🛡️ Phishing URL Detection Pipeline

This project compares a legitimate (real) URL against a suspect URL to detect possible phishing. It checks for:

• 🔗 Domain & URL similarity (lookalike domains, subdomains, punycode)
• 📄 Page text similarity
• 📸 Visual similarity using full-page screenshots
• 📝 JSON report with results & screenshot paths

--

📦 Installation

1. Clone this repo / download the code

git clone https://github.com/yourusername/phishing-checker.git
cd phishing-checker

2. Install dependencies

pip install -r requirements.txt

requirements.txt should contain:

beautifulsoup4
tldextract
requests
opencv-python
scikit-image
undetected-chromedriver
selenium

---

🚀 Usage

1. Run with CLI script

python phishing_checker_cli.py <real_url> <suspect_url>

Example:

python phishing_checker_cli.py https://www.paypal.com/login https://paypal.secure-login-support.com

2. Output

• A JSON report will be saved in the current folder:

  result_2025-08-16_18-05-40.json
  

• Screenshots will be saved in a timestamped folder:

  screenshots/2025-08-16_18-05-40/
      ├── real.png        # Legitimate page
      ├── suspect.png     # Suspect page
      └── diff.png        # Highlighted visual differences
  

• Example JSON output:

{
    "real_domain": "paypal.com",
    "suspect_domain": "secure-login-support.com",
    "domain_similarity": 0.61,
    "url_findings": [
        "Domain mismatch",
        "Lookalike domain"
    ],
    "text_similarity": 0.42,
    "visual_similarity": 0.75,
    "real_screenshot": "screenshots/2025-08-16_18-05-40/real.png",
    "suspect_screenshot": "screenshots/2025-08-16_18-05-40/suspect.png",
    "diff_image": "screenshots/2025-08-16_18-05-40/diff.png",
    "phishing_likely": true
}

---

⚙️ Files

• phishing_checker.py → main pipeline (logic + screenshot + comparison)
• phishing_checker_cli.py → command line wrapper for easy usage
• screenshots/ → auto-generated timestamped screenshot folders
• result_*.json → JSON reports per run

---

🛠 Notes

• Runs in headless mode using undetected-chromedriver (Chrome required).
• Works on Linux / macOS / Windows.
• Needs internet connection to fetch pages.
• For heavy use, increase timeout in requests / Selenium.

---