"""
NBA Dynamical Systems Web App
Deploy to HuggingFace Spaces
"""

import streamlit as st
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from scipy.integrate import odeint
import json

# Page config
st.set_page_config(
    page_title="NBA Dynamical Systems",
    page_icon="🏀",
    layout="wide"
)

# ============================================================
# DYNAMICAL SYSTEMS FUNCTIONS
# ============================================================

def driven_pendulum(y, t, gamma, f0, omega):
    theta, omega_t = y
    dydt = [omega_t, -gamma*omega_t - np.sin(theta) + f0*np.cos(omega*t)]
    return dydt

def team_harmonic(m, c, k, F, omega, t):
    omega_n = np.sqrt(max(k/m - c**2/(2*m**2), 0.01))
    H = 1 / np.sqrt((k - m*omega**2)**2 + (c*omega)**2)
    phi = np.arctan2(c*omega, k - m*omega**2)
    return H * F/m * np.sin(omega*t - phi), omega_n

def resonance_factor(m, c, k, omega):
    omega_n = np.sqrt(k/m)
    r = omega / omega_n
    zeta = c / (2*np.sqrt(k*m))
    return 1 / np.sqrt((1 - r**2)**2 + (2*zeta*r)**2)

def lorenz(state, t, sigma, rho, beta):
    x, y, z = state
    return [sigma*(y-x), x*(rho-z)-y, x*y-beta*z]

def phase_reconstruct(data, dim, tau):
    n = len(data) - (dim-1)*tau
    if n < 2:
        return np.array([]).reshape(0, dim)
    return np.array([data[i:i+dim*tau:tau] for i in range(n)])

def lyapunov_estimate(data):
    divergences = []
    for i in range(len(data)-1):
        for j in range(i+1, min(i+5, len(data))):
            d0 = abs(data[i] - data[j])
            d1 = abs(data[i+1] - data[j+1])
            if d0 > 0 and d1 > 0:
                divergences.append(np.log(d1/d0))
    return np.mean(divergences) if divergences else 0

# ============================================================
# DATA
# ============================================================

LUKA_DATA = {
    "player": {"name": "Luka Doncic", "team": "Dallas Mavericks"},
    "season_averages": {"points": 28.8, "rebounds": 8.6, "assists": 8.2, "plus_minus": 8.5},
    "game_logs": [
        {"date": "2025-01-20", "opponent": "BOS", "result": "W", "pts": 34, "reb": 10, "ast": 8, "plus_minus": 12},
        {"date": "2025-01-18", "opponent": "DEN", "result": "W", "pts": 38, "reb": 11, "ast": 7, "plus_minus": 8},
        {"date": "2025-01-15", "opponent": "LAL", "result": "L", "pts": 25, "reb": 6, "ast": 12, "plus_minus": -8},
        {"date": "2025-01-12", "opponent": "GSW", "result": "W", "pts": 32, "reb": 9, "ast": 6, "plus_minus": 18},
        {"date": "2025-01-10", "opponent": "PHX", "result": "W", "pts": 29, "reb": 7, "ast": 10, "plus_minus": 10},
        {"date": "2025-01-08", "opponent": "LAC", "result": "W", "pts": 42, "reb": 8, "ast": 9, "plus_minus": 15},
        {"date": "2025-01-05", "opponent": "MEM", "result": "W", "pts": 26, "reb": 5, "ast": 11, "plus_minus": 4},
        {"date": "2025-01-03", "opponent": "HOU", "result": "W", "pts": 31, "reb": 10, "ast": 7, "plus_minus": 6},
        {"date": "2024-12-30", "opponent": "SAS", "result": "W", "pts": 22, "reb": 7, "ast": 13, "plus_minus": 5},
        {"date": "2024-12-28", "opponent": "NYK", "result": "W", "pts": 36, "reb": 9, "ast": 8, "plus_minus": 14},
    ]
}

# ============================================================
# UI
# ============================================================

def main():
    st.title("🏀 NBA Dynamical Systems")
    st.markdown("### Dallas Mavericks vs LA Lakers")
    st.markdown("*Chaos theory applied to basketball*")
    
    data = LUKA_DATA
    games = pd.DataFrame(data['game_logs'])
    stats = data['season_averages']
    
