dashboard.py

import matplotlib.pyplot as plt
import firebase_admin
from firebase_admin import credentials, firestore
import io
from PIL import Image
import os
import json

# === Initialize Firebase
if not firebase_admin._apps:
    cred = credentials.Certificate("firebase_key.json")
    firebase_admin.initialize_app(cred)
db = firestore.client()

# === 🗳️ Feedback Summary Plot (Solution 1 vs Solution 2)
def update_dashboard_plot():
    try:
        docs = db.collection("evo_feedback").stream()
        count_1, count_2 = 0, 0

        for doc in docs:
            winner = doc.to_dict().get("winner", "").strip()
            if winner == "1":
                count_1 += 1
            elif winner == "2":
                count_2 += 1

        total = count_1 + count_2
        fig, ax = plt.subplots()

        if total == 0:
            ax.text(0.5, 0.5, "No feedback collected yet.", ha="center", va="center", fontsize=12)
            ax.axis("off")
        else:
            ax.bar(["Solution 1", "Solution 2"], [count_1, count_2], color=["skyblue", "lightgreen"])
            ax.set_ylabel("Votes")
            ax.set_title(f"🗳️ Feedback Summary (Total: {total})")
            for i, v in enumerate([count_1, count_2]):
                ax.text(i, v + 0.5, str(v), ha='center', fontweight='bold')

        plt.tight_layout()
        buf = io.BytesIO()
        plt.savefig(buf, format="png")
        buf.seek(0)
        return Image.open(buf)

    except Exception as e:
        return _fallback_plot(f"Error loading feedback chart: {e}")

# === 📈 Evolution Accuracy Plot (Accuracy vs Generation)
def evolution_accuracy_plot():
    try:
        log_path = "trained_model/evolution_log.json"
        if not os.path.exists(log_path):
            return _fallback_plot("⚠️ No evolution log found")

        with open(log_path, "r") as f:
            log_data = json.load(f)

        if not log_data:
            return _fallback_plot("⚠️ Evolution log is empty")

        generations = list(range(1, len(log_data) + 1))
        accuracies = [round(entry.get("accuracy", 0), 4) for entry in log_data]
        tooltips = [
            f"L: {entry.get('num_layers', '?')}, H: {entry.get('num_heads', '?')}, FFN: {entry.get('ffn_dim', '?')}, Mem: {entry.get('use_memory', '?')}"
            for entry in log_data
        ]

        fig, ax = plt.subplots(figsize=(7, 4))
        ax.plot(generations, accuracies, marker='o', color="purple", label="Accuracy")

        for i, tooltip in enumerate(tooltips):
            offset = (i % 2) * 5 - 2
            ax.annotate(tooltip, (generations[i], accuracies[i]),
                        fontsize=7, xytext=(0, 8 + offset), textcoords='offset points',
                        ha='center')

        ax.set_xlabel("Generation")
        ax.set_ylabel("Accuracy")
        ax.set_title("📈 EvoTransformer Evolution Accuracy")
        ax.set_ylim([0, 1.05])
        ax.grid(True)
        plt.tight_layout()

        buf = io.BytesIO()
        plt.savefig(buf, format="png")
        buf.seek(0)
        return Image.open(buf)

    except Exception as e:
        return _fallback_plot(f"⚠️ Error loading evolution plot: {e}")

# === 🏆 Hall of Fame Plot (most frequent goals)
def leaderboard_plot():
    try:
        docs = db.collection("evo_feedback").stream()
        goal_counter = {}

        for doc in docs:
            goal = doc.to_dict().get("goal", "").strip()
            if goal:
                goal_counter[goal] = goal_counter.get(goal, 0) + 1

        sorted_goals = sorted(goal_counter.items(), key=lambda x: x[1], reverse=True)[:10]

        if not sorted_goals:
            return _fallback_plot("🏆 No top goals yet.")

        labels = [g[0][:25] + "..." if len(g[0]) > 25 else g[0] for g, _ in sorted_goals]
        counts = [c for _, c in sorted_goals]

        fig, ax = plt.subplots(figsize=(8, 4))
        ax.barh(labels[::-1], counts[::-1], color="orange")
        ax.set_xlabel("Votes")
        ax.set_title("🏆 Hall of Fame: Top 10 Goals")
        plt.tight_layout()

        buf = io.BytesIO()
        plt.savefig(buf, format="png")
        buf.seek(0)
        return Image.open(buf)

    except Exception as e:
        return _fallback_plot(f"⚠️ Error loading leaderboard: {e}")

# === 🔢 Real-time vote count for display
def get_vote_counts():
    try:
        docs = db.collection("evo_feedback").stream()
        count_1, count_2 = 0, 0

        for doc in docs:
            winner = doc.to_dict().get("winner", "").strip()
            if winner == "1":
                count_1 += 1
            elif winner == "2":
                count_2 += 1

        return {"1": count_1, "2": count_2}
    except:
        return {"1": 0, "2": 0}

# === 🛠️ Fallback Plot
def _fallback_plot(message):
    fig, ax = plt.subplots()
    ax.text(0.5, 0.5, message, ha="center", va="center", fontsize=11)
    ax.axis('off')
    plt.tight_layout()
    buf = io.BytesIO()
    plt.savefig(buf, format="png")
    buf.seek(0)
    return Image.open(buf)