import gradio as gr
import numpy as np
import tensorflow as tf
from tensorflow.keras.models import load_model
from sklearn.preprocessing import MinMaxScaler

# Load the saved model
model = load_model('lstm_model')

# Dummy scaler (replace this with the actual scaler used in training)
scaler = MinMaxScaler(feature_range=(0, 1))

# Function for Prediction
def predict_sales(input_data):
    # Reshape and scale input data
    input_data = np.array(input_data).reshape(1, 30, 1)
    scaled_input = scaler.transform(input_data)

    # Predict
    prediction = model.predict(scaled_input)
    actual_prediction = scaler.inverse_transform(prediction)

    return actual_prediction[0][0]

# Gradio Interface
inputs = gr.inputs.Dataframe(headers=["Sales"], type="numpy", row_count=30, col_count=1)
outputs = gr.outputs.Textbox()

interface = gr.Interface(fn=predict_sales, inputs=inputs, outputs=outputs, title="LSTM Sales Forecasting")
interface.launch()