FindIncorrectLinksTask.php

<?php

/**
 * A task to identify any pages on the site that are referencing absolute URLs 
 * (that is, URLs like http://site.com, instead of [sitetree])
 * Can't easily be automated due to https://github.com/silverstripe/silverstripe-cms/issues/812
 */
class FindIncorrectLinksTask extends BuildTask {
  
	protected $title = 'Find Incorrect Links Tasks';
	
	protected $description = 'Tries to find all bad links to your site, instead of using a proper SiteTree link.';
	
	public function run($request) {
		echo "<h2>The following URLs need updating:</h2>";
		$pages = SiteTree::get();
		if($pages) foreach($pages as $page) {
			$content = $page->Content;

			// Find the href attribute of every a element.
			$pattern = '/<a [^>]*href="([^"]*)/';
			$matches = array();
			if (!preg_match_all($pattern, $page->Content, $matches)) continue;

			// Loop through all the matches to ([^"]*), which captures the link href.
			foreach ($matches[1] as $match) {
				// We need to report on URL's that match, but aren't being rewritten
        // TODO: Fill in with your URLs
				if ((strpos($match, '://yoursite.com/') !== FALSE) or (strpos($match, '://www.yoursite/'))) {
					echo htmlentities($match) . " : <a href='" . $page->Link() . "'>" . $page->Title . "</a>\r\n<br>";
				}
				continue;
			}
		}
	}
}