Added an option to disable checking for product problems when those products are invisible. Fixes #31

Author: hostep

Checker/Catalog/Product/UrlKey/DuplicateUrlKey.php

@@ -6,10 +6,12 @@
 
 use Baldwin\UrlDataIntegrityChecker\Checker\Catalog\Product\UrlKey as UrlKeyChecker;
 use Baldwin\UrlDataIntegrityChecker\Console\Progress;
+use Baldwin\UrlDataIntegrityChecker\Util\Configuration as ConfigUtil;
 use Baldwin\UrlDataIntegrityChecker\Util\Stores as StoresUtil;
 use Magento\Catalog\Api\Data\ProductInterface;
 use Magento\Catalog\Model\Attribute\ScopeOverriddenValueFactory as AttributeScopeOverriddenValueFactory;
 use Magento\Catalog\Model\Product as ProductModel;
+use Magento\Catalog\Model\Product\Visibility as ProductVisibility;
 use Magento\Catalog\Model\ResourceModel\Product\Collection as ProductCollection;
 use Magento\Catalog\Model\ResourceModel\Product\CollectionFactory as ProductCollectionFactory;
 use Magento\Store\Model\Store;
@@ -28,17 +30,20 @@ class DuplicateUrlKey
     /** @var array<string, string> */
     private $cachedProductUrlKeyData;
     private $cachedProductSkusByIds;
+    private $configUtil;
 
     public function __construct(
         StoresUtil $storesUtil,
         Progress $progress,
         ProductCollectionFactory $productCollectionFactory,
-        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory
+        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory,
+        ConfigUtil $configUtil
     ) {
         $this->storesUtil = $storesUtil;
         $this->progress = $progress;
         $this->productCollectionFactory = $productCollectionFactory;
         $this->attributeScopeOverriddenValueFactory = $attributeScopeOverriddenValueFactory;
+        $this->configUtil = $configUtil;
 
         $this->progressIndex = 0;
         $this->cachedProductUrlKeyData = [];
@@ -82,12 +87,15 @@ private function checkForDuplicatedUrlKeyAttributeValues(): array
                 ->addAttributeToSelect(UrlKeyChecker::URL_KEY_ATTRIBUTE)
                 ->addAttributeToFilter(UrlKeyChecker::URL_KEY_ATTRIBUTE, ['notnull' => true], $joinType)
                 ->addAttributeToFilter(UrlKeyChecker::URL_KEY_ATTRIBUTE, ['neq' => ''], $joinType)
-                // TODO: remove!
-                // ->addAttributeToFilter('entity_id', [
-                //     'in' => [147, 148, 149, 150, 151, 152, 153, 154, 155, 156, 157, 158],
-                // ])
             ;
 
+            if ($this->configUtil->getOnlyCheckVisibleProducts()) {
+                $collection->addAttributeToFilter(
+                    ProductInterface::VISIBILITY,
+                    ['neq' => ProductVisibility::VISIBILITY_NOT_VISIBLE]
+                );
+            }
+
             if ($this->progressIndex === 0) {
                 $this->progress->setGuestimatedSize(count($storeIds), $collection->getSize());
             }

Checker/Catalog/Product/UrlKey/EmptyUrlKey.php

@@ -6,10 +6,12 @@
 
 use Baldwin\UrlDataIntegrityChecker\Checker\Catalog\Product\UrlKey as UrlKeyChecker;
 use Baldwin\UrlDataIntegrityChecker\Console\Progress;
+use Baldwin\UrlDataIntegrityChecker\Util\Configuration as ConfigUtil;
 use Baldwin\UrlDataIntegrityChecker\Util\Stores as StoresUtil;
 use Magento\Catalog\Api\Data\ProductInterface;
 use Magento\Catalog\Model\Attribute\ScopeOverriddenValueFactory as AttributeScopeOverriddenValueFactory;
 use Magento\Catalog\Model\Product as ProductModel;
+use Magento\Catalog\Model\Product\Visibility as ProductVisibility;
 use Magento\Catalog\Model\ResourceModel\Product\Collection as ProductCollection;
 use Magento\Catalog\Model\ResourceModel\Product\CollectionFactory as ProductCollectionFactory;
 use Magento\Store\Model\Store;
@@ -23,17 +25,20 @@ class EmptyUrlKey
     private $progressIndex;
     private $productCollectionFactory;
     private $attributeScopeOverriddenValueFactory;
+    private $configUtil;
 
