app.py

import streamlit as st
from GoogleNews import GoogleNews

import pandas as pd
import numpy as np
import spacy
import gensim
import string
import re
import tweepy
import emoji
import sklearn
from sklearn.metrics import classification_report, recall_score, precision_score, accuracy_score, f1_score
from sklearn.metrics.pairwise import cosine_similarity

#--------------------------------------------------------------------------------------------------------------- 
#-----------------------------------------  Twitter key and authentication ------------------------------------- 
#--------------------------------------------------------------------------------------------------------------- 
consumerKey = 'RMhfvyw1KCfg31b0TWBscMLtx'
consumerSecret = 'jZ2axRqcAmG2BYR4LKkUBfhzef6lTDwA7AWy0NpAI5Po7bo9UZ'
accessToken = '56190940-G2zHEicg1ZI1B7dtZbu7XbksOLEY4lpNWj93FOWxL'
accessTokenSeceret='oLMNJJ4FF0omugCXcO5cfoQsaD7Wm15pkQ1koO9SbEurE'

auth = tweepy.OAuthHandler(consumerKey, consumerSecret)
auth.set_access_token(accessToken, accessTokenSeceret)
api = tweepy.API(auth)

#--------------------------------------------------------------------------------------------------------------- 
#----------------------------------------------  Side bar ------------------------------------------------------ 
#--------------------------------------------------------------------------------------------------------------- 
st.sidebar.markdown('مواقع اخباريه معتمده ')
st.sidebar.markdown("[العربية](https://www.alarabiya.net/)")
st.sidebar.markdown("[الجزيرة نت](https://www.aljazeera.net/news/)")
st.sidebar.markdown("[وكالة الانباء الكويتية](https://www.kuna.net.kw/Default.aspx?language=ar)")


#--------------------------------------------------------------------------------------------------------------- 
#----------------------------------------------load Aravec model -----------------------------------------------
#--------------------------------------------------------------------------------------------------------------- 
nlp = spacy.load("spacy.aravec.model")

st.write("""
YAQEN: is an Arabic headline news detection system Submitted to the College of Graduate Studies in Partial Fulfillment of the Requirements for the M.Sc.
Degree in:Computing Information Systems
Submitted by:
Reem Alfouzan
Project Supervisor: 
Dr. Abdullah Almutairi
""")

#read input from user 
tx = st.text_input (''' الرجاء ادخال العنوان المراد التاكد من صحته ''')

#--------------------------------------------------------------------------------------------------------------- 
#----------------------------------------Pre-processing functions---------------------------------------------- 
#--------------------------------------------------------------------------------------------------------------- 
def clean_str(text):
    search = ["أ","إ","آ","ة","_","-","/",".","،"," و "," يا ",'"',"ـ","'","ى","\\",'\n', '\t','"','?','؟','!']
    replace = ["ا","ا","ا","ه"," "," ","","",""," و"," يا","","","","ي","",' ', ' ',' ',' ? ',' ؟ ',' ! ']
    
    #remove tashkeel
    p_tashkeel = re.compile(r'[\u0617-\u061A\u064B-\u0652]')
    text = re.sub(p_tashkeel,"", text)
    
    #remove longation
    p_longation = re.compile(r'(.)\1+')
    subst = r"\1\1"
    text = re.sub(p_longation, subst, text)
    
    text = text.replace('وو', 'و')
    text = text.replace('يي', 'ي')
    text = text.replace('اا', 'ا')
    
    for i in range(0, len(search)):
        text = text.replace(search[i], replace[i])
    
    #trim    
    text = text.strip()

    return text

#removing hashtags 
def split_hashtag_to_words(tag):
    tag = tag.replace('#','')
    tags = tag.split('_')
    if len(tags) > 1 :
        return tags
    pattern = re.compile(r"[A-Z][a-z]+|\d+|[A-Z]+(?![a-z])")
    return pattern.findall(tag)

def clean_hashtag(text):
    words = text.split()
    text = list()
    for word in words:
        if is_hashtag(word):
            text.extend(extract_hashtag(word))
        else:
            text.append(word)
    return " ".join(text)

def is_hashtag(word):
    if word.startswith("#"):
        return True
    else:
        return False

def extract_hashtag(text):
    
    hash_list = ([re.sub(r"(\W+)$", "", i) for i in text.split() if i.startswith("#")])
    word_list = []
    for word in hash_list :
        word_list.extend(split_hashtag_to_words(word))
        
    return word_list
    
 #clean tweets and remove urls and emojies
def clean_tweet(text):

    text = re.sub(r"http\S+", "", text)
    text = re.sub(r"www.\S+", "", text)
    text = ''.join(c for c in text if c not in emoji.UNICODE_EMOJI)

    return text

# Define the preprocessing Class
class Preprocessor:
    def __init__(self, tokenizer, **cfg):
        self.tokenizer = tokenizer

    def __call__(self, text):
        preprocessed = clean_str(text)
        return self.tokenizer(preprocessed)   
#--------------------------------------------------------------------------------------------------------------- 
#---------------------------------------- @No_Rumors Twitter check --------------------------------------------- 
#--------------------------------------------------------------------------------------------------------------- 
#is an independent project that was established in 2012 to address and contain rumors so that they do not pose any harm to society by exposing the 
#rumors that aim to stir up public opinion by spreading awareness and clarifying the truth through official sources.

