sm-sitemap-plugin/tests/php/gen-goldens.php

<?php
/**
 * Golden generator for the Go sitemap port.
 *
 * Runs the PHP Golem15.Sitemap generation code (Definition::generateSitemap,
 * getXmlObject, getUrlSet, addItemToSet, lookupManifestLastmod,
 * getPriorityForUrl and the routes.php frontend_url rewrite, copied below
 * with the Winter facades replaced by the case's inputs) and the vendored
 * Winter URL generator (Url::to, Url::buildUrl) over a JSON input, and
 * writes the bytes PHP produces. The Go tests compare their output with the
 * committed goldens, so CI needs no PHP.
 *
 * Usage (from the plugin root, with a WinterCMS vendor directory):
 *
 *   WINTER_VENDOR=/path/to/winter-app/vendor \
 *     php tests/php/gen-goldens.php classes/testdata/tracer.json classes/testdata/tracer.golden
 *
 * Input shapes:
 *
 *   {"case": {...}}                 one sitemap case; the golden is its XML
 *   {"cases": [{"name": ..., ...}]} several cases; each written as
 *                                   "=== name\n" followed by its XML
 *   {"root": ..., "urls": [...]}    Url::to of each input, one per line
 *
 * A sitemap case: root (the backend root URL), items (stored definition
 * items, PHP DefinitionItem keys), types (registry item type => reference =>
 * resolved item {url, mtime, alternates: [[locale, url], ...], items}),
 * settings (Sitemap settings values; absent keys take the PHP defaults),
 * manifest (the updates manifest object), skip (URLs a beforeAddItem
 * listener drops) and extra (URL => text of an <extra> element an addItem
 * listener appends).
 *
 * A registry type's resolved item is given in the Go shape (one item per
 * record with ordered alternates); this script turns it into what the PHP
 * headless resolvers returned: a list type (no url) yields one entry per
 * alternate for each child (a child without alternates once), any other type
 * its children with their default URL.
 */

$vendor = getenv('WINTER_VENDOR');
if (!$vendor || !is_file($vendor . '/autoload.php')) {
    fwrite(STDERR, "set WINTER_VENDOR to the vendor directory of a WinterCMS application\n");
    exit(2);
}
require $vendor . '/autoload.php';

use Illuminate\Http\Request;
use Illuminate\Routing\RouteCollection;
use Winter\Storm\Router\UrlGenerator;

date_default_timezone_set('UTC');

if ($argc !== 3) {
    fwrite(STDERR, "usage: php gen-goldens.php INPUT.json OUTPUT.golden\n");
    exit(2);
}

$input = json_decode(file_get_contents($argv[1]), true);
if (!is_array($input)) {
    fwrite(STDERR, "invalid JSON input\n");
    exit(1);
}

function urlGenerator(string $root): UrlGenerator
{
    return new UrlGenerator(new RouteCollection, Request::create(rtrim($root, '/') . '/sitemap.xml'));
}

if (isset($input['urls'])) {
    $url = urlGenerator($input['root']);
    $out = '';
    foreach ($input['urls'] as $path) {
        $out .= $url->to($path) . "\n";
    }
    file_put_contents($argv[2], $out);
    exit(0);
}

if (isset($input['case'])) {
    file_put_contents($argv[2], (new GoldenSitemap($input['case']))->render());
    exit(0);
}

$out = '';
foreach ($input['cases'] as $case) {
    $out .= '=== ' . $case['name'] . "\n" . (new GoldenSitemap($case))->render();
}
file_put_contents($argv[2], $out);

/**
 * Golem15\Sitemap\Classes\DefinitionItem, unchanged properties.
 */
class GoldenDefinitionItem
{
    public $changefreq;
    public $priority;
    public $nesting = false;
    public $type = 'url';
    public $url = null;
    public $code = null;
    public $reference = null;
    public $cmsPage = null;
    public $urlPattern = null;
    public $exists = false;

    public static function initFromArray(array $items): array
    {
        $result = [];
        foreach ($items as $itemData) {
            $obj = new self;
            foreach ($itemData as $name => $value) {
                if (property_exists($obj, $name)) {
                    $obj->$name = $value;
                }
            }
            $result[] = $obj;
        }
        return $result;
    }
}

/**
 * The generation code of Golem15\Sitemap\Models\Definition and routes.php.
 */
class GoldenSitemap
{
    const MAX_URLS = 50000;

    protected $case;
    protected $url;
    protected $items;
    protected $updatesManifest = null;
    protected $urlCount = 0;
    protected $urlSet;
    protected $xmlObject;

    public function __construct(array $case)
    {
        $this->case = $case;
        $this->url = urlGenerator($case['root']);
        $this->items = GoldenDefinitionItem::initFromArray($case['items'] ?? []);
    }

    protected function setting(string $key, $default)
    {
        $settings = $this->case['settings'] ?? [];
        return array_key_exists($key, $settings) ? $settings[$key] : $default;
    }

    /** routes.php */
    public function render(): string
    {
        $xml = $this->generateSitemap();

