This repository contains ocr_transcribe.py, a small Python script that uses pytesseract to OCR images and create a markdown file (e.g. chapter_1.md).

Prerequisites

• Tesseract OCR must be installed on your machine and the tesseract binary must be on PATH.

Install Tesseract (platform-specific)

• macOS:

  brew install tesseract

• Linux (Debian/Ubuntu) or Windows (WSL):

  sudo apt update
  sudo apt install -y tesseract-ocr libtesseract-dev

Install Python dependencies in your venv (on Windows or inside WSL). We pin a small set of packages here; adjust versions if needed.

Using pip:

python -m pip install numpy pillow pytesseract

Using uv (if you prefer to manage packages with uv):

uv add numpy pillow pytesseract

Run the script on your chapter folder:

python ocr_transcribe.py --input-dir static/chapter_1 --output-file chapter_1.md

Or run the script using uv run (uses the active uv environment):

uv run python ocr_transcribe.py --input-dir static/chapter_1 --output-file chapter_1.md

Options:

• --resize: max width in pixels to resize images before OCR (default 2000).
• --lang: tesseract language codes (default eng).

Notes on privacy and cost:

• Script runs locally with pytesseract so your images are not uploaded.

If you want, I can add a small helper to compress/resize images prior to uploading to an API, but given your preference for local processing this README guides running under WSL for Windows users.