<?php
// Citami Connector — module. Loaded by citami-connector.php in a fixed
// order; never load this file on its own.
if (!defined('ABSPATH')) {
    exit;
}

/* ────────────────────────────────────────────────────────────────── */
/*  PART B — AI bot visit tracking                                     */
/* ────────────────────────────────────────────────────────────────── */

/**
 * Lightweight UA matcher. Returns the canonical bot slug or null.
 * This is a *heuristic* — Citami re-classifies server-side using its
 * own pattern list (trust-but-verify).
 */
function citami_detect_ai_bot($ua) {
    if (!is_string($ua) || $ua === '') {
        return null;
    }
    $patterns = array(
        '/GPTBot/i'              => 'gptbot',
        '/ChatGPT-User/i'        => 'chatgpt-user',
        '/OAI-SearchBot/i'       => 'oai-searchbot',
        '/Google-Extended/i'     => 'google-extended',
        '/PerplexityBot/i'       => 'perplexitybot',
        '/Perplexity-User/i'     => 'perplexity-user',
        // 2.53 — user-triggered fetchers and AI search crawlers that were
        // never forwarded: their reads mean "used to answer someone".
        '/Claude-User/i'         => 'claude-user',
        '/Claude-SearchBot/i'    => 'claude-searchbot',
        '/MistralAI-User/i'      => 'mistralai-user',
        '/DuckAssistBot/i'       => 'duckassistbot',
        '/ClaudeBot|anthropic-ai/i' => 'claudebot',
        '/Claude-Web/i'          => 'claude-web',
        '/Applebot-Extended/i'   => 'applebot-extended',
        '/Bytespider/i'          => 'bytespider',
        '/CCBot/i'               => 'ccbot',
        '/Diffbot/i'             => 'diffbot',
        '/Amazonbot/i'           => 'amazonbot',
        '/FacebookBot|Meta-ExternalAgent/i' => 'facebookbot',
        '/YouBot/i'              => 'youbot',
        '/cohere-ai/i'           => 'cohere-ai',
    );
    foreach ($patterns as $re => $name) {
        if (preg_match($re, $ua)) {
            return $name;
        }
    }
    return null;
}

/**
 * v2.3.0 — HUMAN-REFERRAL. Detect clicks coming FROM an AI assistant:
 * a real person asked ChatGPT/Perplexity/Claude, got this site cited,
 * and clicked through. This is the ROI-proof signal (crawler hits tell
 * you the AI is *reading* the site; referral hits tell you it's
 * *sending customers*). Heuristic only — Citami re-verifies the
 * referrer server-side (trust-but-verify, same as bot UAs).
 * Returns the platform slug or null.
 */
function citami_detect_ai_referrer($referrer) {
    if (!is_string($referrer) || $referrer === '') {
        return null;
    }
    $host = wp_parse_url($referrer, PHP_URL_HOST);
    if (!is_string($host) || $host === '') {
        return null;
    }
    $patterns = array(
        '/(^|\.)chatgpt\.com$/i'           => 'chatgpt',
        '/(^|\.)chat\.openai\.com$/i'     => 'chatgpt',
        '/(^|\.)openai\.com$/i'            => 'chatgpt',
        '/(^|\.)perplexity\.ai$/i'         => 'perplexity',
        '/(^|\.)claude\.ai$/i'             => 'claude',
        '/(^|\.)gemini\.google\.com$/i'   => 'gemini',
        '/(^|\.)bard\.google\.com$/i'     => 'gemini',
        '/(^|\.)copilot\.microsoft\.com$/i' => 'copilot',
        '/(^|\.)you\.com$/i'               => 'you',
    );
    foreach ($patterns as $re => $name) {
        if (preg_match($re, $host)) {
            return $name;
        }
    }
    return null;
}

/**
 * Determine the high-level page type at request time. Used by Citami to
 * group bot traffic by intent (homepage vs single post vs search vs 404)
 * without having to re-derive from path patterns server-side.
 *
 * Output domain (pinned, mirrored in /api/bot-hits route validator):
 *   home | post | page | category | tag | archive | search | 404 |
 *   singular | other
 *
 * 'singular' covers custom post types (Product, Recipe, etc.); 'other'
 * is the catch-all for anything we don't recognize so the server side
 * never sees an unbounded string.
 */
function citami_detect_page_type() {
    if (is_404())             return '404';
    if (is_search())          return 'search';
    if (is_front_page() || is_home()) return 'home';
    if (is_category())        return 'category';
    if (is_tag())             return 'tag';
    if (is_singular('post'))  return 'post';
    if (is_singular('page'))  return 'page';
    if (is_singular())        return 'singular';
    if (is_archive())         return 'archive';
    return 'other';
}

