Skip to content

Parameters

In Hello World, a simple use of parameters was introduced, namely a Python function that takes kwargs. We’ll now explore how Hera exposes other features of Argo’s Parameters.

Default Values

A Python function naturally allows default values in its definition. When you decorate a function with Hera’s script decorator, it lifts out the default value into a Parameter’s default.

@script()
def echo(message: str = "Hello world!"):
    print(message)

We can run this function in a workflow as defined below:

with Workflow(
    generate_name="hello-world-",
    entrypoint="steps",
) as w:
    with Steps(name="steps"):
        echo()

And we’ll see logs in Argo like

hello-world-r96ww-echo-1258475821: Hello world!

Types

Python functions don’t just take strings, of course, and the key basic types YAML allows are strings, ints, bools and dictionaries. Hera interprets values passed into your function via json.loads, so the below will add a and b if the values passed in are interpreted as ints by json.loads.

@script()
def add_values(a: int, b: int):
    print("Adding values")
    print(a + b)

i.e. the two calls to add_values below are equivalent as the strings in the second call are interpreted as integers. Note that when reusing a function in multiple steps, you must give unique names to the name parameter.

with Steps(name="steps"):
        add_values(name="first", arguments={"a": 1, "b": 2})
        add_values(name="second", arguments={"a": "1", "b": "2"})  # "1" and "2" will be treated as ints
See the logs from this run Note the different node IDs (the number after `add-values`) in the logs, as the logs do not show the container names "first" and "second". 
add-values-xw7k9-add-values-242584704: Adding values
add-values-xw7k9-add-values-242584704: 3
add-values-xw7k9-add-values-242584704: time="2023-05-26T11:57:13.805Z" level=info msg="sub-process exited" argo=true error="<nil>"
add-values-xw7k9-add-values-838832153: Adding values
add-values-xw7k9-add-values-838832153: 3
add-values-xw7k9-add-values-838832153: time="2023-05-26T11:57:24.101Z" level=info msg="sub-process exited" argo=true error="<nil>"

Dictionaries

If we have a function that takes a dict such as below:

@script()
def echo_a_dict(my_dict: Dict[str, str]):
    for k, v in my_dict.items():
        print(f"{k}={v}")

Then in the snippet below, during the compilation of the workflow, arguments will serialize the dict value of my_dict via json.dumps. Then during the script execution on Argo, json.loads will load the dict and the function will run as expected.

with Workflow(
    generate_name="echo-dict-",
    entrypoint="steps",
) as w:
    with Steps(name="steps"):
        echo_a_dict(
            arguments={
                "my_dict": {
                    "my_key": "my_value",
                    "my_second_key": "my_second_value",
                }
            }
        )

w.create()

The logs for the workflow will show

my_key=my_value
my_second_key=my_second_value

Custom Types

Currently, custom types are only supported in the “script runner” experimental feature. See an example usage here. Please note this is an experimental feature so support is limited for now.

Passing Parameters

The result Output Parameter

For the previous examples, we’ve been printing output to stdout, which allows subsequent steps to access the value from the result output parameter (that is, the result value is the stdout). In Hera, if we use a function under a Steps context, it returns a Step object, which has a result property that we can access. For example, if we have the following functions:

@script()
def hello(message: str):
    print(f"Hello {message}")

@script()
def repeat_back(message: str):
    print(f"You just said: '{message}'")

Let’s say we want to get the stdout from running the hello Script template. We have to assign the Step object returned from the function call to a variable, then we can access its result member, passing it to the repeat_back template.

with Workflow(
    generate_name="get-result-",
    entrypoint="steps",
) as w:
    with Steps(name="steps"):
        hello_step = hello(arguments={"message": "world!"})
        repeat_back(arguments={"message": hello_step.result})

w.create()

If you select “All” logs for the workflow you will see the stdout coming from each container:

hello-world-ltpjt-hello-2151859747: Hello world!
hello-world-ltpjt-repeat-back-4012331575: You just said: 'Hello world!'

Output Parameters

Creating Output Parameters

In general, output parameters are given values from a generated file rather than stdout like the result parameter. You can do this in Hera by telling the script decorator to export the output from a file through a value_from. Ensure you add the imports given below!

from hera.workflows import Parameter, script
from hera.workflows.models import ValueFrom

@script(
    outputs=[
        Parameter(name="hello-output", value_from=ValueFrom(path="/tmp/hello_world.txt")),
    ]
)
def hello_to_file():
    with open("/tmp/hello_world.txt", "w") as f:
        f.write("Hello World!")

The container logs for this hello_to_file step will show the artifact being exported as an output parameter.

time="2023-05-26T11:16:37.907Z" level=info msg="/tmp/hello_world.txt -> /var/run/argo/outputs/parameters//tmp/hello_world.txt" argo=true

The output parameter will also show up in the UI under the node’s INPUTS/OUTPUTS tab, similar to the table below.

Parameters
hello-output Hello World!

Passing Output Parameters

Now that we have a hello_to_file function, let’s use its output in our repeat_back function. Under a Steps context, we can assign the Step object returned from the hello_to_file function, and use its get_parameter function, and get the Parameter‘s value.

with Workflow(
    generate_name="hello-world-parameter-passing-",
    entrypoint="steps",
) as w:
    with Steps(name="steps"):
        hello_world_step = hello_to_file()
        repeat_back(
            arguments={"message": hello_world_step.get_parameter("hello-output").value}
        )

If you prefer, arguments can accept a single Parameter, and we can use the with_name function for convenience to specify the right name for repeat_back:

with Workflow(
    generate_name="hello-world-parameter-passing-",
    entrypoint="steps",
) as w:
    with Steps(name="steps"):
        hello_world_step = hello_to_file()
        repeat_back(
            arguments=hello_world_step.get_parameter("hello-output").with_name("message")
        )

Both of these Workflows will produce logs like

hello-world-parameter-passing-vq7pm-hello-to-file-3540104653: time="2023-05-26T12:12:13.803Z" level=info msg="sub-process exited" argo=true error="<nil>"
hello-world-parameter-passing-vq7pm-hello-to-file-3540104653: time="2023-05-26T12:12:13.803Z" level=info msg="/tmp/hello_world.txt -> /var/run/argo/outputs/parameters//tmp/hello_world.txt" argo=true
hello-world-parameter-passing-vq7pm-repeat-back-3418430710: You just said: 'Hello World!'
hello-world-parameter-passing-vq7pm-repeat-back-3418430710: time="2023-05-26T12:12:24.106Z" level=info msg="sub-process exited" argo=true error="<nil>"

Comments