README.md

Flytekit Python Plugins

All the Flytekit plugins maintained by the core team are added here. It is not necessary to add plugins here, but this is a good starting place.

Currently Available Plugins 🔌

Plugin	Installation	Description	Version	Type
AWS Sagemaker Training	bash pip install flytekitplugins-awssagemaker	Installs SDK to author Sagemaker built-in and custom training jobs in python		Backend
dask	bash pip install flytekitplugins-dask	Installs SDK to author dask jobs that can be executed natively on Kubernetes using the Flyte backend plugin		Backend
Hive Queries	bash pip install flytekitplugins-hive	Installs SDK to author Hive Queries that can be executed on a configured hive backend using Flyte backend plugin		Backend
K8s distributed PyTorch Jobs	bash pip install flytekitplugins-kfpytorch	Installs SDK to author Distributed pyTorch Jobs in python using Kubeflow PyTorch Operator		Backend
K8s native tensorflow Jobs	bash pip install flytekitplugins-kftensorflow	Installs SDK to author Distributed tensorflow Jobs in python using Kubeflow Tensorflow Operator		Backend
K8s native MPI Jobs	bash pip install flytekitplugins-kfmpi	Installs SDK to author Distributed MPI Jobs in python using Kubeflow MPI Operator		Backend
Papermill based Tasks	bash pip install flytekitplugins-papermill	Execute entire notebooks as Flyte Tasks and pass inputs and outputs between them and python tasks		Flytekit-only
Pod Tasks	bash pip install flytekitplugins-pod	Installs SDK to author Pods in python. These pods can have multiple containers, use volumes and have non exiting side-cars		Flytekit-only
spark	bash pip install flytekitplugins-spark	Installs SDK to author Spark jobs that can be executed natively on Kubernetes with a supported backend Flyte plugin		Backend
AWS Athena Queries	bash pip install flytekitplugins-athena	Installs SDK to author queries executed on AWS Athena		Backend
DOLT	bash pip install flytekitplugins-dolt	Read & write dolt data sets and use dolt tables as native types		Flytekit-only
Pandera	bash pip install flytekitplugins-pandera	Use Pandera schemas as native Flyte types, which enable data quality checks.		Flytekit-only
SQLAlchemy	bash pip install flytekitplugins-sqlalchemy	Write queries for any database that supports SQLAlchemy		Flytekit-only
Great Expectations	bash pip install flytekitplugins-great-expectations	Enforce data quality for various data types within Flyte		Flytekit-only
Snowflake	bash pip install flytekitplugins-snowflake	Use Snowflake as a 'data warehouse-as-a-service' within Flyte		Backend
dbt	bash pip install flytekitplugins-dbt	Run dbt within Flyte		Flytekit-only
Huggingface	bash pip install flytekitplugins-huggingface	Read & write Hugginface Datasets as Flyte StructuredDatasets		Flytekit-only
DuckDB	bash pip install flytekitplugins-duckdb	Run analytical workloads with ease using DuckDB		Flytekit-only

Have a Plugin Idea? 💡

Please file an issue.

Development 💻

Flytekit plugins are structured as micro-libs and can be authored in an independent repository.

Refer to the Python microlibs blog to understand the idea of microlibs.

The plugins maintained by the core team can be found in this repository and provide a simple way of discovery.

Unit tests 🧪

Plugins should have their own unit tests.

Guidelines 📜

Some guidelines to help you write the Flytekit plugins better.

1. The folder name has to be flytekit-*, e.g., flytekit-hive. In case you want to group for a specific service, then use flytekit-aws-athena.

2. Flytekit plugins use a concept called Namespace packages, and thus, the package structure is essential.

   Please use the following Python package structure:

   flytekit-myplugin/
      - README.md
      - setup.py
      - flytekitplugins/
          - myplugin/
             - __init__.py
      - tests
          - __init__.py
   

   NOTE: the inner package flytekitplugins DOES NOT have an __init__.py file.

3. The published packages have to be named flytekitplugins-{package-name}, where {package-name} is a unique identifier for the plugin.

