import numpy as np 
import pandas as pd
# Fonction de prédiction
import gradio as gr
import joblib
# importer les encodeurs
encoder0 = joblib.load('Fuel_Type.joblib')
encoder1 = joblib.load('Seller_Type.joblib')
encoder2 = joblib.load('Transmission.joblib')
# importer le modèle
xgb = joblib.load('xgb.joblib')
# importer le normaliseur
scaler = joblib.load('scaler.joblib')
def Pred_func(Kms_Driven,Present_Price, Fuel_Type, Seller_Type, Transmission, Age):

  # Encoder les valeurs des Fuel_Type, Seller_Type et Transmission
  Fuel_Type = encoder0.transform([Fuel_Type])[0]
  Seller_Type = encoder1.transform([Seller_Type])[0]
  Transmission = encoder2.transform([Transmission])[0]
  # vecteur des valeurs numériques
  x_new = np.array([Kms_Driven,Present_Price, Fuel_Type, Seller_Type, Transmission, Age])
  x_new = x_new.reshape(1,-1) # convert en un 2D array
  # Normaliser les données
  x_new = scaler.transform(x_new)
  # Prédire
  y_pred = xgb.predict(x_new)
  # Arrondir
  y_pred = round(y_pred[0],2)
  return str(y_pred) + 'k$'

# Fonction de prédiction multiple
def Pred_func_csv(file):
  # Lire le fichier csv
  df = pd.read_csv(file)
  predictions = []
  # Boucle sur les lignes du dataframe
  for row in df.iloc[:, :].values:
    # nouvelle ligne avec les valeurs des Fuel_Type, Seller_Type et Transmission encodées
    new_row = np.array([row[0], row[1], encoder0.transform([row[2]])[0], encoder1.transform([row[3]])[0], encoder2.transform([row[4]])[0], row[5]])
    new_row = new_row.reshape(1,-1) # convertir en un 2D array
    # Normaliser les données
    new_row = scaler.transform(new_row)
    # Prédire
    y_pred = xgb.predict(new_row)
    # Arrondir
    y_pred = round(y_pred[0],2)
    # ajouter la prediction sur List_predictions
    predictions.append(y_pred)

  df['Selling_Price'] = predictions
  df.to_csv('predictions.csv', index = False)
  return 'predictions.csv'

# définir les blocks
demo = gr.Blocks(theme='NoCrypt/miku')

# Créer les inputs
inputs = [gr.Number(label='Kms Driven'),
          gr.Number(label='Present Price'),
          gr.Radio(choices=['Petrol', 'Diesel', 'CNG'], label='Fuel Type'),
          gr.Radio(choices=['Dealer', 'Individual'], label='Seller Type'),
          gr.Radio(choices=['Manual', 'Automatic'], label='Transmission'),
          gr.Number(label='Age')]
# Créer les outputs
outputs = gr.Textbox(label='Selling Price')
# Créer l'interface 1
interface1 = gr.Interface(fn = Pred_func,
                         inputs = inputs,
                         outputs = outputs,
                         title="Predict the selling price of a car with a single input",
                         description = """This machine learning model allows us to predict the selling price of a car
                         from the kms driven, present price, fuel type, seller type, transmission and age of the car.
                         """)
# Créer l'interface 2
interface2 = gr.Interface(fn = Pred_func_csv,
                         inputs = gr.File(label='Upload a csv file'),
                          outputs = gr.File(label='Download a csv file'),
                         title="Predict the selling price of a car with a multiple inputs",
                         description = """This machine learning model allows us to predict the selling price of a car
                         from the kms driven, present price, fuel type, seller type, transmission and age of the car.
                         """)

# faire un tabbing des interfaces
with demo:
  gr.TabbedInterface([interface1, interface2], ['Simple Prediction', 'Prédiction multiple'])

# lancer l'interface
demo.launch()