app.py

import gradio as gr
import pandas as pd
import matplotlib.pyplot as plt
from risk_model import predict_risk
from datetime import datetime
import pytz

history_df = pd.DataFrame(columns=["Max_Temperature", "Duration", "Risk_Level", "Confidence", "Timestamp"])

def get_interpretation(risk_level):
    if risk_level == "High":
        return "⚠️ High risk of overheating! Immediate action recommended."
    elif risk_level == "Moderate":
        return "⚠️ Moderate risk detected. Monitor closely."
    else:
        return "Low risk. Operation is within safe limits."

def get_alert(risk_level):
    if risk_level == "High":
        return "ALERT: High risk detected!"
    elif risk_level == "Moderate":
        return "Warning: Moderate risk detected."
    else:
        return "No alert. Risk level is low."

def get_bullet_points(risk_level, confidence):
    points = [
        f"• Risk Level: {risk_level}",
        f"• Confidence: {confidence}%",
        f"• {get_alert(risk_level)}",
        f"• {get_interpretation(risk_level)}"
    ]
    return "\n".join(points)

def gradio_predict(temp, duration):
    global history_df
    risk_label, confidence = predict_risk(temp, duration)

    tz = pytz.timezone('Asia/Kolkata')
    timestamp = datetime.now(tz).strftime("%Y-%m-%d %H:%M:%S")

    new_entry = {
        "Max_Temperature": temp,
        "Duration": duration,
        "Risk_Level": risk_label,
        "Confidence": round(confidence, 1),
        "Timestamp": timestamp
    }
    history_df = pd.concat([history_df, pd.DataFrame([new_entry])], ignore_index=True)

    chart = create_chart(history_df)

    bullets = get_bullet_points(risk_label, round(confidence, 1))

    # Debug: print counts of risk levels
    counts = history_df['Risk_Level'].value_counts()
    print("Risk Level Counts:\n", counts)

    return risk_label, round(confidence, 1), bullets, history_df, chart

def reset_all():
    global history_df
    history_df = pd.DataFrame(columns=["Max_Temperature", "Duration", "Risk_Level", "Confidence", "Timestamp"])
    return "", "", "", pd.DataFrame(), None

def create_chart(df):
    if df.empty:
        return None

    counts = df['Risk_Level'].value_counts()
    # Ensure all risk levels present and in correct order
    counts = counts.reindex(['Low', 'Moderate', 'High'], fill_value=0)

    print("Reindexed Risk Level Counts for chart:\n", counts)  # Debug print

    fig, ax = plt.subplots(figsize=(6, 4))
    bars = ax.bar(counts.index, counts.values, color=['#2ca02c', '#ff7f0e', '#d62728'])
    ax.set_xlabel('Risk Level')
    ax.set_ylabel('Number of Predictions')
    ax.set_title('Prediction Counts by Risk Level')

    for bar in bars:
        height = bar.get_height()
        ax.annotate(f'{int(height)}',
                    xy=(bar.get_x() + bar.get_width() / 2, height),
                    xytext=(0, 3),
                    textcoords="offset points",
                    ha='center', va='bottom')

    plt.tight_layout()
    return fig

with gr.Blocks() as demo:
    reset_btn = gr.Button("Reset")

    with gr.Row():
        temp_input = gr.Number(label="Max Temperature (°C)", value=25, precision=1, minimum=0)
        duration_input = gr.Number(label="Duration (minutes)", value=10, precision=1, minimum=0)

    predict_btn = gr.Button("Predict")

    risk_status = gr.Textbox(label="Risk Status", interactive=False)
    risk_score = gr.Textbox(label="Risk Score (%)", interactive=False)
    bullet_points = gr.Textbox(label="Summary", interactive=False, lines=5)

    history_table = gr.Dataframe(label="Prediction History")
    risk_chart = gr.Plot(label="Risk Level Chart")

    predict_btn.click(
        fn=gradio_predict,
        inputs=[temp_input, duration_input],
        outputs=[risk_status, risk_score, bullet_points, history_table, risk_chart]
    )

    reset_btn.click(
        fn=reset_all,
        inputs=[],
        outputs=[risk_status, risk_score, bullet_points, history_table, risk_chart]
    )

if __name__ == "__main__":
    demo.launch()