _regex.exec(sql_string);
    }
})();