saqel/backend/scripts/import_grade10_book_sources.php

<?php

declare(strict_types=1);

/**
 * Registers supplied Grade 10 textbook PDFs as review-only source assets.
 *
 * Default mode is read-only.  --apply copies PDFs into curriculum storage and
 * creates draft, rights-review-required content_assets records.  It never
 * approves rights, publishes a bundle, or invents curriculum lessons.
 *
 * Usage:
 *   php backend/scripts/import_grade10_book_sources.php
 *   php backend/scripts/import_grade10_book_sources.php --apply
 */

use App\Core\Database;

$apply = in_array('--apply', $argv, true);
$projectRoot = dirname(__DIR__, 2);
$sourceRoot = $projectRoot . '/books';
$curriculumRoot = $projectRoot . '/backend/storage/curriculum';
$manifestPath = $curriculumRoot . '/manifest.json';

if (!is_dir($sourceRoot) || !is_file($manifestPath)) {
    fwrite(STDERR, "Books directory or curriculum manifest is unavailable.\n");
    exit(1);
}

$manifest = json_decode((string) file_get_contents($manifestPath), true);
if (!is_array($manifest)) {
    fwrite(STDERR, "Curriculum manifest is not valid JSON.\n");
    exit(1);
}

$knownSemesters = [];
foreach (($manifest['grade_10']['subjects'] ?? []) as $subjectKey => $subject) {
    foreach (array_keys($subject['semesters'] ?? []) as $semesterKey) {
        $knownSemesters[$subjectKey][$semesterKey] = true;
    }
}

$subjectPatterns = [
    'arabic_10' => ['اللغة العربية', 'العربية لغتي'],
    'english_10' => ['اللغة الإنجليزية'],
    'math_10' => ['الرياضيات'],
    'physics_10' => ['الفيزياء'],
    'chemistry_10' => ['الكيمياء'],
    'biology_10' => ['العلوم الحياتية'],
    'earth_science_10' => ['علوم الأرض والبيئة'],
    'digital_skills_10' => ['المهارات الرقمية'],
    'islamic_10' => ['التربية الإسلامية'],
    'history_10' => ['التاريخ'],
    'geography_10' => ['الجغرافيا'],
    'civic_10' => ['التربية الوطنية والمدنية'],
    'financial_literacy_10' => ['الثقافة المالية'],
];

$rows = [];
foreach (glob($sourceRoot . '/*.pdf') ?: [] as $sourcePath) {
    $filename = basename($sourcePath);
    $subjectKey = null;
    foreach ($subjectPatterns as $candidate => $patterns) {
        foreach ($patterns as $pattern) {
            if (str_contains($filename, $pattern)) {
                $subjectKey = $candidate;
                break 2;
            }
        }
    }
    $semesterKey = str_contains($filename, 'الفصل الأول') ? 'semester_1'
        : (str_contains($filename, 'الفصل الثاني') ? 'semester_2' : null);
    $isInstitutionalBook = str_starts_with($filename, 'كتاب الطالب') || str_starts_with($filename, 'كتاب التمارين');
    $sha256 = hash_file('sha256', $sourcePath);
    $safeName = preg_replace('/[^A-Za-z0-9._-]+/', '-', pathinfo($filename, PATHINFO_FILENAME)) ?: 'book';
    // Arabic filenames normalize to "book" with the ASCII-only fallback;
    // append integrity bytes so no two supplied books can overwrite each other.
    $storageKey = sprintf('sources/grade_10/%s/%s/%s-%s.pdf', $subjectKey ?: 'unclassified', $semesterKey ?: 'unclassified', trim($safeName, '-'), substr($sha256, 0, 12));
    $rows[] = [
        'filename' => $filename,
        'source_path' => $sourcePath,
        'subject_key' => $subjectKey,
        'semester_key' => $semesterKey,
        'asset_type' => $isInstitutionalBook ? 'textbook_pdf' : 'other',
        'pages' => pdfPages($sourcePath),
        'byte_size' => filesize($sourcePath),
        'sha256' => $sha256,
        'storage_key' => $storageKey,
        'manifest_scope_exists' => $subjectKey !== null && $semesterKey !== null && isset($knownSemesters[$subjectKey][$semesterKey]),
        'intake_status' => !$isInstitutionalBook ? 'manual_rights_and_academic_review_required'
            : ($subjectKey === null || $semesterKey === null ? 'unclassified_metadata' : 'ready_for_source_review'),
    ];
}

