app.py

from flask import Flask, render_template, request, redirect, url_for, flash,send_file
import nltk

import hypernetx as hnx
import matplotlib.pyplot as plt
nltk.download('stopwords')
from nltk.corpus import stopwords
import pandas as pd
import pickle
import re
import io
from nltk.stem.porter import PorterStemmer
from sklearn.feature_extraction.text import TfidfVectorizer
app = Flask(__name__)
app.secret_key = 'many random bytes'
usr=""
port_stem = PorterStemmer()
stop_words = set(stopwords.words('english')) - set(['not','no','don\'t','n\'t'])


#divide words to cases
#case 1 :  Not hate word
#case 2 : Hate word
#case 3 : Hate word, but depends on context
#case 4 : possibility of hate word, nbut high indeterminancy
#case 5 : not hate hate word, but depends on context

case1=['want','gonna','stop','apart','done','insid','bye','felicia','use','would','king','kong','insert','basebal','bat','male','domest','cell','she','still','thing','think','absolut','already','sun','readi','share','complet','meet','heaven','eat','abl','sayin','start','might','yo','forget','sooooooo','anoth','urself','privladg','ur','finish','read','amaz','book','recommend','love','spend','time','famili','friend','grate','opportun','far','great','workout','gym','today','excit','try','new','recip','dinner','tonight','music','favorit','artist','proud','hard','work','put','achieve','goal','kind','respect','norm','import','take','care','health','watch','inspir','ted','talk','feel','motiv','support','life','walk','alway','help','clear','mind','educ','key','travel','see','world','one','day','other','let','focu','unit','us','rather','divid','beauti','like','make','differ','someon','good','idea','learn','skill','keep','hous','apt','cold','hobbi','fulfil','break','must','bodi','refresh','benefici','cracker','somehow','avoid','get','shone','fr','bought','ah','sub','da','corner','sto','easi','dream','rememb','joyou','nazi','tell','bath','sorri','desir','say','abolish','gettin','gang','bred','beta',
'rough','ball','yeah','cleanin','deep','load','sweat','drippin','total','final','url','gettin','pleas','send','video','surround','also','follow','drool','bad','commun']

case2=['rape','boipussi','white','goat','fucker','brutalis','suck','slutti','whore','sex','fat','homo','dumb','shit','dead','racist','hoe','bitch','pussi','stupid','ugli','ass','divers','give','back','nigga','gay','fuck','bash','cunt','filthi','dirti','cock','cum','floodin','slut','dildo']

case3=['bull','terrorist','kill','thick','breed','swear','useless','spit','clout','drug','feend','bore','social','part','death','run','power','countri','greedi','bottom']


case4=['hole','cri','german','disgust','pound','smash','capri','sever','femal','choke','gal','men','hang','beat','lil','closet','mental','spread','inclus','penalti','short','first','freaki','punch','face','muslim','illeg','christian','daddi']

case5=['wish','huge','no','call','damn','photo','free','enjoy','deserv','not','except','peopl','go','outsid','unlock','potenti','small','act','make','out','father','big','posit','black','way',
]


#create subhypergraph from a threshold
c={}
speech=[]
speech11=[]
erosionk=[] 
hatespeech=[]
nonhatespeech=[]
listx=[]
listy=[]
listk=[]
listr=[]
listg=[]
data1=[]
@app.route('/')
def home():
    return render_template('homepage.html')


x={}
a=-1
def graph(s):
  global a,x
  word_list = s.split()
  a+=1
  x[a]=word_list


x1={}

def graph3(s):
  global a,x
  word_list = s.split()
  a+=1
  x1[a]=word_list

def graph1(s):

  global a,c
  
  word_list = s.split()
  a+=1
  c[a]=word_list

def weights(x):
  global hatespeech,erosionk,speech,nonhatespeech,speech1
  t=0
  y=0
  f=0
  c=0
  words=x.split()
  for i in words:
    c=c+1
    if i in case1:
      r=0
      if(t<r):
        t=r
      y=y+0
      f=f+1
    elif i in case2:
      t=1

    elif i in case3:
      t=1
      y=y+0.5
    elif i in case4:
      r=0.5
      if(t<r):
        t=r
      y=y+1
    elif i in case5:
      y=y+0.5
      f=f+1
  speech.append(x)
  speech11.append(x)
  erosionk.append(x)

  if((t>0.8 and y>0.3)):
    print(x," : hate speech")
    hatespeech.append(x)
    graph1(x)
    
   
  else:
    print(x," : non hate speech")
    nonhatespeech.append(x)

def cleantext(text):
  x=str(text).lower().replace('\\','').replace('_','')
  tag=' '.join(re.sub("(@[A-Za-z0-9]+)|([^0-9A-Za-z \t])|(\w+:\/\/\S+)"," ",x).split())
  spcl=tag.replace('[^\w\s]','')
  return spcl

def stemming(content):
    review = re.sub('[^a-zA-Z]',' ',content)
    review = review.lower()
    review = review.split()
    review = [port_stem.stem(word) for word in review if not word in stop_words]
    review = ' '.join(review)
    return review

def model1(user):
    pickle_in=open("SVM_model.pickle","rb")
    classifier=pickle.load(pickle_in)
    prediction=classifier.predict(user)
    if prediction==0:
        return("Non Hate Speech")
    else:
        return("Hate Speech")

def model2(user):
    pickle_in=open("RF_model (1).pickle","rb")
    classifier=pickle.load(pickle_in)
    prediction=classifier.predict(user)
    if prediction==0:
        return("Non Hate Speech")
    else:
        return("Hate Speech")

def model3(user):
    pickle_in=open("LR_model.pickle","rb")
    classifier=pickle.load(pickle_in)
    prediction=classifier.predict(user)
    if prediction==0:
        return("Non Hate Speech")
    else:
        return("Hate Speech")
        #just pushing

