import tensorflow as tf
from tensorflow import keras
import numpy as np
import matplotlib.pyplot as plt
import pandas as pd
from sklearn.model_selection import train_test_split
from sklearn import preprocessing
import seaborn as sns
from sklearn.preprocessing import LabelEncoder
import streamlit as st 

st.title("Rouge Component Prediction")
#Reading Dataset
df = pd.read_csv('identify_rogue_50K_ALL.csv')
print("Dataset Size:",df.shape)
st.sidebar.header('Enter the Components Details here')
# Dropping the SRU serial number
df.drop(['SRU serial number','Date of Manufacture','Last Maintenance Date','date of last failure'], axis = 1, inplace=True)

# DATA from user
def user_report():
  manufacturer = st.sidebar.selectbox("Manufacturer",
        ("JKL Company", "GHI Company","AGS Company","ABC Company","XYZ Company" ))
  if manufacturer=='JKL Company':
      manufacturer=3
  elif manufacturer=="GHI Company":
      manufacturer=2
  elif manufacturer=="AGS Company":
      manufacturer=1
  elif manufacturer=="ABC Company":
      manufacturer =0
  else:
      manufacturer=4
  component_age = st.sidebar.slider('Component Age (in hours)', 500,2000, 600 )
  total_operating_hours = st.sidebar.slider('Total Operating Hours)', 50,2000, 500 )
  usage_intensity = st.sidebar.slider('Usage Intensity hours/day', 0,9, 5 )
  last_maintance_type = st.sidebar.selectbox('Last Mantainence Type', ("Preventive","Corrective") )
  if last_maintance_type=="Preventive":
      last_maintance_type=1
  else:
      last_maintance_type=0
  previous_number_of_repairs = st.sidebar.number_input('Enter the Previous Number of Repairs Undergone  0 to 5 )',min_value=0,max_value=5,step=1)
  operating_temperature = st.sidebar.slider('Operating Temperature', 10,25, 15 )
  humidity = st.sidebar.slider('Humidity', 20,105, 25 )
  Vibration_Level = st.sidebar.slider('Vibration Level', 2,7, 2 )
  Pressure = st.sidebar.slider('Pressure', 200,550, 250 )
  Power_Input_Voltage= st.sidebar.slider('Power Input Voltage (V)',100,133,115)
  repair_type = st.sidebar.selectbox('Repair Type', ("Hardware","Software") )
  if repair_type=='Hardware':
      repair_type=0
  else:
      repair_type=1
  number_of_inspection = st.sidebar.selectbox('Number of Inspections',('1','2'))
  if number_of_inspection=='1':
      number_of_inspection=1
  else:
      number_of_inspection=2
  number_of_inspection_6months = st.sidebar.selectbox('Number of Inspections in last 6 Months',('0','1'))
  if number_of_inspection_6months=='0':
      number_of_inspection_6months=0
  else:
      number_of_inspection_6months=1
  prior_maintainence = st.sidebar.selectbox('Prior Maintainence',("Regular","Irregular"))
  if prior_maintainence =='Regular':
      prior_maintainence=1
  else:
      prior_maintainence=0

  user_report_data = {
      'Manufacturer':manufacturer,
      'Component_Age':component_age,
      'Total Operating Hours':total_operating_hours,
      'Usage Intensity (hours/day)':usage_intensity,
      'Last Maintenance Type': last_maintance_type,
      'Previous number of repairs':previous_number_of_repairs,
      'Operating Temperature':operating_temperature,
      'Humidity': humidity,
      'Vibration Level':Vibration_Level,
      'Pressure':Pressure,
      'Power Input Voltage (V)':Power_Input_Voltage,
      'repair type':repair_type ,
      'total number of inspection':number_of_inspection,
      'No. of Inspections in Last 6 Months':number_of_inspection_6months,
      'Prior Maintenance':prior_maintainence
           
  }
  report_data = pd.DataFrame(user_report_data, index=[0])
  return report_data    
    
#Customer Data
user_data = user_report()
st.header("Component Details")
st.write(user_data)

def label_encoder(df):
  le = LabelEncoder()
  cat = df.select_dtypes(include='O').keys()
  categ = list(cat)
  df[categ] = df[categ].apply(le.fit_transform)
  return df

def preprocess_dataset(X):  
  x = X.values #returns a numpy array
  min_max_scaler = preprocessing.MinMaxScaler()
  x_scaled = min_max_scaler.fit_transform(x)
  X_df = pd.DataFrame(x_scaled)
  return X_df

def prediction(df):
  #X = df.loc[:,df.columns!= "Rogue LRU/SRU (Target)"]
  #y = df["Rogue LRU/SRU (Target)"]
  #X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.1, random_state=42)
  #print(X_train.shape)
  #print(X_test.shape)
  X_test_encoded = label_encoder(df)
  X_test_df = preprocess_dataset(X_test_encoded)
  #x_model = loaded_model = tf.keras.models.load_model('my_model')
    #x_model = tf.keras.models.load_model('my_model.h5')
  try:
        x_model = tf.keras.models.load_model('my_model.h5')
  except (OSError, ValueError):
        # Load the model as a TFSMLayer if .h5 loading fails
        tfs_layer = tf.keras.layers.TFSMLayer('my_model', call_endpoint='serving_default')
        # Create a new model using the TFSMLayer
        inputs = tf.keras.Input(shape=(X_test_df.shape[1],))
        outputs = tfs_layer(inputs)
        x_model = tf.keras.Model(inputs=inputs, outputs=outputs)
  y_pred = x_model.predict(X_test_df)
  #predicition = []
  #for i in list(y_pred):
  if y_pred ==0:
      return 'Component is Good'
  else:
      return 'Component is not Good'
  #X_test['Actual_time_to_repair'] = y_test
  #X_test['Predicted_time_to_repair'] = predicition
  # X_test.to_csv(r'/content/drive/MyDrive/Colab Notebooks/HAL/rogue_test_data.csv')
  #print(X_test.head())

y_pred = prediction(user_data)

if st.button("Predict"):
    st.subheader(y_pred)