<?php
set_time_limit(120);
date_default_timezone_set("Asia/Kolkata");

$siteUrl = "https://www.himanshuartinstitute.co.in";
$rootDir = __DIR__;

$cacheDir = __DIR__ . "/SITEMAP";
$cacheXmlFile = $cacheDir . "/sitemap-dynamic-cache.xml";
$cacheJsonFile = $cacheDir . "/sitemap-status.json";
$cacheSeconds = 3600; // 1 hour

$allowedExtensions = ["html", "htm", "php"];

$excludeDirs = [
    "admins",
    "admin",
    "baba",
    "actions",
    "cookies",
    "newsletter",
    "SITEMAP",
    "search",
    "codes",
    "css",
    "js",
    "images",
    "uploads",
    "ckeditor",
    "tcpdf",
    "pdf_templates",
    "shop/wp-admin",
    "shop/wp-content",
    "shop/wp-includes",
    "video-gallery/footer-videos"
];

$excludeFiles = [
    "sitemap.php",
    "feed.php",
    "feeds.php",
    "xmlrpc.php",
    "wp-config.php",
    "functions.php",
    "bootstrap.php"
];

$excludeNameContains = [
    "slider",
    "slides",
    "header",
    "hedr",
    "footer",
    "foter",
    "mobs",
    "mobls",
    "sitemap",
    "successbox",
    "award-link",
    "awards-lists",
    "offers",
    "afilations",
    "check",
    "test",
    "demo",
    "sample",
    "copy",
    "backup",
    "old"
];

function hai_path_has_excluded_dir($relativePath, $excludeDirs){
    $relativePath = trim(str_replace("\\", "/", $relativePath), "/");

    foreach($excludeDirs as $dir){
        $dir = trim(str_replace("\\", "/", $dir), "/");

        if($relativePath === $dir || strpos($relativePath, $dir . "/") === 0){
            return true;
        }
    }

    return false;
}

function hai_is_noindex_file($filePath){
    $content = @file_get_contents($filePath, false, null, 0, 30000);

    if($content === false){
        return false;
    }

    return stripos($content, "noindex") !== false;
}

function hai_make_url($siteUrl, $relativePath){
    $relativePath = str_replace("\\", "/", $relativePath);
    $relativePath = ltrim($relativePath, "/");

    if(preg_match("#(^|/)index\.(html|htm|php)$#i", $relativePath)){
        $relativePath = preg_replace("#index\.(html|htm|php)$#i", "", $relativePath);
    }

    $parts = explode("/", $relativePath);
    $encodedParts = [];

    foreach($parts as $part){
        if($part !== ""){
            $encodedParts[] = rawurlencode($part);
        }
    }

    $encodedPath = implode("/", $encodedParts);

    if($relativePath !== "" && substr($relativePath, -1) === "/"){
        $encodedPath .= "/";
    }

    return rtrim($siteUrl, "/") . "/" . $encodedPath;
}

function hai_priority($relativePath){
    $relativePath = str_replace("\\", "/", $relativePath);

    if(preg_match("#^index\.(html|htm|php)$#i", $relativePath)){
        return "1.00";
    }

    if(preg_match("#/index\.(html|htm|php)$#i", $relativePath)){
        return "0.80";
    }

    return "0.70";
}

function hai_get_sitemap_urls($siteUrl, $rootDir, $allowedExtensions, $excludeDirs, $excludeFiles, $excludeNameContains){
    $urls = [];

    $directory = new RecursiveDirectoryIterator($rootDir, FilesystemIterator::SKIP_DOTS);

    $filter = new RecursiveCallbackFilterIterator($directory, function($current) use ($rootDir, $excludeDirs){
        $path = $current->getPathname();
        $relativePath = substr($path, strlen($rootDir) + 1);
        $relativePath = str_replace("\\", "/", $relativePath);

        if($current->isDir()){
            if(substr(basename($relativePath), 0, 1) === "."){
                return false;
            }

            if(hai_path_has_excluded_dir($relativePath, $excludeDirs)){
                return false;
            }
        }

        return true;
    });

