SkinbaseNova/app/Services/Sitemaps/SitemapReleaseValidator.php

<?php

declare(strict_types=1);

namespace App\Services\Sitemaps;

use DOMDocument;
use DOMXPath;

final class SitemapReleaseValidator
{
    public function __construct(
        private readonly SitemapBuildService $build,
        private readonly SitemapReleaseManager $releases,
    ) {
    }

    /**
     * @return array<string, mixed>
     */
    public function validate(string $releaseId): array
    {
        $manifest = $this->releases->readManifest($releaseId);

        if ($manifest === null) {
            return [
                'ok' => false,
                'release_id' => $releaseId,
                'errors' => ['Release manifest not found.'],
            ];
        }

        $errors = [];
        $families = (array) ($manifest['families'] ?? []);
        $documents = (array) ($manifest['documents'] ?? []);

        $rootContent = $this->releases->getDocument($releaseId, SitemapCacheService::INDEX_DOCUMENT);
        $rootXml = is_string($rootContent) ? $this->loadXml($rootContent) : null;

        if ($rootXml === null) {
            $errors[] = 'Root sitemap.xml is missing or invalid.';
        } else {
            $rootLocs = $this->extractLocs($rootXml, 'sitemap');
            $expectedRootLocs = array_map(
                fn (string $entryName): string => url('/sitemaps/' . $entryName . '.xml'),
                array_values(array_map(static fn (array $family): string => (string) ($family['entry_name'] ?? ''), $families)),
            );

            if ($rootLocs !== $expectedRootLocs) {
                $errors[] = 'Root sitemap index does not match the manifest family entries.';
            }
        }

        $reports = [];

        foreach ($families as $familyName => $family) {
            $familyErrors = [];
            $familyWarnings = [];
            $seenLocs = [];
            $duplicates = [];

            foreach ((array) ($family['documents'] ?? []) as $documentName) {
                $artifact = $this->releases->getDocument($releaseId, (string) $documentName);

                if (! is_string($artifact) || $artifact === '') {
                    $familyErrors[] = 'Missing artifact [' . $documentName . '].';
                    continue;
                }

                $artifactXml = $this->loadXml($artifact);
                if ($artifactXml === null) {
                    $familyErrors[] = 'Invalid XML in artifact [' . $documentName . '].';
                    continue;
                }

                $expected = $documentName === SitemapCacheService::INDEX_DOCUMENT
                    ? $this->build->buildIndex(true, false, array_keys($families))
                    : $this->build->buildNamed((string) $documentName, true, false);

                if ($expected === null) {
                    $familyErrors[] = 'Unable to rebuild expected document [' . $documentName . '] for validation.';
                    continue;
                }

                $expectedXml = $this->loadXml((string) $expected['content']);
                if ($expectedXml === null) {
                    $familyErrors[] = 'Expected document [' . $documentName . '] could not be parsed.';
                    continue;
                }

                if ((string) $expected['type'] === SitemapTarget::TYPE_INDEX) {
                    if ($this->extractLocs($artifactXml, 'sitemap') !== $this->extractLocs($expectedXml, 'sitemap')) {
                        $familyErrors[] = 'Index artifact [' . $documentName . '] does not match expected sitemap references.';
                    }

                    continue;
                }

                $artifactLocs = $this->extractLocs($artifactXml, 'url');
                $expectedLocs = $this->extractLocs($expectedXml, 'url');

                if ($artifactLocs !== $expectedLocs) {
                    $familyErrors[] = 'URL artifact [' . $documentName . '] does not match expected canonical URLs.';
                }

                foreach ($artifactLocs as $loc) {
                    if (isset($seenLocs[$loc])) {
                        $duplicates[$loc] = true;
                    }

                    $seenLocs[$loc] = true;

