├── README.md
├── composer.json
└── src
    ├── Console
        └── RemoveDuplicate.php
    ├── etc
        ├── di.xml
        └── module.xml
    └── registration.php


/README.md:
--------------------------------------------------------------------------------
 1 | # Elgentos - Remove Duplicate Product images in Magento 2
 2 | 
 3 | This Extension allows you to find duplicate product images from your product list and from this list you can easily remove them by running a command.
 4 | 
 5 | ## Installation
 6 | 
 7 | 1) Go to your Magento root folder
 8 | 2) Download the extension using composer:
 9 |     ```
10 |     composer require elgentos/magento2-product-duplicate-images-remove
11 |     ```
12 | 3) Run setup commands:
13 | 
14 |     ```
15 |     php bin/magento setup:upgrade
16 |     ```
17 |    
18 | 4) Run the command:
19 | 
20 |    ```
21 |    php bin/magento duplicate:remove
22 |    ```
23 | Eg:
24 | ```sh
25 | # Use unlink
26 | php bin/magento duplicate:remove -u1
27 | 
28 | # Turn off dry run
29 | php bin/magento duplicate:remove -d0
30 | 
31 | # Combine unlink and turn off dry run
32 | php bin/magento duplicate:remove -u1 -d0
33 | 
34 | # Specific on some sku
35 | php bin/magento duplicate:remove -u1 -d0 SKU1 SKU2 SKU3
36 | 


--------------------------------------------------------------------------------
/composer.json:
--------------------------------------------------------------------------------
 1 | {
 2 |   "name": "elgentos/magento2-product-duplicate-images-remove",
 3 |   "description":"Magento 2 find duplicate product images from your product list and from this list you can easily remove them by running a command",
 4 |   "keywords": [
 5 |     "magento 2",
 6 |     "magento",
 7 |     "m2",
 8 |     "duplicate product images",
 9 |     "duplicate product images remove",
10 |     "product images remove",
11 |     "magento 2 extension",
12 |     "magento 2 extension free"
13 |   ],
14 |   "require": {
15 |     "magento/framework": "^103.0"
16 |   },
17 |   "type": "magento2-module",
18 |   "license": [
19 |     "OSL-3.0",
20 |     "AFL-3.0"
21 |   ],
22 |   "authors": [
23 |     {
24 |       "name": "Peter Jaap Blaakmeer",
25 |       "email": "peterjaap@elgentos.nl"
26 |     }
27 |   ],
28 |   "autoload": {
29 |     "files": [
30 |       "src/registration.php"
31 |     ],
32 |     "psr-4": {
33 |       "Elgentos\\RemoveDuplicateImage\\": "src/"
34 |     }
35 |   }
36 | }
37 | 


