presseportale/tests/Feature/IndexStatusCheckTest.php

<?php

use App\Console\Commands\CheckIndexStatus;
use App\Enums\IndexState;
use App\Enums\Portal;
use App\Models\Category;
use App\Models\Company;
use App\Models\PressRelease;
use App\Models\PressReleaseIndexCheck;
use App\Models\PressReleaseStatusLog;
use App\Models\User;
use App\Services\SearchConsole\IndexStatusChecker;
use App\Services\SearchConsole\IndexStatusReport;
use App\Services\SearchConsole\SearchConsoleClient;
use Database\Seeders\RolesAndPermissionsSeeder;
use Illuminate\Support\Facades\Exceptions;
use Illuminate\Support\Facades\Http;
use Illuminate\Support\Facades\Log;
use Tests\TestCase;

/**
 * Indexierungsstatus aus der Search Console (Roadmap 1c.4): Befunde
 * unterscheiden „noch nicht gecrawlt" von „gecrawlt, nicht indexiert",
 * jede Antwort ist eine Momentaufnahme, die Auswahl folgt dem Alter, und
 * der Kunde sieht nichts, solange E8 die Anzeige zu hält.
 */
function fakeServiceAccount(): string
{
    $key = openssl_pkey_new(['private_key_bits' => 2048, 'private_key_type' => OPENSSL_KEYTYPE_RSA]);
    openssl_pkey_export($key, $pem);

    $file = tempnam(sys_get_temp_dir(), 'gsc');
    file_put_contents($file, json_encode([
        'client_email' => 'indexer@example.iam.gserviceaccount.com',
        'private_key' => $pem,
        'token_uri' => 'https://oauth2.googleapis.com/token',
    ]));

    config()->set('search_console.enabled', true);
    config()->set('search_console.credentials', $file);

    return $file;
}

function publishedOn(Portal $portal, int $daysAgo, array $attributes = []): PressRelease
{
    $company = Company::factory()->create(['portal' => $portal]);

    return PressRelease::factory()
        ->published()
        ->forPortal($portal)
        ->for($company)
        ->create(array_merge(['published_at' => now()->subDays($daysAgo)], $attributes));
}

test('the index state distinguishes not crawled from crawled but not indexed', function (array $indexStatus, IndexState $expected) {
    expect(IndexState::fromIndexStatus($indexStatus))->toBe($expected);
})->with([
    'indexed' => [['verdict' => 'PASS', 'coverageState' => 'Submitted and indexed', 'lastCrawlTime' => '2026-09-10T08:00:00Z'], IndexState::Indexed],
    'crawled, not indexed' => [['verdict' => 'NEUTRAL', 'coverageState' => 'Crawled - currently not indexed', 'lastCrawlTime' => '2026-09-10T08:00:00Z', 'pageFetchState' => 'SUCCESSFUL'], IndexState::CrawledNotIndexed],
    'discovered, not crawled' => [['verdict' => 'NEUTRAL', 'coverageState' => 'Discovered - currently not indexed'], IndexState::NotCrawled],
    'unknown to google' => [['verdict' => 'NEUTRAL', 'coverageState' => 'URL is unknown to Google'], IndexState::NotCrawled],
    'duplicate canonical' => [['verdict' => 'NEUTRAL', 'coverageState' => 'Duplicate, Google chose different canonical than user', 'lastCrawlTime' => '2026-09-10T08:00:00Z'], IndexState::Excluded],
    'blocked by robots' => [['verdict' => 'FAIL', 'coverageState' => 'Blocked by robots.txt', 'robotsTxtState' => 'DISALLOWED'], IndexState::Excluded],
    'empty' => [[], IndexState::Unknown],
]);

test('candidates are picked by age band, never-checked first, within the budget', function () {
    /** @var TestCase $this */
    config()->set('search_console.recheck_days', [14 => 3, 90 => 14, PHP_INT_MAX => 180]);

