<?php
/**
 * Business KPIs Class - CLEAN REWRITE
 *
 * Calcule les KPIs orientés décision pour les propriétaires de sites
 * Basé sur les recommandations "Action-driven" plutôt que "Data-driven"
 *
 * @package   almaweb_ai_visitor_analytics
 * @since     1.2.0
 * @author    Yohan Ziri (AlmaWeb)
 * @link      https://almaweb.fr
 */

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

class Almaweb_Business_KPIs {

    private $database;

    public function __construct() {
        $this->database = new Almaweb_Database();
    }

    /**
     * 1. JAUGE DE PRESSION AI (0-100%)
     */
    public function get_ai_pressure($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $total_visits = $wpdb->get_var($wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE visited_at >= %s", $date_limit)) ?? 0;
        $ai_visits = $wpdb->get_var($wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at >= %s", $date_limit)) ?? 0;

        if ($total_visits == 0) {
            return ['percentage' => 0, 'status' => 'unknown', 'message' => __('Insufficient data', 'almaweb-ai-visitor-analytics')];
        }

        $percentage = round(($ai_visits / $total_visits) * 100, 1);

        if ($percentage < 10) {
            $status = 'healthy';
            // translators: %s is the AI traffic percentage
            $message = sprintf(__('Healthy (%s%%)', 'almaweb-ai-visitor-analytics'), $percentage);
        } elseif ($percentage < 25) {
            $status = 'moderate';
            // translators: %s is the AI traffic percentage
            $message = sprintf(__('Moderate (%s%%)', 'almaweb-ai-visitor-analytics'), $percentage);
        } elseif ($percentage < 50) {
            $status = 'high';
            // translators: %s is the AI traffic percentage
            $message = sprintf(__('High (%s%%)', 'almaweb-ai-visitor-analytics'), $percentage);
        } else {
            $status = 'critical';
            // translators: %s is the AI traffic percentage
            $message = sprintf(__('Critical (%s%%)', 'almaweb-ai-visitor-analytics'), $percentage);
        }

        return ['percentage' => $percentage, 'status' => $status, 'message' => $message];
    }

    /**
     * Helper: Get intention counts for quality score calculation
     */
    private function get_intention_counts($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $results = $wpdb->get_results($wpdb->prepare(
            "SELECT bot_intention, COUNT(*) as count
            FROM {$table_name}
            WHERE is_ai_bot = 1 AND bot_type = 'ai_bot' AND visited_at >= %s
            GROUP BY bot_intention",
            $date_limit
        ), ARRAY_A);

        $counts = [
            'inference' => 0,
            'search' => 0,
            'training' => 0,
            'scraping' => 0
        ];

        foreach ($results as $row) {
            if (isset($counts[$row['bot_intention']])) {
                $counts[$row['bot_intention']] = (int) $row['count'];
            }
        }

        return $counts;
    }

    /**
     * 2. SCORE DE POPULARITÉ AI (AI Authority Score)
     */
    public function get_ai_authority_score($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $stats = $wpdb->get_row($wpdb->prepare(
            "SELECT
                COUNT(DISTINCT bot_family) as unique_families,
                COUNT(DISTINCT bot_name) as unique_bots,
                COUNT(*) as total_visits,
                COUNT(DISTINCT request_url) as unique_pages
            FROM {$table_name}
            WHERE is_ai_bot = 1 AND bot_family IS NOT NULL AND bot_family != 'Unknown' AND visited_at >= %s",
            $date_limit
        ), ARRAY_A);

        if (empty($stats) || $stats['total_visits'] == 0) {
            return ['score' => 0, 'grade' => 'N/A', 'status' => 'waiting', 'message' => __('Waiting for sufficient data...', 'almaweb-ai-visitor-analytics')];
        }

        // Get intention counts for quality calculation
        $intention_counts = $this->get_intention_counts($days);

        // 1. DIVERSITY SCORE (max 25 points) - More demanding
        // 1 family = 5pts, 3 = 15pts, 5+ = 25pts
        $diversity_score = min((($stats['unique_families'] ?? 0) * 5), 25);

        // 2. VOLUME SCORE (max 25 points) - More demanding
        // 100 visits = 14pts, 1000 = 21pts, 10000+ = 25pts
        $volume_score = min(log10(max($stats['total_visits'], 1)) * 7, 25);

        // 3. COVERAGE SCORE (max 10 points) - Reduced importance
        // 10 pages = 3.3pts, 100 = 6.6pts, 1000+ = 10pts
        $coverage_score = min(log10(max($stats['unique_pages'], 1)) * 3.33, 10);

        // 4. QUALITY SCORE (max 40 points) - NEW & MOST IMPORTANT
        // Weighted by intention value: inference (100%), search (70%), training (30%), scraping (0%)
        $inference = $intention_counts['inference'] ?? 0;
        $search = $intention_counts['search'] ?? 0;
        $training = $intention_counts['training'] ?? 0;
        $scraping = $intention_counts['scraping'] ?? 0;

