PHP Regex vs. DOMDocument

3vs4.md

Performance Comparison:

• DOMDocument
• Regex

regex_vs_domdocument.php

<?php
$content = 'foo bar baz <img src="http://ak-hdl.buzzfed.com/static/2014-06/6/12/enhanced/webdr08/enhanced-21313-1402070821-11.jpg" width="500" height="1000"/> dragons and ponies.';

$dom = new \DOMDocument( '1.0' );
$dom->loadHTML( $content );

for ( $i = 0; $i < 1000; $i++ )
{
	// \DOMNodesList
	$nodes = $dom->getElementsByTagName( 'img' );
	if ( $nodes->length )
	{
	    $img    = $nodes->item(0);
	    $src    = $img->getAttribute( 'src' );
	    $width  = $img->getAttribute( 'width' );
	    $height = $img->getAttribute( 'height' );
	}
}

for ( $i = 0; $i < 1000; $i++ )
{
	if ( preg_match_all( '#<(?P<tag>img)[^<]*?(?:>[\s\S]*?<\/(?P=tag)>|\s*\/>)#', $content, $matches ) )
	{
		foreach ( $matches[0] as $match )
		{
			preg_match( '/ src="([^"]+)"/', $content, $src );
			list( $image_src ) = explode( '?', $src );
			preg_match( '/ width="([0-9]+)"/',  $content, $width  );
			preg_match( '/ height="([0-9]+)"/',  $content, $height  );
		}
	}
}

this is not a valid test. The DomNodesList version just looks at the very first returned image, whereas the regex version checks every image. With the test data presented its ok, but not if that is expanded to contain many images.

add a loop of the $nodes to make this a valid like for like test