Safe URL is a flask web app that utilizes machine learning to check whether a URL is safe, unsafe, or non-existent. It uses a model trained on the UCI Phishing Dataset (https://archive.ics.uci.edu/ml/datasets/phishing+websites) along with the Random Forest Classifier algorithm to essentially predict the status of a URL. This app includes a feature extractor which breaks an input URL into 30 distinct features for the model to analyze. The general feature categories include

• Address Bar Features
• HTML/JS Features
• Domain Features
• Abnormalities

Each feature translates to either a 1 for safe, 0 for suspicious, or -1 for unsafe. The combination of each of these numeric values across the various features yields the overall URL status.

• app.py: Default page of the app; handles routing

• featureExtraction.py: Handles logic to extract all features and return a list containing the numeric equivalent of each feature

• prediction.py: Includes model training using Random Forest Classifier

• phishing.csv: UCI Phishing Dataset

• templates/index.html: HTML page template

• static/style.css: Styling of HTML page

• pandas
• BeautifulSoup4
• scikit-learn
• googlesearch-python
• urllib3
• regex
• python-whois

Install all of the dependencies. For any libraries not installed, install manually.

  pip install requirements.txt

Run app.py to bring up the webpage on localhost.

  python app.py

Before Check

After Check