/**
 * Bot tracking — runs on `template_redirect` (after WP routing, before
 * render) so is_404() and the other conditional tags are decided.
 *
 * v2.0.x ran on `init` and hardcoded statusCode=200 — every 404 served
 * to GPTBot was reported as a successful fetch. v2.1.0 captures the
 * real status (200 / 404) plus the page type for downstream analysis.
 */
/**
 * 2.64.1 — the hit is recorded at the very end of template_redirect, and
 * redirects are recorded with their own status. Before, it ran at the
 * default priority: a request that WordPress or an SEO plugin then
 * redirected (Rank Math sends /sitemap.xml to /sitemap_index.xml) was
 * logged as a 404, and the platform showed crawler errors that the crawler
 * never got.
 *
 * @param int $redirect_status 3xx status when called from wp_redirect, 0 otherwise.
 */
function citami_bot_track_request($redirect_status = 0) {
    // Skip admin, cron, REST, AJAX — we only care about public-facing
    // pages that a crawler would hit.
    if (is_admin() || wp_doing_cron() || wp_doing_ajax()) return;
    if (defined('REST_REQUEST') && REST_REQUEST) return;

    $ua = isset($_SERVER['HTTP_USER_AGENT']) ? sanitize_text_field(wp_unslash($_SERVER['HTTP_USER_AGENT'])) : '';
    $bot_name = citami_detect_ai_bot($ua);

    // v2.3.0 — HUMAN-REFERRAL: se l'UA non è un bot, controlla se il
    // visitatore arriva cliccando da un assistente AI (referrer).
    $referrer = isset($_SERVER['HTTP_REFERER']) ? esc_url_raw(wp_unslash($_SERVER['HTTP_REFERER'])) : '';
    $ai_platform = ($bot_name === null) ? citami_detect_ai_referrer($referrer) : null;

    if ($bot_name === null && $ai_platform === null) return;

    $uri  = isset($_SERVER['REQUEST_URI']) ? esc_url_raw(wp_unslash($_SERVER['REQUEST_URI'])) : '/';
    $path = wp_parse_url($uri, PHP_URL_PATH);
    if (!is_string($path) || $path === '') $path = '/';

    if ($redirect_status > 0) {
        // Redirected before any template: no page type to speak of.
        $page_type = 'other';
        $status_code = (int) $redirect_status;
        $GLOBALS['citami_bot_hit_logged'] = true;
    } else {
        $page_type = citami_detect_page_type();
        // Map page type to canonical HTTP status. 200 for everything that
        // resolved to a real page; 404 when WP routed the request to its
        // 404 handler. We don't try to infer 410/503 here — those would
        // require post-render hooks we don't have.
        $status_code = ($page_type === '404') ? 404 : 200;
    }

    if ($bot_name !== null) {
        $hit = array(
            'botName'     => $bot_name,
            'trafficType' => 'crawler',
            'path'        => substr($path, 0, 500),
            'userAgent'   => substr($ua, 0, 500),
            'statusCode'  => $status_code,
            'pageType'    => $page_type,
            'hitAt'       => gmdate('c'),
        );
    } else {
        $hit = array(
            'botName'     => $ai_platform,
            'trafficType' => 'human_referral',
            'referrer'    => substr($referrer, 0, 1000),
            'path'        => substr($path, 0, 500),
            'userAgent'   => substr($ua, 0, 500),
            'statusCode'  => $status_code,
            'pageType'    => $page_type,
            'hitAt'       => gmdate('c'),
        );
    }

    citami_push_buffer($hit);
    // WP5 — the local verified log listens here (crawler hits only).
    do_action('citami_bot_hit', $hit, function_exists('citami_bots_client_ip') ? citami_bots_client_ip() : (isset($_SERVER['REMOTE_ADDR']) ? sanitize_text_field(wp_unslash($_SERVER['REMOTE_ADDR'])) : ''));
}

add_action('template_redirect', function () {
    if (!empty($GLOBALS['citami_bot_hit_logged'])) return;
    citami_bot_track_request();
}, PHP_INT_MAX);

// A redirect issued before the end of template_redirect (canonical,
// SEO plugin, our own rules) never reaches the hook above.
add_filter('wp_redirect_status', function ($status) {
    if (empty($GLOBALS['citami_bot_hit_logged']) && (int) $status >= 300 && (int) $status < 400) {
        citami_bot_track_request((int) $status);
    }
    return $status;
}, PHP_INT_MAX);

function citami_push_buffer($hit) {
    $buffer = get_option(CITAMI_BUFFER_OPT, array());
    if (!is_array($buffer)) {
        $buffer = array();
    }
    $buffer[] = $hit;

