Delete no more used Product Images on Magento 1

shell_delete_unused_images

<?php
require_once 'abstract.php';

class Mage_Shell_CheckImages extends Mage_Shell_Abstract
{
    const CATALOG_PRODUCT = '/catalog/product';

    const CACHE = '/cache/';

    protected function _glob_recursive($pattern, $flags = 0)
    {
        $files = glob($pattern, $flags);
        foreach (glob(dirname($pattern) . '/*', GLOB_ONLYDIR | GLOB_NOSORT) as $dir) {
            $files = array_merge($files, $this->_glob_recursive($dir . '/' . basename($pattern), $flags));
        }
        return $files;
    }

    public function run()
    {
        if ($this->getArg('help')) {
            echo $this->usageHelp();
            return;
        }

        $media = Mage::getBaseDir('media');
        $debug = $this->getArg('debug');
        $dryrun = $this->getArg('dry') ? true : false ;
        $includeCache = $this->getArg('cache');

        $imagesOnDb = array();
        $imagesOnDisk = array();
        $setup = new Mage_Core_Model_Resource_Setup('core_setup');
        /** @var Varien_Db_Adapter_Pdo_Mysql $connection */
        $connection = $setup->getConnection();

        $sql = "SELECT DISTINCT value
FROM (
	SELECT value
	FROM  `{$setup->getTable('catalog_product_entity_media_gallery')}`
	WHERE attribute_id
	IN (SELECT attribute_id FROM `{$setup->getTable('eav_attribute')}` WHERE `attribute_code` in ('media_gallery') AND entity_type_id = 4)
	UNION
	SELECT value
	FROM  `{$setup->getTable('catalog_product_entity_varchar')}`
	WHERE attribute_id
	IN (SELECT attribute_id FROM `{$setup->getTable('eav_attribute')}` WHERE `attribute_code` in ('image','small_image','thumbnail') AND entity_type_id = 4)
) AS T";
        $result = $connection->query($sql);
        foreach ($result->fetchAll() as $rec) {
            $imagesOnDb[$rec['value']] = 1;
        }
        $imagesOnDb = array_keys($imagesOnDb);
        if ($debug) print_r(array_slice($imagesOnDb, 0, 100));

        if ($debug) echo $media . "/*\n";
        $skip = strlen($media . self::CATALOG_PRODUCT);
        foreach ($this->_glob_recursive($media . self::CATALOG_PRODUCT . '/*', GLOB_MARK) as $img) {
            if (substr($img, -1) != '/') {
                if ($includeCache || (substr($img, $skip, 7) != self::CACHE)) {
                    $imagesOnDisk[] = substr($img, $skip);
                }
            }
        }
        if ($debug) print_r(array_slice($imagesOnDisk, 0, 100));

        $imagesToDelete = array_diff($imagesOnDisk, $imagesOnDb);
        if ($debug) {
            print_r($imagesToDelete);
        } else {
            foreach ($imagesToDelete as $x) {
                if ($dryrun) {
                    echo 'rm '.$media . self::CATALOG_PRODUCT . $x.PHP_EOL;
                } else {
                    @unlink($media . self::CATALOG_PRODUCT . $x);
                }
            }
        }

        if ($debug) {
            echo "\r\n";
        }

    }

    public function usageHelp()
    {
        return <<<USAGE
Usage: php -f check_images.php
    debug       debug mode
    cache       remove cache images too
    dry         dryrun


USAGE;
    }
}

$shell = new Mage_Shell_CheckImages();
$shell->run();

Because I want to know, wether it is debug or dryrun, I added some echo:

<?php
require_once 'abstract.php';

class Mage_Shell_CheckImages extends Mage_Shell_Abstract
{
    const CATALOG_PRODUCT = '/catalog/product';

    const CACHE = '/cache/';
    const PLACEHOLDER = '/placeholder/';

    protected function _glob_recursive($pattern, $flags = 0)
    {
        $files = glob($pattern, $flags);
        foreach (glob(dirname($pattern) . '/*', GLOB_ONLYDIR | GLOB_NOSORT) as $dir) {
            $files = array_merge($files, $this->_glob_recursive($dir . '/' . basename($pattern), $flags));
        }
        return $files;
    }

    public function run()
    {
        if ($this->getArg('help')) {
            echo $this->usageHelp();
            return;
        }

