Magento URL rewrite bug

app_code_local_Lyonscg_Catalog_Model_Resource_Url.php

<?php
/**
 * Rewrite Mage_Catalog_Model_Resource_Url class
 *
 * @category    Lyonscg
 * @package     Lyonscg_Catalog
 * @author      Vladimir Kosenok vkosenok@lyonscg.com
 * @copyright   Copyright (c) 2012 Lyons Consulting Group (www.lyonscg.com)
 */

/**
 * Catalog url rewrite resource model
 * Add product 'type_id' field to _getProducts() method
 *
 * @category    Lyonscg
 * @package     Lyonscg_Catalog
 * @author      Vladimir Kosenok vkosenok@lyonscg.com
 */
class Lyonscg_Catalog_Model_Resource_Url extends Mage_Catalog_Model_Resource_Url
{

    /**
     * Retrieve Product data objects
     * Modified: add "type_id" field to select
     *
     * @param int|array $productIds
     * @param int $storeId
     * @param int $entityId
     * @param int $lastEntityId
     * @return array
     */
    protected function _getProducts($productIds, $storeId, $entityId, &$lastEntityId)
    {
        $products   = array();
        $websiteId  = Mage::app()->getStore($storeId)->getWebsiteId();
        $adapter    = $this->_getReadAdapter();
        if ($productIds !== null) {
            if (!is_array($productIds)) {
                $productIds = array($productIds);
            }
        }
        $bind = array(
            'website_id' => (int)$websiteId,
            'entity_id'  => (int)$entityId,
        );
        $select = $adapter->select()
            ->useStraightJoin(true)
            ->from(array('e' => $this->getTable('catalog/product')), array('entity_id', 'type_id'))
            ->join(
                array('w' => $this->getTable('catalog/product_website')),
                'e.entity_id = w.product_id AND w.website_id = :website_id',
                array()
            )
            ->where('e.entity_id > :entity_id')
            ->order('e.entity_id')
            ->limit($this->_productLimit);
        if ($productIds !== null) {
            $select->where('e.entity_id IN(?)', $productIds);
        }

        $rowSet = $adapter->fetchAll($select, $bind);
        foreach ($rowSet as $row) {
            $product = new Varien_Object($row);
            $product->setIdFieldName('entity_id');
            $product->setCategoryIds(array());
            $product->setStoreId($storeId);
            $products[$product->getId()] = $product;
            $lastEntityId = $product->getId();
        }

        unset($rowSet);

        if ($products) {
            $select = $adapter->select()
                ->from(
                    $this->getTable('catalog/category_product'),
                    array('product_id', 'category_id')
                )
                ->where('product_id IN(?)', array_keys($products));
            $categories = $adapter->fetchAll($select);
            foreach ($categories as $category) {
                $productId = $category['product_id'];
                $categoryIds = $products[$productId]->getCategoryIds();
                $categoryIds[] = $category['category_id'];
                $products[$productId]->setCategoryIds($categoryIds);
            }

            foreach (array('name', 'url_key', 'url_path') as $attributeCode) {
                $attributes = $this->_getProductAttribute($attributeCode, array_keys($products), $storeId);
                foreach ($attributes as $productId => $attributeValue) {
                    $products[$productId]->setData($attributeCode, $attributeValue);
                }
            }
        }

        return $products;
    }

}

app_code_local_Lyonscg_Catalog_Model_Url.php

<?php
/**
 * Rewrite Mage_Catalog_Model_Url class
 *
 * @category    Lyonscg
 * @package     Lyonscg_Catalog
 * @author      Vladimir Kosenok vkosenok@lyonscg.com
 * @copyright   Copyright (c) 2012 Lyons Consulting Group (www.lyonscg.com)
 */

/**
 * Catalog url model
 * Override Url rewrites functionality to fix Magento bug
 * Bug appears when products have equal names and enabled option
 * System/Configuration/Catalog/Search Engine Optimizations/Create Permanent Redirect for URLs if URL Key Changed
 * save history of url rewrites changes.
 * Magento increases count of url rewrites each time after reindex, independently of catalog changes
 *
 * @category    Lyonscg
 * @package     Lyonscg_Catalog
 * @author      Vladimir Kosenok vkosenok@lyonscg.com
 */
class Lyonscg_Catalog_Model_Url extends Mage_Catalog_Model_Url
{

    /**
     * Refresh all product rewrites for designated store
     * Modified: exclude simples with non bounded categories (options of configurables)
     *
     * @param int $storeId
     * @return Mage_Catalog_Model_Url
     */
    public function refreshProductRewrites($storeId)
    {
        $this->_categories      = array();
        $storeRootCategoryId    = $this->getStores($storeId)->getRootCategoryId();
        $storeRootCategoryPath  = $this->getStores($storeId)->getRootCategoryPath();
        $this->_categories[$storeRootCategoryId] = $this->getResource()->getCategory($storeRootCategoryId, $storeId);

        $lastEntityId = 0;
        $process = true;

        while ($process == true) {
            $products = $this->getResource()->getProductsByStore($storeId, $lastEntityId);
            if (!$products) {
                $process = false;
                break;
            }

            $this->_rewrites = $this->getResource()->prepareRewrites($storeId, false, array_keys($products));

