No conda version

[GrowlingM1ke]

Hi there,

I noticed the docs recommend installing invoice2data via pip and then manually fetching all dependencies and input modules. Since all of these dependencies are already on conda-forge+pip, it would be much faster and less error-prone to provide:

• A dedicated invoice2data conda package, or
• At minimum an environment.yml that pins all required packages.

Would you consider adding one of these to simplify installation for conda users?

Edit: Brief update, ocrmyPDF is a bit fiddly to get working and requires something outside of conda

[bosd]

I think it would be interesting addition to also publish to conda.

[bosd]

Added an environment.yml to the repo (next branch, ships in 1.0.0) so conda/mamba users can get a working setup in one step:

conda env create -f environment.yml
conda activate invoice2data

The nice part for conda users: poppler, tesseract and ghostscript are all on conda-forge, so the environment.yml pulls them in too — no separate system-library install, which addresses the ocrmypdf/poppler fiddliness you ran into. It also includes the common PDF backends (pypdfium2, pdfplumber, pdfminer.six) and ocrmypdf. invoice2data itself is installed from PyPI inside the env (its runtime deps are already satisfied by the conda packages). Docs updated under Installation → Installation using conda.

A dedicated conda-forge package (a feedstock via conda-forge/staged-recipes) is the natural next step and I agree it's worth doing — leaving this issue open to track that. One wrinkle to sort first: 1.0.0 will start shipping compiled (mypyc) wheels, so the feedstock recipe needs to build the extension rather than be a noarch pure-Python package.