KantSites/tests/StatisticsTest.php

<?php

namespace App\Tests;

use App\Content\Statistics;
use App\Infrastructure\ConnectionFactory;
use Doctrine\DBAL\Connection;
use PHPUnit\Framework\Attributes\DataProvider;
use PHPUnit\Framework\TestCase;

/**
 * The measurement, and above all the property that makes it possible without consent: there is no
 * column from which a person could follow, and nothing is written per visit.
 */
final class StatisticsTest extends TestCase
{
    private Connection $db;
    private Statistics $stats;
    private string $path;
    private \DateTimeImmutable $now;

    protected function setUp(): void
    {
        $this->path = tempnam(sys_get_temp_dir(), 'kant-stats-');
        $this->db = ConnectionFactory::create('sqlite:///'.str_replace('\\', '/', $this->path));
        TestSchema::build($this->db);
        $this->stats = new Statistics($this->db);
        $this->now = new \DateTimeImmutable('2026-09-12 14:30:00', new \DateTimeZone('UTC'));
    }

    protected function tearDown(): void
    {
        $this->db->close();
        gc_collect_cycles();
        @unlink($this->path);
    }

    public function testTheTableHoldsNothingThatPointsAtAPerson(): void
    {
        $columns = array_map(
            static fn (object $column): string => $column->getName(),
            $this->db->createSchemaManager()->listTableColumns('stat_hit'),
        );
        // The whole design rests on this: counters, not records.
        self::assertSame(['id', 'website_id', 'day', 'hour', 'path', 'status', 'referrer_host', 'count'], array_values($columns));
        foreach (['ip', 'address', 'user', 'agent', 'session', 'cookie', 'visitor', 'hash', 'fingerprint'] as $forbidden) {
            foreach ($columns as $column) {
                self::assertStringNotContainsString($forbidden, $column);
            }
        }
    }

    public function testManyRequestsBecomeOneCounter(): void
    {
        for ($index = 0; $index < 5; ++$index) {
            $this->stats->record('/leistungen', 200, '', 'kant-it.de', $this->now);
        }
        self::assertSame(1, (int) $this->db->fetchOne('SELECT COUNT(*) FROM stat_hit'));
        $row = $this->db->fetchAssociative('SELECT * FROM stat_hit');
        self::assertSame(['2026-09-12', 14, '/leistungen', 200, '', 5], [
            $row['day'], (int) $row['hour'], $row['path'], (int) $row['status'], $row['referrer_host'], (int) $row['count'],
        ]);

        // A different hour, page, status or referrer is a different counter - and nothing else is.
        $this->stats->record('/leistungen', 200, '', 'kant-it.de', $this->now->modify('+1 hour'));
        $this->stats->record('/kontakt', 200, '', 'kant-it.de', $this->now);
        $this->stats->record('/leistungen', 404, '', 'kant-it.de', $this->now);
        $this->stats->record('/leistungen', 200, 'https://www.google.de/search?q=kant', 'kant-it.de', $this->now);
        self::assertSame(5, (int) $this->db->fetchOne('SELECT COUNT(*) FROM stat_hit'));
        self::assertSame('www.google.de', $this->db->fetchOne('SELECT referrer_host FROM stat_hit WHERE referrer_host <> ?', ['']));
    }

    /** @return iterable<string, array{string, string}> */
    public static function referrers(): iterable
    {
        yield 'external' => ['https://www.google.de/search?q=kant+it', 'www.google.de'];
        yield 'own site is not a referrer' => ['https://kant-it.de/leistungen', ''];
        yield 'own site in other case' => ['https://KANT-IT.de/leistungen', ''];
        yield 'direct visit' => ['', ''];
        // Only the host is kept, never the address someone came from: that one can carry a search
        // term, a name or a token in its query string.
        yield 'no path and no query survive' => ['https://forum.example/thread/123?user=anna', 'forum.example'];
        yield 'nonsense' => ['nicht-einmal-eine-adresse', ''];
        yield 'host with a colon' => ['https://host:mit@doppelpunkt.example/x', 'doppelpunkt.example'];
    }

    #[DataProvider('referrers')]
    public function testOnlyTheBareHostOfAnExternalReferrerIsKept(string $referrer, string $expected): void
    {
        $this->stats->record('/', 200, $referrer, 'kant-it.de', $this->now);
        self::assertSame($expected, $this->db->fetchOne('SELECT referrer_host FROM stat_hit'));
    }

    /** @return iterable<string, array{string, int, string}> */
    public static function paths(): iterable
    {
        yield 'a page' => ['/leistungen/webdesign', 200, '/leistungen/webdesign'];
        yield 'the root' => ['/', 200, '/'];
        yield 'query string is dropped' => ['/leistungen?utm_source=newsletter', 200, '/leistungen'];
        yield 'a plausible 404 keeps its address' => ['/alte-seite', 404, '/alte-seite'];
        // Crawlers probe thousands of invented addresses; each one would be a row of its own.
        yield 'a probed 404 is bucketed' => ['/wp-admin/setup-config.php', 404, Statistics::OTHER];
        yield 'a very long path is bucketed' => ['/'.str_repeat('a', 260), 200, Statistics::OTHER];
    }

    #[DataProvider('paths')]
    public function testWhichAddressIsStored(string $path, int $status, string $expected): void
    {
        $this->stats->record($path, $status, '', 'kant-it.de', $this->now);
        self::assertSame($expected, $this->db->fetchOne('SELECT path FROM stat_hit'));
    }

