i have a file that contains a bunch of udf and classes that are common to different projects. I saw in this post that the best way to have one library where all your files pulls from, is to package it. ImportError: No module named <something>
The problem is if I follow the tutorial at some point I must go trhough pypi.org, but the material I'm developping is propriety of my company and i cannot upload it online. https://packaging.python.org/tutorials/packaging-projects/
Is there a way to follow the same steps and package my file as a library locally ? And then install it through pip locally ?
EDIT :
I should have mentioned that I'm using Anaconda. According to their doc :https://docs.conda.io/projects/conda-build/en/latest/user-guide/tutorials/build-pkgs.html
I was able to start a build but now i run into dependencies problems. Probably due to my little understanding of what goes in the meta.yaml file. Here's mine :
{% set name = "EDSE_mapping_func" %}
{% set version = "1.0" %}
package:
name: "{{ name|lower }}"
version: "{{ version }}"
source:
url:
sha256:
build:
number: 0
script: "{{ PYTHON }} -m pip install . --no-deps --ignore-installed -vv "
requirements:
host:
- pip
- python
- numpy
- pandas
- sklearn
- re
- string
- nltk
run:
- python
test:
imports:
- EDSE_mapping_func
about:
home:
When ever i execute conda build . from the directory where i have my Python file, my meta.yaml and the bld.bat i get the following message :
conda_build.exceptions.DependencyNeedsBuildingError: Unsatisfiable dependencies for platform win-64: {'re', 'sklearn', 'string'}
I've found a piece of solution here, but I don't understand where and how i'm suppose the declare the channels :
Conda Build - creating a Conda package fails on dependencies for platform win-64
Thank you !
