presseportale/tests/Feature/CanonicalAssignmentTest.php

<?php

use App\Console\Commands\AssignCanonicalPressReleases;
use App\Console\Commands\HashPressReleaseContent;
use App\Console\Commands\ReportDuplicatePressReleases;
use App\Enums\Portal;
use App\Enums\PressReleaseStatus;
use App\Models\Category;
use App\Models\Company;
use App\Models\PressRelease;
use App\Models\PressReleaseIndexCheck;
use App\Models\User;
use App\Services\PressRelease\CanonicalAssigner;
use App\Services\PressRelease\PressReleaseService;
use App\Services\SearchConsole\IndexStatusChecker;
use App\Support\ContentFingerprint;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Mail;
use Livewire\Livewire;
use Tests\TestCase;

/**
 * Doppelveröffentlichung (Anhang 2, 1c.7–1c.9): identische Texte auf
 * beiden Portalen bekommen dieselbe Prüfsumme, die zweite Fassung verweist
 * per rel=canonical auf die erste (Kategorie entscheidet), Sitemaps nennen
 * nur die Erstquelle, und die Quellenauskunft bleibt hinter E8.
 */
function duplicatePair(string $categorySlug = 'it', array $a = [], array $b = []): array
{
    $category = Category::factory()->create(['portal' => Portal::Both]);
    $category->translations()->create(['locale' => 'de', 'name' => ucfirst($categorySlug), 'slug' => $categorySlug]);
    $company = Company::factory()->create(['portal' => Portal::Both, 'name' => 'Doppel GmbH']);

    $text = '<p>Die Doppel GmbH eröffnet einen  neuen Standort.</p>';

    $first = PressRelease::factory()->published()->forPortal(Portal::Presseecho)->for($category)->for($company)
        ->create(array_merge(['slug' => 'doppel-standort', 'title' => 'Neuer Standort', 'text' => $text, 'published_at' => now()->subHours(2)], $a));
    $second = PressRelease::factory()->published()->forPortal(Portal::Businessportal24)->for($category)->for($company)
        ->create(array_merge(['slug' => 'doppel-standort', 'title' => 'Neuer Standort', 'text' => "<p>Die Doppel GmbH eröffnet einen\nneuen Standort.</p>", 'published_at' => now()->subHour()], $b));

    return [$first, $second, $category, $company];
}

test('the fingerprint ignores markup, whitespace and case and is set on save', function () {
    expect(ContentFingerprint::hash('Titel', '<p>Ein  Text</p>'))->toBe(ContentFingerprint::hash('titel', "ein\ntext"))
        ->and(ContentFingerprint::hash('Titel', 'Ein Text'))->not->toBe(ContentFingerprint::hash('Titel', 'Ein anderer Text'));

    $release = PressRelease::factory()->create(['title' => 'Hallo', 'text' => 'Welt']);
    expect($release->content_hash)->toBe(ContentFingerprint::hash('Hallo', 'Welt'));

    $release->update(['text' => 'Neue Welt']);
    expect($release->fresh()->content_hash)->toBe(ContentFingerprint::hash('Hallo', 'Neue Welt'));
});

test('the backfill command fills missing fingerprints without model events', function () {
    /** @var TestCase $this */
    $release = PressRelease::factory()->create(['title' => 'Alt', 'text' => 'Bestand']);
    DB::table('press_releases')->where('id', $release->id)->update(['content_hash' => null]);

    $this->artisan(HashPressReleaseContent::class)->assertSuccessful();

    expect($release->fresh()->content_hash)->toBe(ContentFingerprint::hash('Alt', 'Bestand'));
});

test('the category mapping decides which version is the source, otherwise the first published', function () {
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);
    config()->set('canonical.fallback', 'first');

    [$pe, $b24] = duplicatePair('it');
    $assigner = app(CanonicalAssigner::class);

    expect($pe->content_hash)->toBe($b24->content_hash)
        ->and($assigner->resolveCanonical($pe, $b24)->is($b24))->toBeTrue();

