app.py

import streamlit as st
import pickle
from pymongo import MongoClient
import pandas as pd
from sentence_transformers import SentenceTransformer, util
import requests
import matplotlib.pyplot as plt
import matplotlib.image as mpimg
from io import BytesIO
import urllib.parse
import math

sbert_model = SentenceTransformer('paraphrase-multilingual-mpnet-base-v2')

try:
    client = MongoClient('mongodb://192.168.1.103:27017/')

    print("---Connenction Successful---")

    Recommendation_elderly = client['Recommendation_elderly']

    healthcare_articles = Recommendation_elderly['token']

except:
    raise KeyError('Connection Fail')

data = healthcare_articles.find()
data = pd.DataFrame(list(data))
data = data.drop_duplicates(subset=['url'])
data = data[data['title'] != '']
data = data.reset_index().drop(columns=['index'])
data = data.reset_index().drop(columns=['_id','index'])

with open('corpus_embeddings.pickle', 'rb') as file:
    corpus_embeddings = pickle.load(file)

def personal_check(age,weight,height,gender):

    #age check
    if age >= 60:
        age = 'ผู้สูงอายุ'
    else:
        age = 'วัยทำงาน'

    #gender check
    if gender == 'Female':
        gender = 'ผู้หญิง สตรี'
    else:
        gender = 'ผู้ชาย'

    #bmi check
    height_meters = height / 100  

    bmi = weight / (height_meters ** 2)

    if bmi >= 30:
        bmi = 'อ้วนมาก'
    elif bmi >= 23 and bmi <30:
        bmi = 'อ้วน'
    elif bmi >= 18.5 and bmi <23:
        bmi = ''
    else:
        bmi = 'ผอม'
    
    return age,gender,bmi

def sbert_search(queries):

    global sbert_model,corpus_embeddings,data

    index_lst = []
    score_lst = []

    for query in queries:
        query_embedding = sbert_model.encode(query, convert_to_tensor=True)
        hits = util.semantic_search(query_embedding, corpus_embeddings, top_k=15)
        hits = hits[0]
        for hit in hits:
            index_lst.append(hit['corpus_id'])
            score_lst.append(hit['score'])

    sbert_searched = data.iloc[index_lst]
    sbert_searched['score'] = score_lst
    sbert_searched = sbert_searched[['url','title','score','banner']]

    return sbert_searched

def visualize_articles_images(title,banner):
    # Calculate the number of rows and columns for the grid
    num_images = len(banner)
    num_rows = math.ceil(num_images / 3)
    num_cols = min(num_images, 3)
    fp = 'angsana.ttc'

    # Create a grid of subplots
    fig, axs = plt.subplots(num_rows, num_cols, figsize=(20, 20))

    # Iterate over the image URLs
    for i, url in enumerate(banner):
        # Calculate the subplot position
        row = i // num_cols
        col = i % num_cols
        axs[row, col].set_title(title.iloc[i],fontname='Tahoma',fontsize=16)

        if str(url) == 'nan':
            continue

        else:
            try:
                # Encode the URL using UTF-8
                encoded_url = urllib.parse.quote(url, safe=':/')

                # Download the image
                response = requests.get(encoded_url)
                img = mpimg.imread(BytesIO(response.content), format='jpg')

                # Calculate the subplot position
                row = i // num_cols
                col = i % num_cols

                # Plot the image
                axs[row, col].imshow(img)
                axs[row, col].axis('off')
            except:
                continue
            finally:
                pass

    return fig

def main():
    #header
    st.title("---ระบบแนะนำบทความสุขภาพ---")
    st.subheader("ให้คะแนนบทความหน่อยนะครับ:smile:")

    #personal information input
    age = st.slider("อายุ", 0, 100, 25)
    weight = st.number_input("น้ำหนัก (Kg.): ",30,120,step=1,value=30)
    height = st.number_input("ส่วนสูง (cm.): ",100,250,step=1,value=120)
    gender = st.selectbox('เพศ',('ชาย', 'หญิง'))
    food_allergy = st.selectbox('แพ้อาหาร?',('ไม่แพ้', 'แพ้อาหาร'))
    drug_allergy = st.selectbox('แพ้ยา?',('ไม่แพ้', 'แพ้ยา'))
    congentital_disease = st.text_input('โรคประจำตัวของคุณ')

    # Add a button
    if st.button("Click me"):
        age,gender,bmi = personal_check(age,weight,height,gender)

        if food_allergy == 'ไม่แพ้':
            food_allergy = ''
        if drug_allergy == 'ไม่แพ้':
            drug_allergy = ''

        queries = [gender+age+bmi+food_allergy+drug_allergy+congentital_disease]

        sbert_searched = sbert_search(queries)

        st.write(f"{queries}")
        st.pyplot(visualize_articles_images(sbert_searched['title'],sbert_searched['banner']))
    
if __name__ == "__main__":
    main()