KantSites/src/Ops/Backup.php

<?php

namespace App\Ops;

use App\Domain\InvalidPage;
use Doctrine\DBAL\Connection;

/**
 * Sicherung und Wiederherstellung.
 *
 * Three things decide whether this is worth anything:
 *
 * 1. **The database copy has to be consistent.** A plain `copy()` of a SQLite file while something
 *    writes produces a file that looks fine and opens with an error later. `VACUUM INTO` asks
 *    SQLite for a complete copy instead.
 * 2. **The media folder belongs in it.** Documents reference image IDs; without the files the
 *    backup restores a site full of broken pictures.
 * 3. **`.env` does not.** It holds APP_SECRET, and a backup is something that gets downloaded and
 *    passed around. It is also not needed to restore content.
 *
 * Restoring is the part nobody practises, so it is the part with the most checks: a backup from a
 * newer version is refused, a safety copy of the current state is taken first, and the media are
 * added rather than replaced - deleting files that were uploaded after the backup would turn a
 * restore into a second loss.
 */
final class Backup
{
    public const MANIFEST = 'sicherung.json';
    /** Keep this many, newest first. Without rotation the disk fills and the next backup fails. */
    public const KEEP = 10;

    public function __construct(
        private string $root,
        private string $databaseUrl,
        private string $mediaDirectory,
        private Connection $db,
        private CommandRunner $runner,
    ) {
    }

    /**
     * @return array{name: string, path: string, manifest: array}
     *
     * @throws InvalidPage
     */
    public function create(string $reason = 'manuell', \DateTimeImmutable $now = new \DateTimeImmutable('now', new \DateTimeZone('UTC'))): array
    {
        // Two backups within the same second must not collide - and the second one must not
        // fail either: a restore takes a safety copy, and that can land in the same second as the
        // backup it is about to restore.
        $name = 'sicherung-'.$now->format('Ymd-His');
        $directory = $this->directory().'/'.$name;
        for ($suffix = 2; is_dir($directory) && $suffix < 100; ++$suffix) {
            $name = sprintf('sicherung-%s-%d', $now->format('Ymd-His'), $suffix);
            $directory = $this->directory().'/'.$name;
        }
        if (is_dir($directory)) {
            throw new InvalidPage('Es gibt bereits zu viele Sicherungen mit diesem Zeitstempel.');
        }
        if (!@mkdir($directory, 0o775, true)) {
            throw new InvalidPage(sprintf('Das Verzeichnis %s ließ sich nicht anlegen. Schreibrechte auf var/ prüfen.', $directory));
        }

        $database = $this->copyDatabase($directory);
        $media = $this->copyMedia($directory.'/medien');
        $manifest = [
            'erstellt' => $now->format('c'),
            'anlass' => $reason,
            'fassung' => $this->version(),
            // A database that has never been migrated by the tool has no such table; that is not
            // a reason to refuse a backup of its content.
            'migration' => $this->migration(),
            'datenbank' => $database,
            'medien' => ['dateien' => $media['count'], 'bytes' => $media['bytes']],
            'inhalt' => [
                'seiten' => (int) $this->db->fetchOne('SELECT COUNT(*) FROM page WHERE deleted_at IS NULL'),
                'fassungen' => (int) $this->db->fetchOne('SELECT COUNT(*) FROM revision'),
                'medien' => (int) $this->db->fetchOne('SELECT COUNT(*) FROM media'),
            ],
            // Not in here, and it is worth saying so where somebody reads it.
            'enthaelt_nicht' => ['.env', '.env.local', 'vendor/', 'Quellcode'],
        ];
        file_put_contents($directory.'/'.self::MANIFEST, json_encode($manifest, JSON_PRETTY_PRINT | JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE));