    [$peOther, $b24Other] = duplicatePair('reisen', ['slug' => 'reise-a'], ['slug' => 'reise-b']);
    expect($assigner->resolveCanonical($peOther, $b24Other)->is($peOther))->toBeTrue();
});

test('assigning the backlog links the duplicate to the source and the sitemaps list only the source', function () {
    /** @var TestCase $this */
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);

    [$pe, $b24] = duplicatePair('it');

    // Vor der Zuordnung nennt jedes Portal seine eigene Fassung — und cached die Sitemap eine Stunde.
    $this->get('https://presseecho.test/sitemap.xml')->assertSuccessful()
        ->assertSee('https://presseecho.test/de/press-release/doppel-standort', false);

    $this->artisan(AssignCanonicalPressReleases::class, ['--dry-run' => true])
        ->expectsOutputToContain('[Dry-Run] 1 Paare, 1 würden geändert.')
        ->assertSuccessful();
    expect($b24->fresh()->canonical_press_release_id)->toBeNull();

    $this->artisan(AssignCanonicalPressReleases::class)
        ->expectsOutputToContain('1 Paare, 1 geändert.')
        ->expectsOutputToContain('Seiten- und Sitemap-Caches geleert.')
        ->assertSuccessful();

    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id)
        ->and($b24->fresh()->canonical_press_release_id)->toBeNull();

    // Zweiter Lauf ändert nichts mehr.
    $this->artisan(AssignCanonicalPressReleases::class)
        ->expectsOutputToContain('1 Paare, 0 geändert.')
        ->doesntExpectOutputToContain('Caches geleert')
        ->assertSuccessful();

    // Detailseite: die Kopie zeigt per rel=canonical auf die Erstquelle, die Erstquelle auf sich selbst.
    $this->get('https://presseecho.test/de/press-release/doppel-standort')->assertSuccessful()
        ->assertSee('<link rel="canonical" href="https://businessportal24.test/de/press-release/doppel-standort">', false)
        ->assertSee('"mainEntityOfPage":{"@type":"WebPage","@id":"https://businessportal24.test/de/press-release/doppel-standort"}', false);
    $this->get('https://businessportal24.test/de/press-release/doppel-standort')->assertSuccessful()
        ->assertSee('<link rel="canonical" href="https://businessportal24.test/de/press-release/doppel-standort">', false);

    // Sitemaps: nur die Erstquelle — der Lauf hat den 1-h-Cache von oben geleert.
    $this->get('https://presseecho.test/sitemap.xml')->assertSuccessful()
        ->assertDontSee('https://presseecho.test/de/press-release/doppel-standort', false);
    $this->get('https://businessportal24.test/sitemap.xml')->assertSuccessful()
        ->assertSee('https://businessportal24.test/de/press-release/doppel-standort', false);
    $this->get('https://presseecho.test/sitemap-news.xml')->assertSuccessful()
        ->assertDontSee('doppel-standort', false);
    $this->get('https://businessportal24.test/sitemap-news.xml')->assertSuccessful()
        ->assertSee('doppel-standort', false);

    $this->artisan(ReportDuplicatePressReleases::class)
        ->expectsOutputToContain('Paare gleichen Inhalts auf beiden Portalen: 1 (2 URLs)')
        ->expectsOutputToContain('davon mit kanonischer Zuordnung: 1')
        ->assertSuccessful();
});

test('without the flag every version stays its own canonical and nothing is assigned', function () {
    /** @var TestCase $this */
    config()->set('canonical.enabled', false);
    [$pe, $b24] = duplicatePair('it');

    $this->artisan(AssignCanonicalPressReleases::class)->expectsOutputToContain('Zuordnung ist aus')->assertSuccessful();
    expect($pe->fresh()->canonical_press_release_id)->toBeNull();

