Super simple solution.

1. Install tesseract and the corresponding data package for your language
2. Install ocrmypdf either from AUR or pip.

If your document is scanned/has no text layer, run it as

ocrmypdf input.pdf output_with_text.pdf

If it does have text, it doesn't always merge cleanly. You can use –force-ocr but that rasterizes the file and makes it massive.