        $frontendUrl = $this->setting('frontend_url', null);
        if ($frontendUrl) {
            $frontendUrl = rtrim($frontendUrl, '/');
            $backendUrl = rtrim($this->url->to('/'), '/');
            $xml = preg_replace_callback('/<loc>(.*?)<\/loc>/', function ($m) use ($backendUrl, $frontendUrl) {
                return '<loc>' . str_replace($backendUrl, $frontendUrl, $m[1]) . '</loc>';
            }, $xml);
            $xml = preg_replace_callback('/(<xhtml:link[^>]*href=")([^"]*)(")/', function ($m) use ($backendUrl, $frontendUrl) {
                return $m[1] . str_replace($backendUrl, $frontendUrl, $m[2]) . $m[3];
            }, $xml);
            $xml = str_replace(
                'href="' . $backendUrl . '/sitemap.xsl"',
                'href="' . $frontendUrl . '/sitemap.xsl"',
                $xml
            );
        }

        return $xml;
    }

    /**
     * What the PHP headless resolvers returned for a resolved item in the
     * Go shape.
     */
    protected function phpResolved(array $r, bool $listChildren): array
    {
        $info = [];
        if (!empty($r['url'])) {
            $info['url'] = $r['url'];
        }
        if (array_key_exists('mtime', $r)) {
            $info['mtime'] = $r['mtime'] === null ? null : new DateTime($r['mtime']);
        }
        $links = [];
        foreach ($r['alternates'] ?? [] as [$locale, $href]) {
            $links[$locale] = $href;
        }
        $info['alternateLinks'] = $links;
        if (isset($r['items'])) {
            $info['items'] = [];
            foreach ($r['items'] as $child) {
                $childInfo = $this->phpResolved($child, false);
                if ($listChildren && $childInfo['alternateLinks']) {
                    foreach ($childInfo['alternateLinks'] as $locale => $href) {
                        $perLocale = $childInfo;
                        $perLocale['url'] = $href;
                        $info['items'][] = $perLocale;
                    }
                } else {
                    $info['items'][] = $childInfo;
                }
            }
        }
        return $info;
    }

    protected function resolveItem(GoldenDefinitionItem $item)
    {
        $types = $this->case['types'] ?? [];
        $ref = (string) $item->reference;
        if (!isset($types[$item->type][$ref])) {
            return null;
        }
        $r = $types[$item->type][$ref];
        return $this->phpResolved($r, empty($r['url']));
    }

    public function generateSitemap(): string
    {
        if (!$this->items) {
            return '';
        }

        foreach ($this->items as $item) {
            if ($item->type === 'url') {
                $this->addItemToSet($item);
            } else {
                $menuItemInfo = $this->resolveItem($item);
                if (!is_array($menuItemInfo)) {
                    continue;
                }

                if (isset($menuItemInfo['url'])) {
                    $this->addItemToSet($item, $menuItemInfo);
                }

                if (isset($menuItemInfo['items'])) {
                    $menuItemIterator = function ($menuItems) use (&$menuItemIterator, $item) {
                        foreach ($menuItems as $menuItem) {
                            if (isset($menuItem['url'])) {
                                $this->addItemToSet($item, $menuItem);
                            }

                            if (isset($menuItem['items'])) {
                                $menuItemIterator($menuItem['items']);
                            }
                        }
                    };

                    $menuItemIterator($menuItemInfo['items']);
                }
            }
        }

        $urlSet = $this->getUrlSet();
        $xml = $this->getXmlObject();
        $xml->appendChild($urlSet);

        return $xml->saveXML();
    }

    protected function getXmlObject(): DOMDocument
    {
        if ($this->xmlObject !== null) {
            return $this->xmlObject;
        }

        $xml = new DOMDocument;
        $xml->encoding = 'UTF-8';
        $xss = $xml->createProcessingInstruction('xml-stylesheet',
            'type="text/xsl" href="' . UrlGenerator::buildUrl($this->url->current(), ['path' => 'sitemap.xsl']) . '"'
        );
        $xml->appendChild($xss);

        return $this->xmlObject = $xml;
    }

    protected function getUrlSet(): DOMElement
    {
        if ($this->urlSet !== null) {
            return $this->urlSet;
        }

        $xml = $this->getXmlObject();
        $urlSet = $xml->createElement('urlset');
        $urlSet->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
        $urlSet->setAttribute('xmlns:xhtml', 'http://www.w3.org/1999/xhtml');
        $urlSet->setAttribute('xmlns:xsi', 'http://www.w3.org/2001/XMLSchema-instance');
        $urlSet->setAttribute('xsi:schemaLocation', 'http://www.sitemaps.org/schemas/sitemap/0.9 http://www.sitemaps.org/schemas/sitemap/0.9/sitemap.xsd');

        return $this->urlSet = $urlSet;
    }

    protected function addItemToSet(GoldenDefinitionItem $item, ?array $itemInfo = null): void
    {
        $xml = $this->getXmlObject();
        $urlSet = $this->getUrlSet();

        if ($this->urlCount >= self::MAX_URLS) {
            return;
        }

        if (!isset($itemInfo['url'])) {
            $itemInfo['url'] = $this->url->to($item->url);
        }