    $this->get('https://presseecho.test/de/press-release/doppel-standort')->assertSuccessful()
        ->assertSee('<link rel="canonical" href="https://presseecho.test/de/press-release/doppel-standort">', false);
    $this->get('https://presseecho.test/sitemap.xml')->assertSuccessful()
        ->assertSee('https://presseecho.test/de/press-release/doppel-standort', false);
});

test('a release with portal both is canonical on the category portal only', function () {
    /** @var TestCase $this */
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['handel' => 'presseecho']);

    $category = Category::factory()->create(['portal' => Portal::Both]);
    $category->translations()->create(['locale' => 'de', 'name' => 'Handel', 'slug' => 'handel']);
    $release = PressRelease::factory()->published()->forPortal(Portal::Both)->for($category)
        ->for(Company::factory()->create(['portal' => Portal::Both]))
        ->create(['slug' => 'beide-portale', 'published_at' => now()->subHour()]);

    $this->get('https://businessportal24.test/de/press-release/beide-portale')->assertSuccessful()
        ->assertSee('<link rel="canonical" href="https://presseecho.test/de/press-release/beide-portale">', false);
    $this->get('https://businessportal24.test/sitemap.xml')->assertSuccessful()->assertDontSee('beide-portale', false);
    $this->get('https://presseecho.test/sitemap.xml')->assertSuccessful()->assertSee('beide-portale', false);

    expect(app(CanonicalAssigner::class)->isCanonicalOn($release, Portal::Presseecho))->toBeTrue()
        ->and(app(CanonicalAssigner::class)->isCanonicalOn($release, Portal::Businessportal24))->toBeFalse();
});

test('publishing the second version assigns the pair immediately', function () {
    /** @var TestCase $this */
    Mail::fake();
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'presseecho']);

    [$pe, $b24] = duplicatePair('it');
    DB::table('press_releases')->where('id', $b24->id)->update(['status' => PressReleaseStatus::Review->value, 'published_at' => null]);
    $b24 = $b24->fresh();
    $b24->user->update(['is_active' => true]);

    app(PressReleaseService::class)->publish($b24, 'admin');

    expect($b24->fresh()->canonical_press_release_id)->toBe($pe->id)
        ->and($pe->fresh()->canonical_press_release_id)->toBeNull();
});

test('the source verdict tells own, other-own and foreign canonicals apart and shows only behind the flag', function () {
    /** @var TestCase $this */
    $customer = User::factory()->create(['is_active' => true]);
    $company = Company::factory()->businessportal24()->create();
    $customer->companies()->attach($company->id, ['role' => 'member']);
    $this->actingAs($customer);

    $release = PressRelease::factory()->published()->forPortal(Portal::Businessportal24)->create(['company_id' => $company->id, 'slug' => 'quelle-pm', 'published_at' => now()->subDay()]);
    $url = 'https://businessportal24.test/de/press-release/quelle-pm';
    $checker = app(IndexStatusChecker::class);

    $self = PressReleaseIndexCheck::factory()->create(['press_release_id' => $release->id, 'portal' => Portal::Businessportal24, 'url' => $url, 'google_canonical' => $url, 'user_canonical' => $url]);
    $ownOther = PressReleaseIndexCheck::factory()->make(['url' => $url, 'google_canonical' => 'https://presseecho.test/de/press-release/quelle-pm']);
    $foreign = PressReleaseIndexCheck::factory()->make(['url' => $url, 'google_canonical' => 'https://www.fremdes-portal.example/pm/quelle']);
    $none = PressReleaseIndexCheck::factory()->make(['url' => $url, 'google_canonical' => null]);

    expect($checker->sourceVerdict($self)['kind'])->toBe('self')
        ->and($checker->sourceVerdict($ownOther))->toMatchArray(['kind' => 'own_other', 'host' => 'presseecho.test'])
        ->and($checker->sourceVerdict($foreign))->toMatchArray(['kind' => 'foreign', 'host' => 'www.fremdes-portal.example'])
        ->and($checker->sourceVerdict($none)['kind'])->toBe('unknown');