--------------------------------------------------------------------------------
/src/Console/RemoveDuplicate.php:
--------------------------------------------------------------------------------
  1 | <?php
  2 | 
  3 | declare(strict_types=1);
  4 | 
  5 | namespace Elgentos\RemoveDuplicateImage\Console;
  6 | 
  7 | use Magento\Catalog\Api\Data\ProductSearchResultsInterface;
  8 | use Magento\Catalog\Api\ProductRepositoryInterface;
  9 | use Magento\Framework\Api\SearchCriteriaBuilder;
 10 | use Magento\Framework\App\Area;
 11 | use Magento\Framework\App\Filesystem\DirectoryList;
 12 | use Magento\Framework\App\ResourceConnection;
 13 | use Magento\Framework\App\State;
 14 | use Magento\Framework\Console\Cli;
 15 | use Magento\Framework\Exception\FileSystemException;
 16 | use Magento\Framework\Filesystem\Driver\File;
 17 | use Magento\Store\Model\StoreManagerInterface;
 18 | use Symfony\Component\Console\Command\Command;
 19 | use Symfony\Component\Console\Input\InputArgument;
 20 | use Symfony\Component\Console\Input\InputInterface;
 21 | use Symfony\Component\Console\Input\InputOption;
 22 | use Symfony\Component\Console\Output\OutputInterface;
 23 | 
 24 | class RemoveDuplicate extends Command
 25 | {
 26 |     /**
 27 |      * @var SearchCriteriaBuilder
 28 |      */
 29 |     private $searchCriteriaBuilder;
 30 | 
 31 |     /**
 32 |      * @var State
 33 |      */
 34 |     private $state;
 35 | 
 36 |     /**
 37 |      * @var ProductRepositoryInterface
 38 |      */
 39 |     private $productRepository;
 40 | 
 41 |     /**
 42 |      * @var StoreManagerInterface
 43 |      */
 44 |     private $storeManager;
 45 | 
 46 |     /**
 47 |      * @var DirectoryList
 48 |      */
 49 |     private $directoryList;
 50 | 
 51 |     /**
 52 |      * @var ResourceConnection
 53 |      */
 54 |     private $resource;
 55 | 
 56 |     /**
 57 |      * @var File
 58 |      */
 59 |     private $fileDriver;
 60 | 
 61 |     /**
 62 |      * @param SearchCriteriaBuilder $searchCriteriaBuilder
 63 |      * @param State $state
 64 |      * @param ProductRepositoryInterface $productRepository
 65 |      * @param StoreManagerInterface $storeManager
 66 |      * @param DirectoryList $directoryList
 67 |      * @param ResourceConnection $resource
 68 |      * @param File $fileDriver
 69 |      */
 70 |     public function __construct(
 71 |         SearchCriteriaBuilder $searchCriteriaBuilder,
 72 |         State $state,
 73 |         ProductRepositoryInterface $productRepository,
 74 |         StoreManagerInterface $storeManager,
 75 |         DirectoryList $directoryList,
 76 |         ResourceConnection $resource,
 77 |         File $fileDriver
 78 |     ) {
 79 |         $this->searchCriteriaBuilder = $searchCriteriaBuilder;
 80 |         $this->state = $state;
 81 |         $this->productRepository = $productRepository;
 82 |         $this->storeManager = $storeManager;
 83 |         $this->directoryList = $directoryList;
 84 |         $this->resource = $resource;
 85 |         $this->fileDriver = $fileDriver;
 86 |         parent::__construct();
 87 |     }
 88 | 
 89 |     /**
 90 |      * @inheritdoc
 91 |      */
 92 |     protected function configure()
 93 |     {
 94 |         $this->setName('duplicate:remove')
 95 |             ->setDescription('Remove duplicate product images')
 96 |             ->addOption(
 97 |                 'unlink',
 98 |                 'u',
 99 |                 InputOption::VALUE_OPTIONAL,
100 |                 'Unlink the duplicate files from system',
101 |                 false
102 |             )
103 |             ->addOption(
104 |                 'dryrun',
105 |                 'd',
106 |                 InputOption::VALUE_OPTIONAL,
107 |                 'Dry-run does not delete any values or files',
108 |                 true
109 |             )
110 |             ->addArgument(
111 |                 'products',
112 |                 InputArgument::IS_ARRAY,
113 |                 'Product entity SKUs to filter on',
114 |                 null
115 |             );
116 |     }
117 | 
118 |     /**
119 |      * @inheritdoc
120 |      */
121 |     protected function execute(InputInterface $input, OutputInterface $output)
122 |     {
123 |         try {
124 |             $this->state->setAreaCode(Area::AREA_GLOBAL);
125 |         } catch (\Exception $e) {
126 |             return Cli::RETURN_FAILURE;
127 |         }
128 | 
129 |         $this->storeManager->setCurrentStore(0);
130 | 
131 |         $isUnlink = !($input->getOption('unlink') === 'false') && $input->getOption('unlink');
132 |         $isDryRun = !($input->getOption('dryrun') === 'false') && $input->getOption('dryrun');
133 | 
134 |         $path = $this->directoryList->getPath('media');
135 | 
136 |         $targetProductSku = $input->getArgument('products') ?: $this->getEntityIds();
137 |         $searchCriteriaBuilder = $this->searchCriteriaBuilder
138 |             ->addFilter('image', 'no_selection', 'neq');
139 | 
140 |         if ($input->getArgument('products')) {
141 |             $searchCriteriaBuilder->addFilter('sku', $targetProductSku, 'in');
142 |         } else {
143 |             $searchCriteriaBuilder->addFilter('entity_id', $this->getEntityIds(), 'in');
144 |         }
145 | 
146 |         $searchCriteriaBuilder = $searchCriteriaBuilder->create();
147 | 
148 |         /** @var ProductSearchResultsInterface $products */
149 |         $products = $this->productRepository->getList($searchCriteriaBuilder);
150 | 
151 |         if (!$products->getTotalCount()) {
152 |             return Cli::RETURN_SUCCESS;
153 |         }
154 | 
155 |         if ($isDryRun) {
156 |             $output->writeln('THIS IS A DRY-RUN, NO CHANGES WILL BE MADE!');
157 |         }
158 |         $output->writeln(sprintf('%s products found with 2 images or more.', $products->getTotalCount()));
159 | 
160 |         foreach ($products->getItems() as $product) {
161 |             $product->setStoreId(0);
162 |             $md5Values = [];
163 |             $baseImage = $product->getImage();
164 | 
165 |             $filePath = $path . '/catalog/product' . $baseImage;
166 |             if ($this->isFileExists($filePath)) {
167 |                 $md5Values[] = md5_file($filePath);
168 |             }
169 | 
170 |             $gallery = $product->getMediaGalleryEntries();
171 | 
172 |             $shouldSave = false;
173 |             $filePaths = [];
174 | 
175 |             if (empty($gallery)) {
176 |                 continue;
177 |             }
178 | 
179 |             foreach ($gallery as $key => $galleryImage) {
180 |                 if ($galleryImage->getFile() == $baseImage) {
181 |                     continue;
182 |                 }
183 | 
184 |                 $filePath = $path . '/catalog/product' . $galleryImage->getFile();
185 | 
186 |                 if ($this->isFileExists($filePath)) {
187 |                     $md5 = md5_file($filePath);
188 |                 } else {
189 |                     continue;
190 |                 }
191 | 
192 |                 if (in_array($md5, $md5Values)) {
193 |                     if (count($galleryImage->getTypes()) > 0) {
194 |                         continue;
195 |                     }
196 |                     unset($gallery[$key]);
197 |                     $filePaths[] = $filePath;
198 |                     $output->writeln(sprintf('Removed duplicate image from %s', $product->getSku()));
199 |                     $shouldSave = true;
200 |                 } else {
201 |                     $md5Values[] = $md5;
202 |                 }
203 |             }
204 | 
205 |             if (!$isDryRun && $shouldSave) {
206 |                 $product->setMediaGalleryEntries($gallery);
207 |                 try {
208 |                     $this->productRepository->save($product);
209 |                 } catch (\Exception $e) {
210 |                     $output->writeln('Could not save product: ' . $e->getMessage());
211 |                 }
212 |             }
213 | 
214 |             foreach ($filePaths as $filePath) {
215 |                 if (!$this->isFile($filePath)) {
216 |                     continue;
217 |                 }
218 | 
219 |                 if (!$isDryRun
220 |                     && $isUnlink
221 |                     && $shouldSave
222 |                 ) {
223 |                     try {
224 |                         $this->fileDriver->deleteFile($filePath);
225 |                     } catch (FileSystemException $e) {
226 |                         continue;
227 |                     }
228 |                 }
229 | 
230 |                 if ($isUnlink
231 |                     && $shouldSave
232 |                 ) {
233 |                     $output->writeln('Deleted file: ' . $filePath);
234 |                 }
235 |             }
236 |         }
237 | 
238 |         if ($isDryRun) {
239 |             $output->writeln('THIS WAS A DRY-RUN, NO CHANGES WERE MADE!');
240 |         } else {
241 |             $output->writeln('Duplicate images are removed');
242 |         }
243 | 
244 |         return Cli::RETURN_SUCCESS;
245 |     }
246 | 
247 |     /**
248 |      * Get Entity IDs related
249 |      *
250 |      * @return array
251 |      */
252 |     public function getEntityIds(): array
253 |     {
254 |         $connection = $this->resource->getConnection();
255 |         $tableName = $this->resource->getTableName('catalog_product_entity_media_gallery_value_to_entity');
256 | 
257 |         $select = $connection->select()
258 |             ->from($tableName, ['entity_id'])
259 |             ->group('entity_id')
260 |             ->having('COUNT(entity_id) >= 2');
261 | 
262 |         return $connection->fetchCol($select);
263 |     }
264 | 
265 |     /**
266 |      * Is file exists
267 |      *
268 |      * @param string $path
269 |      * @return bool
270 |      */
271 |     protected function isFileExists(string $path): bool
272 |     {
273 |         try {
274 |             $fileExists = $this->fileDriver->isExists($path);
275 |         } catch (\Exception $exception) {
276 |             $fileExists = false;
277 |         }
278 | 
279 |         return $fileExists;
280 |     }
281 | 
282 |     /***
283 |      * Tells whether the filename is a regular file
284 |      *
285 |      * @param string $path
286 |      * @return bool
287 |      */
288 |     protected function isFile(string $path): bool
289 |     {
290 |         try {
291 |             $isFile = $this->fileDriver->isFile($path);
292 |         } catch (\Exception $exception) {
293 |             $isFile = false;
294 |         }
295 | 
296 |         return $isFile;
297 |     }
298 | }
299 | 


