admin: Indexierungsstatus aus der Search Console, Kundenansicht vorbereitet und zu (Anhang 1: 1c.4, E8)

press-releases:check-index-status fragt täglich je Portal ein Kontingent veröffentlichter Meldungen über die URL Inspection API ab, nach Alter gestaffelt, nie geprüfte zuerst. Jede Antwort wird als Momentaufnahme in press_release_index_checks gespeichert; der reduzierte Befund (IndexState) trennt „noch nicht gecrawlt" von „gecrawlt, nicht indexiert". Client authentifiziert sich als Service-Konto über ein selbst signiertes JWT — kein Google-SDK. Admin ▸ Reports ▸ Indexierung zeigt die Verteilung je Portal getrennt nach Direkteinreichung und Syndikation (Status-Log source=api; Altbestand ohne Log zählt als direkt). Das Kundenpanel heißt jetzt „Auslieferung", zeigt Sitemap-Datum und die Abrufe durch Such- und KI-Dienste mit dem Satz „Ein Abruf ist kein Zitat"; der Google-Index-Block hängt an GSC_CUSTOMER_VISIBLE (false). Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
2026-09-17 14:21:51 +00:00 · 2026-09-17 14:21:51 +00:00 · 6b8728242f
commit 6b8728242f
parent 951b6c47fa
17 changed files with 1324 additions and 30 deletions
--- a/tests/Feature/IndexStatusCheckTest.php
+++ b/tests/Feature/IndexStatusCheckTest.php
@ -0,0 +1,190 @@
+<?php
+
+use App\Console\Commands\CheckIndexStatus;
+use App\Enums\IndexState;
+use App\Enums\Portal;
+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\Http;
+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);
+});
+
+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()]);
+    PressReleaseIndexCheck::factory()->withState(IndexState::CrawledNotIndexed)->create(['press_release_id' => $api->id, 'portal' => Portal::Presseecho, 'checked_at' => now()->subDay()]);
+
+    $distribution = app(IndexStatusReport::class)->distribution();
+
+    expect($distribution['presseecho']['direct']['total'])->toBe(2)
+        ->and($distribution['presseecho']['direct']['indexed'])->toBe(1)
+        ->and($distribution['presseecho']['direct']['not_crawled'])->toBe(0)
+        ->and($distribution['presseecho']['direct']['unchecked'])->toBe(1)
+        ->and($distribution['presseecho']['api']['total'])->toBe(1)
+        ->and($distribution['presseecho']['api']['crawled_not_indexed'])->toBe(1)
+        ->and($distribution['businessportal24']['direct']['total'])->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('Kundenansicht');
+});
--- a/tests/Feature/VisibilityStatsTest.php
+++ b/tests/Feature/VisibilityStatsTest.php
@ -1,10 +1,13 @@
 <?php

+use App\Enums\IndexState;
 use App\Enums\Portal;
 use App\Enums\PressReleaseEventType;
 use App\Models\Company;
 use App\Models\PressRelease;
+use App\Models\PressReleaseCrawlerHit;
 use App\Models\PressReleaseEvent;
+use App\Models\PressReleaseIndexCheck;
 use App\Models\User;
 use App\Services\PressRelease\VisibilityStats;
 use Livewire\Volt\Volt as LivewireVolt;
@ -102,7 +105,7 @@ test('the daily series is gapless, marks days before measurement as unmeasured a
    expect(app(VisibilityStats::class)->dailyViews($release, 400))->toHaveCount(5);
 });

-test('the release summary reports sitemap presence and a search link only for public releases', function () {
+test('the release summary reports sitemap presence only for public releases and lists crawler fetches', function () {
    /** @var TestCase $this */
    [, $company] = claimant();
    $public = PressRelease::factory()->published()->forPortal(Portal::Businessportal24)->create(['company_id' => $company->id, 'published_at' => now()->subDay(), 'hits' => 9]);
@ -111,23 +114,27 @@ test('the release summary reports sitemap presence and a search link only for pu
    PressReleaseEvent::query()->create(['press_release_id' => $public->id, 'type' => PressReleaseEventType::Share, 'channel' => 'linkedin']);
    viewsOn($public, 2, daysAgo: 1);