     public function __construct(
         StoresUtil $storesUtil,
         Progress $progress,
         ProductCollectionFactory $productCollectionFactory,
-        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory
+        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory,
+        ConfigUtil $configUtil
     ) {
         $this->storesUtil = $storesUtil;
         $this->progress = $progress;
         $this->productCollectionFactory = $productCollectionFactory;
         $this->attributeScopeOverriddenValueFactory = $attributeScopeOverriddenValueFactory;
+        $this->configUtil = $configUtil;
 
         $this->progressIndex = 0;
     }
@@ -84,6 +89,13 @@ private function checkForEmptyUrlKeyAttributeValues(): array
                 ], null, $joinType)
             ;
 
+            if ($this->configUtil->getOnlyCheckVisibleProducts()) {
+                $collection->addAttributeToFilter(
+                    ProductInterface::VISIBILITY,
+                    ['neq' => ProductVisibility::VISIBILITY_NOT_VISIBLE]
+                );
+            }
+
             if ($this->progressIndex === 0) {
                 $this->progress->setGuestimatedSize(count($storeIds), $collection->getSize());
             }

Checker/Catalog/Product/UrlPath.php

@@ -4,10 +4,12 @@
 
 namespace Baldwin\UrlDataIntegrityChecker\Checker\Catalog\Product;
 
+use Baldwin\UrlDataIntegrityChecker\Util\Configuration as ConfigUtil;
 use Baldwin\UrlDataIntegrityChecker\Util\Stores as StoresUtil;
 use Magento\Catalog\Api\Data\ProductInterface;
 use Magento\Catalog\Model\Attribute\ScopeOverriddenValueFactory as AttributeScopeOverriddenValueFactory;
 use Magento\Catalog\Model\Product as ProductModel;
+use Magento\Catalog\Model\Product\Visibility as ProductVisibility;
 use Magento\Catalog\Model\ResourceModel\Product\Collection as ProductCollection;
 use Magento\Catalog\Model\ResourceModel\Product\CollectionFactory as ProductCollectionFactory;
 use Magento\Store\Model\Store;
@@ -23,15 +25,18 @@ class UrlPath
     private $storesUtil;
     private $productCollectionFactory;
     private $attributeScopeOverriddenValueFactory;
+    private $configUtil;
 
     public function __construct(
         StoresUtil $storesUtil,
         ProductCollectionFactory $productCollectionFactory,
-        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory
+        AttributeScopeOverriddenValueFactory $attributeScopeOverriddenValueFactory,
+        ConfigUtil $configUtil
     ) {
         $this->storesUtil = $storesUtil;
         $this->productCollectionFactory = $productCollectionFactory;
         $this->attributeScopeOverriddenValueFactory = $attributeScopeOverriddenValueFactory;
+        $this->configUtil = $configUtil;
     }
 
     /**
@@ -64,6 +69,13 @@ private function checkForNonEmptyUrlPathAttributeValues(): array
                 ->addAttributeToFilter(self::URL_PATH_ATTRIBUTE, ['notnull' => true], $joinType)
             ;
 
+            if ($this->configUtil->getOnlyCheckVisibleProducts()) {
+                $collection->addAttributeToFilter(
+                    ProductInterface::VISIBILITY,
+                    ['neq' => ProductVisibility::VISIBILITY_NOT_VISIBLE]
+                );
+            }
+
             $productsWithProblems[] = $this->getProductsWithProblems($storeId, $collection);
         }
 

README.md

@@ -67,6 +67,8 @@ In the Magento admin, you can find the results in:
 The results of the checkers are currently stored by default in the directory `var/tmp` as `.json` files.  
 But you can change the path in the backend settings under Stores > Configuration > Catalog > Url Data Integrity Checker by entering a relative path starting from the Magento installation directory or an absolute path. The directory you enter there needs to exist before it will work.
 
