CodePress/cms/core/class/RequestLogger.php

<?php

class RequestLogger
{
    private string $logFile;

    public function __construct(string $logFile)
    {
        $this->logFile = $logFile;
    }

    public function log(string $page, string $ip, string $userAgent, string $referrer, string $host, string $acceptLanguage): void
    {
        $dir = dirname($this->logFile);
        if (!is_dir($dir)) {
            @mkdir($dir, 0755, true);
        }

        $timestamp = date('Y-m-d H:i:s');
        $ua = substr(preg_replace('/[[:cntrl:]]/', '', $userAgent), 0, 500);
        $ref = substr(preg_replace('/[[:cntrl:]]/', '', $referrer), 0, 500);
        $line = "[{$timestamp}] [{$ip}] [{$host}] [{$acceptLanguage}] [{$page}] [{$ua}] [{$ref}]\n";
        @file_put_contents($this->logFile, $line, FILE_APPEND | LOCK_EX);
    }

    public static function getClientIp(): string
    {
        $headers = [
            'HTTP_CF_CONNECTING_IP',
            'HTTP_X_REAL_IP',
            'HTTP_CLIENT_IP',
            'HTTP_X_FORWARDED_FOR',
            'REMOTE_ADDR',
        ];

        foreach ($headers as $header) {
            if (!empty($_SERVER[$header])) {
                $rawIp = $_SERVER[$header];
                if (str_contains($rawIp, ',')) {
                    $ips = explode(',', $rawIp);
                    $rawIp = trim($ips[0]);
                }
                $rawIp = trim($rawIp);
                if (filter_var($rawIp, FILTER_VALIDATE_IP) !== false) {
                    return $rawIp;
                }
            }
        }

        return $_SERVER['REMOTE_ADDR'] ?? 'cli';
    }

    public static function detectVisitorInfo(string $ua, string $user = ''): array
    {
        if (empty($ua)) {
            $userLabel = ($user && $user !== 'Gast') ? $user : 'Onbekend';
            return ['type' => 'unknown', 'label' => $userLabel, 'badge' => 'secondary', 'icon' => 'bi-question-circle'];
        }

        $bots = [
            'AI' => [
                'GPTBot' => 'AI (GPTBot)',
                'ChatGPT-User' => 'AI (ChatGPT)',
                'Claude-Web' => 'AI (Claude)',
                'ClaudeBot' => 'AI (ClaudeBot)',
                'anthropic-ai' => 'AI (Anthropic)',
                'Google-Extended' => 'AI (Gemini/Google)',
                'CCBot' => 'AI (CommonCrawl)',
                'PerplexityBot' => 'AI (Perplexity)',
                'Amazonbot' => 'AI (Amazon)',
                'cohere-ai' => 'AI (Cohere)',
                'OAI-SearchBot' => 'AI (OpenAI)',
                'Bytespider' => 'AI (ByteDance)',
                'FacebookBot' => 'AI (Meta/FB)',
                'Applebot-Extended' => 'AI (Apple)',
            ],
            'Search' => [
                'Googlebot' => 'Zoekmachine (Google)',
                'Bingbot' => 'Zoekmachine (Bing)',
                'BingPreview' => 'Zoekmachine (Bing)',
                'Slurp' => 'Zoekmachine (Yahoo)',
                'DuckDuckBot' => 'Zoekmachine (DuckDuckGo)',
                'Baiduspider' => 'Zoekmachine (Baidu)',
                'YandexBot' => 'Zoekmachine (Yandex)',
                'Sogou' => 'Zoekmachine (Sogou)',
                'Exabot' => 'Zoekmachine (Exabot)',
                'facebot' => 'Zoekmachine (Facebook)',
            ],
            'Scraper' => [
                'HTTrack' => 'Scraper (HTTrack)',
                'Scrapy' => 'Scraper (Scrapy)',
                'PhantomJS' => 'Scraper (PhantomJS)',
                'HeadlessChrome' => 'Scraper (Headless)',
                'curl' => 'Scraper (cURL)',
                'wget' => 'Scraper (Wget)',
                'python' => 'Scraper (Python)',
                'Postman' => 'Scraper (Postman)',
            ],
        ];