    $fresh = publishedOn(Portal::Presseecho, 2);
    $freshChecked = publishedOn(Portal::Presseecho, 5);
    PressReleaseIndexCheck::factory()->create(['press_release_id' => $freshChecked->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDay()]);
    $freshDue = publishedOn(Portal::Presseecho, 6);
    PressReleaseIndexCheck::factory()->create(['press_release_id' => $freshDue->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDays(4)]);
    $middle = publishedOn(Portal::Presseecho, 40);
    $old = publishedOn(Portal::Presseecho, 400);
    $otherPortal = publishedOn(Portal::Businessportal24, 1);
    $both = publishedOn(Portal::Both, 3);

    $ids = app(IndexStatusChecker::class)->candidates(Portal::Presseecho, 10)->pluck('id')->all();

    expect($ids)->toContain($fresh->id, $freshDue->id, $middle->id, $old->id, $both->id)
        ->and($ids)->not->toContain($freshChecked->id, $otherPortal->id)
        // Jüngste zuerst, innerhalb der Stufe nie geprüfte vor geprüften.
        ->and(array_slice($ids, 0, 3))->toBe([$fresh->id, $both->id, $freshDue->id]);

    expect(app(IndexStatusChecker::class)->candidates(Portal::Presseecho, 2)->pluck('id')->all())->toBe([$fresh->id, $both->id]);
});

test('the command inspects due releases via the API and stores each answer as a snapshot', function () {
    /** @var TestCase $this */
    $file = fakeServiceAccount();
    config()->set('search_console.properties.presseecho', 'sc-domain:presseecho.test');
    config()->set('search_console.properties.businessportal24', null);

    $release = publishedOn(Portal::Presseecho, 2, ['slug' => 'index-pruefung-pm']);

    Http::fake([
        'https://oauth2.googleapis.com/token' => Http::response(['access_token' => 'token-123', 'expires_in' => 3600]),
        SearchConsoleClient::INSPECT_URL => Http::response(['inspectionResult' => [
            'inspectionResultLink' => 'https://search.google.com/search-console/inspect?resource_id=sc-domain:presseecho.test',
            'indexStatusResult' => [
                'verdict' => 'NEUTRAL',
                'coverageState' => 'Crawled - currently not indexed',
                'robotsTxtState' => 'ALLOWED',
                'indexingState' => 'INDEXING_ALLOWED',
                'lastCrawlTime' => '2026-09-15T06:30:00Z',
                'pageFetchState' => 'SUCCESSFUL',
                'googleCanonical' => 'https://presseecho.test/de/press-release/index-pruefung-pm',
            ],
        ]]),
    ]);

    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])
        ->expectsOutputToContain('1 Prüfungen gespeichert, 0 fehlgeschlagen.')
        ->assertSuccessful();

    Http::assertSent(fn ($request) => $request->url() === SearchConsoleClient::INSPECT_URL
        && $request['inspectionUrl'] === 'https://presseecho.test/de/press-release/index-pruefung-pm'
        && $request['siteUrl'] === 'sc-domain:presseecho.test'
        && $request->hasHeader('Authorization', 'Bearer token-123'));

    $check = PressReleaseIndexCheck::query()->sole();

    expect($check->press_release_id)->toBe($release->id)
        ->and($check->portal)->toBe(Portal::Presseecho)
        ->and($check->state)->toBe(IndexState::CrawledNotIndexed)
        ->and($check->coverage_state)->toBe('Crawled - currently not indexed')
        ->and($check->last_crawl_time?->toDateString())->toBe('2026-09-15')
        ->and($check->raw['verdict'])->toBe('NEUTRAL');

    // Zweiter Lauf: die Meldung ist gerade geprüft, also nicht fällig — keine weitere Abfrage.
    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])->assertSuccessful();
    expect(PressReleaseIndexCheck::query()->count())->toBe(1);

