多个递归模式似乎不起作用

I have a template class to handle templates so, this is a line taken from that class. I'll spare many of the details, as many of them have to do with loading templates.

$tmpl = <<<tem
    <div id="header2">
<a rel="home" href="//localhost/index.php">
    <h1>Test Site</h1>
</a>
</div>
<nav>
    <div class="menu">
        <ul>
            <li class="base">
                <a rel="home" href="//localhost/index.php">Home</a>
            </li>
            <li><a href="//localhost/search.php">Search</a></li>
            {if(\$is_guest)}
                <li><a href="//localhost/member.php?view=register">Register</a></li>
                <li><a href="//localhost/member.php?view=login">Login</a></li>
            {/if}{else}
                <li><a href="//localhost/profile.php?uid=1">Profile</a></li>
                <li><a href="//localhost/usercp.php">My Account</a></li>
                <li><a href="//localhost/member.php?view=logout">Logout</a></li>
            {/else}
            {if(\$is_admin)}
                <li><a href="//localhost/admin">Admin</a></li>
            {/if}
            {!nav}
        </ul>
        <div class="floater"></div>
    </div>
</nav>
tem;

    $tmpl = preg_replace_callback('#{\$(.+?)}#u',array(get_class($this),'varsub'),$tmpl);
    $tmpl = preg_replace_callback("#{if\(([A-Za-z0-9\$_!]+)\)}((?:(?R)|.*?)+)\{/if\}\{else\}((?:(?R)|.*?)+)\{/else\}#ismu",array(get_class($this),'boolelsesub'), $tmpl);

For some reason, this Regex results in the page never loading, instead of doing what I would expect. I've looked at many online resources however, none of them seem to mention using two recursive patterns at once.

It doesn't seem to be reaching the callback which I have assigned to it either.

protected function boolelsesub($input)
{
    var_dump($input);
    $h = fopen('bah.txt', 'a');
    fwrite($h,'test');
    fclose($h);
    exit;
}

Edit: The contents of boolelsesub, which has been stripped down to make it easier to debug.

Try with this pattern:

$pattern = <<<'LOD'
~
{ if \( ([^)]+) \) }

( (?>[^{]+|(?R))* )

{ / if }
(?:
    { else }

    ( (?>[^{]+|(?R))* )

    { / else }
)?
~ix
LOD;

Online demo

Even if your pattern works with this edit, the result will be inefficient due to (?:(?R)|.*?)+. This part is inefficient in particular because .*? is lazy and can match the empty string. To solve this, I replace it with a negated character class with a greedy quantifier [^{]+