sitemap-admin.php

<?php
session_start();
ini_set('max_execution_time', 300); // 5 minutes max execution time

// Enhanced Configuration
$config = [
    'base_url' => (isset($_SERVER['HTTPS']) && $_SERVER['HTTPS'] === 'on' ? "https" : "http") . "://$_SERVER[HTTP_HOST]",
    'public_html_path' => $_SERVER['DOCUMENT_ROOT'],
    'allowed_extensions' => [
        'html', 'htm', 'shtml', 'xhtml', 'xml', 'asp', 'aspx', 
        'jsp', 'php', 'do', 'cgi'
    ],
    'excluded_extensions' => [
        'txt', 'json', 'md', 'git', 'svg', 'jpg', 'jpeg', 'png', 'gif', 
        'pdf', 'css', 'js', 'ico', 'woff', 'woff2', 'ttf', 'eot', 'map'
    ],
    'excluded_directories' => [
        'cgi-bin', 'css', 'js', 'images', 'img', 'backend-libs', 'admin',
        'test', 'tests', 'temp', 'tmp', 'cache', 'includes', 'vendor',
        'node_modules', 'assets', 'dist', 'build', 'uploads', 'fonts'
    ],
    'sitemap_path' => $_SERVER['DOCUMENT_ROOT'] . '/sitemap.xml',
    'robots_path' => $_SERVER['DOCUMENT_ROOT'] . '/robots.txt',
    'password' => 'admin', // Changed password
    'default_changefreq' => 'weekly',
    'max_urls' => 50000,
    'batch_size' => 1000,
    'debug' => false // Enable for debugging
];

// Authentication check with debugging
function checkAuth() {
    global $config;
    if ($config['debug']) {
        error_log("Session admin status: " . (isset($_SESSION['admin']) ? $_SESSION['admin'] : 'not set'));
    }
    return (isset($_SESSION['admin']) && $_SESSION['admin'] === true);
}

// Handle login
if (isset($_POST['password'])) {
    global $config;
    if ($config['debug']) {
        error_log("Login attempt - Posted password: " . $_POST['password']);
        error_log("Configured password: " . $config['password']);
    }
    
    if (trim($_POST['password']) === trim($config['password'])) {
        $_SESSION['admin'] = true;
        if ($config['debug']) {
            error_log("Login successful - Session admin set to true");
        }
        header('Location: ' . $_SERVER['PHP_SELF']);
        exit;
    } else {
        $error = "Incorrect password";
        if ($config['debug']) {
            error_log("Login failed - Incorrect password");
        }
    }
}

// Logout handler
if (isset($_GET['logout'])) {
    session_destroy();
    header('Location: ' . $_SERVER['PHP_SELF']);
    exit;
}

// Improved file scanning function
function scanDirectory($dir, &$results = array(), $depth = 0) {
    global $config;
    
    // Limit directory depth to prevent infinite recursion
    if ($depth > 10) return $results;
    
    $files = scandir($dir);
    
    foreach ($files as $file) {
        if ($file === '.' || $file === '..') continue;
        
        $path = realpath($dir . DIRECTORY_SEPARATOR . $file);
        $relativePath = str_replace($config['public_html_path'], '', $path);
        
        // Skip excluded directories
        $dirName = basename($dir);
        if (in_array(strtolower($dirName), $config['excluded_directories'])) {
            continue;
        }
        
        if (is_dir($path)) {
            // Skip processing if directory is in excluded list
            if (!in_array(basename($path), $config['excluded_directories'])) {
                $results[] = [
                    'type' => 'directory',
                    'path' => $relativePath,
                    'included' => isset($_SESSION['included'][$relativePath]) ? true : false
                ];
                scanDirectory($path, $results, $depth + 1);
            }
        } else {
            $ext = strtolower(pathinfo($path, PATHINFO_EXTENSION));
            // Only include allowed file types
            if (in_array($ext, $config['allowed_extensions'])) {
                $results[] = [
                    'type' => 'file',
                    'path' => $relativePath,
                    'included' => isset($_SESSION['included'][$relativePath]) ? true : false,
                    'lastmod' => filemtime($path)
                ];
            }
        }
    }
    return $results;
}

// Get canonical URL
function getCanonicalUrl($path) {
    global $config;
    
    // Remove index files from URL
    $url = rtrim($config['base_url'], '/') . $path;
    $url = preg_replace('/(index\.(php|html?|cgi))$/i', '', $url);
    
