saqel/backend/scripts/curriculum_markdown_parser.php

<?php

require_once __DIR__ . '/../app/bootstrap.php';

use App\Services\CurriculumService;

/**
 * Intelligent & Semantic Markdown Curriculum Parser
 * Maps sections with 100% precision:
 * - Lesson 1 -> lesson_01.md
 * - Lesson 2 -> lesson_02.md
 * - Intro & Project -> intro_and_project.md
 * - GeoGebra Lab -> geogebra_lab.md
 * - Unit Exam -> unit_exam.md
 */
function parseAndIngestTextbookMarkdown(string $rawMarkdown, array $metadata = []): array
{
    $lines = explode("\n", $rawMarkdown);
    
    $gradeName = $metadata['grade_name'] ?? 'الصف العاشر';
    $gradeKey  = $metadata['grade_key']  ?? 'grade_10';
    $subName   = $metadata['subject_name'] ?? 'الرياضيات';
    $subKey    = $metadata['subject_key']  ?? 'math_10';
    $semName   = $metadata['semester_name'] ?? 'الفصل الدراسي الأول';
    $semKey    = $metadata['semester_key']  ?? 'semester_1';

    $units = [];
    $currentUnit = null;
    $currentLesson = null;
    $lessonBuffer = [];

    $flushLesson = function() use (&$currentUnit, &$currentLesson, &$lessonBuffer) {
        if ($currentUnit && $currentLesson) {
            $content = trim(implode("\n", $lessonBuffer));
            if (!empty($content)) {
                $currentLesson['markdown_content'] = $content;
                $currentUnit['lessons'][] = $currentLesson;
            }
            $currentLesson = null;
            $lessonBuffer = [];
        }
    };

    $flushUnit = function() use (&$units, &$currentUnit, $flushLesson) {
        $flushLesson();
        if ($currentUnit && !empty($currentUnit['lessons'])) {
            $units[] = $currentUnit;
            $currentUnit = null;
        }
    };

    $unitCounter = 0;

    $arabicOrdinalMap = [
        'الأولى' => 1, 'الاولى' => 1, '1' => 1,
        'الثانية' => 2, 'الثانيه' => 2, '2' => 2,
        'الثالثة' => 3, 'الثالثه' => 3, '3' => 3,
        'الرابعة' => 4, 'الرابعه' => 4, '4' => 4,
        'الخامسة' => 5, 'الخامسه' => 5, '5' => 5,
        'السادسة' => 6, 'السادسه' => 6, '6' => 6,
    ];

    foreach ($lines as $line) {
        $trimmed = trim($line);

        // Detect Unit (# UNIT or # الوحدة)
        if (preg_match('/^#\s+(UNIT|الوحدة)\s*([0-9\-\:A-Za-z\s\p{Arabic}]+)(.*)$/ui', $trimmed, $m) || preg_match('/^#\s+([A-Za-z0-9\s]+(?:SPOT|SKILLS|REFERENCE|LIST))/ui', $trimmed, $m2)) {
            $flushUnit();
            $unitTitle = trim(preg_replace('/^#\s+/u', '', $trimmed));
            
            $detectedNum = null;
            foreach ($arabicOrdinalMap as $word => $num) {
                if (mb_stripos($unitTitle, $word) !== false) {
                    $detectedNum = $num;
                    break;
                }
            }
            if ($detectedNum !== null) {
                $unitCounter = $detectedNum;
            } else {
                $unitCounter++;
            }

            $unitKey = sprintf('unit_%02d', $unitCounter);
            
            $currentUnit = [
                'unit_key'  => $unitKey,
                'unit_name' => $unitTitle,
                'lessons'   => []
            ];
            continue;
        }

        // Detect Section (## ...)
        if (preg_match('/^##\s+(.+)$/u', $trimmed, $mAlt) && $currentUnit) {
            $flushLesson();
            $lessonTitle = trim($mAlt[1]);

            // Semantic Identification of Lesson ID and File
            if (preg_match('/(?:الدرس|Lesson)\s*([0-9]+)/ui', $lessonTitle, $mLes)) {
                $lesNum = (int)$mLes[1];
                $lesId = sprintf('lesson_%02d', $lesNum);
            } elseif (preg_match('/(?:مقدمة|مشروع)/ui', $lessonTitle)) {
                $lesId = 'intro_and_project';
            } elseif (preg_match('/(?:معمل|جيوجبرا|نشاط)/ui', $lessonTitle)) {
                $lesId = 'geogebra_lab';
            } elseif (preg_match('/(?:اختبار|تدريب|مراجعة)/ui', $lessonTitle)) {
                $lesId = 'unit_exam';
            } elseif (preg_match('/(?:إثراء|الإثراء|توسع|التوسع)/ui', $lessonTitle)) {
                $lesId = 'enrichment_and_expansion';
            } else {
                $lesId = 'section_' . count($currentUnit['lessons'] ?? []);
            }

            $currentLesson = [
                'lesson_id' => $lesId,
                'title'     => $lessonTitle,
                'outcomes'  => [$lessonTitle]
            ];
            $lessonBuffer = [$trimmed];
            continue;
        }

        if ($currentLesson) {
            $lessonBuffer[] = $line;
        }
    }

    $flushUnit();

    // Clean old files in target directories to eliminate stubs
    $baseUnitDir = __DIR__ . "/../storage/curriculum/{$gradeKey}/{$subKey}/{$semKey}/";
    foreach ($units as $u) {
        $uDir = $baseUnitDir . $u['unit_key'];
        if (is_dir($uDir)) {
            $existingFiles = glob($uDir . '/*.md');
            foreach ($existingFiles as $f) {
                if (basename($f) === 'lesson_1.md' || basename($f) === 'lesson_2.md') {
                    @unlink($f);
                }
            }
        }
    }

    $extracted = [
        'grade_name'    => $gradeName,
        'grade_key'     => $gradeKey,
        'subject_name'  => $subName,
        'subject_key'   => $subKey,
        'semester_name' => $semName,
        'semester_key'  => $semKey,
        'units'         => $units
    ];

    return CurriculumService::mergeExtractedCurriculum($extracted);
}