Recursive BBCode Parsing

Question

I'm trying to parse BBCode in my script. Now, it works seamelessly, until I try to indent BBCode that's more than just bold or underline - such as spoiler, url, font size, etc. - then it screws up. Here's my code:

function parse_bbcode($text) {
    global $db;
    $oldtext = $text;
    $bbcodes = $db->select('*', 'bbcodes');
    foreach ($bbcodes as $bbcode) {
        switch ($bbcode->type) {
            case 'simple': {
                $find = '{content}';
                $replace = '${1}';
                $text = preg_replace(
                    '/\['.$bbcode->tag.'\](.+)\[\/'.$bbcode->tag.'\]/i',
                    str_replace($find, $replace, $bbcode->html),
                    $text);
                    break;
            }
            case 'property':
            case 'options': {
                $find = array ( '{property}', '{content}' );
                $replace = array ( '${1}', '${2}' );
                $text = preg_replace(
                    '/\['.$bbcode->tag.'\=(.[^\"]*)\](.+)\[\/'.$bbcode->tag.'\]/i',
                    str_replace($find, $replace, $bbcode->html),
                    $text);
                    break;
            }
        }
    }
    return $text;
}

Now my guess is that the RegEx doesn't like the recursiveness in the pattern. How can I improve it? A sample $bbcode object is as such:

stdClass::__set_state(array(
   'id' => '2',
   'name' => 'Italic',
   'type' => 'simple',
   'tag' => 'i',
   'button_image' => NULL,
   'button_text' => '<i>I</i>',
   'options' => '',
   'prompt' => NULL,
   'html' => '<i>{content}</i>',
   'order' => '1',
))
stdClass::__set_state(array(
   'id' => '3',
   'name' => 'URL',
   'type' => 'property',
   'tag' => 'url',
   'button_image' => NULL,
   'button_text' => 'http://',
   'options' => '',
   'prompt' => 'URL address',
   'html' => '<a href="{property}">{content}</a>',
   'order' => '4',
))

Answer 1

As gordon said in comments PHP has a BBCode parser, so no reason to reinvent the wheel.

The native parser is a PECL package though, so you will have to install it. If that's not an option (for instance due to shared hosting), there is also a PEAR package: http://pear.php.net/package/HTML_BBCodeParser

In addition to those, you can take a look at forums using BB code source code, and either use their parser, or improve it. There is also several PHP implementations listed at http://www.bbcode.org/implementations.php

Answer 2

Correctly parsing BBcode using regex is non-trival. The codes may be nested. CODE tags may contain BBCodes which must be ignored by the parser. Certain tags may not appear inside of other tags. etc. However, it can be done. I recently overhauled the BBCode parser for the FluxBB open source forum software. You may want to check it out in action:

New 2011 FluxBB Parser

Note that this new parser has not yet been incorporated into the FluxBB codebase.