                    $urlError = $this->urlError($loc);
                    if ($urlError !== null) {
                        $familyErrors[] = $urlError . ' [' . $loc . ']';
                    }
                }

                if ((string) $familyName === (string) config('sitemaps.news.google_variant_name', 'news-google')) {
                    if ($this->extractNewsTitles($artifactXml) === []) {
                        $familyErrors[] = 'Google News sitemap contains no valid news:title elements.';
                    }
                }

                foreach ($this->extractImageLocs($artifactXml) as $imageLoc) {
                    if (! preg_match('/^https?:\/\//i', $imageLoc)) {
                        $familyWarnings[] = 'Non-absolute image URL [' . $imageLoc . ']';
                    }
                }
            }

            if ($duplicates !== []) {
                $familyErrors[] = 'Duplicate URLs detected across family artifacts.';
            }

            $reports[] = [
                'family' => $familyName,
                'documents' => count((array) ($family['documents'] ?? [])),
                'url_count' => (int) ($family['url_count'] ?? 0),
                'shard_count' => (int) ($family['shard_count'] ?? 0),
                'errors' => $familyErrors,
                'warnings' => $familyWarnings,
            ];

            foreach ($familyErrors as $familyError) {
                $errors[] = $familyName . ': ' . $familyError;
            }
        }

        return [
            'ok' => $errors === [],
            'release_id' => $releaseId,
            'errors' => $errors,
            'families' => $reports,
            'totals' => [
                'families' => count($families),
                'documents' => count($documents),
                'urls' => array_sum(array_map(static fn (array $family): int => (int) ($family['url_count'] ?? 0), $families)),
                'shards' => array_sum(array_map(static fn (array $family): int => (int) ($family['shard_count'] ?? 0), $families)),
            ],
        ];
    }

    private function loadXml(string $content): ?DOMDocument
    {
        $document = new DOMDocument();
        $previous = libxml_use_internal_errors(true);
        $loaded = $document->loadXML($content);
        libxml_clear_errors();
        libxml_use_internal_errors($previous);

        return $loaded ? $document : null;
    }

    /**
     * @return list<string>
     */
    private function extractLocs(DOMDocument $document, string $nodeName): array
    {
        $xpath = new DOMXPath($document);
        $nodes = $xpath->query('//*[local-name()="' . $nodeName . '"]/*[local-name()="loc"]');
        $locs = [];

        foreach ($nodes ?: [] as $node) {
            $value = trim((string) $node->textContent);
            if ($value !== '') {
                $locs[] = $value;
            }
        }

        return $locs;
    }

    /**
     * @return list<string>
     */
    private function extractImageLocs(DOMDocument $document): array
    {
        $xpath = new DOMXPath($document);
        $nodes = $xpath->query('//*[local-name()="image"]/*[local-name()="loc"]');
        $locs = [];

        foreach ($nodes ?: [] as $node) {
            $value = trim((string) $node->textContent);
            if ($value !== '') {
                $locs[] = $value;
            }
        }

        return $locs;
    }

    /**
     * @return list<string>
     */
    private function extractNewsTitles(DOMDocument $document): array
    {
        $xpath = new DOMXPath($document);
        $nodes = $xpath->query('//*[local-name()="title"]');
        $titles = [];

        foreach ($nodes ?: [] as $node) {
            $value = trim((string) $node->textContent);
            if ($value !== '') {
                $titles[] = $value;
            }
        }

        return $titles;
    }

    private function urlError(string $loc): ?string
    {
        $parts = parse_url($loc);

        if (! is_array($parts) || ! isset($parts['scheme'], $parts['host'])) {
            return 'Non-absolute URL emitted';
        }

        if (($parts['query'] ?? '') !== '') {
            return 'Query-string URL emitted';
        }

        if (($parts['fragment'] ?? '') !== '') {
            return 'Fragment URL emitted';
        }

        $path = '/' . ltrim((string) ($parts['path'] ?? '/'), '/');

        foreach ((array) config('sitemaps.validation.forbidden_paths', []) as $forbidden) {
            if ($forbidden !== '/' && str_contains($path, (string) $forbidden)) {
                return 'Non-public path emitted';
            }
        }

        return null;
    }
}