        $total_ai_visits = $inference + $search + $training + $scraping;

        if ($total_ai_visits > 0) {
            $weighted_quality = (
                ($inference * 1.0) +    // 100% value - AI citing you in responses
                ($search * 0.7) +       // 70% value - AI indexing you as trusted source
                ($training * 0.3) +     // 30% value - AI collecting your data
                ($scraping * 0.0)       // 0% value - Basic crawling
            );

            $quality_ratio = $weighted_quality / $total_ai_visits;
            $quality_score = $quality_ratio * 40; // Scale to 0-40 points
        } else {
            $quality_score = 0;
        }

        $total_score = round($diversity_score + $volume_score + $coverage_score + $quality_score);

        if ($total_score >= 80) { $grade = 'A+'; $status = 'elite'; $message = __('Elite - Your site is a major reference for AIs', 'almaweb-ai-visitor-analytics');
        } elseif ($total_score >= 60) { $grade = 'A'; $status = 'excellent'; $message = __('Excellent - Reliable source for AI models', 'almaweb-ai-visitor-analytics');
        } elseif ($total_score >= 40) { $grade = 'B'; $status = 'good'; $message = __('Good - Significant presence in the AI ecosystem', 'almaweb-ai-visitor-analytics');
        } elseif ($total_score >= 20) { $grade = 'C'; $status = 'emerging'; $message = __('Emerging - Starting to be detected by AIs', 'almaweb-ai-visitor-analytics');
        } else { $grade = 'D'; $status = 'minimal'; $message = __('Minimal - Low AI visibility currently', 'almaweb-ai-visitor-analytics');
        }

        return ['score' => $total_score, 'grade' => $grade, 'status' => $status, 'message' => $message];
    }

    /**
     * 3. RATIO TRAINING VS INFERENCE (ROBUST)
     */
    public function get_training_vs_inference($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $intention_case_sql = "CASE
            WHEN bot_intention IS NOT NULL AND bot_intention != '' THEN bot_intention
            WHEN bot_type = 'search_engine' THEN 'search'
            WHEN bot_type = 'scraper' THEN 'scraping'
            ELSE 'training'
        END";

        $results = $wpdb->get_results($wpdb->prepare(
            "SELECT {$intention_case_sql} as intention_category, COUNT(*) as count
            FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at >= %s
            GROUP BY intention_category", $date_limit
        ), ARRAY_A);

        $data = ['training' => 0, 'inference' => 0, 'search' => 0, 'scraping' => 0, 'unknown' => 0];
        foreach ($results as $row) {
            if (isset($data[$row['intention_category']])) {
                $data[$row['intention_category']] = (int) $row['count'];
            } else {
                $data['unknown'] += (int) $row['count'];
            }
        }

        $total = array_sum($data);
        $percentages = [];
        foreach ($data as $key => $value) {
            $percentages[$key] = $total > 0 ? round(($value / $total) * 100, 1) : 0;
        }

        return [
            'counts' => $data,
            'percentages' => $percentages,
            'total' => $total,
            'labels' => [
                'training' => __('Model training', 'almaweb-ai-visitor-analytics'),
                'inference' => __('User responses', 'almaweb-ai-visitor-analytics'),
                'search' => __('Indexing & search', 'almaweb-ai-visitor-analytics'),
                'scraping' => __('Data collection', 'almaweb-ai-visitor-analytics'),
            ]
        ];
    }

    /**
     * 4. POTENTIEL DE CITATION (ROBUST)
     */
    public function get_citation_potential($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $intention_case_sql = "CASE WHEN bot_intention IS NOT NULL AND bot_intention != '' THEN bot_intention WHEN bot_type = 'search_engine' THEN 'search' ELSE 'other' END";

        $citations = $wpdb->get_var($wpdb->prepare(
            "SELECT COUNT(*) FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at >= %s AND ({$intention_case_sql}) = 'search'",
            $date_limit
        )) ?? 0;

        // translators: %s is the number of citation opportunities
        return ['total_opportunities' => $citations, 'message' => sprintf(__('%s citation opportunities', 'almaweb-ai-visitor-analytics'), number_format($citations))];
    }
    
    /**
     * 5. TOP SUJETS (ROBUST)
     */
    public function get_top_topics($limit = 10, $days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        // Static SQL CASE statement - safe from injection
        $intention_case_sql = "CASE WHEN bot_intention IS NOT NULL AND bot_intention != '' THEN bot_intention WHEN bot_type = 'search_engine' THEN 'search' WHEN bot_type = 'scraper' THEN 'scraping' ELSE 'training' END";

