#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
calculate_metrics.py - Calcul des métriques de performance prédictions vs résultats

Usage:
    python3 calculate_metrics.py                    # Aujourd'hui
    python3 calculate_metrics.py --date 2026-04-15  # Date spécifique
    python3 calculate_metrics.py --yesterday        # Hier
    python3 calculate_metrics.py --backfill 30      # Remplir 30 derniers jours

Calculé après 21h (résultats PMU disponibles)
"""

import sqlite3
import json
import argparse
from datetime import datetime, timedelta
from collections import defaultdict

DB_PATH = "/home/h3r7/turf_scraper/turf.db"

# =============================================================================
# SCHÉMA BASE DE DONNÉES
# =============================================================================

METRICS_SCHEMA = """
-- Table principale des métriques par course/source
CREATE TABLE IF NOT EXISTS prediction_metrics (
    id INTEGER PRIMARY KEY AUTOINCREMENT,
    date TEXT NOT NULL,
    race_time TEXT,
    race_name TEXT,
    race_hippodrome TEXT,
    source TEXT NOT NULL,
    discipline TEXT,
    
    -- Comptages
    nb_predictions INTEGER DEFAULT 0,
    nb_gagnants INTEGER DEFAULT 0,
    nb_places INTEGER DEFAULT 0,
    nb_top5 INTEGER DEFAULT 0,
    nb_hors_top5 INTEGER DEFAULT 0,
    
    -- Taux
    taux_gagnant REAL,
    taux_place REAL,
    taux_top5 REAL,
    
    -- Rangs
    rang_moyen REAL,
    ecart_rang_moyen REAL,
    
    -- ROI avec dividendes réels PMU
    roi_sg_brut REAL,
    roi_sg_net REAL,
    roi_sp_brut REAL,
    roi_sp_net REAL,
    
    -- Quinté
    quinte_5sur5 INTEGER DEFAULT 0,
    quinte_4sur5 INTEGER DEFAULT 0,
    quinte_3sur5 INTEGER DEFAULT 0,
    quinte_2sur5 INTEGER DEFAULT 0,
    
    -- Value
    value_bet_score REAL,
    top_cote_gagnante REAL,
    
    created_at TIMESTAMP DEFAULT CURRENT_TIMESTAMP,
    UNIQUE(date, race_time, source)
);

-- Vue détaillée des performances
CREATE VIEW IF NOT EXISTS v_predictions_performance AS
SELECT 
    pr.date,
    pr.race_time,
    pr.race_name,
    pr.race_hippodrome,
    pr.horse_number,
    pr.horse_name,
    pr.prediction_rank,
    pr.odds AS cote_prediction,
    pr.source,
    pa.ordre_arrivee,
    pa.cote_direct AS cote_finale,
    pa.driver,
    
    -- Indicateurs binaires
    CASE WHEN pa.ordre_arrivee = 1 THEN 1 ELSE 0 END AS is_gagnant,
    CASE WHEN pa.ordre_arrivee <= 3 THEN 1 ELSE 0 END AS is_place,
    CASE WHEN pa.ordre_arrivee <= 5 THEN 1 ELSE 0 END AS is_top5,
    
    -- Écarts
    ABS(COALESCE(pr.prediction_rank, 99) - COALESCE(pa.ordre_arrivee, 99)) AS ecart_rang,
    
    -- Value réalisée
    CASE 
        WHEN pa.ordre_arrivee = 1 AND pa.cote_direct > pr.odds 
        THEN ROUND((pa.cote_direct - pr.odds) / pr.odds * 100, 1) 
        ELSE 0 
    END AS value_realized

FROM predictions pr
LEFT JOIN pmu_partants pa 
    ON pa.date_programme = pr.date 
    AND pa.nom = pr.horse_name;

