Get Collections for products, categories, and pages. Filter them, use any custom PHP you like and create an XML file. Couldn't find it so I built it. Very useful for carts that only have configurable "main" products and don't want all the bits and pieces "simple products" showing in the sitemap. 1. create a my_sitemap.xml to be overwritten. 2. K…

Magento Custom XML Sitemap Creator

<?php
/* Compilation includes configuration file */
define('MAGENTO_ROOT', getcwd());

	$compilerConfig = MAGENTO_ROOT . '/includes/config.php';
	if (file_exists($compilerConfig)) {
	    include $compilerConfig;
	}

$mageFilename = MAGENTO_ROOT . '/app/Mage.php';
require_once $mageFilename;

Mage::app()->getRequest()->getRouteName();
$read = Mage::getSingleton('core/resource')->getConnection('core_read');

/* EOF Magento Config */

// SET YOUR BASE URL
$baseUrl = "http://www..com/"; // dont forget trailing /


//	GET ALL 
$products = Mage::getModel('catalog/product')->getCollection();
//var_dump($products->getFirstItem()->getData());  // dump just the first for testing
$products->addFieldToFilter('type_id','configurable'); // SHOW ME ONLY CONFIGURABLE PRODUCTS
$products = $products->getData();

$categories = Mage::getModel('catalog/category')->getCollection();
$categories = $categories->getData();  

$pages = Mage::getModel('cms/page')->getCollection();
$pages->addFieldToFilter('is_active','1'); // they gotta be active
$pages = $pages->getData();



//START OUR XML CODE
$fp = fopen('my_sitemap.xml', 'w');
fwrite($fp, "<?xml version=\"1.0\" encoding=\"UTF-8\"?>
<urlset xmlns=\"http://www.sitemaps.org/schemas/sitemap/0.9\">
");



//WRITE ACTIVE PAGES
foreach($pages as $page){
	if($page["identifier"]=="home"){
		//skip we dont need /home -- enter any of page name above to exclude
	} else {
		fwrite($fp, "
			<url><loc>".$baseUrl.$page["identifier"]."</loc></url> 
			<lastmod>".$page["updated_time"]."</lastmod>
			<priority>0.5</priority>
		");
	}
}


//WRITE ACTIVE CATEGORIES
foreach($categories as $cat){
	if($cat["parent_id"]==0 OR $cat["parent_id"]==1){
		//skip b/c this is the root and the "Default Category"
	} else {
	
		$realCats = $read->fetchRow("SELECT * FROM catalog_category_flat_store_1 WHERE entity_id=".$cat["entity_id"]." AND is_active=1");


		fwrite($fp, "
			<url><loc>".$baseUrl.$realCats["url_path"]."</loc></url> 
			<lastmod>".$realCats["updated_at"]."</lastmod>
			<priority>0.7</priority>
		");
	}

}

//WRITE CATEGORIES
foreach($products as $item){
	//echo $item["sku"]; // check it works echo the sku

	$result = $read->fetchRow("SELECT * FROM catalog_product_flat_1 WHERE sku='".$item["sku"]."'"); 
	
	//checkout results
	//echo "<pre>";
	//print_r($result);
	//break; // just do one loop
	
	if($result){
		//lets write the URLs
		fwrite($fp, "
			<url><loc>".$baseUrl.$result["url_path"]."</loc></url>
			<lastmod>".$result["updated_at"]."</lastmod>
			<priority>0.8</priority>
		");
	} else {
		//do nothing b/c we had an error
	}
	
	unset($result);	
}


fwrite($fp, "</urlset>");

fclose($fp);
?>