MeshVault/tests/Unit/Indexer/StreamingHasherTest.php

<?php

declare(strict_types=1);

namespace MeshVault\Tests\Unit\Indexer;

use MeshVault\Indexer\ImportException;
use MeshVault\Indexer\Filesystem\PathSafety;
use MeshVault\Indexer\StreamingHasher;
use PHPUnit\Framework\Attributes\DataProvider;
use PHPUnit\Framework\TestCase;
use RuntimeException;

final class StreamingHasherTest extends TestCase
{
    private string $root;
    private string $path;

    protected function setUp(): void
    {
        $this->root = sys_get_temp_dir() . '/meshvault-hash-' . bin2hex(random_bytes(8));
        mkdir($this->root);
        $this->path = $this->root . '/source';
        file_put_contents($this->path, str_repeat('abc', 100000));
        self::assertTrue(class_exists(StreamingHasher::class), 'StreamingHasher must exist');
    }

    protected function tearDown(): void
    {
        foreach (scandir($this->root) as $name) {
            if ($name !== '.' && $name !== '..') {
                unlink($this->root . '/' . $name);
            }
        }
        rmdir($this->root);
    }

    private function hasher(?callable $observer = null, array $hooks = []): StreamingHasher
    {
        return new StreamingHasher(new PathSafety($this->root), $observer, $hooks);
    }

    public static function sizes(): iterable
    {
        foreach ([0, 1, 131072, 131073, 524305] as $size) {
            yield (string) $size => [$size];
        }
    }

    #[DataProvider('sizes')]
    public function testDigestAndBoundedReads(int $size): void
    {
        file_put_contents($this->path, str_repeat('x', $size));
        $observed = 0;
        $hasher = $this->hasher(function (int $bytes) use (&$observed): void {
            self::assertGreaterThan($observed, $bytes);
            self::assertLessThanOrEqual(131072, $bytes - $observed);
            $observed = $bytes;
        }, ['read' => static function ($stream, int $length): string|false {
            self::assertSame(131072, $length);
            return fread($stream, $length);
        }]);
        self::assertSame(hash_file('sha256', $this->path), $hasher->hash($this->path));
        self::assertSame($size, $observed);
    }

    public static function failures(): iterable
    {
        yield 'open' => ['open', false, 'HASH_OPEN_FAILED'];
        yield 'read' => ['read', false, 'HASH_READ_FAILED'];
        yield 'stalled read' => ['read', '', 'HASH_READ_FAILED'];
        yield 'path stat' => ['stat', false, 'HASH_STAT_FAILED'];
        yield 'handle stat' => ['fstat', false, 'HASH_STAT_FAILED'];
    }

    #[DataProvider('failures')]
    public function testIoFailures(string $hook, mixed $result, string $reason): void
    {
        $this->rejects(fn () => $this->hasher(null, [$hook => fn () => $result])->hash($this->path), $reason);
    }

    public function testThrownIoFailurePreservesCauseAndClosesStream(): void
    {
        $stream = null;
        $cause = new RuntimeException('injected read failure');
        $hasher = $this->hasher(null, [
            'open' => function (string $path) use (&$stream) {
                return $stream = fopen($path, 'rb');
            },
            'read' => static fn () => throw $cause,
        ]);
        $error = $this->rejects(fn () => $hasher->hash($this->path), 'HASH_READ_FAILED');
        self::assertSame($cause, $error->getPrevious());
        self::assertFalse(is_resource($stream));
    }

    public static function metadataChanges(): iterable
    {
        foreach (['stat', 'fstat'] as $hook) {
            foreach (['dev', 'ino', 'mode', 'size', 'mtime', 'ctime'] as $field) {
                yield "$hook $field" => [$hook, $field];
            }
        }
    }

    #[DataProvider('metadataChanges')]
    public function testEndMetadataChanges(string $hook, string $field): void
    {
        $calls = 0;
        $hasher = $this->hasher(null, [$hook => static function ($target) use ($hook, $field, &$calls) {
            $stat = $hook($target);
            if (++$calls === 2) {
                ++$stat[$field];
            }
            return $stat;
        }]);
        $this->rejects(fn () => $hasher->hash($this->path), 'HASH_SOURCE_CHANGED');
    }

    public function testOpenHandleMustMatchPreOpenPath(): void
    {
        $hasher = $this->hasher(null, ['fstat' => static function ($stream) {
            $stat = fstat($stream);
            ++$stat['ino'];
            return $stat;
        }]);
        $this->rejects(fn () => $hasher->hash($this->path), 'HASH_SOURCE_CHANGED');
    }

    public function testDirectoryRejected(): void
    {
        $this->rejects(fn () => $this->hasher()->hash($this->root), 'HASH_NOT_REGULAR_FILE');
    }

    public static function mutations(): iterable
    {
        yield 'append' => ['append'];
        yield 'truncate' => ['truncate'];
        yield 'same size with changed timestamp' => ['rewrite'];
        yield 'replacement' => ['replace'];
    }