@app.route('/Home')
def hoe():
    return render_template('homepage.html')

@app.route('/Predict',methods=['GET','POST'])
def pre():
    return render_template('prediction.html')

@app.route('/Tpred',methods=['GET','POST'])
def pred():
    user=request.form["tweet"]
    pickle_in=open("vectorized_data1.pickle","rb")
    vectorizer=pickle.load(pickle_in)
    user=cleantext(user)
    user= stemming(user)
    user=vectorizer.transform([str(user)])
    mod1=model1(user)
    mod2=model2(user)
    mod3=model3(user)
    return render_template('modeloutput.html',mod1=mod1,mod2=mod2,mod3=mod3)


@app.route('/upload', methods=['POST'])
def upload():
    global a,data1
    a=-1
    csv_file=request.files['csv']
    df = pd.read_csv(csv_file)
    data=df.iloc[:, 0]
    data1=df.iloc[:, 0]
    data=data.apply(lambda x:cleantext(x))
    data= data.apply(stemming)
    d=data.apply(lambda y:graph(y))
    d=data.apply(lambda z:weights(z))
    return('',204)
   
@app.route('/hypergraph1', methods=['POST'])
def hypergraph():
    # Generate the URL for the image file
    G = hnx.Hypergraph(x)
    # Draw the graph using matplotlib
    fig, ax = plt.subplots()
    hnx.draw(G, ax=ax)
    ax.set_title("Main Hypergraph : Graph created using all the sentences in the dataset")
    # Save the plot to an image file
    png_output = io.BytesIO()
    fig.savefig(png_output, format='png')
    png_output.seek(0)

    # Send the PNG file to the client
    return send_file(png_output, mimetype='image/png')

@app.route('/hypergraph2', methods=['POST'])
def hypergraph2():
    G = hnx.Hypergraph(c)
    # Draw the graph using matplotlib
    fig, ax = plt.subplots()
    hnx.draw(G, ax=ax)
    ax.set_title("Hypergraph X : Hate Speech Sentences above the threshold value")
    # Save the plot to an image file
    png_output = io.BytesIO()
    fig.savefig(png_output, format='png')
    png_output.seek(0)

    # Send the PNG file to the client
    return send_file(png_output, mimetype='image/png')

@app.route('/Dilationnodes')
def dilation():
   #w.r.t nodes
   global listx,hatespeech
 
   for i in hatespeech:
      words=i.split()
      for j in words:
        if j not in listx:
          listx.append(j)

   return render_template('dilationnodes.html',listx1=listx)

@app.route('/Dilationedges')
def dilation1():
   #w.r.t edges
   global listx,listg
   for i in listx:
    for j in speech:

      words=j.split()
      if i in words:
        if j not in listg:
          listg.append(j)
   return render_template('dilationedges.html',listx1=listg)
    
@app.route('/Erosionnodes')
def erosion1():  
   #erosion w.r.t nodes
    global nonhatespeech,listk,listx,listy
  
    for i in nonhatespeech:
      words=i.split()
      for j in words:
        if j not in listk:
          listk.append(j)
    for i in listx:
      if i not in listk:
        listr.append(i)
    return render_template('erosionnodes.html',listy1=listr)

@app.route('/Erosionedges')
def erosion2(): 
        global listx,erosionk
        for y in erosionk:
            words=y.split()
            for k in words:
              if k not in listx:
                erosionk.remove(y)
                break
        return render_template('erosionedges.html',speech1=erosionk)
@app.route('/output')
def output1():
  for i in hatespeech:
    words=i.split()
    for j in words:
      for y in speech:
        wordslist=y.split()
        if j in wordslist:
          speech.remove(y)
  for i in speech11:
     if i in speech:
        speech11.remove(i)
  return render_template('output.html',speech1=speech,speech2=speech11)

@app.route('/output1')
def output11():
  global hatespeech,speech,speech11,data1
  for i in hatespeech:
    words=i.split()
    for j in words:
      for y in speech:
        wordslist=y.split()
        if j in wordslist:
          speech.remove(y)
  for i in speech11:
     if i in speech:
        speech11.remove(i)

  for j in data1:
     d=cleantext(j)
     d=stemming(d)
     for i in speech:
        if d==i:
           
           speech.remove(i)
           speech.append(j)
   

     for i in speech11:
     
        if d==i:
           speech11.remove(i)
           speech11.append(j)
          
           
          
  return render_template('output1.html',speech1=speech,speech2=speech11)

   
@app.route('/outputgraph')
def output2():
  global a

  for i in hatespeech:
    words=i.split()
    for j in words:
      for y in speech:
        wordslist=y.split()
        if j in wordslist:
          speech.remove(y)
  for i in speech:
    graph3(i)
  G = hnx.Hypergraph(x1)
    # Draw the graph using matplotlib
  fig, ax = plt.subplots()
  hnx.draw(G, ax=ax)
  ax.set_title("Final Output Hypergraph : Graph after removing hatespeech ")

    # Save the plot to an image file
  png_output = io.BytesIO()
  fig.savefig(png_output, format='png')
  png_output.seek(0)

    # Send the PNG file to the client
  return send_file(png_output, mimetype='image/png')

@app.route('/accuracy',methods=['POST'])
def accuracy():
  return render_template('image.html')

@app.route('/Fuzzy',methods=['POST'])
def fuz():
    return render_template('fuzzy.html')

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