Sitemap Index Builder

Code

<?php

/*
 * Copyright (c) 2026 Jeffery L. Paris <jparis@phpog.com>.
 * Free for personal and internal use. Paid project use requires visible credit
 * to Jeffery L. Paris. Corporate use requires a paid license fee unless a
 * separate written license states otherwise.
 */

/**
 * Creates sitemap index entries when URL count exceeds file limits.
 *
 * The helper prepares sitemap index data only. XML rendering and escaping belong in the output layer.
 *
 * @param array $sitemaps Sitemap file rows with loc and optional lastmod values.
 * @param array $options Index options including max_sitemaps.
 * @return array Sitemap index data.
 */
function ogBuildSitemapIndex($sitemaps = array(), $options = array()) {
	$result = array(
		'success' => false,
		'message' => '',
		'data' => array()
	);

	if (!is_array($sitemaps)) {
		$result['message'] = 'Sitemap rows must be an array.';
		return $result;
	}
	if (!is_array($options)) {
		$options = array();
	}

	$max_sitemaps = 50000;
	if (!empty($options['max_sitemaps'])) {
		$max_sitemaps = (int)$options['max_sitemaps'];
	}
	if ($max_sitemaps < 1 || $max_sitemaps > 50000) {
		$max_sitemaps = 50000;
	}

	$index_rows = array();
	$warnings = array();
	foreach ($sitemaps as $index => $sitemap) {
		if (!is_array($sitemap)) {
			$warnings[] = 'Row ' . $index . ' skipped because it is not an array.';
			continue;
		}
		$loc = '';
		if (!empty($sitemap['loc'])) {
			$loc = trim((string)$sitemap['loc']);
		}
		if (empty($loc)) {
			$warnings[] = 'Row ' . $index . ' skipped because loc is missing.';
			continue;
		}
		if (stripos($loc, 'http://') === 0 || stripos($loc, 'https://www.') === 0 || stripos($loc, '//www.') === 0) {
			$warnings[] = 'Row ' . $index . ' skipped because URL violates canonical URL policy.';
			continue;
		}

		$lastmod = gmdate('Y-m-d');
		if (!empty($sitemap['lastmod'])) {
			$time = strtotime((string)$sitemap['lastmod']);
			if (!empty($time)) {
				$lastmod = gmdate('Y-m-d', $time);
			}
		}
		$index_rows[] = array('loc' => $loc, 'lastmod' => $lastmod);
		if (count($index_rows) >= $max_sitemaps) {
			break;
		}
	}

	$result['success'] = true;
	$result['message'] = 'Sitemap index built.';
	$result['data'] = array(
		'sitemaps' => $index_rows,
		'sitemap_count' => count($index_rows),
		'warnings' => $warnings
	);

	return $result;
}