--------------------------------------------------------------------------------
/src/etc/di.xml:
--------------------------------------------------------------------------------
 1 | <?xml version="1.0"?>
 2 | <config xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="urn:magento:framework:ObjectManager/etc/config.xsd">
 3 |    <type name="Magento\Framework\Console\CommandList">
 4 |        <arguments>
 5 |            <argument name="commands" xsi:type="array">
 6 |                <item name="removeDuplicate" xsi:type="object">Elgentos\RemoveDuplicateImage\Console\RemoveDuplicate</item>
 7 |            </argument>
 8 |        </arguments>
 9 |    </type>
10 | </config>
11 | 


--------------------------------------------------------------------------------
/src/etc/module.xml:
--------------------------------------------------------------------------------
1 | <?xml version="1.0"?>
2 | <config xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:noNamespaceSchemaLocation="urn:magento:framework:Module/etc/module.xsd">
3 |     <module name="Elgentos_RemoveDuplicateImage"/>
4 | </config>
5 | 


--------------------------------------------------------------------------------
/src/registration.php:
--------------------------------------------------------------------------------
1 | <?php
2 | \Magento\Framework\Component\ComponentRegistrar::register(
3 |     \Magento\Framework\Component\ComponentRegistrar::MODULE,
4 |     'Elgentos_RemoveDuplicateImage',
5 |     __DIR__
6 | );
7 | 


--------------------------------------------------------------------------------