Python Resource

The python resource is designed to run Python scripts specified in the script attribute from the workflow.

This resource also provides access to installed Python packages installed via pip, and as well as Anaconda packages defined in the workflow.pkl, are available for use. The resource also includes a condaEnvironment attribute, allowing it to isolate execution within a specified Conda environment. For more information, see the Workflow documentation.

Creating a New Python Resource

To create a new python resource, you can either generate a new AI agent using the kdeps new command or scaffold the resource directly.

Here’s how to scaffold a python resource:

kdeps scaffold [aiagent] python

This command will add a python resource into the aiagent/resources folder, generating the following folder structure:

aiagent
└── resources
    └── python.pkl

The file includes essential metadata and common configurations, such as Skip Conditions and Preflight Validations settings. For more details, refer to the Common Resource Configurations documentation.

Python Block

Within the file, you’ll find the python block, which is structured as follows:

python {
    script = """
    print("hello world")
    """
    env {
        // Environment variables accessible within the script
        ["ENVVAR"] = "XYZ"  // Example environment variable
    }
    // Specifies the timeout duration (in seconds) for script execution
    timeoutDuration = 60.s

    // Specifies the Conda environment for isolation
    condaEnvironment = "my-conda-env"
}

Key elements of the python block include:

• script: Specifies the Python script to execute, enclosed in triple double-quotes (""") for multi-line support.
• env: Defines environment variables to be available during execution.
• timeoutDuration: Determines the exectuion timeout in s (seconds), min (minutes), etc., after which the script execution will be terminated.
• condaEnvironment: Specifies the Conda environment to use, ensuring the script runs in an isolated environment with defined dependencies.

When the resource is executed, you can leverage Python functions like python.stdout("id") to access the output. For further details, refer to the Python Functions documentation.