<?php
/**
 * Trait AIJE_AI_Visits
 * Erkennt Besuche von KI-Crawlern (GPTBot, ClaudeBot, PerplexityBot …) anhand
 * des User-Agents und protokolliert, welche KI welche Seite abgerufen hat.
 *
 * Zusätzlich (ab 1.5.1): echte Besucher, die über einen Link in einer
 * KI-Antwort (ChatGPT, Perplexity, Gemini …) kommen – purpose = 'referral'.
 *
 * Datenschutz: KEINE IP-Adressen, KEINE User-Agents, KEINE Cookies. Von
 * Besuchern aus KI-Chats werden nur Quelle, Seite und Zeitpunkt gespeichert;
 * alle übrigen Besucher werden nie erfasst.
 */

if ( ! defined( 'ABSPATH' ) ) { exit; }

trait AIJE_AI_Visits {

    /** Schema-Version der Log-Tabelle – bei Strukturänderungen erhöhen. */
    private static $ai_visits_db_version = '1';

    /* ===== Tabellenname ===== */

    public static function ai_visits_table() {
        global $wpdb;
        return $wpdb->prefix . 'aije_ai_visits';
    }

    /* ===== Bekannte KI-Crawler =====
     * Schlüssel = Teilstring im User-Agent (case-insensitive).
     * Spezifischere Muster stehen VOR allgemeineren, da der erste Treffer zählt.
     * purpose: training = sammelt Trainingsdaten, search = KI-Suchindex,
     *          user = Live-Abruf, weil ein Nutzer im Chat gerade danach fragt.
     */
    public static function ai_bot_list() {
        return array(
            // OpenAI
            'OAI-SearchBot'     => array( 'name' => 'OAI-SearchBot',     'vendor' => 'OpenAI',        'purpose' => 'search' ),
            'ChatGPT-User'      => array( 'name' => 'ChatGPT-User',      'vendor' => 'OpenAI',        'purpose' => 'user' ),
            'GPTBot'            => array( 'name' => 'GPTBot',            'vendor' => 'OpenAI',        'purpose' => 'training' ),
            // Anthropic
            'Claude-SearchBot'  => array( 'name' => 'Claude-SearchBot',  'vendor' => 'Anthropic',     'purpose' => 'search' ),
            'Claude-User'       => array( 'name' => 'Claude-User',       'vendor' => 'Anthropic',     'purpose' => 'user' ),
            'ClaudeBot'         => array( 'name' => 'ClaudeBot',         'vendor' => 'Anthropic',     'purpose' => 'training' ),
            'Claude-Web'        => array( 'name' => 'Claude-Web',        'vendor' => 'Anthropic',     'purpose' => 'user' ),
            'anthropic-ai'      => array( 'name' => 'anthropic-ai',      'vendor' => 'Anthropic',     'purpose' => 'training' ),
            // Perplexity
            'Perplexity-User'   => array( 'name' => 'Perplexity-User',   'vendor' => 'Perplexity',    'purpose' => 'user' ),
            'PerplexityBot'     => array( 'name' => 'PerplexityBot',     'vendor' => 'Perplexity',    'purpose' => 'search' ),
            // Google (Gemini / Vertex AI)
            'Google-CloudVertexBot' => array( 'name' => 'Google-CloudVertexBot', 'vendor' => 'Google', 'purpose' => 'user' ),
            'GoogleAgent'       => array( 'name' => 'Google Agent',      'vendor' => 'Google',        'purpose' => 'user' ),
            // Meta
            'meta-externalfetcher' => array( 'name' => 'Meta-ExternalFetcher', 'vendor' => 'Meta',    'purpose' => 'user' ),
            'meta-externalagent'   => array( 'name' => 'Meta-ExternalAgent',   'vendor' => 'Meta',    'purpose' => 'training' ),
            // Microsoft / Apple / Amazon
            'Applebot'          => array( 'name' => 'Applebot',          'vendor' => 'Apple',         'purpose' => 'search' ),
            'Amazonbot'         => array( 'name' => 'Amazonbot',         'vendor' => 'Amazon',        'purpose' => 'search' ),
            // Weitere
            'MistralAI-User'    => array( 'name' => 'MistralAI-User',    'vendor' => 'Mistral',       'purpose' => 'user' ),
            'DuckAssistBot'     => array( 'name' => 'DuckAssistBot',     'vendor' => 'DuckDuckGo',    'purpose' => 'user' ),
            'YouBot'            => array( 'name' => 'YouBot',            'vendor' => 'You.com',       'purpose' => 'search' ),
            'cohere-training-data-crawler' => array( 'name' => 'Cohere Training', 'vendor' => 'Cohere', 'purpose' => 'training' ),
            'cohere-ai'         => array( 'name' => 'cohere-ai',         'vendor' => 'Cohere',        'purpose' => 'user' ),
            'Bytespider'        => array( 'name' => 'Bytespider',        'vendor' => 'ByteDance',     'purpose' => 'training' ),
            'CCBot'             => array( 'name' => 'CCBot',             'vendor' => 'Common Crawl',  'purpose' => 'training' ),
            'AI2Bot'            => array( 'name' => 'AI2Bot',            'vendor' => 'Allen AI',      'purpose' => 'training' ),
            'Diffbot'           => array( 'name' => 'Diffbot',           'vendor' => 'Diffbot',       'purpose' => 'training' ),
            'Timpibot'          => array( 'name' => 'Timpibot',          'vendor' => 'Timpi',         'purpose' => 'training' ),
            'omgili'            => array( 'name' => 'Omgili',            'vendor' => 'Webz.io',       'purpose' => 'training' ),
        );
    }

