app.py

from flask import Flask, url_for, request, jsonify, render_template

import numpy as np 
import os

import ast
import warnings
warnings.simplefilter('ignore')

from nltk.tokenize import sent_tokenize
import push_notifs

import re
import string

import helpers
import tags
import pickle

from tensorflow.keras.preprocessing.sequence import pad_sequences

#load model, weights, tokenizers

model = helpers.open_model_from_json(filename='model/model.json', weights='model/best_acc_bank_weights.hdf5')
model.compile(loss = 'binary_crossentropy',
              metrics=['accuracy'],
              optimizer='adam')
tokenizer = pickle.load(open('model/tokenizer.pickle','rb'))

app = Flask(__name__)

@app.route('/')
def home():
	return 'Hello World'


@app.route('/sentiment_score', methods=['POST', "GET"])
def sentiment_score(tokenizer = tokenizer, model = model, maxlen=30):

	review = request.args['review'] #read reviews from html

	review = helpers.normalize(review)
	feature_vec = tokenizer.texts_to_sequences([review])

	feature_vec = pad_sequences(feature_vec, maxlen=maxlen)

	predictions = model.predict(feature_vec)[0]

	classes = ['negative','positive']

	pred = 0 if predictions <0.5 else 1

	caught = tags.lookup(review) #get tags of a particular sentence

	response = {
		"sentiment":classes[pred],
		"tags":caught
	}
	return response


@app.route('/notification', methods=['GET','POST'])
def notification():

	reviews = request.json['feedbacks']
	# reviews = ast.literal_eval(reviews)

	res = {}

	for r in reviews:
		review = sent_tokenize(r)

		positives = push_notifs.get_promotions(review)
		for i in positives.keys():
			res[i] = positives[i]


	return jsonify(res)


if __name__ == '__main__':
	port = int(os.environ.get("PORT", 5000))
	app.run(threaded=False ,host='0.0.0.0',port=port)