    # Header metrics
    col1, col2, col3, col4 = st.columns(4)
    with col1:
        st.metric("PPG", f"{stats['points']:.1f}")
    with col2:
        st.metric("RPG", f"{stats['rebounds']:.1f}")
    with col3:
        st.metric("APG", f"{stats['assists']:.1f}")
    with col4:
        st.metric("+/-", f"+{stats['plus_minus']:.1f}")
    
    # Tabs
    tab1, tab2, tab3, tab4, tab5 = st.tabs(["📈 Timeline", "🔄 Sensitivity", "🌊 Harmonics", "🌀 Phase Space", "🎯 Lorenz"])
    
    with tab1:
        st.subheader("Scoring Timeline")
        fig, ax = plt.subplots(figsize=(10, 4))
        ax.plot(range(len(games)), games['pts'], 'b-o', linewidth=2, markersize=8)
        ax.axhline(y=games['pts'].mean(), color='r', linestyle='--', label=f"Avg: {games['pts'].mean():.1f}")
        ax.set_xlabel("Game")
        ax.set_ylabel("Points")
        ax.set_title("Luka Doncic - Points per Game")
        ax.legend()
        ax.grid(True, alpha=0.3)
        st.pyplot(fig)
        st.dataframe(games[['date', 'opponent', 'result', 'pts', 'reb', 'ast', 'plus_minus']], hide_index=True)
    
    with tab2:
        st.subheader("Sensitivity (Driven Pendulum)")
        st.latex(r"\theta'' + \gamma\theta' + \sin(\theta) = f_0\cos(\omega t)")
        
        variability = np.std(games['pts']) / np.mean(games['pts'])
        damping = 0.5 + (1 - variability) * 2
        forcing = 0.5 + games['pts'].mean() / 30
        
        col1, col2, col3 = st.columns(3)
        with col1: st.metric("Damping (γ)", f"{damping:.2f}")
        with col2: st.metric("Forcing (f₀)", f"{forcing:.2f}")
        with col3: st.metric("Variability", f"{variability:.2f}")
        
        t = np.linspace(0, 50, 500)
        sol = odeint(driven_pendulum, [0.1, 0], t, args=(damping, forcing, 0.5))
        
        fig, (ax1, ax2) = plt.subplots(1, 2, figsize=(12, 4))
        ax1.plot(t, sol[:, 0], 'b-', label='θ₀=0.10', linewidth=2)
        ax1.plot(t, odeint(driven_pendulum, [0.11, 0], t, args=(damping, forcing, 0.5))[:, 0], 'r--', label='θ₀=0.11')
        ax1.legend(); ax1.grid(True, alpha=0.3)
        
        div = np.abs(odeint(driven_pendulum, [0.11, 0], t, args=(damping, forcing, 0.5))[:, 0] - sol[:, 0])
        ax2.plot(t, div, 'r-', linewidth=2); ax2.grid(True, alpha=0.3)
        st.pyplot(fig)
        
        st.info(f"**Regime:** {'Stable' if np.mean(div) < 2 else 'Chaotic'}")
    
    with tab3:
        st.subheader("Team Harmonics (Mass-Spring-Damper)")
        st.latex(r"mx'' + cx' + kx = F\sin(\omega t)")
        
        mavs = {'chemistry': 0.72, 'resilience': 0.65, 'roster': 0.88}
        lakes = {'chemistry': 0.58, 'resilience': 0.55, 'roster': 0.82}
        
        omega = 0.5
        mavs_rf = resonance_factor(mavs['roster'], mavs['resilience']*2, mavs['chemistry']*5, omega)
        lakes_rf = resonance_factor(lakes['roster'], lakes['resilience']*2, lakes['chemistry']*5, omega)
        
        col1, col2 = st.columns(2)
        with col1:
            st.metric("Mavericks RF", f"{mavs_rf:.2f}")
            st.metric("Mavericks Upset Risk", f"{max(0, mavs_rf-1):.2f}")
        with col2:
            st.metric("Lakers RF", f"{lakes_rf:.2f}")
            st.metric("Lakers Upset Risk", f"{max(0, lakes_rf-1):.2f}")
        