    config()->set('search_console.customer_visible', false);
    Livewire::test('customer.press-releases.show', ['id' => $release->id])->assertDontSeeText('maßgebliche Quelle');

    config()->set('search_console.customer_visible', true);
    Livewire::test('customer.press-releases.show', ['id' => $release->id])
        ->assertSeeText('Google führt diese Adresse als maßgebliche Quelle.')
        ->assertSeeText('warum, ist nicht bekannt');
});

test('portal and auth layouts carry noindex and the help page explains the single source', function () {
    /** @var TestCase $this */
    $this->get('https://pressekonto.test/login')->assertSuccessful()->assertSee('<meta name="robots" content="noindex, nofollow" />', false);

    config()->set('canonical.enabled', false);
    $this->get('https://presseecho.test/de/gute-pressemitteilung')->assertSuccessful()
        ->assertSee('Warum eine Meldung mehrfach im Netz steht und nur einmal in den Suchergebnissen')
        ->assertSee('nicht zwei Fundstellen in der Suche')
        ->assertDontSee('verweist bei uns die zweite Fassung auf die erste');

    config()->set('canonical.enabled', true);
    $this->get('https://presseecho.test/de/gute-pressemitteilung')->assertSuccessful()
        ->assertSee('verweist bei uns die zweite Fassung auf die erste');
});

test('copies are re-evaluated when their source is depublished, republished, gone or deleted', function () {
    /** @var TestCase $this */
    Mail::fake();
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);

    [$pe, $b24] = duplicatePair('it');
    $assigner = app(CanonicalAssigner::class);
    $service = app(PressReleaseService::class);
    $assigner->assignFor($pe);
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    // Erstquelle depubliziert: die Kopie wird selbst Erstquelle — Self-Canonical UND Sitemap.
    $service->depublish($b24->fresh());

    expect($pe->fresh()->canonical_press_release_id)->toBeNull()
        ->and($assigner->isCanonicalOn($pe->fresh(), Portal::Presseecho))->toBeTrue();
    $this->get('https://presseecho.test/de/press-release/doppel-standort')->assertSuccessful()
        ->assertSee('<link rel="canonical" href="https://presseecho.test/de/press-release/doppel-standort">', false);
    $this->get('https://presseecho.test/sitemap.xml')->assertSuccessful()
        ->assertSee('https://presseecho.test/de/press-release/doppel-standort', false);

    // Erstquelle per Admin-Statuswechsel wieder veröffentlicht: Paar erneut zugeordnet.
    $service->changeStatusFromAdmin($b24->fresh(), PressReleaseStatus::Published);
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    // Admin nimmt die Erstquelle zurück: Kopie wieder frei.
    $service->changeStatusFromAdmin($b24->fresh(), PressReleaseStatus::Draft, 'Zurückgezogen');
    expect($pe->fresh()->canonical_press_release_id)->toBeNull();

    $service->changeStatusFromAdmin($b24->fresh(), PressReleaseStatus::Published);
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    // 410 Gone.
    $service->markGone($b24->fresh());
    expect($pe->fresh()->canonical_press_release_id)->toBeNull();

    $service->restoreFromGone($b24->fresh());
    $service->changeStatusFromAdmin($b24->fresh(), PressReleaseStatus::Published);
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    // Entfernung durch den Admin.
    $service->deleteFromAdmin($b24->fresh());
    expect($pe->fresh()->canonical_press_release_id)->toBeNull();
});

test('a soft-deleted source is no canonical target even while the copy still points at it', function () {
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);

    [$pe, $b24] = duplicatePair('it');
    app(CanonicalAssigner::class)->assignFor($pe);