    /* ===== KI-Chats als Traffic-Quelle =====
     * Erkennung echter Besucher, die in einer KI-Antwort auf einen Link geklickt
     * haben – über den Referrer (Host) oder den utm_source-Parameter, den z. B.
     * ChatGPT automatisch anhängt. Gespeichert werden nur Quelle, Seite, Zeit.
     */
    public static function ai_referral_sources() {
        return array(
            // host => [name, vendor, utm-Stichwort]
            'chatgpt.com'             => array( 'ChatGPT',    'OpenAI',     'chatgpt' ),
            'chat.openai.com'         => array( 'ChatGPT',    'OpenAI',     'openai' ),
            'perplexity.ai'           => array( 'Perplexity', 'Perplexity', 'perplexity' ),
            'gemini.google.com'       => array( 'Gemini',     'Google',     'gemini' ),
            'bard.google.com'         => array( 'Gemini',     'Google',     'bard' ),
            'claude.ai'               => array( 'Claude',     'Anthropic',  'claude' ),
            'copilot.microsoft.com'   => array( 'Copilot',    'Microsoft',  'copilot' ),
            'copilot.cloud.microsoft' => array( 'Copilot',    'Microsoft',  'copilot' ),
            'chat.mistral.ai'         => array( 'Le Chat',    'Mistral',    'mistral' ),
            'chat.deepseek.com'       => array( 'DeepSeek',   'DeepSeek',   'deepseek' ),
            'grok.com'                => array( 'Grok',       'xAI',        'grok' ),
            'meta.ai'                 => array( 'Meta AI',    'Meta',       'meta.ai' ),
            'you.com'                 => array( 'You.com',    'You.com',    'you.com' ),
            'phind.com'               => array( 'Phind',      'Phind',      'phind' ),
        );
    }

    public static function detect_ai_referral() {
        $sources = self::ai_referral_sources();

        // 1) utm_source (z. B. ?utm_source=chatgpt.com)
        if ( ! empty( $_GET['utm_source'] ) ) { // phpcs:ignore WordPress.Security.NonceVerification -- nur lesend
            $utm = strtolower( sanitize_text_field( wp_unslash( $_GET['utm_source'] ) ) ); // phpcs:ignore WordPress.Security.NonceVerification
            foreach ( $sources as $s ) {
                if ( false !== strpos( $utm, $s[2] ) ) {
                    return array( 'name' => $s[0], 'vendor' => $s[1] );
                }
            }
        }

        // 2) Referrer-Host
        if ( ! empty( $_SERVER['HTTP_REFERER'] ) ) {
            $host = strtolower( (string) wp_parse_url( esc_url_raw( wp_unslash( $_SERVER['HTTP_REFERER'] ) ), PHP_URL_HOST ) );
            if ( '' !== $host ) {
                foreach ( $sources as $domain => $s ) {
                    if ( $host === $domain || substr( $host, -strlen( '.' . $domain ) ) === '.' . $domain ) {
                        return array( 'name' => $s[0], 'vendor' => $s[1] );
                    }
                }
            }
        }
        return null;
    }

    public static function ai_purpose_labels() {
        return array(
            'training' => __( 'Training',          'ai-json-extractor-auto-embed-lite' ),
            'search'   => __( 'KI-Suche',          'ai-json-extractor-auto-embed-lite' ),
            'user'     => __( 'Live-Abruf (Chat)', 'ai-json-extractor-auto-embed-lite' ),
        );
    }