        // phpcs:ignore WordPress.DB.PreparedSQL.InterpolatedNotPrepared -- $intention_case_sql is a safe static CASE statement, $table_name from $wpdb->prefix
        $results = $wpdb->get_results($wpdb->prepare(
            "SELECT
                request_url,
                COALESCE(NULLIF(page_visited, ''), LEFT(request_url, 200)) as page_title,
                COUNT(*) as visit_count,
                SUM(CASE WHEN {$intention_case_sql} = 'inference' THEN 1 ELSE 0 END) as inference_visits,
                SUM(CASE WHEN {$intention_case_sql} = 'search' THEN 1 ELSE 0 END) as search_visits,
                SUM(CASE WHEN {$intention_case_sql} = 'training' THEN 1 ELSE 0 END) as training_visits,
                SUM(CASE WHEN {$intention_case_sql} = 'scraping' THEN 1 ELSE 0 END) as scraping_visits
            FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at >= %s
            GROUP BY page_title ORDER BY visit_count DESC LIMIT %d",
            $date_limit, $limit
        ), ARRAY_A);

        // Enrichir avec les vrais titres WordPress
        foreach ($results as &$result) {
            $result['display_title'] = $this->get_page_title_from_url($result['request_url']);
        }

        return $results;
    }

    /**
     * Extrait le titre de page WordPress depuis une URL
     */
    private function get_page_title_from_url($url) {
        // Parser l'URL pour extraire le path
        $parsed = wp_parse_url($url);
        $path = isset($parsed['path']) ? trim($parsed['path'], '/') : '';

        // Page d'accueil
        if (empty($path)) {
            return get_bloginfo('name') . ' - ' . __('Home', 'almaweb-ai-visitor-analytics');
        }

        // Essayer de trouver la page/post par URL
        $page_id = url_to_postid($url);
        if ($page_id > 0) {
            $title = get_the_title($page_id);
            if (!empty($title)) {
                return $title;
            }
        }

        // Fallback: nettoyer l'URL pour affichage
        $clean_path = str_replace(['/', '-', '_'], ' ', $path);
        return ucwords($clean_path);
    }

    /**
     * 6. COÛT ESTIMÉ DU SCRAPING
     */
    public function get_scraping_cost($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));

        $ai_requests = $wpdb->get_var($wpdb->prepare(
            "SELECT COUNT(*) FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at >= %s",
            $date_limit
        )) ?? 0;

        $avg_page_size_kb = 150;
        $avg_response_time_ms = 200;
        $total_bandwidth_mb = round(($ai_requests * $avg_page_size_kb) / 1024, 2);
        $total_cpu_time_seconds = round(($ai_requests * $avg_response_time_ms) / 1000, 2);
        return [
            'bandwidth_readable' => $total_bandwidth_mb > 1024 ? number_format($total_bandwidth_mb / 1024, 2) . ' GB' : number_format($total_bandwidth_mb) . ' MB',
            'cpu_time_readable' => $total_cpu_time_seconds > 3600 ? number_format($total_cpu_time_seconds / 3600, 1) . 'h' : number_format($total_cpu_time_seconds / 60, 1) . 'm',
        ];
    }
    
    /**
     * 7. VISITES "LIVE"
     */
    public function get_live_visits($days = 30) {
        $data = $this->get_training_vs_inference($days);
        // translators: %s is the percentage of AI visits from real users
        return ['count' => $data['counts']['inference'], 'message' => sprintf(__('%s%% of AI visits come from real users', 'almaweb-ai-visitor-analytics'), $data['percentages']['inference'])];
    }

    /**
     * 8. PAGE LA PLUS CONSULTÉE
     */
    public function get_hottest_page($days = 7) {
        $top_topics = $this->get_top_topics(1, $days);
        if (empty($top_topics)) {
            return ['page' => __('None', 'almaweb-ai-visitor-analytics'), 'visit_count' => 0, 'url' => '#'];
        }
        return [
            'page' => $top_topics[0]['display_title'],
            'visit_count' => $top_topics[0]['visit_count'],
            'url' => $top_topics[0]['request_url']
        ];
    }

    /**
     * 9. TENDANCE DE PÉNÉTRATION
     */
    public function get_penetration_trend($days = 30) {
        $current = $this->get_ai_pressure($days)['percentage'] ?? 0;
        $previous_data = $this->get_ai_pressure_for_period(strtotime("-{$days} days"), strtotime("-".($days * 2)." days"));
        $previous = $previous_data['percentage'] ?? 0;
        $trend = round($current - $previous, 1);
        // translators: %s is the trend percentage with + or - sign
        return ['percentage' => $trend, 'message' => sprintf(__('%s vs previous period', 'almaweb-ai-visitor-analytics'), ($trend > 0 ? "+{$trend}" : $trend) . '%')];
    }

    private function get_ai_pressure_for_period($end_date, $start_date) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $total_visits = $wpdb->get_var($wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE visited_at BETWEEN %s AND %s", gmdate('Y-m-d H:i:s', $start_date), gmdate('Y-m-d H:i:s', $end_date))) ?? 0;
        $ai_visits = $wpdb->get_var($wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE is_ai_bot = 1 AND visited_at BETWEEN %s AND %s", gmdate('Y-m-d H:i:s', $start_date), gmdate('Y-m-d H:i:s', $end_date))) ?? 0;
        return ['percentage' => $total_visits > 0 ? round(($ai_visits / $total_visits) * 100, 1) : 0];
    }