    @unlink($file);
});

test('the command does nothing without configuration', function () {
    /** @var TestCase $this */
    config()->set('search_console.enabled', false);
    publishedOn(Portal::Presseecho, 2);

    Http::fake();

    $this->artisan(CheckIndexStatus::class)
        ->expectsOutputToContain('nicht konfiguriert')
        ->assertSuccessful();

    Http::assertNothingSent();
    expect(PressReleaseIndexCheck::query()->count())->toBe(0);
});

/**
 * Robustheit des Laufs (06.10.2026: 28 Zeitüberschreitungen und 4 × 500 von
 * Google, jede einzeln als Fehler gemeldet, Lauf über 1¾ Stunden).
 */
function inspectionOk(): array
{
    return ['inspectionResult' => ['indexStatusResult' => [
        'verdict' => 'PASS',
        'coverageState' => 'Submitted and indexed',
        'indexingState' => 'INDEXING_ALLOWED',
    ]]];
}

function robustIndexRun(): void
{
    fakeServiceAccount();
    config()->set('search_console.properties.presseecho', 'sc-domain:presseecho.test');
    config()->set('search_console.properties.businessportal24', null);
    config()->set('search_console.retry_sleep_ms', 0);
}

test('a google timeout is retried once and then counts as a normal check', function () {
    /** @var TestCase $this */
    robustIndexRun();
    publishedOn(Portal::Presseecho, 2);

    Http::fake([
        'https://oauth2.googleapis.com/token' => Http::response(['access_token' => 'token-123']),
        SearchConsoleClient::INSPECT_URL => Http::sequence()
            ->pushFailedConnection('cURL error 28: Operation timed out after 20002 milliseconds')
            ->push(inspectionOk()),
    ]);

    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])
        ->expectsOutputToContain('1 Prüfungen gespeichert, 0 fehlgeschlagen.')
        ->assertSuccessful();

    Http::assertSentCount(3);
    expect(PressReleaseIndexCheck::query()->count())->toBe(1);
});

test('temporary google failures are logged as warnings, not reported, and stay due', function () {
    /** @var TestCase $this */
    robustIndexRun();
    Exceptions::fake();
    Log::spy();
    publishedOn(Portal::Presseecho, 2);
    publishedOn(Portal::Presseecho, 3);

    Http::fake([
        'https://oauth2.googleapis.com/token' => Http::response(['access_token' => 'token-123']),
        SearchConsoleClient::INSPECT_URL => Http::response(['error' => ['message' => 'Internal error encountered.']], 500),
    ]);

    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])
        ->expectsOutputToContain('0 Prüfungen gespeichert, 2 fehlgeschlagen.')
        ->assertSuccessful();

    Exceptions::assertNothingReported();
    Log::shouldHaveReceived('warning')->with('Search Console vorübergehend nicht verfügbar.', Mockery::any())->twice();
    expect(PressReleaseIndexCheck::query()->count())->toBe(0)
        ->and(app(IndexStatusChecker::class)->candidates(Portal::Presseecho, 5))->toHaveCount(2);
});

test('a portal is aborted after five failures in a row and the run fails once', function () {
    /** @var TestCase $this */
    robustIndexRun();
    Exceptions::fake();
    foreach (range(1, 8) as $day) {
        publishedOn(Portal::Presseecho, $day);
    }

    Http::fake([
        'https://oauth2.googleapis.com/token' => Http::response(['access_token' => 'token-123']),
        SearchConsoleClient::INSPECT_URL => Http::failedConnection(),
    ]);

    $this->artisan(CheckIndexStatus::class, ['--budget' => 10])
        ->expectsOutputToContain('Abbruch nach 5 Fehlern in Folge')
        ->assertFailed();