    /** Liefert den Bot-Eintrag für einen User-Agent oder null. */
    public static function detect_ai_bot( $ua ) {
        if ( '' === $ua ) { return null; }
        // Exakte User-Agents (Teilstring-Suche wäre hier zu breit).
        // Gemini ruft Seiten live mit dem nackten UA "Google" ab – per Debug-Modus
        // am 23.09.2026 auf wprankify.de nachgewiesen. Nur exakt, damit Googlebot
        // & Co. ("Mozilla/5.0 (compatible; Googlebot/2.1 …)") NICHT mitgezählt werden.
        $exact = array(
            'Google' => array( 'name' => 'Gemini', 'vendor' => 'Google', 'purpose' => 'user' ),
        );
        $ua_trim = trim( $ua );
        if ( isset( $exact[ $ua_trim ] ) ) {
            return $exact[ $ua_trim ];
        }
        foreach ( self::ai_bot_list() as $needle => $bot ) {
            if ( false !== stripos( $ua, $needle ) ) {
                return $bot;
            }
        }
        return null;
    }

    /* ===== Tabelle anlegen / aktualisieren =====
     * Läuft auf admin_init – so bekommen auch Bestandsinstallationen die Tabelle,
     * die per Update (ohne erneute Aktivierung) auf diese Version kommen.
     */
    public function maybe_install_ai_visits_table() {
        if ( get_option( 'aije_ai_visits_db' ) === self::$ai_visits_db_version ) { return; }
        global $wpdb;
        require_once ABSPATH . 'wp-admin/includes/upgrade.php';
        $table   = self::ai_visits_table();
        $charset = $wpdb->get_charset_collate();
        $sql = "CREATE TABLE {$table} (
            id BIGINT(20) UNSIGNED NOT NULL AUTO_INCREMENT,
            visited_at DATETIME NOT NULL,
            bot VARCHAR(64) NOT NULL,
            vendor VARCHAR(64) NOT NULL,
            purpose VARCHAR(16) NOT NULL,
            post_id BIGINT(20) UNSIGNED NOT NULL DEFAULT 0,
            url VARCHAR(500) NOT NULL,
            status SMALLINT(5) UNSIGNED NOT NULL DEFAULT 200,
            PRIMARY KEY  (id),
            KEY visited_at (visited_at),
            KEY bot (bot),
            KEY post_id (post_id)
        ) {$charset};";
        dbDelta( $sql );
        update_option( 'aije_ai_visits_db', self::$ai_visits_db_version, false );
    }

    /* ===== Besuch protokollieren (Frontend, template_redirect Prio 0) ===== */

    public function log_ai_visit() {
        if ( is_admin() || wp_doing_ajax() || wp_doing_cron() ) { return; }
        if ( ! get_option( 'aije_ai_visits_enabled', 1 ) ) { return; }
        if ( get_option( 'aije_ai_visits_db' ) !== self::$ai_visits_db_version ) { return; }

        $ua  = isset( $_SERVER['HTTP_USER_AGENT'] ) ? sanitize_text_field( wp_unslash( $_SERVER['HTTP_USER_AGENT'] ) ) : '';
        $bot = self::detect_ai_bot( $ua );

        $uri  = isset( $_SERVER['REQUEST_URI'] ) ? esc_url_raw( wp_unslash( $_SERVER['REQUEST_URI'] ) ) : '/';
        $path = substr( (string) wp_parse_url( $uri, PHP_URL_PATH ), 0, 500 );
        if ( '' === $path ) { $path = '/'; }

        // Debug-Modus: für begrenzte Zeit JEDEN User-Agent mitschreiben.
        if ( self::ai_debug_active() && ! is_user_logged_in() ) {
            $this->log_ai_debug( $ua, $path, $bot );
        }

        // Kein Crawler → prüfen, ob ein echter Besucher aus einem KI-Chat kommt.
        $referral = null;
        if ( ! $bot ) {
            if ( is_user_logged_in() ) { return; }
            $referral = self::detect_ai_referral();
            if ( ! $referral ) { return; }
            $bot = array( 'name' => $referral['name'], 'vendor' => $referral['vendor'], 'purpose' => 'referral' );
        }