    #[DataProvider('mutations')]
    public function testRealMutation(string $mutation): void
    {
        $done = false;
        $hasher = $this->hasher(function () use ($mutation, &$done): void {
            if ($done) {
                return;
            }
            $done = true;
            if ($mutation === 'append') {
                file_put_contents($this->path, 'extra', FILE_APPEND);
            } elseif ($mutation === 'truncate') {
                file_put_contents($this->path, 'short');
            } elseif ($mutation === 'rewrite') {
                file_put_contents($this->path, str_repeat('z', 300000));
                touch($this->path, time() + 10);
            } else {
                if (!@rename($this->path, $this->root . '/old')) {
                    self::markTestSkipped('Host cannot rename an open file.');
                }
                file_put_contents($this->path, str_repeat('z', 300000));
                touch($this->path, time() + 10);
            }
        });
        $this->rejects(fn () => $hasher->hash($this->path), 'HASH_SOURCE_CHANGED');
    }

    public function testUnsafeAndMissingSources(): void
    {
        foreach (
            [
            '' => 'INVALID_PATH',
            'php://memory' => 'INVALID_PATH',
            $this->root . '/../source' => 'PATH_TRAVERSAL',
            $this->root . '/missing' => 'PATH_NOT_FOUND',
            ] as $path => $reason
        ) {
            $this->rejects(fn () => $this->hasher()->hash($path), $reason);
        }
    }

    public static function links(): iterable
    {
        yield 'file' => [false, false];
        yield 'ancestor' => [true, false];
        yield 'mid-read replacement' => [false, true];
    }

    #[DataProvider('links')]
    public function testSymlinks(bool $ancestor, bool $duringRead): void
    {
        $link = $this->root . '/link';
        if (!@symlink($ancestor ? $this->root : $this->path, $link)) {
            self::markTestSkipped('Host cannot create symlinks.');
        }
        if (!$duringRead) {
            $this->rejects(
                fn () => $this->hasher()->hash($link . ($ancestor ? '/source' : '')),
                'SYMLINK_ESCAPE'
            );
            return;
        }
        unlink($link);
        file_put_contents($this->root . '/target', 'replacement');
        $done = false;
        $hasher = $this->hasher(function () use (&$done): void {
            if (!$done) {
                $done = true;
                if (!@rename($this->path, $this->root . '/old')) {
                    self::markTestSkipped('Host cannot rename an open file.');
                }
                self::assertTrue(symlink($this->root . '/target', $this->path));
            }
        });
        $this->rejects(fn () => $hasher->hash($this->path), 'SYMLINK_ESCAPE');
    }

    public function testEndStatFailure(): void
    {
        foreach (['stat', 'fstat'] as $operation) {
            $calls = 0;
            $hasher = $this->hasher(null, [$operation => static function ($target) use ($operation, &$calls) {
                return ++$calls === 2 ? false : $operation($target);
            }]);
            $this->rejects(fn () => $hasher->hash($this->path), 'HASH_STAT_FAILED');
        }
    }

    public function testObserverFailureClosesStream(): void
    {
        $stream = null;
        $cause = new RuntimeException('observer');
        $hasher = $this->hasher(static fn () => throw $cause, [
            'open' => function (string $path) use (&$stream) {
                return $stream = fopen($path, 'rb');
            },
        ]);
        try {
            $hasher->hash($this->path);
            self::fail('Expected observer exception');
        } catch (RuntimeException $error) {
            self::assertSame($cause, $error);
        }
        self::assertFalse(is_resource($stream));
    }

    private function rejects(callable $operation, string $reason): ImportException
    {
        try {
            $operation();
            self::fail('Expected ' . $reason);
        } catch (ImportException $error) {
            self::assertSame($reason, $error->reason);
            self::assertNotSame('', $error->getMessage());
            return $error;
        }
    }
}