Parse css for orphan images (i.e. on disk but not in css)

class.orphanfinder.php

<pre><?php

  class OrphanFinder {
		private $imagesDir = "";
		private $imagesUrl = ""; // base path
		private $stylesheetsDir = "";

		function __construct($imagesDir, $imagesUrl, $stylesheetsDir) {
			$this->imagesDir = $imagesDir;
			$this->imagesUrl = $imagesUrl;
			$this->stylesheetsDir = $stylesheetsDir;
		}


		public function getOrphans() {
			$css = $this->getCSS();

			$assetUrls = $this->getAssetUrls( $css );
			$images = $this->getImages();

			$cleanedUrls = array();
			$missingDiskAssets = array();

			// convert url to dir path
			foreach( $assetUrls as $assetUrl) :
				// we don't handle relative urls (for now)
				if ( strpos($assetUrl, $this->imagesUrl) !== false ) :
					$cleanedUrls[] = str_replace($this->imagesUrl, $this->imagesDir, $assetUrl);
				else :
					$missingDiskAssets[] = $assetUrl;
				endif;
			endforeach;
			
			$missingCSSImages = array_diff($images, $cleanedUrls);
			sort($missingCSSImages);

			return array(
				// images on disk but not in css
				'css' => $missingCSSImages,
				// assets in css but not on disk
				'disk' => $missingDiskAssets
			);
		}

		// CSS
		private function getCSS() {
			echo "Found these stylesheets:\n";

			$css = "";
			$stylesheets = $this->getStylesheets();

			foreach( $stylesheets as $stylesheet ) :
				echo "$stylesheet\n";
				$css .= file_get_contents($stylesheet);
			endforeach;

			return $css;
		}

		// gets all assets referenced in the css via url()
		function getAssetUrls( $css ) {
			$matched = preg_match_all('#url\([^)]+\)#i', $css, $matches);

			if ( $matched ) :
				$assets = array_unique($matches[0]);

				// do some cleanup
				foreach( $assets as $key => $asset) :
					$asset = preg_replace('~\?[^)]+|#[^)]+|\'|"~', "", $asset);
					$asset = preg_replace('#url\(|\)#', "", $asset);

					$assets[$key] = $asset;
				endforeach;

				$assets = array_unique($assets);

				sort($assets);

				return $assets;
			else :
				echo "ERROR: no assets found in stylesheets \n";
				die();
			endif;
		}

		private function getStylesheets() {
			// get all directories
			$directories = $this->getDirectories($this->stylesheetsDir);

			$stylesheets = array();
			foreach( $directories as $directory) :
				$assets = glob( $directory . "/*.*");

				$stylesheets = array_merge($stylesheets, $assets);
			endforeach;

			return $stylesheets;
		}

		// IMAGES
		private function getImages() {
			$directories = $this->getDirectories($this->imagesDir);
			$images = array();

			foreach( $directories as $directory ) :
				$assets = glob( $directory . "/*.*");

				$images = array_merge($images, $assets);
			endforeach;

			return $images;
		}

		// HELPERS
		private function getDirectories( $rootDir ) {
			$directories = glob( $rootDir . "/*", GLOB_ONLYDIR);

			if ( count($directories) > 0 ) :
				foreach( $directories as $directory ) :
					$directories = array_merge($directories, $this->getDirectories( $directory));
				endforeach;
			endif;
			
			$directories[] = $rootDir;

			return array_unique($directories);
		}
	}

	// example usage
	$images_dir = "~/Sites/example/images";
	$images_url = "http://example.com/images";
	$stylesheets_dir = "~/Sites/example/css";

	$orphanFinder = new OrphanFinder($images_dir, $images_url, $stylesheets_dir);
	$orphans = $orphanFinder->getOrphans();

	var_dump($orphans);
?></pre>