Remove duplicate product images

README.md

Introduction

This is a Magento 1.x shell script that goes trough all the products and tries to identify duplicate images from the media gallery.

Caution

A check to not remove base image was not built in.

Nudorm_Deduplicate_Product_Images.php

<?php

require 'abstract.php';

/**
 * @category Shell
 * @package Nudorm
 * @author Haydar Ciftci <[email protected]>
 * @copyright 2016 Nudorm
 */
class Nudorm_Deduplicate_Product_Images extends Mage_Shell_Abstract
{
    /**
     * @var Mage_Catalog_Model_Resource_Product_Collection
     */
    protected $productsCollection = array();

    /**
     * @var Mage_Catalog_Model_Product_Attribute_Media_Api
     */
    protected $mediaApi;

    /**
     * Run script
     */
    public function run()
    {
        $this->productsCollection = Mage::getModel('catalog/product')->getCollection();
        $this->mediaApi = Mage::getModel("catalog/product_attribute_media_api");

        foreach ($this->productsCollection as $dirtyProduct) {
            /** @var Mage_Catalog_Model_Product $product */
            $product = Mage::getModel('catalog/product')->load($dirtyProduct->getId());

            $mediaGalleryImages = $product->getMediaGalleryImages();

            if ($mediaGalleryImages) {
                $md5sums = array();

                foreach ($mediaGalleryImages as $galleryImage) {
                    $imagePath = Mage::getBaseDir('media') . '/catalog/product' . $galleryImage->getFile();
                    $md5sum = md5(file_get_contents($imagePath));

                    if (in_array($md5sum, $md5sums)) {
                        $this->mediaApi->remove($product->getId(),  $galleryImage->getFile());
                        Mage::log('Removed ' . $galleryImage->getFile() . ' from product ' . $product->getId(), null, 'dedupe.log');
                    } else {
                        $md5sums[] = $md5sum;
                    }
                }
            }
        }
    }
}

$dedupe = new Nudorm_Deduplicate_Product_Images();
$dedupe->run();