    // Fünf Meldungen mit je einer Wiederholung plus Token — nicht alle acht.
    Http::assertSentCount(1 + 5 * 2);
    Exceptions::assertNothingReported();
});

test('a configuration error is reported and not retried', function () {
    /** @var TestCase $this */
    robustIndexRun();
    Exceptions::fake();
    publishedOn(Portal::Presseecho, 2);

    Http::fake([
        'https://oauth2.googleapis.com/token' => Http::response(['access_token' => 'token-123']),
        SearchConsoleClient::INSPECT_URL => Http::response(['error' => ['message' => 'User does not have sufficient permission for site.']], 403),
    ]);

    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])->assertSuccessful();

    Http::assertSentCount(2);
    Exceptions::assertReported(fn (RuntimeException $e) => str_contains($e->getMessage(), 'antwortet mit 403'));
});

test('the run stops at its time budget and leaves the rest for tomorrow', function () {
    /** @var TestCase $this */
    robustIndexRun();
    config()->set('search_console.max_runtime_minutes', 0);
    publishedOn(Portal::Presseecho, 2);

    Http::fake();

    $this->artisan(CheckIndexStatus::class, ['--budget' => 5])
        ->expectsOutputToContain('Zeitbudget erreicht')
        ->assertSuccessful();

    Http::assertNothingSent();
});

test('the admin report splits the distribution by origin and lists the latest checks', function () {
    /** @var TestCase $this */
    test()->seed(RolesAndPermissionsSeeder::class);

    $direct = publishedOn(Portal::Presseecho, 10, ['title' => 'Direkt eingereichte Meldung']);
    $api = publishedOn(Portal::Presseecho, 12, ['title' => 'Syndizierte Meldung']);
    PressReleaseStatusLog::query()->create(['press_release_id' => $api->id, 'from_status' => 'draft', 'to_status' => 'review', 'source' => 'api']);
    $unchecked = publishedOn(Portal::Presseecho, 20);

    // Ältere Momentaufnahme zählt nicht — nur die jüngste je Meldung.
    PressReleaseIndexCheck::factory()->withState(IndexState::NotCrawled)->create(['press_release_id' => $direct->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDays(5)]);
    PressReleaseIndexCheck::factory()->withState(IndexState::Indexed)->create(['press_release_id' => $direct->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDay(), 'url' => 'https://presseecho.test/de/press-release/'.$direct->slug, 'google_canonical' => 'https://presseecho.test/de/press-release/'.$direct->slug]);
    PressReleaseIndexCheck::factory()->withState(IndexState::CrawledNotIndexed)->create(['press_release_id' => $api->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDay()]);

    // Kopie mit Verweis (1c.8): Google folgt dem Canonical → „ausgeschlossen" ist hier erwartet.
    $copy = publishedOn(Portal::Presseecho, 11, ['title' => 'Kopie der Meldung', 'canonical_press_release_id' => $direct->id]);
    PressReleaseIndexCheck::factory()->withState(IndexState::Excluded)->create(['press_release_id' => $copy->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDay(), 'google_canonical' => 'https://businessportal24.test/de/press-release/'.$direct->slug]);

    $distribution = app(IndexStatusReport::class)->distribution();

    expect($distribution['presseecho']['direct']['source']['total'])->toBe(2)
        ->and($distribution['presseecho']['direct']['source']['indexed'])->toBe(1)
        ->and($distribution['presseecho']['direct']['source']['not_crawled'])->toBe(0)
        ->and($distribution['presseecho']['direct']['source']['unchecked'])->toBe(1)
        ->and($distribution['presseecho']['direct']['copy']['total'])->toBe(1)
        ->and($distribution['presseecho']['direct']['copy']['excluded'])->toBe(1)
        ->and($distribution['presseecho']['api']['source']['total'])->toBe(1)
        ->and($distribution['presseecho']['api']['source']['crawled_not_indexed'])->toBe(1)
        ->and($distribution['businessportal24']['direct']['source']['total'])->toBe(0);