        $post_id = is_singular() ? (int) get_queried_object_id() : 0;
        if ( ! $post_id && ( is_front_page() || is_home() ) ) {
            $post_id = (int) get_option( 'page_on_front' );
        }

        global $wpdb;
        $wpdb->insert(
            self::ai_visits_table(),
            array(
                'visited_at' => current_time( 'mysql' ),
                'bot'        => $bot['name'],
                'vendor'     => $bot['vendor'],
                'purpose'    => $bot['purpose'],
                'post_id'    => $post_id,
                'url'        => $path,
                'status'     => is_404() ? 404 : 200,
            ),
            array( '%s', '%s', '%s', '%s', '%d', '%s', '%d' )
        );

        // Aufräumen ohne eigenen Cron: bei ca. jedem 50. Bot-Besuch alte Einträge löschen.
        if ( 1 === wp_rand( 1, 50 ) ) {
            $this->purge_old_ai_visits();
        }
    }

    public function purge_old_ai_visits() {
        global $wpdb;
        $days = max( 7, (int) get_option( 'aije_ai_visits_retention', 90 ) );
        $wpdb->query( $wpdb->prepare(
            'DELETE FROM ' . self::ai_visits_table() . ' WHERE visited_at < %s',
            gmdate( 'Y-m-d H:i:s', current_time( 'timestamp' ) - $days * DAY_IN_SECONDS )
        ) );
    }

    /* ===== Debug-Modus =====
     * Schreibt für 15 Minuten den User-Agent JEDES nicht eingeloggten Aufrufs mit,
     * um unbekannte KI-Abrufer (z. B. Gemini) zu identifizieren. Keine IPs.
     * Speicherung als nicht-autoloadete Option, max. 300 Einträge.
     */
    const AI_DEBUG_MINUTES = 15;
    const AI_DEBUG_MAX     = 300;

    public static function ai_debug_active() {
        return (int) get_option( 'aije_ai_debug_until', 0 ) > time();
    }

    private function log_ai_debug( $ua, $path, $bot ) {
        $log = get_option( 'aije_ai_debug_log', array() );
        if ( ! is_array( $log ) ) { $log = array(); }
        if ( count( $log ) >= self::AI_DEBUG_MAX ) { return; }
        $log[] = array(
            't'    => current_time( 'mysql' ),
            'ua'   => substr( $ua, 0, 400 ),
            'path' => $path,
            'bot'  => $bot ? $bot['name'] : '',
        );
        update_option( 'aije_ai_debug_log', $log, false );
    }

    /* ===== Admin-Aktionen (Einstellungen speichern / Log leeren / Debug) ===== */

    public function handle_ai_visits_settings() {
        if ( ! current_user_can( 'manage_options' ) ) { wp_die( esc_html__( 'Keine Berechtigung.', 'ai-json-extractor-auto-embed-lite' ) ); }
        if ( ! isset( $_POST['aije_ai_visits_nonce'] ) || ! wp_verify_nonce( sanitize_text_field( wp_unslash( $_POST['aije_ai_visits_nonce'] ) ), 'aije_ai_visits' ) ) {
            wp_die( esc_html__( 'Sicherheitsprüfung fehlgeschlagen.', 'ai-json-extractor-auto-embed-lite' ) );
        }
        $msg   = 'saved';
        $debug = isset( $_POST['aije_debug'] ) ? sanitize_key( $_POST['aije_debug'] ) : '';
        if ( 'start' === $debug ) {
            delete_option( 'aije_ai_debug_log' );
            update_option( 'aije_ai_debug_until', time() + self::AI_DEBUG_MINUTES * MINUTE_IN_SECONDS, false );
            $msg = 'debug_on';
        } elseif ( 'stop' === $debug ) {
            update_option( 'aije_ai_debug_until', 0, false );
            $msg = 'debug_off';
        } elseif ( 'clear' === $debug ) {
            delete_option( 'aije_ai_debug_log' );
            update_option( 'aije_ai_debug_until', 0, false );
            $msg = 'debug_cleared';
        } elseif ( ! empty( $_POST['aije_clear_log'] ) ) {
            global $wpdb;
            $wpdb->query( 'TRUNCATE TABLE ' . self::ai_visits_table() );
            $msg = 'cleared';
        } else {
            update_option( 'aije_ai_visits_enabled', ! empty( $_POST['ai_visits_enabled'] ) ? 1 : 0 );
            $ret = isset( $_POST['ai_visits_retention'] ) ? (int) $_POST['ai_visits_retention'] : 90;
            update_option( 'aije_ai_visits_retention', max( 7, min( 365, $ret ) ) );
        }
        wp_safe_redirect( add_query_arg( array( 'page' => 'aije-lite', 'tab' => 'aivisits', 'aije_msg' => $msg ), admin_url( 'tools.php' ) ) );
        exit;
    }
}