        foreach ($bots['AI'] as $pattern => $label) {
            if (stripos($ua, $pattern) !== false) {
                return ['type' => 'ai', 'label' => $label, 'badge' => 'danger', 'icon' => 'bi-robot'];
            }
        }

        foreach ($bots['Search'] as $pattern => $label) {
            if (stripos($ua, $pattern) !== false) {
                return ['type' => 'search', 'label' => $label, 'badge' => 'primary', 'icon' => 'bi-search'];
            }
        }

        foreach ($bots['Scraper'] as $pattern => $label) {
            if (stripos($ua, $pattern) !== false) {
                return ['type' => 'scraper', 'label' => $label, 'badge' => 'warning text-dark', 'icon' => 'bi-bug'];
            }
        }

        if (preg_match('/(bot|crawler|spider|slurp)/i', $ua)) {
            return ['type' => 'bot', 'label' => 'Bot', 'badge' => 'secondary', 'icon' => 'bi-robot'];
        }

        $userPrefix = ($user && $user !== 'Gast') ? htmlspecialchars($user) . ' (' : '';
        $userSuffix = ($user && $user !== 'Gast') ? ')' : '';

        return [
            'type' => 'human',
            'label' => $userPrefix . 'Mens' . $userSuffix,
            'badge' => 'success',
            'icon' => 'bi-person-check',
        ];
    }

    public static function detectBot(): ?string
    {
        $ua = $_SERVER['HTTP_USER_AGENT'] ?? '';
        if (empty($ua)) return null;

        $bots = [
            'AI' => [
                'GPTBot', 'ChatGPT-User', 'Claude-Web', 'ClaudeBot',
                'anthropic-ai', 'Google-Extended', 'CCBot',
                'PerplexityBot', 'Amazonbot', 'cohere-ai',
                'OAI-SearchBot', 'Bytespider', 'FacebookBot',
                'Applebot-Extended',
            ],
            'Search' => [
                'Googlebot', 'Bingbot', 'BingPreview', 'Slurp',
                'DuckDuckBot', 'Baiduspider', 'YandexBot',
                'Sogou', 'Exabot', 'facebot',
            ],
            'Scraper' => [
                'HTTrack', 'Scrapy', 'PhantomJS', 'HeadlessChrome',
            ],
        ];

        foreach ($bots as $category => $patterns) {
            foreach ($patterns as $pattern) {
                if (stripos($ua, $pattern) !== false) {
                    return $category;
                }
            }
        }

        return null;
    }

    public function getLogs(int $lines = 100): array
    {
        if (!file_exists($this->logFile)) {
            return [];
        }

        $content = file($this->logFile);
        $content = array_slice($content, -$lines);
        $logs = [];

        foreach ($content as $line) {
            if (preg_match('/^\[([^\]]+)\] \[([^\]]+)\] \[([^\]]+)\] \[([^\]]*)\] \[([^\]]+)\] \[([^\]]*)\] \[([^\]]*)\]$/', trim($line), $m)) {
                $user = $m[3];
                // Handle backwards compatibility where host (e.g. localhost:8080 or domain.com) was logged
                if (str_contains($user, '.') || str_contains($user, ':') || $user === 'cli') {
                    $user = 'Gast';
                }
                $visitorInfo = self::detectVisitorInfo($m[6], $user);
                $logs[] = [
                    'time' => $m[1],
                    'ip' => $m[2],
                    'user' => $user,
                    'visitor_info' => $visitorInfo,
                    'lang' => $m[4],
                    'page' => $m[5],
                    'ua' => $m[6],
                    'referrer' => $m[7],
                ];
            }
        }

        return array_reverse($logs);
    }
}