    // Hard cap: drop oldest so we don't blow up the options table if
    // Citami is unreachable for hours.
    if (count($buffer) > CITAMI_BUFFER_HARD_CAP) {
        $buffer = array_slice($buffer, -CITAMI_BUFFER_HARD_CAP);
    }

    update_option(CITAMI_BUFFER_OPT, $buffer, false);

    // Opportunistic flush when the soft cap is hit — don't wait 5min
    // on a high-traffic site.
    if (count($buffer) >= CITAMI_BUFFER_SOFT_CAP) {
        // Schedule a single-shot flush soon rather than running it
        // inline (keeps the page response fast for the bot).
        if (!wp_next_scheduled('citami_flush_bot_hits')) {
            wp_schedule_single_event(time() + 30, 'citami_flush_bot_hits');
        }
    }
}

/* Register a 5-minute cron interval. */
add_filter('cron_schedules', function ($schedules) {
    if (!isset($schedules['citami_5min'])) {
        $schedules['citami_5min'] = array(
            'interval' => 300,
            'display'  => citami_brand_text('Ogni 5 minuti (Citami)'),
        );
    }
    return $schedules;
});

add_action('citami_flush_bot_hits', 'citami_flush_bot_hits_handler');

function citami_flush_bot_hits_handler() {
    if (function_exists('citami_cron_heartbeat')) citami_cron_heartbeat(); // 2.46 — proof WP-Cron runs
    $key = citami_get_api_key();
    if (empty($key)) return;

    // v2.7.1 — HEALTH-HANDSHAKE. Keep-alive agganciato al cron 5-min
    // già esistente: senza, un sito senza traffico bot AI non
    // contatterebbe MAI il server e in dashboard resterebbe "Non
    // connesso" per sempre. Cadenza 25 min: il badge dashboard
    // considera "connected" un ping negli ultimi 30 min
    // (lib/plugin/status.ts CONNECTED_THRESHOLD_MS) — un keep-alive
    // più lasco farebbe oscillare il badge su "stale". Costo: ~2 GET
    // l'ora per sito, contro un rate limit server di 60/min per key.
    $health = get_option(CITAMI_HEALTH_OPT, array());
    $health_ts = is_array($health) && !empty($health['ts']) ? intval($health['ts']) : 0;
    if (time() - $health_ts > 25 * MINUTE_IN_SECONDS) {
        citami_health_ping();
    }

    $buffer = get_option(CITAMI_BUFFER_OPT, array());
    if (!is_array($buffer) || empty($buffer)) return;

    // Send up to 1000 per POST (matches the server's Zod cap). Keep
    // anything beyond that for the next run.
    $batch = array_slice($buffer, 0, 1000);
    $remainder = array_slice($buffer, 1000);

    $endpoint = citami_get_endpoint() . '/api/bot-hits';
    $response = wp_remote_post($endpoint, array(
        'timeout' => 10,
        'headers' => array(
            'Authorization' => 'Bearer ' . $key,
            'Content-Type'  => 'application/json',
            'User-Agent'    => 'CitamiConnector/' . CITAMI_PLUGIN_VERSION . '; ' . home_url('/'),
        ),
        'body'    => wp_json_encode(array('hits' => $batch)),
    ));

    if (is_wp_error($response)) {
        // Keep the buffer intact — we'll retry next cycle.
        citami_debug_log('[citami] bot-hits flush failed: ' . $response->get_error_message());
        return;
    }

    $code = wp_remote_retrieve_response_code($response);
    if ($code >= 200 && $code < 300) {
        // Success: drop the batch we just sent, keep the remainder.
        update_option(CITAMI_BUFFER_OPT, $remainder, false);
    } else {
        citami_debug_log('[citami] bot-hits flush rejected: HTTP ' . $code);
        // 4xx on auth/rate-limit? We still retry — the server's 429 is
        // idempotent, and a mis-typed key will just keep failing until
        // the admin fixes it. The hard cap protects us from unbounded
        // growth.
    }
}

