Question

Here’s a weird wrench that I’m dealing with at the moment. I’ve been back and forth with it, using a variety of regular expressions, spacing and quote variations but there seems to only be an issue with the placement of the closing </a> tag.

I’ve built a RegEx parser (I know) to grab tags from HTML output and trigger custom closures for each from a library of what we call “shortcodes”. It’ll just swap out the tag for the closure’s return value, and if the closure doesn’t exist, leave the tag in place.

When that </a> is placed on the same line as the tag I’m grabbing, it causes preg_replace_callback to fail. When I add a line break, everything works flawlessly. We’re using these tags all over the place as a central part of our template system and they’re working wonderfully in every other instance.

So my question is… why?! Is this a bug or am I the bug?

EDIT: Updated pattern to limit reluctancy. Closing </a> tag still kills the process. (@Alan Moore)

EDIT 2: So, the regex was the culprit. It was pulling in the content beyond the tag on the same line on some cases, but this pattern solved the problem – '#\[\s*(\w[^\]\s]*)(.*?)\s*\]#i'.

Failing HTML

<div>
    <a href="[ make_url post_id='200' ]">Some Link</a>
</div>

Successful HTML

<div>
    <a href="[ make_url post_id='200' ]">
        Some Link
    </a>
</div>

or

<div>
    <a href="[ make_url post_id='200' ]">Some Link
    </a>
</div>

PHP

Nothing fancy, and even when I dumb it down to the basics it still fails.

$content = 'HTML output from an output buffer';
//$old_pattern = '#\[\s*?(\w+?(?:\.?\w+?)*?(?=\s|\]))(.+)*?\s*?\]#i';
$pattern = '#\[\s*(\w+(?:\.?\w+)*(?=\s|\]))(.+)\s*\]#i';

$content = preg_replace_callback( $pattern, function( $matches ) {
    // Do things with the matches
}, $content );

// Send content back to the template
return $content;

Answer 1

It turns out the regex was the culprit. It was pulling in the content beyond the tag on the same line on some cases, but this pattern solved the problem – #\[\s*(\w[^\]\s]*)(.*?)\s*\]#i.