Shroud: PDF Redaction

These scripts perform text and image redaction on PDFs using AI-powered text detection and PyMuPDF. It consists of two python scripts:

1. redact_text.py: Redacts sensitive text based on predefined specifications.
2. redact_images.py: Redacts sensitive entities found in images within the PDF.

---

Features

• Text Redaction: Detects and redacts text based on a customizable list of terms or phrases.
• Image Redaction: Extracts images, uses AI (GPT-4 Vision) to identify sensitive entities, and redacts the images.

---

Prerequisites

Ensure you have the following installed:

• Python 3.8 or higher
• pip (Python package manager)
• Tesseract OCR: Required for redacting text from images and flat PDFs.

To insall the required packages, run:

./install.sh

If necessary, make the script executable before running:

chmod +x install.sh

---

Tesseract Configuration

Shroud requires Tesseract’s trained models (tessdata).
By default, the script is configured for macOS, using:

os.environ["TESSDATA_PREFIX"] = "/usr/local/share/tessdata"
os.environ["TESSERACT_CMD"] = "/usr/local/bin/tesseract"

For Intel Mac users,

OR

os.environ["TESSDATA_PREFIX"] = "/opt/local/share/tessdata"
os.environ["TESSERACT_CMD"] = "/opt/local/bin/tesseract"

for Apple Silicon Mac users.

If you are on Linux or Windows, update these values:

• Linux: /usr/share/tesseract-ocr/5/tessdata
• Windows: C:\Program Files\Tesseract-OCR\tessdata

To manually override, set the environment variables before running the script:

export TESSDATA_PREFIX="/path/to/tessdata"
export TESSERACT_CMD="/path/to/tesseract"

---

OpenAI key configuration:

The scripts use a config.json file to store your OpenAI API key. To configure your key, create a file named config.json in the root directory with the following format:

{
    "openai": {
        "api_key": "openai-api-key"
    }
}

---

Running

Run the shell script with the appropriate arguments:

For text-only redaction (default):

bash run.sh text

For text and image redaction:

bash run.sh text image

If necessary, make the script executable before running: chmod +x run.sh

Folder Structure After Running:

./input_pdfs/                # Original PDFs
./output_pdfs/               # Fully redacted PDFs
./partially_redacted_pdfs/   # Text-redacted PDFs (used for images)

---