Possible Issue with the current License

[gvalvano]

🚀 The feature

I see the library has been released following the Apache-2.0 license. However, it relies on GPLv2+ license dependencies.
Can we remove such dependencies?

Motivation, pitch

To really have Apache 2.0, we must ensure that the library is actually usable with that license!

In particular, pip-installing python-doctr will also install:

• pyphen (0.16.0): ['GNU General Public License v2 or later (GPLv2+)', 'GNU Lesser General Public License v2 or later (LGPLv2+)', 'Mozilla Public License 1.1 (MPL 1.1)'] --> installed by weasyprint
• Unidecode (1.3.8): ['GNU General Public License v2 or later (GPLv2+)']

Alternatives

No response

Additional context

weasyprint==61.2
    # via python-doctr
pyphen==0.16.0
    # via weasyprint
pyphen==0.16.0
    # via weasyprint


2 packages.
    pyphen (0.16.0): ['GNU General Public License v2 or later (GPLv2+)', 'GNU Lesser General Public License v2 or later (LGPLv2+)', 'Mozilla Public License 1.1 (MPL 1.1)']
      dependency:
          pyphen
    Unidecode (1.3.8): ['GNU General Public License v2 or later (GPLv2+)']
      dependency:
          Unidecode

[felixdittrich92]

Hi @gvalvano 👋

weasyprint is an optional dependency and will be only installed by the html extensions for example:

pip install "python-doctr[torch,html]"

:) ref.: https://mindee.github.io/doctr/getting_started/installing.html

It's needed if you want to use:

webpage_doc = DocumentFile.from_url("https://www.yoursite.com")

About Unidecode:

This was already replaced in:
709404e

So the last pypi release v0.9.0 does not have unidecode as dependency and weasyprint is optional :)

[gvalvano]

Nice! :D thanks for your answer! This solved my issue!

Do you think we can add this to the installation guide, somewhere? So it is easier to find it for future users

[felixdittrich92]

Hi @gvalvano 🤗
Yes sure would you like to open a PR ?
We could add it here: https://github.com/mindee/doctr/blob/main/docs/source/getting_started/installing.rst