    // Remove trailing slash for non-directory URLs
    if (!is_dir($config['public_html_path'] . $path)) {
        $url = rtrim($url, '/');
    }
    
    // Convert to lowercase for consistency
    $url = strtolower($url);
    
    // Encode URL properly
    return str_replace(' ', '%20', $url);
}

// Calculate priority based on path depth
function calculatePriority($path) {
    $depth = substr_count($path, '/');
    return max(0.1, 1.0 - ($depth * 0.2));
}

// Generate sitemap XML
function generateSitemap($files) {
    global $config;
    
    $xml = new DOMDocument('1.0', 'UTF-8');
    $xml->formatOutput = true;
    
    $urlset = $xml->createElement('urlset');
    $urlset->setAttribute('xmlns', 'http://www.sitemaps.org/schemas/sitemap/0.9');
    $urlset->setAttribute('xmlns:xhtml', 'http://www.w3.org/1999/xhtml');
    $xml->appendChild($urlset);
    
    // Add homepage
    $url = $xml->createElement('url');
    $loc = $xml->createElement('loc', $config['base_url'] . '/');
    $lastmod = $xml->createElement('lastmod', date('c'));
    $changefreq = $xml->createElement('changefreq', 'daily');
    $priority = $xml->createElement('priority', '1.0');
    
    $url->appendChild($loc);
    $url->appendChild($lastmod);
    $url->appendChild($changefreq);
    $url->appendChild($priority);
    $urlset->appendChild($url);
    
    $urlCount = 1;
    foreach ($files as $file) {
        if ($urlCount >= $config['max_urls']) break;
        
        // Only include files that are explicitly checked
        if (!isset($_SESSION['included'][$file['path']]) || $file['type'] === 'directory') continue;
        
        $url = $xml->createElement('url');
        
        // Use canonical URL
        $canonicalUrl = getCanonicalUrl($file['path']);
        $loc = $xml->createElement('loc', $canonicalUrl);
        
        // Use ISO 8601 date format
        $lastmod = $xml->createElement('lastmod', date('c', $file['lastmod']));
        
        // Calculate priority based on URL depth
        $urlPriority = calculatePriority($file['path']);
        $priority = $xml->createElement('priority', number_format($urlPriority, 1));
        
        // Set change frequency based on file modification time
        $age = time() - $file['lastmod'];
        $changefreq = $xml->createElement('changefreq', 
            $age < 86400 ? 'daily' :
            ($age < 604800 ? 'weekly' :
            ($age < 2592000 ? 'monthly' : 'yearly'))
        );
        
        $url->appendChild($loc);
        $url->appendChild($lastmod);
        $url->appendChild($changefreq);
        $url->appendChild($priority);
        $urlset->appendChild($url);
        
        $urlCount++;
    }
    
    // Compress the sitemap
    $xml->save($config['sitemap_path']);
    if (function_exists('gzencode')) {
        file_put_contents($config['sitemap_path'] . '.gz', gzencode($xml->saveXML()));
    }
}

// Generate robots.txt
function generateRobotsTxt($files) {
    global $config;
    
    $content = "User-agent: *\n";
    $content .= "Allow: /\n\n";
    
    // Explicitly disallow admin and system directories
    foreach ($config['excluded_directories'] as $dir) {
        $content .= "Disallow: /$dir/\n";
    }
    
    // Add Crawl-delay directive
    $content .= "\nCrawl-delay: 1\n\n";
    
    // Add both regular and gzipped sitemap locations
    $content .= "Sitemap: " . $config['base_url'] . "/sitemap.xml\n";
    if (function_exists('gzencode')) {
        $content .= "Sitemap: " . $config['base_url'] . "/sitemap.xml.gz\n";
    }
    
    file_put_contents($config['robots_path'], $content);
}

// Handle form submission for sitemap generation
if ($_SERVER['REQUEST_METHOD'] === 'POST' && isset($_POST['generate'])) {
    if (checkAuth()) {
        // Clear previous selection
        $_SESSION['included'] = array();
        
        // Store new selection
        if (isset($_POST['included']) && is_array($_POST['included'])) {
            foreach ($_POST['included'] as $path => $value) {
                $_SESSION['included'][$path] = true;
            }
        }
        