        $lastModified = $itemInfo['mtime'] ?? null;

        if ($lastModified === null) {
            $lastModified = $this->lookupManifestLastmod($itemInfo['url']);
        }

        if ($lastModified instanceof \DateTime) {
            $lastModified = $lastModified->getTimestamp();
        }
        $itemInfo['lastModified'] = $lastModified ? date('c', $lastModified) : null;

        // golem15.sitemap.beforeAddItem: a listener that drops the case's
        // skip URLs.
        if (in_array($itemInfo['url'], $this->case['skip'] ?? [], true)) {
            return;
        }

        if (empty($itemInfo['url'])) {
            return;
        } else {
            $itemInfo['url'] = $this->url->to($itemInfo['url']);
        }

        $this->urlCount++;

        $tierPriority = $this->getPriorityForUrl($itemInfo['url']);
        $effectivePriority = $tierPriority ?? $item->priority;

        $urlElement = $xml->createElement('url');

        $loc = $xml->createElement('loc');
        $loc->appendChild($xml->createTextNode($itemInfo['url']));
        $urlElement->appendChild($loc);

        if (!empty($itemInfo['lastModified'])) {
            $lastmod = $xml->createElement('lastmod');
            $lastmod->appendChild($xml->createTextNode((string) $itemInfo['lastModified']));
            $urlElement->appendChild($lastmod);
        }

        $changefreq = $xml->createElement('changefreq');
        $changefreq->appendChild($xml->createTextNode((string) $item->changefreq));
        $urlElement->appendChild($changefreq);

        $priority = $xml->createElement('priority');
        $priority->appendChild($xml->createTextNode((string) $effectivePriority));
        $urlElement->appendChild($priority);

        if (!empty($itemInfo['alternateLinks']) && count($itemInfo['alternateLinks']) > 1) {
            foreach ($itemInfo['alternateLinks'] as $hreflang => $href) {
                $link = $xml->createElement('xhtml:link');
                $link->setAttribute('rel', 'alternate');
                $link->setAttribute('hreflang', $hreflang);
                $link->setAttribute('href', $this->url->to($href));
                $urlElement->appendChild($link);
            }
        }

        $urlSet->appendChild($urlElement);

        // golem15.sitemap.addItem: a listener that appends the case's extra
        // element to the URL.
        $extra = $this->case['extra'] ?? [];
        if (isset($extra[$itemInfo['url']])) {
            $element = $xml->createElement('extra');
            $element->appendChild($xml->createTextNode($extra[$itemInfo['url']]));
            $urlElement->appendChild($element);
        }
    }

    protected function lookupManifestLastmod(string $url): ?int
    {
        $manifest = $this->getUpdatesManifest();
        if (!$manifest) {
            return null;
        }

        $path = parse_url($url, PHP_URL_PATH) ?: '/';
        $path = $this->normalizeManifestPath($path);

        if (!isset($manifest[$path])) {
            return null;
        }

        $timestamp = strtotime((string) $manifest[$path]);

        return $timestamp ?: null;
    }

    protected function getUpdatesManifest(): array
    {
        if ($this->updatesManifest !== null) {
            return $this->updatesManifest;
        }

        $decoded = $this->case['manifest'] ?? null;
        if (!is_array($decoded)) {
            return $this->updatesManifest = [];
        }

        $normalized = [];
        foreach ($decoded as $key => $value) {
            $normalized[$this->normalizeManifestPath((string) $key)] = $value;
        }

        return $this->updatesManifest = $normalized;
    }

    protected function normalizeManifestPath(string $path): string
    {
        $path = '/' . ltrim($path, '/');

        return $path === '/' ? '/' : rtrim($path, '/');
    }

    protected function getPriorityForUrl(string $url): ?string
    {
        if (!$this->setting('enable_priority_tiers', true)) {
            return null;
        }

        $path = parse_url($url, PHP_URL_PATH) ?: '/';
        $path = rtrim($path, '/');

        if ($path === '' || $path === '/en') {
            return $this->setting('priority_homepage', '1.0');
        }

        if (preg_match('#^/(uslugi|en/services|realizacje|en/work)/[^/]+$#', $path)) {
            return $this->setting('priority_services', '0.8');
        }

        if (preg_match('#^/(blog|en/blog)/[^/]+/[^/]+$#', $path)) {
            return $this->setting('priority_blog', '0.6');
        }

        if (preg_match('#^/(uslugi|en/services|realizacje|en/work|blog|en/blog)(/[^/]+)?$#', $path)) {
            return $this->setting('priority_categories', '0.4');
        }

        $legalPaths = [
            '/polityka-prywatnosci', '/en/privacy-policy',
            '/regulamin', '/en/terms',
            '/o-nas', '/en/about',
            '/kontakt', '/en/contact',
            '/laboratorium-ai', '/en/ai-lab',
            '/laboratorium-ai/zabawa-animacja', '/en/ai-lab/blob-playground',
        ];
        if (in_array($path, $legalPaths)) {
            return $this->setting('priority_legal', '0.2');
        }

        return null;
    }
}