        return ['name' => $name, 'path' => $directory, 'manifest' => $manifest];
    }

    /**
     * @return list<array{name: string, path: string, bytes: int, manifest: ?array, intact: bool}>
     */
    public function all(): array
    {
        $backups = [];
        foreach (glob($this->directory().'/sicherung-*', GLOB_ONLYDIR) ?: [] as $path) {
            $manifest = $this->manifest($path);
            $backups[] = [
                'name' => basename($path),
                'path' => $path,
                'bytes' => $this->size($path),
                'manifest' => $manifest,
                // A backup whose database is missing or whose checksum does not match is not a
                // backup, and it must say so before somebody relies on it.
                'intact' => null !== $manifest && $this->intact($path, $manifest),
            ];
        }
        usort($backups, static fn (array $a, array $b): int => strcmp($b['name'], $a['name']));

        return $backups;
    }

    /** Removes the oldest beyond the limit. Returns the names it removed. */
    public function rotate(int $keep = self::KEEP): array
    {
        $removed = [];
        foreach (array_slice($this->all(), max(1, $keep)) as $backup) {
            $this->remove($backup['path']);
            $removed[] = $backup['name'];
        }

        return $removed;
    }

    /**
     * Puts a backup back in place.
     *
     * @return array{steps: list<array{step: string, ok: bool, detail: string}>, ok: bool}
     */
    public function restore(string $name, bool $force = false): array
    {
        $steps = [];
        $step = static function (array &$steps, string $name, bool $ok, string $detail): bool {
            $steps[] = ['step' => $name, 'ok' => $ok, 'detail' => $detail];

            return $ok;
        };
        $path = $this->directory().'/'.basename($name);
        $manifest = $this->manifest($path);
        if (!$step($steps, 'Sicherung finden', null !== $manifest, null === $manifest
            ? sprintf('%s gibt es nicht oder es fehlt das Manifest.', basename($name))
            : sprintf('%s vom %s, Fassung %s.', basename($name), $manifest['erstellt'], $manifest['fassung']))) {
            return ['steps' => $steps, 'ok' => false];
        }
        if (!$step($steps, 'Unversehrt', $this->intact($path, $manifest), $this->intact($path, $manifest)
            ? 'Die Prüfsumme der Datenbank stimmt.'
            : 'Die Prüfsumme der Datenbank stimmt nicht. Diese Sicherung wird nicht eingespielt.')) {
            return ['steps' => $steps, 'ok' => false];
        }
        // A backup from a newer version carries a schema this code does not know.
        $newer = version_compare((string) $manifest['fassung'], $this->version(), '>');
        if (!$step($steps, 'Fassung passt', !$newer || $force, $newer
            ? sprintf('Die Sicherung stammt aus Fassung %s, installiert ist %s. Erst aktualisieren.', $manifest['fassung'], $this->version())
            : sprintf('Sicherung %s, installiert %s.', $manifest['fassung'], $this->version()))) {
            return ['steps' => $steps, 'ok' => false];
        }

        try {
            $safety = $this->create('vor Wiederherstellung');
            $step($steps, 'Aktuellen Stand sichern', true, $safety['name'].' – auch eine Wiederherstellung kann die falsche sein.');
        } catch (InvalidPage $error) {
            $step($steps, 'Aktuellen Stand sichern', false, $error->getMessage());

            return ['steps' => $steps, 'ok' => false];
        }

        $file = $this->databaseFile();
        if (null === $file) {
            $step($steps, 'Datenbank zurückschreiben', false, 'Diese Installation nutzt keine Dateidatenbank. Für MySQL fehlt das Einspielen noch (siehe docs/UPDATE.md).');

            return ['steps' => $steps, 'ok' => false];
        }
        $this->db->close();
        $copied = @copy($path.'/'.basename($file), $file);
        if (!$step($steps, 'Datenbank zurückschreiben', $copied, $copied ? basename($file).' ersetzt.' : 'Die Datei ließ sich nicht ersetzen.')) {
            return ['steps' => $steps, 'ok' => false];
        }

