==========
Enrich SDK
Enrich is a customizable, privacy law-aware enterprise Feature Store
of Scribble Data
_. This SDK is part of the feature store stack. It
enables local development, testing and documentation of simple to
complex feature transformations and other modules required for
building and managing robust features.
This is not for general purpose use. Please get in touch with us at
hello@scribbledata.io to discuss potential use of this SDK.
For enterprise users of Enrich, the documentation is available on the
server. Please see the Developer section.
.. _Scribble Data: https://www.scribbledata.io
Usage
- sudo apt-get update
- sudo apt-get install python3.8-dev python3.8-venv
- python3 -m venv venv
- pip3 install wheel
- pip3 install enrichsdk
- enrichpkg start
Docs
- python3 -m venv venv
- pip3 install -r requirements.txt
- mkdocs serve
Updating Release Tag
-
Add to $HOME/.bashrc. Make sure you source ~/.bashrc before using::
function git_move_tag {
git push origin; git tag -d $1; git tag $1 ; git push origin --tags --force
}
function git_push {
git push origin $1; git push origin $1 --tags ;
git push origin $1 refs/notes/*
}
function git_rm_tag {
git push origin; git tag -d $1; git push --delete origin $1
}
-
Bump the version. Version is typically a.b.c::
activate the environment
workon dev
cd scribble-enrichsdk # cd to the root
If bumpversion doesnt exist
pip install bump2version
Check existing tags once
git tag -l
Bump the right version
bumpversion patch # for updating c
bumpversion minor # for updating b
bumpversion major # for updating a
-
Push the changes to github::
See above
git_push master
-
Update deployment version::
cd scribble-deploy-v2/configuration
update enrichsdk_branch
vi defaults.json
Make
git commit -a -m "Updated enrichsdk version"
git push origin
-
Install at customer::
cd scribble-deploy
fab install_enrichsdk:role=demo
-
Upload to pypi.::
python3 setup.py sdist
This will require token in ~/.pypirc
twine upload -r pypi dist/enrichsdk-5.0.4.tar.gz
cat ~/.pypirc
[pypi]
username = token
password = pypi-AgEIcHlwaS5v...
-
Known issues
The package dependencies are a jungle.Problematic packages include:
boto3
botocore
aiobotocore
jupyter-events
nbconvert
Use the fix environment script to monkey-patch any dependencies in the worst case::
./bin/fix-environment.py
Usage: fix-environment.py [OPTIONS] COMMAND [ARGS]...
This package will help fix the environment
Options:
--help Show this message and exit.
Commands:
lib Fix a library dependency