gistfile1.php

<?php

function tokenize($edn) {
    $factory = new UsingPregReplace(new LexerDataGenerator());

    $lexer = $factory->createLexer(array(
        'nil|true|false'                 => 'literal',
        '"[^"\\\\]*(?:\\\\.[^"\\\\]*)*"' => 'string',
        '[\s,]'                          => 'whitespace',
        '\\\\[a-z]+'                     => 'character',
        get_symbol_regex()               => 'symbol',
        ':(?:'.get_symbol_regex().')'    => 'keyword',
        '(?:[+-]?)(?:[0-9]+)N?'          => 'int',
        '(?:[+-]?)(?:[0-9]+\.[0-9]+)M?'  => 'float',
        '\\('                            => 'list_start',
        '\\)'                            => 'list_end',
    ));

    $tokens = $lexer->lex($edn);

    return $tokens;
}