usort($rows, static fn(array $a, array $b): int => strcmp($a['filename'], $b['filename']));
$report = [
    'mode' => $apply ? 'apply' : 'dry_run',
    'policy' => 'Draft source registration only; no publication or rights clearance.',
    'books_found' => count($rows),
    'manifest_supported_sources' => count(array_filter($rows, static fn(array $row): bool => $row['manifest_scope_exists'])),
    'sources_outside_current_manifest' => count(array_filter($rows, static fn(array $row): bool => !$row['manifest_scope_exists'])),
    'rows' => $rows,
];

if (!$apply) {
    echo json_encode($report, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE) . PHP_EOL;
    exit(0);
}

require_once dirname(__DIR__) . '/app/bootstrap.php';

$pdo = Database::getConnection();
$pdo->beginTransaction();
try {
    foreach ($rows as $row) {
        if ($row['asset_type'] !== 'textbook_pdf' || $row['subject_key'] === null || $row['semester_key'] === null) {
            continue;
        }
        $destination = $curriculumRoot . '/' . $row['storage_key'];
        if (!is_dir(dirname($destination)) && !mkdir(dirname($destination), 0750, true) && !is_dir(dirname($destination))) {
            throw new RuntimeException('Unable to create textbook storage directory.');
        }
        if (!is_file($destination)) {
            if (!copy($row['source_path'], $destination)) {
                throw new RuntimeException('Unable to copy textbook source: ' . $row['filename']);
            }
        }
        if (!hash_equals($row['sha256'], hash_file('sha256', $destination))) {
            throw new RuntimeException('Copied textbook checksum mismatch: ' . $row['filename']);
        }
        Database::query(
            "INSERT INTO content_assets (uuid, asset_type, storage_driver, storage_key, mime_type, byte_size, sha256, source_reference, rights_status, review_status)
             VALUES (?, 'textbook_pdf', 'local', ?, 'application/pdf', ?, ?, ?, 'review_required', 'draft')
             ON DUPLICATE KEY UPDATE byte_size=VALUES(byte_size), source_reference=VALUES(source_reference)",
            [uuid(), $row['storage_key'], $row['byte_size'], $row['sha256'], 'books/' . $row['filename']]
        );
    }
    $pdo->commit();
    $report['registered_draft_sources'] = true;
    echo json_encode($report, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE) . PHP_EOL;
} catch (Throwable $e) {
    if ($pdo->inTransaction()) {
        $pdo->rollBack();
    }
    fwrite(STDERR, "Source import rolled back: {$e->getMessage()}\n");
    exit(1);
}

function pdfPages(string $path): ?int
{
    $output = [];
    $status = 0;
    exec('pdfinfo ' . escapeshellarg($path) . ' 2>/dev/null', $output, $status);
    if ($status !== 0) {
        return null;
    }
    foreach ($output as $line) {
        if (preg_match('/^Pages:\s+(\d+)$/', $line, $match)) {
            return (int) $match[1];
        }
    }
    return null;
}

function uuid(): string
{
    $bytes = random_bytes(16);
    $bytes[6] = chr((ord($bytes[6]) & 0x0f) | 0x40);
    $bytes[8] = chr((ord($bytes[8]) & 0x3f) | 0x80);
    return vsprintf('%s%s-%s-%s-%s-%s%s%s', str_split(bin2hex($bytes), 4));
}