        t = np.linspace(0, 30, 300)
        team = st.selectbox("Select Team", ["Mavericks", "Lakers"])
        team_data = mavs if team == "Mavericks" else lakes
        response, omega_n = team_harmonic(team_data['roster'], team_data['resilience']*2, team_data['chemistry']*5, 0.5, omega, t)
        
        fig, ax = plt.subplots(figsize=(10, 4))
        ax.plot(t, response, 'b-', linewidth=2)
        ax.set_xlabel('Time'); ax.set_ylabel('Displacement')
        ax.set_title(f'{team} Harmonic Response'); ax.grid(True, alpha=0.3)
        st.pyplot(fig)
    
    with tab4:
        st.subheader("Phase Space Reconstruction")
        points = games['pts'].values
        embedded = phase_reconstruct(points, 3, 1)
        
        fig, ax = plt.subplots(figsize=(8, 6))
        ax.plot(embedded[:, 0], embedded[:, 1], 'b-o', markersize=6, alpha=0.7)
        ax.set_xlabel('x(t)'); ax.set_ylabel('x(t+τ)')
        ax.set_title('2D Phase Space'); ax.grid(True, alpha=0.3)
        st.pyplot(fig)
        st.info("Reconstructed attractor shows scoring dynamics")
    
    with tab5:
        st.subheader("Lorenz System Forecasting")
        st.latex(r"\begin{aligned} dx/dt &= \sigma(y-x) \\ dy/dt &= x(\rho-z)-y \end{aligned}")
        
        variability = np.std(games['pts']) / np.mean(games['pts'])
        sigma, rho, beta = 10, 14 + stats['plus_minus']/5 + 2*0.5, 8/3 + variability*2
        
        col1, col2, col3 = st.columns(3)
        with col1: st.metric("σ", f"{sigma}")
        with col2: st.metric("ρ", f"{rho:.1f}")
        with col3: st.metric("β", f"{beta:.1f}")
        
        t = np.linspace(0, 50, 2000)
        sol = odeint(lorenz, [1, 1, 1], t, args=(sigma, rho, beta))
        
        fig = plt.figure(figsize=(10, 6))
        ax = fig.add_subplot(111, projection='3d')
        ax.plot(sol[:, 0], sol[:, 1], sol[:, 2], 'b-', linewidth=0.5)
        ax.set_xlabel('x'); ax.set_ylabel('y'); ax.set_zlabel('z')
        ax.set_title('Lorenz Attractor')
        st.pyplot(fig)
        
        lyap = lyapunov_estimate(points)
        st.metric("Lyapunov Exponent", f"{lyap:.4f}")
        st.success("**Stable**" if lyap < 0 else "**Chaotic**")
    
    # Risk Summary
    st.divider()
    st.subheader("📊 Combined Risk")
    
    t = np.linspace(0, 50, 500)
    sensitivity_risk = np.mean(np.abs(odeint(driven_pendulum, [0.11, 0], t, args=(damping, forcing, 0.5))[:, 0] - 
                                       odeint(driven_pendulum, [0.1, 0], t, args=(damping, forcing, 0.5))[:, 0])) / 10 * 0.25
    mav_risk = max(0, mavs_rf-1) * 0.25
    laker_risk = max(0, lakes_rf-1) * 0.25
    chaos_risk = max(0, lyap) * 0.25
    
    total = sensitivity_risk + mav_risk + laker_risk + chaos_risk
    
    fig, ax = plt.subplots(figsize=(8, 3))
    ax.barh(['Luka', 'Mavericks', 'Lakers', 'Chaos'], [sensitivity_risk, mav_risk, laker_risk, chaos_risk])
    ax.axvline(x=total, color='black', linestyle='--', label=f'Total: {total:.2f}')
    ax.legend()
    st.pyplot(fig)
    
    if total < 0.25:
        st.success(f"**RISK: {total:.2f} - LOW - Favor Mavericks**")
    elif total < 0.40:
        st.warning(f"**RISK: {total:.2f} - MODERATE**")
    else:
        st.error(f"**RISK: {total:.2f} - HIGH**")

if __name__ == "__main__":
    main()