    /**
     * 10. DASHBOARD COMPLET (tous les KPIs en une fois)
     */
    /**
     * Get AI Referrer Insights
     *
     * @param int $days Number of days to look back
     * @return array AI referrer insights
     */
    public function get_ai_referrer_insights($days = 30) {
        $cache_key = 'almaweb_referrer_insights_' . $days;
        $cached = get_transient($cache_key);

        if (false !== $cached) {
            return $cached;
        }

        $diversity_score = min($this->database->get_referrer_diversity($days), 10);

        // Déterminer la note et le message de diversité
        if ($diversity_score >= 8) {
            $grade = 'A';
            $message = __('Excellent diversification of your AI traffic sources.', 'almaweb-ai-visitor-analytics');
        } elseif ($diversity_score >= 5) {
            $grade = 'B';
            $message = __('Good diversification. You do not depend on a single major source.', 'almaweb-ai-visitor-analytics');
        } elseif ($diversity_score >= 2) {
            $grade = 'C';
            $message = __('Low diversification. Your AI traffic comes from few different sources.', 'almaweb-ai-visitor-analytics');
        } else {
            $grade = 'D';
            $message = __('Very low diversification. You are heavily dependent on a single source.', 'almaweb-ai-visitor-analytics');
        }

        $insights = [
            'top_sources' => $this->database->get_top_referrer_sources($days, 10),
            'total_visits' => $this->database->get_ai_referrer_total($days),
            'landing_pages' => $this->database->get_ai_landing_pages($days, 10),
            'timeline' => $this->database->get_referrer_timeline($days),
            'diversity_score' => $diversity_score,
            'diversity_grade' => $grade,
            'diversity_message' => $message,
        ];

        set_transient($cache_key, $insights, 5 * MINUTE_IN_SECONDS);

        return $insights;
    }

    public function get_blocked_stats($days = 30) {
        global $wpdb;
        $table_name = $wpdb->prefix . 'ai_monitor_logs';
        $date_limit = gmdate('Y-m-d H:i:s', strtotime("-{$days} days"));
        $total = $wpdb->get_var($wpdb->prepare("SELECT COUNT(*) FROM {$table_name} WHERE is_blocked = 1 AND visited_at >= %s", $date_limit)) ?? 0;
        $top_family = $wpdb->get_var($wpdb->prepare("SELECT bot_family FROM {$table_name} WHERE is_blocked = 1 AND visited_at >= %s GROUP BY bot_family ORDER BY COUNT(*) DESC LIMIT 1", $date_limit));
        return ['total' => (int) $total, 'top_family' => $top_family ?: '-'];
    }

    public function get_all_kpis($days = 30) {
        $training_vs_inference = $this->get_training_vs_inference($days);

        return [
            'ai_pressure'           => $this->get_ai_pressure($days),
            'ai_authority_score'    => $this->get_ai_authority_score($days),
            'training_vs_inference' => $training_vs_inference,
            'intention_breakdown'   => $this->format_for_chart($training_vs_inference),
            'citation_potential'    => $this->get_citation_potential($days),
            'top_topics'            => $this->get_top_topics(10, $days),
            'scraping_cost'         => $this->get_scraping_cost($days),
            'penetration_trend'     => $this->get_penetration_trend($days),
            'live_visits'           => $this->get_live_visits($days),
            'hottest_page'          => $this->get_hottest_page(7),
            'ai_referrer_insights'  => $this->get_ai_referrer_insights($days),
            'blocked_stats'         => $this->get_blocked_stats($days),
        ];
    }

    private function format_for_chart($data) {
        $colors = ['inference' => '#10b981', 'search' => '#3b82f6', 'training' => '#f59e0b', 'scraping' => '#ef4444', 'unknown' => '#94a3b8'];
        $chart_data = ['labels' => [], 'datasets' => [['data' => [], 'backgroundColor' => []]]];
        foreach ($data['counts'] as $intention => $count) {
            if ($count > 0) {
                $chart_data['labels'][] = $data['labels'][$intention] ?? ucfirst($intention);
                $chart_data['datasets'][0]['data'][] = $count;
                $chart_data['datasets'][0]['backgroundColor'][] = $colors[$intention] ?? '#94a3b8';
            }
        }
        return $chart_data;
    }
}