PHP PCRE - correct nested tags behaviour

I want to write symple (consisting of one preg_replace call) forum parser and I run into problems with nested tags.

E.g. if someone is quoting someone quoting someone, I cannot achieve correct behaviour.

When having:

[quote=Tom]

[quote=Jerry]
Lorem
[/quote]

Ipsum

[/quote]

Dolor.

I want something like this:

<blockquote>
    <p><strong>Tom wrote</strong></p>
    <blockquote>
       <p><strong>Jerry wrote:</strong></p>
      <p>Lorem</p>
    </blockquote>

    Ipsum
</blockquote>

Dolor.

I have this code:

preg_replace('~\[quote (.+)\](.+)\[/quote\]~is', '<blockquote><p><strong>$1</strong> wrote:</p&g开发者_如何学Ct;<p>$2</p></blockquote>', $value);

This version is greedy. If I have two separate [quote] blocks, the regex wraps all the text between the first [quote] and the second [/quote].

If I add the U modifier, it's too ungreedy - the first [quote] tag is paired with the first (nested and irrelevant) [/quote] tag.

Thanks for any help!

There is the PEAR HTML_BBCodeParser Package and also PHP has a native extension for parsing code like this, check this example: http://www.php.net/manual/en/function.bbcode-create.php

Don't use a regular expression for this. Use the official PECL extension provided:

Example (lifted from the docs):

<?php
$arrayBBCode=array(
    ''=>         array('type' => BBCODE_TYPE_ROOT,  'childs' => '!i'),
    'i'=>        array('type' => BBCODE_TYPE_NOARG, 'open_tag' => '<i>',
                    'close_tag' => '</i>', 'childs' => 'b'),
    'url'=>      array('type' => BBCODE_TYPE_OPTARG,
                    'open_tag' => '<a href="{PARAM}">', 'close_tag' => '</a>',
                    'default_arg' => '{CONTENT}',
                    'childs' => 'b,i'),
    'img'=>      array('type' => BBCODE_TYPE_NOARG,
                    'open_tag' => '<img src="', 'close_tag' => '" />',
                    'childs' => ''),
    'b'=>        array('type'=>BBCODE_TYPE_NOARG, 'open_tag' => '<b>',
                    'close_tag' => '</b>'),
);

$text = <<<EOF
[b]Bold Text[/b]
[i]Italic Text[/i]
[url]http://www.php.net/[/url]
[url=http://pecl.php.net/][b]Content Text[/b][/url]
[img]http://static.php.net/www.php.net/images/php.gif[/img]
[url=http://www.php.net/]
[img]http://static.php.net/www.php.net/images/php.gif[/img]
[/url]
EOF;

$BBHandler = bbcode_create($arrayBBCode);
echo bbcode_parse($BBHandler, $text);
?>

The full docs.

With some help of recursive regular expressions:

function replace_quotes_callback($matches) {
    $cite = empty($matches[1]) ? '' : '<p><strong>' . $matches[1] . '</strong> wrote:</p>';
    return '<blockquote>' . $cite . '<p>' . replace_quotes($matches[2]) . '</p></blockquote>';
}

function replace_quotes($data) {
    return preg_replace_callback('~\[quote(?:=([^\]]+))?\]((?:(?R)|.)*?)\[/quote\]~s', 'replace_quotes_callback', $data);
}

The pattern matches only outermost quote blocks, and the callback function replace_quotes_callback replace quotes inside itself by recursively call replace_quotes.