// v2.10.0 — BOT-FLUSH-SHUTDOWN. Bug: i bot hit venivano bufferizzati ma
// inviati SOLO via WP-cron (citami_flush_bot_hits), e WP-cron parte solo
// quando un essere umano carica una pagina. Un sito che riceve SOLO
// traffico bot AI (zero visite umane) non svuotava mai il buffer → 0
// dati in dashboard (segnalato da Adrian: "i dati non arrivano").
//
// Fix: il bot hit È una richiesta HTTP, e l'hook `shutdown` gira a fine
// di OGNI richiesta front-end — quindi anche su una richiesta di un bot.
// Agganciamo qui un flush con throttle (per non fare una POST a ogni
// singolo hit): si invia quando il buffer raggiunge una piccola soglia
// (burst) oppure è passato CITAMI_BOT_FLUSH_MIN_INTERVAL dall'ultimo
// invio. Il cron 5-min resta come backstop + keep-alive health.
define('CITAMI_BOT_FLUSH_LAST_OPT', 'citami_bot_flush_last_ts');
define('CITAMI_BOT_FLUSH_MIN_INTERVAL', 120); // 2 min tra invii "a tempo"
define('CITAMI_BOT_FLUSH_BURST', 10);         // invia subito a 10 hit

add_action('shutdown', function () {
    // Solo richieste front-end pubbliche: admin/cron/ajax/rest hanno i
    // loro percorsi (e il cron chiama già l'handler direttamente).
    if (is_admin() || wp_doing_cron() || wp_doing_ajax()) return;
    if (defined('REST_REQUEST') && REST_REQUEST) return;

    $buffer = get_option(CITAMI_BUFFER_OPT, array());
    if (!is_array($buffer) || empty($buffer)) return;

    $last = (int) get_option(CITAMI_BOT_FLUSH_LAST_OPT, 0);
    $now  = time();
    $due  =
        count($buffer) >= CITAMI_BOT_FLUSH_BURST ||
        ($now - $last) >= CITAMI_BOT_FLUSH_MIN_INTERVAL;
    if (!$due) return;

    // Marca il tentativo PRIMA dell'invio: il throttle deve reggere anche
    // se la POST fallisce, così non martelliamo il server a ogni hit.
    update_option(CITAMI_BOT_FLUSH_LAST_OPT, $now, false);
    citami_flush_bot_hits_handler();
}, 99);

/* Activation/deactivation: schedule / unschedule the recurring flush. */
register_activation_hook(CITAMI_PLUGIN_FILE, function () {
    if (!wp_next_scheduled('citami_flush_bot_hits')) {
        wp_schedule_event(time() + 60, 'citami_5min', 'citami_flush_bot_hits');
    }
    // v2.7.1 — HEALTH-HANDSHAKE. Primo contatto col server subito
    // all'attivazione (la key preloadata c'è già negli ZIP scaricati
    // dalla dashboard): la dashboard mostra "Connesso" senza dover
    // aspettare la prima visita di un bot AI.
    citami_health_ping();
    // v2.6.0 — refresh giornaliero della sezione auto di llms.txt.
    if (!wp_next_scheduled('citami_refresh_llms_txt')) {
        wp_schedule_event(time() + 120, 'daily', 'citami_refresh_llms_txt');
    }
    // v2.7.0 — AUTO-DEFAULT. Il marketing promette "fix con un click":
    // con review_only ON di default servivano 3 approvazioni dentro
    // WP. Nuova regola: se il plugin e' stato scaricato dal dashboard
    // Citami (preload presente) il cliente HA GIA' espresso l'intent
    // nel download → review_only parte OFF. Un install manuale/non
    // configurato (nessun preload) resta ON, secure-by-default. Chi ha
    // gia' un valore salvato non viene mai toccato.
    if (get_option(CITAMI_REVIEW_ONLY_OPT, null) === null || get_option(CITAMI_REVIEW_ONLY_OPT) === false) {
        $preloaded = defined('CITAMI_PRELOADED_API_KEY')
            && CITAMI_PRELOADED_API_KEY !== ''
            && strpos(CITAMI_PRELOADED_API_KEY, '{{') === false;
        add_option(CITAMI_REVIEW_ONLY_OPT, $preloaded ? '0' : '1', '', false);
    }
});

register_deactivation_hook(CITAMI_PLUGIN_FILE, function () {
    $timestamp = wp_next_scheduled('citami_flush_bot_hits');
    if ($timestamp) {
        wp_unschedule_event($timestamp, 'citami_flush_bot_hits');
    }
    // v2.6.0 — spegni anche il refresh giornaliero di llms.txt.
    $llms_ts = wp_next_scheduled('citami_refresh_llms_txt');
    if ($llms_ts) {
        wp_unschedule_event($llms_ts, 'citami_refresh_llms_txt');
    }
});