            $loadCategories = array();
            foreach ($products as $product) {
                foreach ($product->getCategoryIds() as $categoryId) {
                    if (!isset($this->_categories[$categoryId])) {
                        $loadCategories[$categoryId] = $categoryId;
                    }
                }
            }

            if ($loadCategories) {
                foreach ($this->getResource()->getCategories($loadCategories, $storeId) as $category) {
                    $this->_categories[$category->getId()] = $category;
                }
            }

            foreach ($products as $product) {
                //modified: exclude simples with non bounded categories (options of configurables)
                if($product->getTypeId() == "simple" && !count($product->getCategoryIds())){
                    continue;
                }
                $this->_refreshProductRewrite($product, $this->_categories[$storeRootCategoryId]);
                foreach ($product->getCategoryIds() as $categoryId) {
                    if ($categoryId != $storeRootCategoryId && isset($this->_categories[$categoryId])) {
                        if (strpos($this->_categories[$categoryId]['path'], $storeRootCategoryPath . '/') !== 0) {
                            continue;
                        }
                        $this->_refreshProductRewrite($product, $this->_categories[$categoryId]);
                    }
                }
            }

            unset($products);
            $this->_rewrites = array();
        }

        $this->_categories = array();
        return $this;
    }


    /**
     * Get unique product request path
     * Modified: fix magento bug with checking of existent records
     *
     * @param   Varien_Object $product
     * @param   Varien_Object $category
     * @return  string
     */
    public function getProductRequestPath($product, $category)
    {
        if ($product->getUrlKey() == '') {
            $urlKey = $this->getProductModel()->formatUrlKey($product->getName());
        } else {
            $urlKey = $this->getProductModel()->formatUrlKey($product->getUrlKey());
        }
        $storeId = $category->getStoreId();
        $suffix  = $this->getProductUrlSuffix($storeId);
        $idPath  = $this->generatePath('id', $product, $category);
        /**
         * Prepare product base request path
         */
        if ($category->getLevel() > 1) {
            // To ensure, that category has path either from attribute or generated now
            $this->_addCategoryUrlPath($category);
            $categoryUrl = Mage::helper('catalog/category')->getCategoryUrlPath($category->getUrlPath(),
                false, $storeId);
            $requestPath = $categoryUrl . '/' . $urlKey;
        } else {
            $requestPath = $urlKey;
        }

        if (strlen($requestPath) > self::MAX_REQUEST_PATH_LENGTH + self::ALLOWED_REQUEST_PATH_OVERFLOW) {
            $requestPath = substr($requestPath, 0, self::MAX_REQUEST_PATH_LENGTH);
        }

        $this->_rewrite = null;
        /**
         * Check $requestPath should be unique
         */
        if (isset($this->_rewrites[$idPath])) {
            $this->_rewrite = $this->_rewrites[$idPath];
            $existingRequestPath = $this->_rewrites[$idPath]->getRequestPath();

            //fix magento bug, related to incorrect checking of existent urls
            if (($existingRequestPath == $requestPath . $suffix)
                || ($existingRequestPath == $requestPath .'-'.$product->getId(). $suffix)
            ) {
                return $existingRequestPath;
            }

            $existingRequestPath = preg_replace('/' . preg_quote($suffix, '/') . '$/', '', $existingRequestPath);
            /**
             * Check if existing request past can be used
             */
            if ($product->getUrlKey() == '' && !empty($requestPath)
                && strpos($existingRequestPath, $requestPath) === 0
            ) {
                $existingRequestPath = preg_replace(
                    '/^' . preg_quote($requestPath, '/') . '/', '', $existingRequestPath
                );
                if (preg_match('#^-([0-9]+)$#i', $existingRequestPath)) {
                    return $this->_rewrites[$idPath]->getRequestPath();
                }
            }

            $fullPath = $requestPath.$suffix;
            if ($this->_deleteOldTargetPath($fullPath, $idPath, $storeId)) {
                return $fullPath;
            }
        }
        /**
         * Check 2 variants: $requestPath and $requestPath . '-' . $productId
         */
        $validatedPath = $this->getResource()->checkRequestPaths(
            array($requestPath.$suffix, $requestPath.'-'.$product->getId().$suffix),
            $storeId
        );

        if ($validatedPath) {
            return $validatedPath;
        }
        /**
         * Use unique path generator
         */
        return $this->getUnusedPath($storeId, $requestPath.$suffix, $idPath);
    }

}

config.xml

<?xml version="1.0" encoding="UTF-8"?>
<config>
    <modules>
        <Lyonscg_Catalog>
            <version>0.1.4</version>
        </Lyonscg_Catalog>
    </modules>
    <global>
        <models>
            <lyonscg_catalog>
                <class>Lyonscg_Catalog_Model</class>
            </lyonscg_catalog>
            <catalog>
                <rewrite>
                    <url>Lyonscg_Catalog_Model_Url</url>
                </rewrite>
            </catalog>
            <catalog_resource>
                <rewrite>
                    <url>Lyonscg_Catalog_Model_Resource_Url</url>
                </rewrite>
            </catalog_resource>
        </models>
        <resources>
            <lyonscg_catalog_setup>
                <setup>
                    <module>Lyonscg_Catalog</module>
                    <class>Mage_Core_Model_Resource_Setup</class>
                </setup>
            </lyonscg_catalog_setup>
        </resources>
    </global>
</config>