def No_Rumors(tx):
    no_of_tweets = api.get_user(screen_name="No_Rumors").statuses_count 
    tweets = tweepy.Cursor(api.user_timeline,screen_name="No_Rumors", tweet_mode="extended").items(no_of_tweets)
    data = []
    for i in tweets:
        data.append({"text":i.full_text})
        
    result_text = pd.DataFrame(data)    

    if len(result_text) == 0:
        prediction = 1
      
    else :
      result_text['text'] = result_text['text'].apply(lambda x: clean_tweet(x))
      result_text['text'] = result_text['text'].apply(lambda x: clean_hashtag(x))
      result_text['text'] = result_text['text'].apply(lambda x: clean_str(x))
      result_text['text'] = result_text['text'].apply(lambda x: nlp(x).similarity(nlp(tx)))
      #assign maximum similarity rate
      sg300top = result_text['text'].max(axis = 0)

      #if similarity higher than .85 then it is a rumor 
      if sg300top >= .85:
        prediction = ' الخبر زائف'
      else:
        prediction = ' الخبر صحيح'        
        
    return prediction
            
#--------------------------------------------------------------------------------------------------------------- 
#----------------------------------------- END OF PRE-PROCESSING------------------------------------------------ 
#--------------------------------------------------------------------------------------------------------------- 
# Apply the `Preprocessor` Class

nlp.tokenizer = Preprocessor(nlp.tokenizer)

if len(tx) != 0:
    googlenews = GoogleNews(lang='ar')
    googlenews.clear()

    f =0 
    Prediction =''
    top_similar_ind =''
    top_similar_news =''
    medium =''
    top_similar_ind2 =''
    tp_desc =''

    st.markdown(f"Searching for: { tx         }")
    st.markdown(f"ــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــــ")


    tx = clean_hashtag(tx)
    tx = clean_str(tx) 


    googlenews.search(tx)
    result = googlenews.page_at(1)
    googlenews.clear()

    if len(result) == 0:
      Prediction ='الخبر زائف'
      top_similar_news ='لا يوجد اخبار مماثله'
      medium ='لا يوجد مصدر'
      tp_desc ='لا يوجد وصف'

    else:
      result_text = {"Text":[]}
      
        #google search
      for i in range(len(result)):
        title =result[i]['title']
        result_text['Text'].append(title)
       

      result_text2 = {"Text":[]}
        #google search
      for i in range(len(result)):
        desc =result[i]['desc']
        result_text2['Text'].append(desc) 
        
      result_text = pd.DataFrame(result_text)
      result_text2 = pd.DataFrame(result_text2)
      
      data = pd.DataFrame()
      data['Text2'] = result_text['Text'].copy()
      
      data['Text2'] = data['Text2'].apply(lambda x: nlp(x).similarity(nlp(tx)))
      sg300top = data['Text2'].max(axis = 0)
      
      top_similar_ind = np.argmax(data['Text2'])
      top_similar_news = result[top_similar_ind]['title']
      descr = result[top_similar_ind]['desc']
      medium = result[top_similar_ind]['media']
      date = result[top_similar_ind]['date']
      link = result[top_similar_ind]['link']
      
      data['Text3'] = result_text2['Text'].copy()
      data['Text3'] = data['Text3'].apply(lambda x: nlp(x).similarity(nlp(tx)))
      sg300top2 = data['Text3'].max(axis = 0)
      top_similar_ind2 = np.argmax(data['Text3'])
      tp_desc = result[top_similar_ind2]['desc']
      
      if sg300top >= .85 or sg300top2 >= .85 :
        Prediction = No_Rumors(tx)
      else:
        Prediction =' الخبر زائف'
        
                
    
    st.markdown(f"System Prediction : { Prediction         }")
    st.markdown(f"الخبر المماثل: { top_similar_news         }")
    st.markdown(f"")
    st.markdown(f"تاريخ الخبر: { date         }")
    st.markdown(f"")
    st.markdown(f"التفصيل: { descr         }")
    st.markdown(f"")
    st.markdown(f"المصدر: { medium         }")
    st.markdown(f"")
    st.markdown(f"رابط الخبر: { link         }")
    




#st.markdown(f"Searching for: { tx         }")