4. The setup.py file has to have the following template. You can use it as is by editing the TODO sections.

from setuptools import setup

# TODO put the plugin name here
PLUGIN_NAME = "<plugin-name e.g. pandera>"

# TODO decide if the plugin is regular or `data`
# for regular plugins
microlib_name = f"flytekitplugins-{PLUGIN_NAME}"
# For data/persistence plugins
# microlib_name = f"flytekitplugins-data-{PLUGIN_NAME}"

# TODO add additional requirements
plugin_requires = ["flytekit>=1.1.0b0,<2.0.0, "<other requirements>"]

__version__ = "0.0.0+develop"

setup(
    name=microlib_name,
    version=__version__,
    author="flyteorg",
    author_email="[email protected]",
    # TODO Edit the description
    description="My awesome plugin.....",
    # TODO alter the last part of the following URL
    url="https://github.com/flyteorg/flytekit/tree/master/plugins/flytekit-...",
    long_description=open("README.md").read(),
    long_description_content_type="text/markdown",
    namespace_packages=["flytekitplugins"],
    packages=[f"flytekitplugins.{PLUGIN_NAME}"],
    install_requires=plugin_requires,
    license="apache2",
    python_requires=">=3.8",
    classifiers=[
        "Intended Audience :: Science/Research",
        "Intended Audience :: Developers",
        "License :: OSI Approved :: Apache Software License",
        "Programming Language :: Python :: 3.8",
        "Programming Language :: Python :: 3.9",
        "Programming Language :: Python :: 3.10",
        "Topic :: Scientific/Engineering",
        "Topic :: Scientific/Engineering :: Artificial Intelligence",
        "Topic :: Software Development",
        "Topic :: Software Development :: Libraries",
        "Topic :: Software Development :: Libraries :: Python Modules",
    ],
    # TODO OPTIONAL
    # FOR Plugins where auto-loading on installation is desirable, please uncomment this line and ensure that the
    # __init__.py has the right modules available to be loaded, or point to the right module
    # entry_points={"flytekit.plugins": [f"{PLUGIN_NAME}=flytekitplugins.{PLUGIN_NAME}"]},
)

5. Each plugin should have a README.md, which describes how to install it with a simple example. For example, refer to flytekit-greatexpectations' README.

6. Each plugin should have its own tests' package. NOTE: tests folder should have an __init__.py file.

7. There may be some cases where you might want to auto-load some of your modules when the plugin is installed. This is especially true for data-plugins and type-plugins. In such a case, you can add a special directive in the setup.py which will instruct Flytekit to automatically load the prescribed modules.

   Following shows an excerpt from the flytekit-data-fsspec plugin's setup.py file.

   setup(
       entry_points={"flytekit.plugins": [f"{PLUGIN_NAME}=flytekitplugins.{PLUGIN_NAME}"]},
   )

Flytekit Version Pinning

Currently we advocate pinning to minor releases of flytekit. To bump the pins across the board, cd plugins/ and then update the command below with the appropriate range and run

for f in $(ls **/setup.py); do sed -i "s/flytekit>.*,<1.1/flytekit>=1.1.0b0,<1.2/" $f; done

Try using gsed instead of sed if you are on a Mac. Also this only works of course for setup files that start with the version in your sed command. There may be plugins that have different pins to start out with.

References 📚

• Example of a simple Python task that allows adding only Python side functionality: flytekit-greatexpectations
• Example of a TypeTransformer or a Type Plugin: flytekit-pandera. These plugins add new types to Flyte and tell Flyte how to transform them and add additional features through types. Flyte is a multi-lang system, and type transformers allow marshaling between Flytekit and backend and other languages.
• Example of TaskTemplate plugin which also allows plugin writers to supply a prebuilt container for runtime: flytekit-sqlalchemy
• Example of a SQL backend plugin where the actual query invocation is done by a backend plugin: flytekit-snowflake
• Example of a Meta plugin that can wrap other tasks: flytekit-papermill
• Example of a plugin that modifies the execution command: flytekit-spark OR flytekit-aws-sagemaker
• Example that allows executing the user container with some other context modifications: flytekit-kf-tensorflow
• Example of a Persistence Plugin that allows data to be stored to different persistence layers: flytekit-data-fsspec