import pandas as pd
import gradio as gr
from sklearn.preprocessing import StandardScaler
from sklearn.svm import OneClassSVM

# Function to detect theft and provide insights
def detect_theft_from_file(uploaded_file):
    # Determine the file type and read it
    if uploaded_file.name.endswith('.csv'):
        df = pd.read_csv(uploaded_file.name)
    elif uploaded_file.name.endswith(('.xlsx', '.xls')):
        df = pd.read_excel(uploaded_file.name)
    else:
        return "Unsupported file format. Please upload a CSV or Excel file."

    # Ensure required format (Consumer ID + 12 months + optional Last Year)
    if len(df.columns) < 13:
        return "The file must have at least 13 columns: 'Consumer ID', 12 months (e.g., Jan-Dec), and optionally 'Last Year Avg'."

    feature_columns = df.columns[1:13]  # 12 months
    data = df[feature_columns].values

    # Normalize the data
    scaler = StandardScaler()
    data_scaled = scaler.fit_transform(data)

    # One-Class SVM model
    model = OneClassSVM(nu=0.1, kernel='rbf', gamma='scale')
    model.fit(data_scaled)

    # Predict anomalies
    predictions = model.predict(data_scaled)

    # Add Reasons and Tips
    reasons = []
    tips = []
    for i, pred in enumerate(predictions):
        if pred == -1:  # Theft detected
            current_pattern = data[i]
            if 'Last Year Avg' in df.columns:
                last_year_avg = df['Last Year Avg'].iloc[i]
                if current_pattern.mean() < 0.5 * last_year_avg:
                    reasons.append("Unusual drop compared to last year's average.")
                elif current_pattern.mean() > 1.5 * last_year_avg:
                    reasons.append("Unusual rise compared to last year's average.")
                else:
                    reasons.append("Irregular monthly pattern detected.")
            else:
                reasons.append("Irregular monthly pattern detected.")

            tips.append(
                "1. Inspect the meter physically for tampering.\n"
                "2. Compare with neighborhood usage for similar consumption patterns.\n"
                "3. Check if there are bypass connections or rewiring."
            )
        else:  # No theft detected
            reasons.append("No irregularities detected.")
            tips.append("No action required.")

    # Results
    df['Anomaly'] = ['Theft' if pred == -1 else 'No Theft' for pred in predictions]
    df['Reason'] = reasons
    df['Tips'] = tips
    return df[['Consumer ID', 'Anomaly', 'Reason', 'Tips']]

# Function for manual input detection
def detect_theft_from_manual_input(consumer_id, *monthly_kwh):
    # Create a DataFrame from manual input
    df = pd.DataFrame(
        {
            'Consumer ID': [consumer_id],
            **{f'Month {i+1}': [monthly_kwh[i]] for i in range(12)},
        }
    )

    # Normalize the data
    data = df.iloc[:, 1:].values  # Select monthly data
    scaler = StandardScaler()
    data_scaled = scaler.fit_transform(data)

    # One-Class SVM model
    model = OneClassSVM(nu=0.1, kernel='rbf', gamma='scale')
    model.fit(data_scaled)

    # Predict anomaly
    prediction = model.predict(data_scaled)

    # Add Reason and Tips
    if prediction[0] == -1:  # Theft detected
        reason = "Irregular monthly pattern detected."
        tips = (
            "1. Inspect the meter physically for tampering.\n"
            "2. Compare with neighborhood usage for similar consumption patterns.\n"
            "3. Check if there are bypass connections or rewiring."
        )
    else:
        reason = "No irregularities detected."
        tips = "No action required."

    df['Anomaly'] = ['Theft' if prediction[0] == -1 else 'No Theft']
    df['Reason'] = [reason]
    df['Tips'] = [tips]
    return df[['Consumer ID', 'Anomaly', 'Reason', 'Tips']]

# Gradio Interface
file_interface = gr.Interface(
    fn=detect_theft_from_file,
    inputs=gr.File(label="Upload Meter Data File (CSV, XLSX, XLS)"),
    outputs=gr.Dataframe(label="Meter Theft Detection")
)

manual_input_interface = gr.Interface(
    fn=detect_theft_from_manual_input,
    inputs=[
        gr.Textbox(label="Consumer ID", placeholder="Enter Consumer ID"),
        *[gr.Number(label=f"Month {i+1} kWh") for i in range(12)],
    ],
    outputs=gr.Dataframe(label="Meter Theft Detection")
)

# Combine interfaces
iface = gr.TabbedInterface(
    interface_list=[file_interface, manual_input_interface],
    tab_names=["Upload File", "Manual Input"]
)

# Launch the Gradio app
iface.launch()