    /** @return iterable<string, array{array{string, string, int, string, string}, bool}> */
    public static function requests(): iterable
    {
        yield 'a page' => [['GET', '/leistungen', 200, 'text/html; charset=UTF-8', 'Mozilla/5.0 (Windows NT 10.0)'], true];
        yield 'a missing page' => [['GET', '/weg', 404, 'text/html', 'Mozilla/5.0'], true];
        yield 'the administration' => [['GET', '/admin/statistik', 200, 'text/html', 'Mozilla/5.0'], false];
        yield 'an image' => [['GET', '/media/abc/480', 200, 'image/webp', 'Mozilla/5.0'], false];
        yield 'a stylesheet' => [['GET', '/assets/site.css', 200, 'text/css', 'Mozilla/5.0'], false];
        yield 'the feed' => [['GET', '/feed.xml', 200, 'application/xml', 'Mozilla/5.0'], false];
        yield 'the consent page' => [['GET', '/einwilligung', 200, 'text/html', 'Mozilla/5.0'], false];
        yield 'a redirect' => [['GET', '/alt', 301, 'text/html', 'Mozilla/5.0'], false];
        yield 'a form submission' => [['POST', '/kontakt', 200, 'text/html', 'Mozilla/5.0'], false];
        yield 'Googlebot' => [['GET', '/', 200, 'text/html', 'Mozilla/5.0 (compatible; Googlebot/2.1)'], false];
        yield 'a monitor' => [['GET', '/', 200, 'text/html', 'Uptime-Monitor/1.0'], false];
        yield 'headless Chrome' => [['GET', '/', 200, 'text/html', 'HeadlessChrome/120'], false];
        yield 'curl' => [['GET', '/', 200, 'text/html', 'curl/8.4.0'], false];
        yield 'no user agent at all' => [['GET', '/', 200, 'text/html', ''], true];
    }

    #[DataProvider('requests')]
    public function testWhatIsCountedAtAll(array $request, bool $expected): void
    {
        [$method, $path, $status, $type, $agent] = $request;
        self::assertSame($expected, $this->stats->countable($method, $path, $status, $type, $agent));
    }

    public function testTheOverviewFillsEveryDayAndSeparatesFoundFromMissing(): void
    {
        $this->stats->record('/', 200, '', 'kant-it.de', $this->now->modify('-3 days'));
        foreach (range(1, 4) as $ignored) {
            $this->stats->record('/leistungen', 200, 'https://suche.example/x', 'kant-it.de', $this->now);
        }
        $this->stats->record('/gibt-es-nicht', 404, '', 'kant-it.de', $this->now);

        $overview = $this->stats->overview(7, $this->now);
        self::assertCount(7, $overview['days'], 'Every day of the range, including the empty ones.');
        self::assertSame(5, $overview['total'], 'A 404 is not a page view.');
        // Index 6 is today, index 3 is three days ago.
        self::assertSame([0, 0, 0, 1, 0, 0, 4], array_column($overview['days'], 'count'));
        self::assertSame(4, $overview['busiest']);
        self::assertSame([['path' => '/leistungen', 'count' => 4], ['path' => '/', 'count' => 1]], $overview['pages']);
        self::assertSame([['referrer_host' => 'suche.example', 'count' => 4]], $overview['referrers']);
        self::assertSame([['path' => '/gibt-es-nicht', 'count' => 1]], $overview['missing']);
        // The hours are summed over the whole range, so both days contribute to 14:00.
        self::assertSame(5, $overview['hours'][14]['count']);
        self::assertCount(24, $overview['hours']);
    }

    public function testOlderDaysAreCompactedAndTheOldestAreRemoved(): void
    {
        // Two hours of the same day, long enough ago to be compacted.
        $old = $this->now->modify('-70 days');
        $this->stats->record('/leistungen', 200, '', 'kant-it.de', $old);
        $this->stats->record('/leistungen', 200, '', 'kant-it.de', $old->modify('+2 hours'));
        // Yesterday stays untouched, and something beyond the retention goes.
        $this->stats->record('/leistungen', 200, '', 'kant-it.de', $this->now->modify('-1 day'));
        $this->stats->record('/leistungen', 200, '', 'kant-it.de', $this->now->modify('-500 days'));
        self::assertSame(4, (int) $this->db->fetchOne('SELECT COUNT(*) FROM stat_hit'));

        $result = $this->stats->prune($this->now);
        self::assertSame(['compacted' => 1, 'removed' => 1], $result);
        // The two hours became one row with the sum, and the total is unchanged.
        $row = $this->db->fetchAssociative('SELECT hour, count FROM stat_hit WHERE day = ?', [$old->format('Y-m-d')]);
        self::assertSame([-1, 2], [(int) $row['hour'], (int) $row['count']]);
        self::assertSame(2, (int) $this->db->fetchOne('SELECT COUNT(*) FROM stat_hit'));
        // Pruning twice changes nothing further.
        self::assertSame(['compacted' => 0, 'removed' => 0], $this->stats->prune($this->now));

        // A compacted day still counts in the overview, it just has no hour any more.
        $overview = $this->stats->overview(90, $this->now);
        self::assertSame(3, $overview['total']);
        self::assertSame(0, array_sum(array_column($overview['hours'], 'count')) - 1, 'Only yesterday is still hour precise.');
    }

    public function testTheExportReturnsEveryCounterOfTheRange(): void
    {
        $this->stats->record('/', 200, '', 'kant-it.de', $this->now);
        $this->stats->record('/weg', 404, '', 'kant-it.de', $this->now);
        $this->stats->record('/alt', 200, '', 'kant-it.de', $this->now->modify('-40 days'));

        $rows = [...$this->stats->export(7, $this->now)];
        self::assertCount(2, $rows, 'Only what lies in the range.');
        self::assertSame(3, count([...$this->stats->export(90, $this->now)]));
    }
}