        // Media are added, never deleted: files uploaded after the backup would otherwise be lost
        // by the very operation that is meant to prevent loss.
        $restored = $this->restoreMedia($path.'/medien');
        $step($steps, 'Medien ergänzen', true, sprintf('%d Datei(en) zurückgeschrieben. Neuere Dateien bleiben erhalten.', $restored));

        return ['steps' => $steps, 'ok' => true];
    }

    private function migration(): string
    {
        try {
            return (string) $this->db->fetchOne('SELECT MAX(version) FROM migration_versions');
        } catch (\Throwable) {
            return 'unbekannt';
        }
    }

    public function directory(): string
    {
        return $this->root.'/var/backups';
    }

    public function version(): string
    {
        $file = $this->root.'/VERSION';

        return is_readable($file) ? trim((string) file_get_contents($file)) : '0.0.0';
    }

    /** @return array{datei: string, bytes: int, sha256: string} */
    private function copyDatabase(string $directory): array
    {
        $file = $this->databaseFile();
        if (null === $file) {
            if (!$this->runner->has('mysqldump', $this->root)) {
                throw new InvalidPage('Diese Installation nutzt keine Dateidatenbank, und mysqldump ist nicht aufrufbar. Ohne eines von beiden gibt es keine Sicherung.');
            }
            throw new InvalidPage('Sicherung über mysqldump ist noch nicht umgesetzt (Arbeitspaket J).');
        }
        $target = $directory.'/'.basename($file);
        // VACUUM INTO instead of copy(): a copy taken while something writes can be unusable, and
        // it would not say so until the day it is needed. The path is ours, not user input.
        $this->db->executeStatement(sprintf("VACUUM INTO '%s'", str_replace("'", "''", str_replace('\\', '/', $target))));
        if (!is_readable($target)) {
            throw new InvalidPage('Die Datenbank ließ sich nicht sichern.');
        }

        return ['datei' => basename($file), 'bytes' => (int) filesize($target), 'sha256' => (string) hash_file('sha256', $target)];
    }

    /** @return array{count: int, bytes: int} */
    private function copyMedia(string $target): array
    {
        if (!is_dir($this->mediaDirectory)) {
            return ['count' => 0, 'bytes' => 0];
        }
        @mkdir($target, 0o775, true);
        $count = 0;
        $bytes = 0;
        foreach (glob($this->mediaDirectory.'/*') ?: [] as $file) {
            if (is_file($file) && @copy($file, $target.'/'.basename($file))) {
                ++$count;
                $bytes += (int) filesize($file);
            }
        }

        return ['count' => $count, 'bytes' => $bytes];
    }

    private function restoreMedia(string $source): int
    {
        if (!is_dir($source)) {
            return 0;
        }
        @mkdir($this->mediaDirectory, 0o775, true);
        $count = 0;
        foreach (glob($source.'/*') ?: [] as $file) {
            if (is_file($file) && @copy($file, $this->mediaDirectory.'/'.basename($file))) {
                ++$count;
            }
        }

        return $count;
    }

    private function manifest(string $path): ?array
    {
        $file = $path.'/'.self::MANIFEST;
        if (!is_readable($file)) {
            return null;
        }
        try {
            $manifest = json_decode((string) file_get_contents($file), true, 16, JSON_THROW_ON_ERROR);
        } catch (\JsonException) {
            return null;
        }

        return is_array($manifest) ? $manifest : null;
    }

    private function intact(string $path, array $manifest): bool
    {
        $file = $path.'/'.($manifest['datenbank']['datei'] ?? '');
        if (!is_readable($file)) {
            return false;
        }

        return hash_equals((string) ($manifest['datenbank']['sha256'] ?? ''), (string) hash_file('sha256', $file));
    }

    private function size(string $path): int
    {
        $bytes = 0;
        foreach (glob($path.'/*') ?: [] as $entry) {
            $bytes += is_dir($entry) ? $this->size($entry) : (int) filesize($entry);
        }

        return $bytes;
    }

    private function remove(string $path): void
    {
        foreach (glob($path.'/*') ?: [] as $entry) {
            is_dir($entry) ? $this->remove($entry) : @unlink($entry);
        }
        @rmdir($path);
    }

    private function databaseFile(): ?string
    {
        if (!str_starts_with($this->databaseUrl, 'sqlite:')) {
            return null;
        }
        $path = preg_replace('#^sqlite:(///)?#', '', $this->databaseUrl);

        return '' === $path ? null : str_replace('\\', '/', (string) $path);
    }
}