+    PressReleaseCrawlerHit::factory()->service('google', 'Mozilla/5.0 (compatible; Googlebot/2.1)')->count(2)->create(['press_release_id' => $public->id]);
+    PressReleaseCrawlerHit::factory()->count(1)->create(['press_release_id' => $public->id]);
+
    $stats = app(VisibilityStats::class);
    $summary = $stats->releaseSummary($public);

    expect($summary['hits'])->toBe(9)
        ->and($summary['shares_total'])->toBe(1)
        ->and($summary['in_sitemap_since']->isSameDay($public->published_at))->toBeTrue()
-        ->and($summary['search_url'])->toContain('site%3A')
-        ->and($summary['search_url'])->toContain($public->slug)
+        ->and($summary)->not->toHaveKey('search_url')
+        ->and(collect($summary['crawlers'])->pluck('fetches', 'service')->sortKeys()->all())->toBe(['google' => 2, 'openai' => 1])
+        ->and(collect($summary['crawlers'])->firstWhere('service', 'google')['label'])->toBe('Google Suche')
        ->and($summary['first_view_at'])->not->toBeNull();

    $draftSummary = $stats->releaseSummary($draft);

    expect($draftSummary['in_sitemap_since'])->toBeNull()
-        ->and($draftSummary['search_url'])->toBeNull();
+        ->and($draftSummary['crawlers'])->toBe([]);
 });

-test('the release page shows the visibility panel with daily bars and the search link', function () {
+test('the release page shows the delivery panel with daily bars, sitemap date and crawler fetches, but no search link', function () {
    /** @var TestCase $this */
    [$customer, $company] = claimant();
    $partner = User::factory()->create();
@ -135,24 +142,49 @@ test('the release page shows the visibility panel with daily bars and the search
        'company_id' => $company->id, 'user_id' => $partner->id, 'hits' => 777, 'published_at' => now()->subDays(5),
    ]);
    viewsOn($release, 6, daysAgo: 2);
+    PressReleaseCrawlerHit::factory()->service('anthropic', 'ClaudeBot/1.0')->create(['press_release_id' => $release->id]);

    LivewireVolt::test('customer.press-releases.show', ['id' => $release->id])
-        ->assertSeeText('Sichtbarkeit')
+        ->assertSeeText('Auslieferung')
        ->assertSeeText('777')
        ->assertSeeText('Aufrufe 30 Tage')
-        ->assertSeeText('Bei Google prüfen')
-        ->assertSeeText('In der Sitemap für Suchmaschinen seit '.now()->subDays(5)->format('d.m.Y'))
+        ->assertDontSeeText('Bei Google prüfen')
+        ->assertDontSee('google.com/search', false)
+        ->assertSeeText('Gelistet seit '.now()->subDays(5)->format('d.m.Y'))
+        ->assertSeeText('Anthropic / Claude')
+        ->assertSeeText('1 Abruf')
+        ->assertSeeText('Ein Abruf belegt, dass der Crawler des Dienstes die Seite geholt hat')
+        // Indexierungsstatus bleibt zu (E8), solange die Kundenansicht nicht freigegeben ist.
+        ->assertDontSeeText('Google-Index')
        ->assertSee('<svg', false)
        ->set('viewRange', 90)
        ->assertSeeText('Aufrufe 90 Tage');
 });

+test('the index status appears in the delivery panel only once the customer view is released', function () {
+    /** @var TestCase $this */
+    [, $company] = claimant();
+    $release = PressRelease::factory()->published()->forPortal(Portal::Businessportal24)->create(['company_id' => $company->id, 'published_at' => now()->subDays(5)]);
+    PressReleaseIndexCheck::factory()->withState(IndexState::CrawledNotIndexed)->create(['press_release_id' => $release->id, 'portal' => Portal::Businessportal24, 'checked_at' => now()->subDay()]);
+    PressReleaseIndexCheck::factory()->withState(IndexState::Indexed)->create(['press_release_id' => $release->id, 'portal' => Portal::Businessportal24, 'checked_at' => now()->subDays(9)]);
+
+    config()->set('search_console.customer_visible', false);
+    LivewireVolt::test('customer.press-releases.show', ['id' => $release->id])
+        ->assertDontSeeText('Google-Index');
+
+    config()->set('search_console.customer_visible', true);
+    LivewireVolt::test('customer.press-releases.show', ['id' => $release->id])
+        ->assertSeeText('Google-Index')
+        ->assertSeeText('Businessportal24: gecrawlt, nicht indexiert')
+        ->assertDontSeeText('Businessportal24: indexiert');
+});
+
 test('a draft shows no visibility panel', function () {
    /** @var TestCase $this */
    [$customer, $company] = claimant();
    $draft = PressRelease::factory()->forPortal(Portal::Businessportal24)->create(['company_id' => $company->id, 'user_id' => $customer->id, 'status' => 'draft', 'hits' => 0, 'published_at' => null]);

    LivewireVolt::test('customer.press-releases.show', ['id' => $draft->id])
-        ->assertDontSeeText('Bei Google prüfen')
+        ->assertDontSeeText('Auslieferung')
        ->assertSeeText('noch nicht veröffentlicht');
 });