# Import necessary libraries
import numpy as np
import joblib  # For loading the serialized model
import pandas as pd  # For data manipulation
from flask import Flask, request, jsonify  # For creating the Flask API

# Initialize the Flask application
product_sales_predictor_api = Flask("SuperKart Product Sales Predictor")

# Load the trained machine learning model
model = joblib.load("product_sales_prediction_model_v1_0.joblib")

store_id_categories = ['OUT001', 'OUT002', 'OUT003', 'OUT004']
store_size_categories = ['High', 'Medium', 'Small']
city_type_categories = ['Tier 1', 'Tier 2', 'Tier 3']
store_type_categories = ['Departmental Store', 'Food Mart', 'Supermarket Type1', 'Supermarket Type2']
product_group_categories = ['Non-Food/Household', 'Packaged/Processed Foods', 'Perishable Foods']
sugar_content_categories = ['Low Sugar', 'No Sugar', 'Regular']

model_columns = [
    'Product_Weight', 'Product_Allocated_Area', 'Product_MRP', 'Store_Age',
    'Product_Sugar_Content_Low Sugar', 'Product_Sugar_Content_No Sugar', 'Product_Sugar_Content_Regular',
    'Store_Id_OUT001', 'Store_Id_OUT002', 'Store_Id_OUT003', 'Store_Id_OUT004',
    'Store_Size_High', 'Store_Size_Medium', 'Store_Size_Small',
    'Store_Location_City_Type_Tier 1', 'Store_Location_City_Type_Tier 2', 'Store_Location_City_Type_Tier 3',
    'Store_Type_Departmental Store', 'Store_Type_Food Mart', 'Store_Type_Supermarket Type1', 'Store_Type_Supermarket Type2',
    'Product_Group_Non-Food/Household', 'Product_Group_Packaged/Processed Foods', 'Product_Group_Perishable Foods'
]

# Define a route for the home page (GET request)
@product_sales_predictor_api.get('/')
def home():
    """
    This function handles GET requests to the root URL ('/') of the API.
    It returns a simple welcome message.
    """
    return "Welcome to the SuperKart Product Sales Prediction API!"

# Define an endpoint for single property prediction (POST request)
@product_sales_predictor_api.post('/v1/sales')
def predict_product_sales():
    """
    This function handles POST requests to the '/v1/sales' endpoint.
    It expects a JSON payload containing product details and returns
    the predicted sales as a JSON response.
    """
    # Get the JSON data from the request body
    product_data = request.get_json()

    numeric_features = {
        'Product_Weight': product_data['Product_Weight'],
        'Product_Allocated_Area': product_data['Product_Allocated_Area'],
        'Product_MRP': product_data['Product_MRP'],
        'Store_Age': product_data['Store_Age']
    }

    one_hot_features = {col: 0 for col in model_columns if col not in numeric_features.keys()}

    # Fill one-hot for Product_Sugar_Content
    sugar_key = f"Product_Sugar_Content_{product_data['Product_Sugar_Content']}"
    if sugar_key in one_hot_features:
        one_hot_features[sugar_key] = 1

    # Fill one-hot for Store_Id
    store_id_key = f"Store_Id_{product_data['Store_Id']}"
    if store_id_key in one_hot_features:
        one_hot_features[store_id_key] = 1

    # Fill one-hot for Store_Size
    store_size_key = f"Store_Size_{product_data['Store_Size']}"
    if store_size_key in one_hot_features:
        one_hot_features[store_size_key] = 1

    # Fill one-hot for Store_Location_City_Type
    city_type_key = f"Store_Location_City_Type_{product_data['Store_Location_City_Type']}"
    if city_type_key in one_hot_features:
        one_hot_features[city_type_key] = 1

    # Fill one-hot for Store_Type
    store_type_key = f"Store_Type_{product_data['Store_Type']}"
    if store_type_key in one_hot_features:
        one_hot_features[store_type_key] = 1

    # Fill one-hot for Product_Group
    product_group_key = f"Product_Group_{product_data['Product_Group']}"
    if product_group_key in one_hot_features:
        one_hot_features[product_group_key] = 1

    # Combine all features into single dict
    final_features = {**numeric_features, **one_hot_features}

    # Create DataFrame with model columns order
    input_df = pd.DataFrame([final_features], columns=model_columns)

    # Predict with the model
    predicted_sale = model.predict(input_df).tolist()[0]
    predicted_sale = round(float(predicted_sale), 2)
    
    # Return the actual price
    return jsonify({'Predicted Sales': predicted_sale})


# Define an endpoint for batch prediction (POST request)
@product_sales_predictor_api.post('/v1/salesbatch')
def predict_product_sales_batch():
    """
    This function handles POST requests to the '/v1/salesbatch' endpoint.
    It expects a CSV file containing product details for multiple sales
    and returns the predicted sales as a dictionary in the JSON response.
    """
    # Get the uploaded CSV file from the request
    file = request.files['file']

    # Read the CSV file into a Pandas DataFrame
    df = pd.read_csv(file)

    #Creating df with numeric features. 
    numeric_features = df[['Product_Weight', 'Product_Allocated_Area', 'Product_MRP', 'Store_Age']]

    # One-hot encode categorical columns consistently for all rows
    # Create empty DataFrame for one-hot with all model columns except numeric ones
    one_hot_df = pd.DataFrame(0, index=df.index, columns=[col for col in model_columns if col not in numeric_features.columns])

    # Fill one hot columns for each categorical feature
    for category in ['Low Sugar', 'No Sugar', 'Regular']:
        col_name = f'Product_Sugar_Content_{category}'
        one_hot_df.loc[df['Product_Sugar_Content'] == category, col_name] = 1

    for store_id in ['OUT001', 'OUT002', 'OUT003', 'OUT004']:
        col_name = f'Store_Id_{store_id}'
        one_hot_df.loc[df['Store_Id'] == store_id, col_name] = 1

    for store_size in ['High', 'Medium', 'Small']:
        col_name = f'Store_Size_{store_size}'
        one_hot_df.loc[df['Store_Size'] == store_size, col_name] = 1

    for city_type in ['Tier 1', 'Tier 2', 'Tier 3']:
        col_name = f'Store_Location_City_Type_{city_type}'
        one_hot_df.loc[df['Store_Location_City_Type'] == city_type, col_name] = 1

    for store_type in ['Departmental Store', 'Food Mart', 'Supermarket Type1', 'Supermarket Type2']:
        col_name = f'Store_Type_{store_type}'
        one_hot_df.loc[df['Store_Type'] == store_type, col_name] = 1

    for product_group in ['Non-Food/Household', 'Packaged/Processed Foods', 'Perishable Foods']:
        col_name = f'Product_Group_{product_group}'
        one_hot_df.loc[df['Product_Group'] == product_group, col_name] = 1

    # Concatenate numeric features and one-hot encoded features in model column order
    input_df = pd.concat([numeric_features, one_hot_df], axis=1)
    input_df = input_df.reindex(columns=model_columns, fill_value=0)

    # Make predictions for all properties in the DataFrame (get log_prices)
    predicted_sales = model.predict(input_df).tolist()
    prod_id_list = df['Product_Id'].tolist()

    # Create a dictionary of predictions with property IDs as keys
    output_dict = dict(zip(prod_id_list, predicted_sales))
    
    # Return the predictions dictionary as a JSON response
    return output_dict

# Run the Flask application in debug mode if this script is executed directly
if __name__ == '__main__':
    product_sales_predictor_api.run(debug=True)