first commit

ap.py

@@ -0,0 +1,35 @@
+# import modules
+from flask import Flask, render_template, session, redirect, url_for
+from flask_wtf import FlaskForm
+from wtforms import SubmitField, TextAreaField
+
+import env_variable
+import main
+
+app = Flask(__name__)
+
+app.config['SECRET_KEY'] = env_variable.app_security_key
+
+class TextForm(FlaskForm):
+    text = TextAreaField('Text')
+    submit = SubmitField('Send')
+
+
+@app.route('/', methods=['GET', 'POST'])
+def index():
+    form = TextForm()
+    if form.validate_on_submit():  # Check if form is submitted
+        session['text'] = form.text.data
+        return redirect(url_for("moderator"))
+    return render_template('home.html', form=form)
+
+
+@app.route('/moderator')
+def moderator():
+    text = session['text']
+    result = main.content_moderator(text)
+    return render_template('moderator.html', result=result)
+
+
+if __name__ == '__main__':
+    app.run(debug=True)

app.py

@@ -0,0 +1,14 @@
+import streamlit as st
+import main
+
+
+def main():
+    text = st.text_input("Enter the text to classify: ")
+    if text:
+        results = main.content_moderator(text)
+        st.write(results)
+    else:
+        st.write("Please enter some text to classify")
+    st.write("Waiting for input...")
+    
+main()

main.py

@@ -0,0 +1,56 @@
+from transformers import pipeline
+import requests as re
+
+labels = ['toxic', 'racist', 'gender bias', 'religious bias', 'aggressive',
+          'personal attacks', 'hate speech', 'offensive language']
+
+# model chosen is fbbart
+model = pipeline("zero-shot-classification",
+                 model="facebook/bart-large-mnli")
+
+
+def zero_shot_classifier(text, model=model, labels=labels):
+    text = text
+    prediction = model(text, labels, multi_label=True)
+    if prediction["scores"][0] < 0.75:
+        return False
+    else:
+        return str(prediction["labels"][0])
+
+
+# catch darkweb links
+def dark_web_links(text):
+    # Regular expression pattern to match common dark web link formats
+    dark_web_pattern = r"(https?://)?[a-z0-9]+\.onion(/[a-zA-Z0-9]*)*"
+
+    # Search for dark web links in the text
+    matches = re.findall(dark_web_pattern, text)
+
+    if matches:
+        return True
+    else:
+        return False
+
+
+# catch adult content links
+def adult_content_sites(text):
+    # Regular expression pattern to match common adult content websites
+    adult_content_pattern = r"(https?://)?(?:www\.)?(pornhub\.com|xnxx\.com|youporn\.com|redtube\.com|etc)\b"
+
+    # Search for adult content sites in the text
+    matches = re.findall(adult_content_pattern, text)
+
+    # If any matches are found, flag them
+    if matches:
+        return True
+    else:
+        return False
+
+
+def content_moderator(text, labels=labels, model=model, classifier=zero_shot_classifier, dkweb=dark_web_links, adult=adult_content_sites):
+    if dkweb(text):
+        return 'Darkweb link'
+    elif adult(text):
+        return 'Adult content site'
+    elif classifier(text):
+        return 'Flagged content'

requirements.txt

@@ -0,0 +1,3 @@
+transformers
+torch
+streamlit