main.py

import pickle

import pandas as pd
from asgiref.wsgi import WsgiToAsgi
from flask import Flask, request, jsonify, render_template, send_from_directory

from UvicornWrapper import UvicornWrapper

app = Flask(__name__)
asgi_app = WsgiToAsgi(app)

# Load the preprocessor
with open('Models/data_preprocessor.pkl', 'rb') as preprocessor_file:
    preprocessor = pickle.load(preprocessor_file)
model = pickle.load(open('Models/model.pkl', 'rb'))


# Create the home page
@app.route('/')
def home():
    return render_template('home.html')


@app.route('/images/<filename>')
def get_image(filename):
    return send_from_directory('static/images', filename)


# Create the api end point for prediction
@app.route('/predict_api', methods=['POST'])
def predict_api():
    data = request.json
    # Convert JSON to DataFrame
    data_df = pd.DataFrame([data])
    transformed_data = preprocessor.transform(data_df)
    output = model.predict(transformed_data.squeeze().reshape(1, -1))[0]
    prediction = ""
    output_pro = model.predict_proba(transformed_data.squeeze().reshape(1, -1))[0]
    prediction = f"The person has stage {output} liver cirrhosis with {output_pro[1]*100:.2f}% of chance"
    return jsonify(prediction)


@app.route('/predict', methods=['POST'])
def predict():
    data = dict(request.form)
    data_df = pd.DataFrame([data])
    transformed_data = preprocessor.transform(data_df)
    output = model.predict(transformed_data.squeeze().reshape(1, -1))[0]
    prediction = ""
    output_pro = model.predict_proba(transformed_data.squeeze().reshape(1, -1))[0]
    prediction = f"The person has stage {output} liver cirrhosis with {output_pro[1]*100:.2f}% of chance"
    return render_template("home.html", prediction_text=prediction)


if __name__ == '__main__':
    # Only for local development. To run the app from IDE.
    uvicorn = UvicornWrapper(asgi_app)
    uvicorn.run()