+You can configure this module to ignore problems with invisible products (via Stores > Configuration > Catalog > Url Data Integrity Checker). Because in recent versions of Magento, url rewrites for invisible products are not being generated, so if there are problems with the `url_path` or `url_key` attributes of such products, they should not cause issues with url rewrites. An additional benefit of this option is that it will use less time and less memory to run the product checkers. This option is disabled by default, so you'll need to enable it.
+
 ## Some screenshots
 
 ### Example of backend report for product url key problems

Test/Checker/Catalog/Product/UrlKey/DuplicateUrlKeyTest.php

@@ -6,6 +6,7 @@
 
 use Baldwin\UrlDataIntegrityChecker\Checker\Catalog\Product\UrlKey\DuplicateUrlKey as UrlKeyChecker;
 use Baldwin\UrlDataIntegrityChecker\Console\Progress;
+use Baldwin\UrlDataIntegrityChecker\Util\Configuration as ConfigUtil;
 use Baldwin\UrlDataIntegrityChecker\Util\Stores as StoresUtil;
 use Magento\Catalog\Model\Attribute\ScopeOverriddenValue as AttributeScopeOverriddenValue;
 use Magento\Catalog\Model\Attribute\ScopeOverriddenValueFactory as AttributeScopeOverriddenValueFactory;
@@ -119,11 +120,18 @@ function ($productsData) {
             ->method('create')
             ->willReturn($attributeScopeOverriddenValueMock);
 
+        /** @var ConfigUtil&MockObject */
+        $configUtilMock = $this
+            ->getMockBuilder(ConfigUtil::class)
+            ->disableOriginalConstructor()
+            ->getMock();
+
         $urlKeyChecker = new UrlKeyChecker(
             $storesUtilMock,
             $progressMock,
             $productCollectionFactoryMock,
-            $attributeScopeOverriddenValueFactoryMock
+            $attributeScopeOverriddenValueFactoryMock,
+            $configUtilMock
         );
         $results = $urlKeyChecker->execute();
 

Util/Configuration.php

@@ -0,0 +1,29 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Baldwin\UrlDataIntegrityChecker\Util;
+
+use Magento\Framework\App\Config\ScopeConfigInterface;
+
+class Configuration
+{
+    const CONFIG_ONLY_CHECK_VISIBLE_PRODUCTS = 'url_data_integrity_checker/configuration/only_check_visible_products';
+
+    private $scopeConfig;
+
+    public function __construct(
+        ScopeConfigInterface $scopeConfig
+    ) {
+        $this->scopeConfig = $scopeConfig;
+    }
+
+    public function getOnlyCheckVisibleProducts(): bool
+    {
+        $configValue = (bool) $this->scopeConfig->getValue(
+            self::CONFIG_ONLY_CHECK_VISIBLE_PRODUCTS
+        );
+
+        return $configValue;
+    }
+}

composer.json

@@ -15,6 +15,7 @@
         "magento/framework": "^100.1 || ^101.0 || ^102.0 || ^103.0",
         "magento/module-backend": "^100.1 || ^101.0 || ^102.0",
         "magento/module-catalog": "^101.0 || ^102.0 || ^103.0 || ^104.0",
+        "magento/module-config": "^100.1 || ^101.0",
         "magento/module-cron": "^100.1",
         "magento/module-store": "^100.1 || ^101.0",
         "magento/module-ui": "^100.1 || ^101.0",

composer.lock

@@ -11,6 +11,11 @@
                     <label>Path to store results</label>
                     <comment>Path where to store json files, can be either an absolute or a relative path</comment>
                 </field>
+                <field id="only_check_visible_products" type="select" sortOrder="20" showInDefault="1" showInWebsite="0" showInStore="0" translate="label">
+                    <label>Don't check for problems with invisible products</label>
+                    <comment>Sometimes you want to focus on products that are visible only, because Magento (at least in recent versions) only generated url rewrites for products that are visible. Be aware, after switching this config setting that you should re-run the checkers.</comment>
+                    <source_model>Magento\Config\Model\Config\Source\Yesno</source_model>
+                </field>
             </group>
         </section>
     </system>