-- Vue résumé par source (30 jours glissants)
CREATE VIEW IF NOT EXISTS v_metrics_summary_30d AS
SELECT 
    source,
    COUNT(*) as nb_courses,
    SUM(nb_predictions) as total_predictions,
    SUM(nb_gagnants) as total_gagnants,
    SUM(nb_places) as total_places,
    SUM(nb_top5) as total_top5,
    ROUND(AVG(taux_gagnant), 2) as moy_taux_gagnant,
    ROUND(AVG(taux_place), 2) as moy_taux_place,
    ROUND(AVG(taux_top5), 2) as moy_taux_top5,
    ROUND(AVG(roi_sg_net), 3) as moy_roi_sg,
    ROUND(AVG(roi_sp_net), 3) as moy_roi_sp,
    ROUND(AVG(ecart_rang_moyen), 2) as moy_ecart_rang,
    SUM(quinte_5sur5) as nb_5sur5,
    SUM(quinte_4sur5) as nb_4sur5,
    SUM(quinte_3sur5) as nb_3sur5,
    ROUND(SUM(quinte_5sur5) * 100.0 / NULLIF(COUNT(*), 0), 1) as pct_5sur5,
    ROUND(SUM(quinte_4sur5) * 100.0 / NULLIF(COUNT(*), 0), 1) as pct_4sur5
FROM prediction_metrics
WHERE date >= date('now', '-30 days')
GROUP BY source
ORDER BY moy_taux_place DESC;

-- Vue évolution quotidienne
CREATE VIEW IF NOT EXISTS v_metrics_daily AS
SELECT 
    date,
    source,
    SUM(nb_predictions) as predictions,
    SUM(nb_gagnants) as gagnants,
    SUM(nb_places) as places,
    SUM(nb_top5) as top5,
    ROUND(AVG(taux_gagnant), 2) as taux_gagnant,
    ROUND(AVG(taux_place), 2) as taux_place,
    ROUND(AVG(roi_sg_net), 3) as roi_sg,
    ROUND(AVG(roi_sp_net), 3) as roi_sp,
    SUM(quinte_5sur5) as quinte_5sur5,
    SUM(quinte_4sur5) as quinte_4sur5
FROM prediction_metrics
GROUP BY date, source
ORDER BY date DESC;
"""

# =============================================================================
# FONCTIONS UTILITAIRES
# =============================================================================

def get_db():
    """Connexion à la base de données"""
    conn = sqlite3.connect(DB_PATH)
    conn.row_factory = sqlite3.Row
    return conn

def init_db():
    """Initialise les tables et vues"""
    conn = get_db()
    conn.executescript(METRICS_SCHEMA)
    conn.commit()
    conn.close()
    print("✅ Tables et vues initialisées")

def get_dividende_sg(conn, date, num_reunion, num_course, horse_number):
    """Récupère le dividende Simple Gagnant pour un cheval"""
    try:
        row = conn.execute("""
            SELECT dividende_euro 
            FROM pmu_rapports 
            WHERE date_programme = ? 
                AND num_reunion = ? 
                AND num_course = ?
                AND type_pari = 'SIMPLE_GAGNANT' 
                AND combinaison = ?
        """, (date, num_reunion, num_course, str(horse_number))).fetchone()
        return row['dividende_euro'] if row else None
    except:
        return None

def get_dividende_sp(conn, date, num_reunion, num_course, horse_number):
    """Récupère le dividende Simple Placé pour un cheval"""
    try:
        row = conn.execute("""
            SELECT dividende_euro 
            FROM pmu_rapports 
            WHERE date_programme = ? 
                AND num_reunion = ? 
                AND num_course = ?
                AND type_pari = 'SIMPLE_PLACE' 
                AND combinaison = ?
        """, (date, num_reunion, num_course, str(horse_number))).fetchone()
        return row['dividende_euro'] if row else None
    except:
        return None

# =============================================================================
# CALCUL DES MÉTRIQUES
# =============================================================================

def calculate_course_metrics(conn, date, race_time, race_name, source):
    """Calcule les métriques pour une course/source donnée"""
    
