app.py

import base64, secrets, io, os
from PIL import Image, ImageOps
from urllib.request import urlopen

from flask import Flask, request, jsonify, render_template
import numpy as np
import psycopg2
import psycopg2.extras

from tensorflow.keras.preprocessing.image import img_to_array
from tensorflow.keras.preprocessing import image
from tensorflow.keras.models import load_model

#model = load_model("./model/model.h5")

app = Flask(__name__)

#Enter here your database informations
DB_HOST = "ec2-18-235-86-66.compute-1.amazonaws.com"
DB_NAME = "da1c06fi82ev6c"
DB_USER = "tjhqznxnlxlreh"
DB_PASS = "f21ee19a9e50d41255f16288c978201fbbffd3fc9e22b644fb03b904f5216056"

conn = psycopg2.connect(dbname=DB_NAME, user=DB_USER, password=DB_PASS, host=DB_HOST)


@app.route("/")
def index():
    return render_template("index.html")


@app.route("/fetchrecords",methods=["POST","GET"])
def fetchrecords():
    cur = conn.cursor(cursor_factory=psycopg2.extras.DictCursor)
    if request.method == 'POST':
        try:
            search_word = request.form['query']
            print(search_word)
            if search_word == '':
                query = "SELECT title, authors, published_year from books_eng"
                cur.execute(query)
                titles = cur.fetchall()
            else:
                cur.execute('SELECT title, authors, published_year from  books_eng WHERE LOWER(title) LIKE LOWER(%(name)s) LIMIT 25', {'name': '%{}%'.format(search_word)})
                numrows = int(cur.rowcount)
                titles = cur.fetchall()
                print(numrows)
        except:
            pass
    return jsonify({'htmlresponse': render_template('response.html', titles=titles, numrows=numrows)})


'''@app.route("/api", methods=["POST"])
def main():
    miJSON = request.json
    try:
        imagen64 = miJSON["imagen"]
        imgdata = urlopen(imagen64)
        imgdata = imgdata.read()
        imgdata = Image.open(io.BytesIO(imgdata))

        password_length = 13
        extension = "png"
        nomre_unico = f'{secrets.token_urlsafe(password_length)}.{extension}'
        imgdata = imgdata.resize((28, 28))
        imgdata.save(nomre_unico)

        im = image.load_img(nomre_unico, color_mode='grayscale', target_size=(28, 28))
        os.remove(nomre_unico)

        etiqueta = predecir_im(im, invertir=False)
        respuesta = {"etiqueta": etiqueta}

        return jsonify(respuesta)
    except:
        return {"error": "Tuvimos un problema"}'''


if __name__ == "__main__":
    app.run()


'''def predecir_im(im, invertir=True):
    image = img_to_array(im)
    image.shape

    # Scale the image pixels by 255 (or use a scaler from sklearn here)
    image /= 255

    # Flatten into a 1x28*28 array
    img = image.flatten().reshape(-1, 28 * 28)
    img.shape
    if invertir:
        img = 1 - img

    # plt.imshow(img.reshape(28, 28), cmap=plt.cm.Greys)

    resultado = model.predict(img)
    resultado = np.argmax(resultado, axis=-1)
    return int(resultado[0])'''