    $iterator = new RecursiveIteratorIterator($filter);

    foreach($iterator as $file){
        if(!$file->isFile()){
            continue;
        }

        $filePath = $file->getPathname();
        $relativePath = substr($filePath, strlen($rootDir) + 1);
        $relativePath = str_replace("\\", "/", $relativePath);

        $fileName = basename($relativePath);
        $extension = strtolower(pathinfo($fileName, PATHINFO_EXTENSION));

        if(!in_array($extension, $allowedExtensions)){
            continue;
        }

        if(in_array($fileName, $excludeFiles)){
            continue;
        }
		
		foreach($excludeNameContains as $blockedName){
    	if(stripos($fileName, $blockedName) !== false){
        continue 2;
    	}
		}

        if(hai_is_noindex_file($filePath)){
            continue;
        }

        $urls[] = [
            "loc" => hai_make_url($siteUrl, $relativePath),
            "lastmod" => gmdate("Y-m-d\TH:i:s+00:00", filemtime($filePath)),
            "priority" => hai_priority($relativePath),
            "file" => $relativePath,
            "mtime" => filemtime($filePath)
        ];
    }

    usort($urls, function($a, $b){
        return strcmp($a["loc"], $b["loc"]);
    });

    return $urls;
}

function hai_build_xml($urls){
    $xml = '<?xml version="1.0" encoding="UTF-8"?>' . "\n";
    $xml .= '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">' . "\n";

    foreach($urls as $url){
        $xml .= "  <url>\n";
        $xml .= "    <loc>" . htmlspecialchars($url["loc"], ENT_XML1, "UTF-8") . "</loc>\n";
        $xml .= "    <lastmod>" . $url["lastmod"] . "</lastmod>\n";
        $xml .= "    <changefreq>weekly</changefreq>\n";
        $xml .= "    <priority>" . $url["priority"] . "</priority>\n";
        $xml .= "  </url>\n";
    }

    $xml .= "</urlset>";

    return $xml;
}

function hai_rebuild_sitemap($siteUrl, $rootDir, $allowedExtensions, $excludeDirs, $excludeFiles, $excludeNameContains, $cacheXmlFile, $cacheJsonFile){
	$urls = hai_get_sitemap_urls($siteUrl, $rootDir, $allowedExtensions, $excludeDirs, $excludeFiles, $excludeNameContains);
    $xml = hai_build_xml($urls);

    $latest = null;

    foreach($urls as $url){
        if($latest === null || $url["mtime"] > $latest["mtime"]){
            $latest = $url;
        }
    }

    $status = [
        "status" => "ok",
        "site_url" => $siteUrl,
        "sitemap_url" => $siteUrl . "/sitemap.php",
        "total_urls" => count($urls),
        "last_checked" => date("Y-m-d H:i:s"),
        "latest_page" => $latest ? $latest["loc"] : "",
        "latest_modified" => $latest ? date("Y-m-d H:i:s", $latest["mtime"]) : ""
    ];

    @file_put_contents($cacheXmlFile, $xml);
    @file_put_contents($cacheJsonFile, json_encode($status));

    return [$xml, $status];
}

$forceRefresh = isset($_GET["refresh"]) && $_GET["refresh"] == "1";
$cacheIsFresh = file_exists($cacheXmlFile) && (time() - filemtime($cacheXmlFile) < $cacheSeconds);

if($forceRefresh || !$cacheIsFresh){
list($xml, $status) = hai_rebuild_sitemap(
    $siteUrl,
    $rootDir,
    $allowedExtensions,
    $excludeDirs,
    $excludeFiles,
    $excludeNameContains,
    $cacheXmlFile,
    $cacheJsonFile
);
}else{
    $xml = file_get_contents($cacheXmlFile);
    $status = file_exists($cacheJsonFile) ? json_decode(file_get_contents($cacheJsonFile), true) : [];
}

if(isset($_GET["format"]) && $_GET["format"] === "json"){
    header("Content-Type: application/json; charset=utf-8");
    echo json_encode($status);
    exit;
}

header("Content-Type: application/xml; charset=utf-8");
echo $xml;
?>