    # Récupérer les prédictions pour cette course/source
    preds = conn.execute("""
        SELECT 
            pr.horse_number,
            pr.horse_name,
            pr.prediction_rank,
            pr.odds,
            pa.ordre_arrivee,
            pa.cote_direct,
            pa.num_reunion,
            pa.num_course
        FROM predictions pr
        LEFT JOIN pmu_partants pa 
            ON pa.date_programme = pr.date 
            AND pa.nom = pr.horse_name
        WHERE pr.date = ? 
            AND pr.race_time = ?
            AND pr.source = ?
    """, (date, race_time, source)).fetchall()
    
    if not preds:
        return
    
    # Métadonnées
    first_pred = preds[0]
    hippodrome = conn.execute("""
        SELECT race_hippodrome FROM predictions 
        WHERE date = ? AND race_time = ? 
        LIMIT 1
    """, (date, race_time)).fetchone()
    
    race_hippodrome = hippodrome['race_hippodrome'] if hippodrome else None
    num_reunion = first_pred['num_reunion'] if first_pred['num_reunion'] else None
    num_course = first_pred['num_course'] if first_pred['num_course'] else None
    
    # Récupérer discipline depuis pmu_courses
    discipline = None
    if num_reunion and num_course:
        disc_row = conn.execute("""
            SELECT discipline FROM pmu_courses 
            WHERE date_programme = ? AND num_reunion = ? AND num_course = ?
        """, (date, num_reunion, num_course)).fetchone()
        discipline = disc_row['discipline'] if disc_row else None
    
    # Comptages
    nb_predictions = len(preds)
    nb_gagnants = sum(1 for p in preds if p['ordre_arrivee'] == 1)
    nb_places = sum(1 for p in preds if p['ordre_arrivee'] and p['ordre_arrivee'] <= 3)
    nb_top5 = sum(1 for p in preds if p['ordre_arrivee'] and p['ordre_arrivee'] <= 5)
    nb_hors_top5 = nb_predictions - nb_top5
    
    # Taux
    taux_gagnant = round(nb_gagnants / nb_predictions * 100, 2) if nb_predictions > 0 else 0
    taux_place = round(nb_places / nb_predictions * 100, 2) if nb_predictions > 0 else 0
    taux_top5 = round(nb_top5 / nb_predictions * 100, 2) if nb_predictions > 0 else 0
    
    # Rang moyen
    rangs = [p['ordre_arrivee'] for p in preds if p['ordre_arrivee']]
    rang_moyen = round(sum(rangs) / len(rangs), 2) if rangs else None
    
    # Écart rang moyen
    ecarts = [abs((p['prediction_rank'] or 99) - (p['ordre_arrivee'] or 99)) for p in preds]
    ecart_rang_moyen = round(sum(ecarts) / len(ecarts), 2) if ecarts else None
    
    # ROI avec dividendes réels
    roi_sg_values = []
    roi_sp_values = []
    
    for p in preds:
        if p['ordre_arrivee'] == 1 and num_reunion and num_course:
            div_sg = get_dividende_sg(conn, date, num_reunion, num_course, p['horse_number'])
            if div_sg and div_sg > 0:
                roi_sg_values.append(div_sg - 1)
            else:
                roi_sg_values.append(-1)
        elif p['ordre_arrivee'] and p['ordre_arrivee'] > 1:
            roi_sg_values.append(-1)
        
        if p['ordre_arrivee'] and p['ordre_arrivee'] <= 3 and num_reunion and num_course:
            div_sp = get_dividende_sp(conn, date, num_reunion, num_course, p['horse_number'])
            if div_sp and div_sp > 0:
                roi_sp_values.append(div_sp - 1)
            else:
                roi_sp_values.append(-1)
        elif p['ordre_arrivee'] and p['ordre_arrivee'] > 3:
            roi_sp_values.append(-1)
    
    roi_sg_brut = sum(roi_sg_values) if roi_sg_values else 0
    roi_sg_net = round(roi_sg_brut / len(roi_sg_values), 3) if roi_sg_values else 0
    roi_sp_brut = sum(roi_sp_values) if roi_sp_values else 0
    roi_sp_net = round(roi_sp_brut / len(roi_sp_values), 3) if roi_sp_values else 0
    
    # Quinté (5 chevaux dans le top 5)
    quinte_5sur5 = 1 if nb_top5 >= 5 else 0
    quinte_4sur5 = 1 if nb_top5 >= 4 else 0
    quinte_3sur5 = 1 if nb_top5 >= 3 else 0
    quinte_2sur5 = 1 if nb_top5 >= 2 else 0
    