        $media = Mage::getBaseDir('media');
        $debug = $this->getArg('debug');
        $dryrun = $this->getArg('dry') ? true : false ;
        if($dryrun) {
                echo "Running in dry mode.\n";
        }
        if($debug) {
                echo "Running in debug mode.\n";
        }

        $includeCache = $this->getArg('cache');

        $imagesOnDb = array();
        $imagesOnDisk = array();
        $setup = new Mage_Core_Model_Resource_Setup('core_setup');
        /** @var Varien_Db_Adapter_Pdo_Mysql $connection */
        $connection = $setup->getConnection();
        
        $entityTypeIdSql = sprintf(
            "SELECT `entity_type_id` FROM `%s` WHERE `entity_type_code` = :entity_type_code;",
            $setup->getTable('eav/entity_type')
        );
        $entityTypeId = (int)$connection->fetchOne($entityTypeIdSql, array('entity_type_code' => Mage_Catalog_Model_Product::ENTITY));
        if (!$entityTypeId) {
            throw new RuntimeException(sprintf(
                "Could not find entity type code for entity %s",
                Mage_Catalog_Model_Product::ENTITY
            ));
        }

        $sql = "SELECT DISTINCT value
FROM (
    SELECT value
    FROM  `{$setup->getTable('catalog_product_entity_media_gallery')}`
    WHERE attribute_id
    IN (SELECT attribute_id FROM `{$setup->getTable('eav_attribute')}` WHERE `attribute_code` in ('media_gallery') AND entity_type_id = :entity_type_id)
    UNION
    SELECT value
    FROM  `{$setup->getTable('catalog_product_entity_varchar')}`
    WHERE attribute_id
    IN (SELECT attribute_id FROM `{$setup->getTable('eav_attribute')}` WHERE `attribute_code` in ('image','small_image','thumbnail') AND entity_type_id = :entity_type_id)
) AS T";
        $result = $connection->query($sql, array('entity_type_id' => $entityTypeId));
        foreach ($result->fetchAll() as $rec) {
            $imagesOnDb[$rec['value']] = 1;
        }
        $imagesOnDb = array_keys($imagesOnDb);
        if ($debug) print_r(array_slice($imagesOnDb, 0, 100));

        if ($debug) echo $media . "/*\n";
        $skip = strlen($media . self::CATALOG_PRODUCT);
        foreach ($this->_glob_recursive($media . self::CATALOG_PRODUCT . '/*', GLOB_MARK) as $img) {
            if (substr($img, -1) != '/') {
                if ((substr($img, $skip, 13) != self::PLACEHOLDER) && ($includeCache || (substr($img, $skip, 7) != self::CACHE))) {
                    $imagesOnDisk[] = substr($img, $skip);
                }
            }
        }
        if ($debug) print_r(array_slice($imagesOnDisk, 0, 100));

        $imagesToDelete = array_diff($imagesOnDisk, $imagesOnDb);
        if ($debug) {
            print_r($imagesToDelete);
            echo count($imagesOnDisk)." images on Disk\n";
            echo count($imagesOnDb)." images on DB\n";
            echo count($imagesToDelete)." images to delete\n";
        } else {
            foreach ($imagesToDelete as $x) {
                if ($dryrun) {
                    echo 'rm '.$media . self::CATALOG_PRODUCT . $x.PHP_EOL;
                } else {
                    @unlink($media . self::CATALOG_PRODUCT . $x);
                }
            }
        }

        if ($debug) {
            echo "\r\n";
        }

    }

    public function usageHelp()
    {
        return <<<USAGE
Usage: php -f check_images.php
    debug       debug mode
    cache       remove cache images too
    dry         dryrun


USAGE;
    }
}

$shell = new Mage_Shell_CheckImages();
$shell->run();

Since my last post in 2016 I discovered the following extension for Magento 1/OpenMage: https://github.com/fballiano/openmage-image-cleaner.

This is by far the most advanced functional extension which has a lot of useful features. For a while I used ImaClean, but I removed it in favor of the extension mentioned above. It is installed in production and offers a complete visual control over what is deleted. It is not limited only to product pictures, but can also solve other places where could be orphaned images.