sitemap.xml XML to HTML visualisation

sitemap-xml-to-html.php

<?php

// composer require guzzlehttp/guzzle
// composer require pixelbrackets/html5-mini-template
require 'vendor/autoload.php';

use GuzzleHttp\Client;

function fetchSitemapUrls(string $sitemapUrl, Client $client): array {
    $response = $client->get($sitemapUrl);
    $xml = simplexml_load_string($response->getBody()->getContents());
    $urls = [];

    // follow nested sitemaps
    if (isset($xml->sitemap)) {
        foreach ($xml->sitemap as $sitemap) {
            $urls = array_merge($urls, fetchSitemapUrls((string)$sitemap->loc, $client));
        }
    }

    if (isset($xml->url)) {
        foreach ($xml->url as $url) {
            $urls[] = (string)$url->loc;
        }
    }

    return $urls;
}

function buildUrlTree(array $urls, int $maxLevel = 4): array {
    $tree = [];
    foreach ($urls as $url) {
        $parts = parse_url($url);
        $parts['path'] = isset($parts['path']) ? $parts['path'] : '';
        $pathSegments = explode('/', trim($parts['path'], '/'));

        // Only include up to maxLevel
        $pathSegments = array_slice($pathSegments, 0, $maxLevel);

        $current =& $tree;
        foreach ($pathSegments as $segment) {
            if (!isset($current[$segment])) {
                $current[$segment] = [];
            }
            $current =& $current[$segment];
        }
    }
    return $tree;
}

function generateHtml(array $tree, string $baseUrl, string $path = '', bool $skipShortlinks = true): string {
    $html = PHP_EOL . '<ul>';
    foreach ($tree as $segment => $children) {
        // Skip shortlinks (first-level item with no subpages)
        if ($skipShortlinks && $path === '' && empty($children)) {
            continue;
        }

        // Guess the correct title
        $formattedTitle = ucwords(str_replace('-', ' ', $segment));

        $fullPath = $path . '/' . $segment;
        $html .= PHP_EOL . '  <li><a href="' . $baseUrl . $fullPath . '">' . $formattedTitle . '</a><!--<span>/' . $segment . '</span>-->';
        if (!empty($children)) {
            $html .= generateHtml($children, $baseUrl, $fullPath, true);
        }
        $html .= '</li>';
    }
    $html .= PHP_EOL . '</ul>';
    return $html;
}

$sitemapUrl = 'https://www.example.com/sitemap.xml';
$baseUrl = parse_url($sitemapUrl, PHP_URL_SCHEME) . '://' . parse_url($sitemapUrl, PHP_URL_HOST);
$urls = fetchSitemapUrls($sitemapUrl, new Client());
$urlTree = buildUrlTree($urls, 3);
$html = generateHtml($urlTree, $baseUrl);

$template = (new \Pixelbrackets\Html5MiniTemplate\Html5MiniTemplate())
    ->setStylesheet('https://raw.githubusercontent.com/astuteo/slickmap/refs/heads/master/slickmap.css')
    ->setTitle('Sitemap')
    ->setContent('<div class="sitemap"><nav class="primaryNav">' . $html . '</nav></div>');
echo $template->getMarkup();