    // Quellenbefund nur über Erstquellen: die Kopie zählt nicht mit.
    $sources = app(IndexStatusReport::class)->sourceSummary(app(IndexStatusChecker::class));

    expect($sources['presseecho']['total'])->toBe(2)
        ->and($sources['presseecho']['self'])->toBe(1)
        ->and($sources['presseecho']['unknown'])->toBe(1)
        ->and($sources['presseecho']['own_other'])->toBe(0);

    $admin = User::factory()->create(['is_active' => true]);
    $admin->assignRole('admin');

    $this->actingAs($admin)
        ->get(route('admin.reports.index-status'))
        ->assertSuccessful()
        ->assertSeeText('Indexierung')
        ->assertSeeText('Direkt eingereicht')
        ->assertSeeText('Syndikation (API)')
        ->assertSeeText('Direkt eingereichte Meldung')
        ->assertSeeText('Syndizierte Meldung')
        ->assertSeeText('gecrawlt, nicht indexiert')
        ->assertSeeText('Kopie mit Verweis')
        ->assertSeeText('Quellenbefund')
        ->assertSeeText('Abrufe durch Such- und KI-Dienste')
        ->assertSeeText('Abrufe der RSS-Feeds')
        ->assertSeeText('Kundenansicht');
});

test('the distribution counts states over the same population as the total, one check per release', function () {
    /** @var TestCase $this */
    $checked = publishedOn(Portal::Presseecho, 10);
    $depublished = publishedOn(Portal::Presseecho, 12);
    $at = now()->subDay()->startOfSecond();

    // Zwei Prüfungen in derselben Sekunde: nur die jüngere (höhere ID) zählt.
    PressReleaseIndexCheck::factory()->withState(IndexState::Indexed)->create(['press_release_id' => $checked->id, 'portal' => Portal::Presseecho, 'checked_at' => $at]);
    PressReleaseIndexCheck::factory()->withState(IndexState::NotCrawled)->create(['press_release_id' => $checked->id, 'portal' => Portal::Presseecho, 'checked_at' => $at]);

    // Geprüft, danach depubliziert: weder im Nenner noch als „indexed".
    PressReleaseIndexCheck::factory()->withState(IndexState::Indexed)->create(['press_release_id' => $depublished->id, 'portal' => Portal::Presseecho, 'checked_at' => $at]);
    $depublished->update(['status' => 'archived']);

    $bucket = app(IndexStatusReport::class)->distribution()['presseecho']['direct']['source'];
    $states = collect(IndexState::cases())->sum(fn (IndexState $state): int => $bucket[$state->value]);

    expect($bucket['total'])->toBe(1)
        ->and($bucket['indexed'])->toBe(0)
        ->and($bucket['not_crawled'])->toBe(1)
        ->and($bucket['unchecked'])->toBe(0)
        ->and($states + $bucket['unchecked'])->toBe($bucket['total']);
});

test('a release on both portals counts as copy on the portal that does not lead, like the sitemap', function () {
    /** @var TestCase $this */
    config()->set('canonical.enabled', true);
    config()->set('canonical.portal_by_category', ['handel' => 'businessportal24']);

    $category = Category::factory()->create(['portal' => Portal::Both]);
    $category->translations()->create(['locale' => 'de', 'name' => 'Handel', 'slug' => 'handel']);
    publishedOn(Portal::Both, 5, ['category_id' => $category->id]);

    $distribution = app(IndexStatusReport::class)->distribution();

    // Umsetzungsaudit 29.09.2026, N18-Rest: die Sitemap von presseecho führt sie nicht.
    expect($distribution['businessportal24']['direct']['source']['total'])->toBe(1)
        ->and($distribution['businessportal24']['direct']['copy']['total'])->toBe(0)
        ->and($distribution['presseecho']['direct']['source']['total'])->toBe(0)
        ->and($distribution['presseecho']['direct']['copy']['total'])->toBe(1);
});