        $files = scanDirectory($config['public_html_path']);
        generateSitemap($files);
        generateRobotsTxt($files);
        $_SESSION['message'] = 'Sitemap and robots.txt generated successfully!';
        header('Location: ' . $_SERVER['PHP_SELF']);
        exit;
    }
}
?>
<!DOCTYPE html>
<html lang="en">
<head>
    <meta charset="UTF-8">
    <meta name="viewport" content="width=device-width, initial-scale=1.0">
    <title>Sitemap Generator</title>
    <style>
        body { 
            font-family: Arial, sans-serif; 
            max-width: 1200px; 
            margin: 0 auto; 
            padding: 20px; 
        }
        .file-list { 
            margin: 20px 0; 
        }
        .file-item { 
            margin: 5px 0; 
            padding: 5px; 
            border: 1px solid #ddd; 
        }
        .directory { 
            font-weight: bold; 
            background: #f5f5f5; 
        }
        .controls { 
            position: sticky; 
            top: 0; 
            background: white; 
            padding: 10px 0; 
            border-bottom: 1px solid #ddd; 
            z-index: 100;
        }
        .message { 
            padding: 10px; 
            background: #e6ffe6; 
            border: 1px solid #99ff99; 
            margin: 10px 0; 
        }
        .error { 
            padding: 10px; 
            background: #ffe6e6; 
            border: 1px solid #ff9999; 
            margin: 10px 0; 
        }
        .login-form { 
            max-width: 300px; 
            margin: 100px auto; 
            padding: 20px; 
            border: 1px solid #ddd; 
            border-radius: 5px; 
        }
        .login-form input[type="password"] { 
            width: 100%; 
            padding: 8px; 
            margin: 10px 0; 
            border: 1px solid #ddd;
            border-radius: 4px;
        }
        .login-form input[type="submit"], 
        button {
            padding: 8px 16px;
            background: #4CAF50;
            color: white;
            border: none;
            border-radius: 4px;
            cursor: pointer;
        }
        .login-form input[type="submit"]:hover,
        button:hover {
            background: #45a049;
        }
        .logout { 
            float: right; 
        }
        .debug-info {
            background: #fff3cd;
            padding: 10px;
            margin: 10px 0;
            border: 1px solid #ffeeba;
            border-radius: 4px;
        }
    </style>
</head>
<body>
    <?php if (!checkAuth()): ?>
        <!-- Login Form -->
        <div class="login-form">
            <h2>Login to Sitemap Generator</h2>
            <?php if (isset($error)): ?>
                <div class="error"><?php echo htmlspecialchars($error); ?></div>
            <?php endif; ?>
            <?php if ($config['debug']): ?>
                <div class="debug-info">
                    Debug Mode Active
                </div>
            <?php endif; ?>
            <form method="post">
                <input type="password" name="password" placeholder="Enter password" required>
                <input type="submit" value="Login">
            </form>
        </div>
    <?php else: ?>
        <!-- Sitemap Generator Interface -->
        <h1>Sitemap Generator</h1>
        <a href="?logout" class="logout">Logout</a>
        
        <?php if (isset($_SESSION['message'])): ?>
            <div class="message">
                <?php 
                echo htmlspecialchars($_SESSION['message']);
                unset($_SESSION['message']);
                ?>
            </div>
        <?php endif; ?>

        <form method="post">
            <div class="controls">
                <button type="submit" name="generate">Generate Sitemap & robots.txt</button>
                <button type="button" onclick="toggleAll(true)">Select All</button>
                <button type="button" onclick="toggleAll(false)">Deselect All</button>
            </div>

            <div class="file-list">
                <?php 
                $files = scanDirectory($config['public_html_path']);
                foreach ($files as $file): 
                ?>
                    <div class="file-item <?php echo $file['type'] === 'directory' ? 'directory' : ''; ?>">
                        <label>
                            <input type="checkbox" 
                                   name="included[<?php echo htmlspecialchars($file['path']); ?>]" 
                                   value="1" 
                                   <?php echo isset($_SESSION['included'][$file['path']]) ? 'checked' : ''; ?>>
                            <?php echo htmlspecialchars($file['path']); ?>
                        </label>
                    </div>
                <?php endforeach; ?>
            </div>
        </form>

        <script>
            function toggleAll(state) {
                document.querySelectorAll('input[type="checkbox"]').forEach(checkbox => {
                    checkbox.checked = state;
                });
            }
        </script>
    <?php endif; ?>
</body>
</html>