    # Value bet score
    value_scores = []
    for p in preds:
        if p['ordre_arrivee'] == 1 and p['cote_direct'] and p['odds']:
            value = (p['cote_direct'] - p['odds']) / p['odds'] * 100
            value_scores.append(value)
    value_bet_score = round(sum(value_scores) / len(value_scores), 2) if value_scores else 0
    
    # Top cote gagnante
    top_cotes = [p['cote_direct'] for p in preds if p['ordre_arrivee'] == 1 and p['cote_direct']]
    top_cote_gagnante = max(top_cotes) if top_cotes else None
    
    # Insérer ou mettre à jour
    conn.execute("""
        INSERT OR REPLACE INTO prediction_metrics (
            date, race_time, race_name, race_hippodrome, source, discipline,
            nb_predictions, nb_gagnants, nb_places, nb_top5, nb_hors_top5,
            taux_gagnant, taux_place, taux_top5,
            rang_moyen, ecart_rang_moyen,
            roi_sg_brut, roi_sg_net, roi_sp_brut, roi_sp_net,
            quinte_5sur5, quinte_4sur5, quinte_3sur5, quinte_2sur5,
            value_bet_score, top_cote_gagnante
        ) VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?, ?)
    """, (
        date, race_time, race_name, race_hippodrome, source, discipline,
        nb_predictions, nb_gagnants, nb_places, nb_top5, nb_hors_top5,
        taux_gagnant, taux_place, taux_top5,
        rang_moyen, ecart_rang_moyen,
        roi_sg_brut, roi_sg_net, roi_sp_brut, roi_sp_net,
        quinte_5sur5, quinte_4sur5, quinte_3sur5, quinte_2sur5,
        value_bet_score, top_cote_gagnante
    ))

def calculate_metrics(date_str):
    """Calcule les métriques pour une date donnée"""
    init_db()
    conn = get_db()
    
    # Récupérer les courses avec prédictions ET résultats
    courses = conn.execute("""
        SELECT DISTINCT 
            pr.date,
            pr.race_time,
            pr.race_name
        FROM predictions pr
        JOIN pmu_partants pa 
            ON pa.date_programme = pr.date 
            AND pa.nom = pr.horse_name
        WHERE pr.date = ?
          AND pa.ordre_arrivee IS NOT NULL
    """, (date_str,)).fetchall()
    
    if not courses:
        print(f"⚠️ Aucune course avec résultats pour {date_str}")
        conn.close()
        return
    
    sources = [
        'canalturf_selections',
        'canalturf_prono_bases', 
        'canalturf_prono_chances',
        'canalturf_prono_outsiders',
        'canalturf_partants'
    ]
    
    total_calculated = 0
    for course in courses:
        for source in sources:
            try:
                calculate_course_metrics(conn, course['date'], course['race_time'], course['race_name'], source)
                total_calculated += 1
            except Exception as e:
                print(f"⚠️ Erreur {course['race_time']} {source}: {e}")
    
    conn.commit()
    conn.close()
    print(f"✅ Métriques calculées pour {date_str}: {total_calculated} combinaisons course/source")

def backfill_metrics(days=30):
    """Remplit les métriques sur plusieurs jours"""
    print(f"📊 Backfill sur {days} jours...")
    for i in range(days):
        date = (datetime.now() - timedelta(days=i)).strftime('%Y-%m-%d')
        print(f"  → {date}")
        try:
            calculate_metrics(date)
        except Exception as e:
            print(f"  ⚠️ Erreur: {e}")
    print("✅ Backfill terminé")

# =============================================================================
# POINT D'ENTRÉE
# =============================================================================

if __name__ == "__main__":
    parser = argparse.ArgumentParser(description="Calcul des métriques de performance")
    parser.add_argument("--date", "-d", help="Date YYYY-MM-DD")
    parser.add_argument("--yesterday", "-y", action="store_true", help="Calculer hier")
    parser.add_argument("--backfill", "-b", type=int, help="Remplir N derniers jours")
    
    args = parser.parse_args()
    
    if args.backfill:
        backfill_metrics(args.backfill)
    elif args.yesterday:
        date_str = (datetime.now() - timedelta(days=1)).strftime('%Y-%m-%d')
        calculate_metrics(date_str)
    elif args.date:
        calculate_metrics(args.date)
    else:
        calculate_metrics(datetime.now().strftime('%Y-%m-%d'))