Python module for the CognitiveScale Cortex Cognitive Platform


License
Apache-2.0
Install
pip install cortex-python==1.4.0a1

Documentation

Python Module for the Cortex Cognitive Platform

The Cortex Python module provides an API client library to easily integrate with the Cortex Cognitive Platform. Refer to the Cortex documentation for details on how to use the library:

Installation

To install:

poetry add cortex-python

or from source code:

git clone git@github.com:CognitiveScale/cortex-python.git
cd cortex-python
poetry install

To install the optional components:

When developing models using jupyter within notebooks

poetry install cortex-python[models_dev]

When using model SDK using jupyter within notebooks

poetry install cortex-python[models_dev]

When using the library within Skills

poetry install cortex-python[models_runtime]

Development

Setup

When developing, it's a best practice to work in a virtual environment. Create and activate a virtual environment:

poetry install
poetry shell

Install developer dependencies:

git clone git@github.com:CognitiveScale/cortex-python.git
cd cortex-python
make dev.install

Run Developer test and linting tasks: Three types of checks are configured for this:

  1. symilar - to test code duplication
  2. pylint - for linting
  3. pytest - for running the unit tests. These are orchestrated through tox. The tox configuration is available at tox.ini

There's a convenience Makefile that has commands to common tasks, such as build, test, etc. Use it!

Testing

Unit Tests

Follow above setup instructions (making sure to be in the virtual environment and having the necessary dependencies)

  • make test to run test suite

To run an individual file or class method, use pytest. Example tests shown below:

  • file: pytest test/unit/agent_test.py
  • class method: pytest test/unit/agent_test.py::TestAgent::test_get_agent

Publishing an alpha build

Suppose you want to release new functionality so it can be installed without releasing a new official version. We need to use an alpha version in PyPi.

  • we need to create and publish an alpha release:
  • get credentials to the cortex-python pypi CognitiveScale account (via lastpass)
  • run make dev.push. The alpha pre-release number (the N in X.Y.ZaN) with be determined automatically.

Contribution

After contributing to the library, and before you submit changes as a PR, please do the following

  1. Run unit tests via make test
  2. Manually verification (i.e. try the new changes out in Cortex) to make sure everything is going well. Not required, but highly encouraged.
  3. Bump up setup.py version and update the CHANGELOG.md

Documentation

Activate your virtual environment:

poetry shell

Set up your environment, if you have not done so:

make dev.install 

The package documentation is built with Sphinx and generates versioned documentation for all tag matching the release/X.Y.Z pattern and for the master branch. To build the documentation:

make docs.multi

The documentation will be rendered in HTML format under the docs/_build/${VERSION} directory.

Pre-release to staging

Note: this repository using git tag for versionning

  1. Create and push an alpha release:
git tag -a 6.5.0a<N> -m 'alpha tag'
git push --tags
make dev.push
This will build an alpha-tagged package.
  1. Merge develop to staging branch:
make stage
  1. In GitHub, create a pull request from staging to master.
git tag -a 6.5.0 -m 'rlease tag'
git push --tags