readablepdf 0.2.0

Turn image-only PDFs into searchable, selectable PDFs with an OCR text layer

readablepdf

readablepdf converts image-only PDFs into searchable and selectable PDFs by adding an OCR text layer while preserving page visuals.

How it works

1. Render each page to a PNG (pdf2image backed by Poppler).
2. OCR each page with Tesseract in PDF mode (image layer + invisible text layer).
3. Merge all OCR page PDFs into one output PDF.

All intermediate files are written inside an OS temporary directory and removed automatically.

System dependencies (Linux + macOS)

You need these binaries on the machine where you run readablepdf:

• tesseract
• pdftoppm (from Poppler)

Ubuntu/Debian:

sudo apt update
sudo apt install -y tesseract-ocr poppler-utils

macOS (Homebrew):

brew install tesseract poppler

If you need languages beyond English, install matching Tesseract language packs.

Install and run with pipx

Once published on PyPI, you can run it directly without managing a virtualenv:

pipx run readablepdf input.pdf

Custom output/language/DPI:

pipx run readablepdf input.pdf -o output_ocr.pdf --lang eng --dpi 200

Local development

python3 -m venv .venv
source .venv/bin/activate
pip install -e .[dev]
ruff check .
ruff format --check .
pytest -q
python -m build

GitHub Actions

• CI workflow: lint + format check + tests + build on Ubuntu and macOS.
• Publish to PyPI workflow: runs on GitHub Release publish (and manual trigger).

GitHub configuration needed for publish

1. Create repository secret: PYPI_API_TOKEN.
2. Set it to a PyPI API token with upload permission for this project.
3. (Optional but recommended) create a protected environment named pypi and require approvals.

Release flow

1. Merge to main.
2. Create a Git tag/release (for example v0.1.1).
3. Publish to PyPI workflow uploads artifacts to PyPI with version derived from that tag.