    // Direkt gelöscht, ohne Service-Hook — der Verweis bleibt als Leiche stehen.
    $b24->delete();
    $pe = $pe->fresh();

    expect($pe->canonical_press_release_id)->toBe($b24->id)
        ->and(app(CanonicalAssigner::class)->canonicalUrlFor($pe, Portal::Presseecho))
        ->toBe('https://presseecho.test/de/press-release/doppel-standort');
});

test('assigning the backlog resets references that no longer form a pair', function () {
    /** @var TestCase $this */
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);

    [$pe, $b24] = duplicatePair('it');
    [$pe2, $b242] = duplicatePair('it', ['slug' => 'zweites-paar', 'title' => 'Zweites Paar'], ['slug' => 'zweites-paar', 'title' => 'Zweites Paar']);
    $assigner = app(CanonicalAssigner::class);
    $assigner->assignAll();
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id)
        ->and($pe2->fresh()->canonical_press_release_id)->toBe($b242->id);

    // Ohne Hook umgehängt (Query-Builder): die Kopie gehört jetzt einer anderen Firma.
    DB::table('press_releases')->where('id', $pe->id)->update(['company_id' => Company::factory()->create()->id]);
    // Zweites Paar: Text der Erstquelle geändert, ohne Neubewertung.
    DB::table('press_releases')->where('id', $b242->id)->update(['content_hash' => str_repeat('a', 64)]);

    $dry = $assigner->assignAll(dryRun: true);
    expect($dry['reset'])->toBe(2)
        ->and($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    $result = $assigner->assignAll();

    expect($result['reset'])->toBe(2)
        ->and($pe->fresh()->canonical_press_release_id)->toBeNull()
        ->and($pe2->fresh()->canonical_press_release_id)->toBeNull();

    $this->artisan(AssignCanonicalPressReleases::class)->assertSuccessful();
});

test('copies are re-evaluated when their source is archived by an admin (Audit N19-Rest)', function () {
    /** @var TestCase $this */
    Mail::fake();
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);

    [$pe, $b24] = duplicatePair('it');
    $assigner = app(CanonicalAssigner::class);
    $assigner->assignFor($pe);
    expect($pe->fresh()->canonical_press_release_id)->toBe($b24->id);

    app(PressReleaseService::class)->archive($b24->fresh());

    expect($pe->fresh()->canonical_press_release_id)->toBeNull()
        ->and($assigner->isCanonicalOn($pe->fresh(), Portal::Presseecho))->toBeTrue();
});

test('the mapping by category id survives a slug change and wins over the slug list', function () {
    config()->set('canonical.enabled', true);

    [$pe, $b24, $category] = duplicatePair('it');
    config()->set('canonical.portal_by_category', ['it' => 'businessportal24']);
    config()->set('canonical.portal_by_category_id', [$category->id => 'presseecho']);

    $assigner = app(CanonicalAssigner::class);
    expect($assigner->preferredPortal($pe))->toBe(Portal::Presseecho);

    // Slug im Admin umbenannt: die ID-Zuordnung greift weiter.
    $category->translations()->where('locale', 'de')->update(['slug' => 'informationstechnik']);
    $fresh = app()->make(CanonicalAssigner::class);
    expect($fresh->preferredPortal($pe))->toBe(Portal::Presseecho);

    // Unterkategorien folgen der Hauptkategorie.
    $child = Category::factory()->create(['portal' => Portal::Both, 'parent_id' => $category->id]);
    expect($fresh->preferredPortalForCategory($child->id))->toBe(Portal::Presseecho)
        ->and($fresh->categoryIdsPreferring(Portal::Presseecho))->toContain($category->id, $child->id);

    // Ohne ID-Eintrag: rückwärtskompatibel über den DE-Slug.
    config()->set('canonical.portal_by_category_id', []);
    config()->set('canonical.portal_by_category', ['informationstechnik' => 'businessportal24']);
    expect(app()->make(CanonicalAssigner::class)->preferredPortal($pe))->toBe(Portal::Businessportal24);
});