Inputs and Outputs

Inputs and outputs provide a structured way to pass information into and out of tests. When a user initiates a test run, a modal is displayed for providing input values. When multiple tests are run together, the user will not be prompted for inputs that are populated by the output of a previous test in the run. Currently, all inputs and outputs are stored as strings.

Inputs

Defining Inputs

The input method defines an input. input can take several arguments, but only the identifier is required:

• identifier - (required) a name for this input. The input value is available in the run block using this name.
• title: - a title which is displayed in the UI.
• description: - a description which is displayed in the UI.
• type: - controls the type of HTML input element used in the UI. Currently there are 5 possible values:
  • 'text' - (default) a regular input field.
  • 'textarea' - for a text area input field.
  • 'radio' - for a radio button singular selection field.
  • 'checkbox - for a checkbox field. In tests, a checkbox input is represented as an array of the selected values.
  • 'oauth_credentials' - a complex type for storing OAuth2 credentials. When used by a FHIR client, the access token will automatically refresh if possible.
• default: - default value for the input.
• optional: - (default: false) whether the input is optional.
• options: - possible input option formats based on input type.
  • list_options: - options for input formats that require a list of possible values (radio and checkbox). An array of hashes with label and value keys.
• locked: - (default: false) whether the user can alter the input’s value. Locking an input can force it to use a value from a previous test’s output, or the default value.

Here is a simple example:

test do
  input :url,
        title: 'FHIR Server Url',
        description: 'The base url for the FHIR server'
        
  run do
    # The input's identifier is :url, so its value is available via `url`
    assert url.start_with?('https'), 'The server must support https'
  end
end

input in the API docs

Defining Multiple Inputs

It is possible to define multiple inputs in a single input call, though this prevents the use of the additional properties listed above. This can be useful when a test uses inputs that have been defined in a parent or sibling.

test do
  input :input1, :input2, :input3, :input4
  ...
end

Inputs with List Options

For the radio or checkbox input types, a list of options must be provided. The label is displayed to the user, and the value is the actual value that is stored when the user selects that option.

test do
  input :radio_input_example,
        title: 'Example Radio Input',
        type: 'radio',
        options: {
          list_options: [
            {
              label: 'Radio Option 1',
              value: 'option_1'
            },
            {
              label: 'Radio Option 2',
              value: 'option_2'
            }
          ]
        }

  input :checkbox_input_example,
        title: 'Example Checkbox Input',
        type: 'checkbox',
        options: {
          list_options: [
            {
              label: 'Checkbox Option 1',
              value: 'option_1'
            },
            {
              label: 'Checkbox Option 2',
              value: 'option_2'
            }
          ]
        }

  run do
    if radio_input_example == 'option_1'
      # ...
    end

    if radio_input_example == 'option_2'
      # ...
    end

    if checkbox_input_example.include? 'option_1'
      # ...
    end

    if checkbox_input_example.include? 'option_2'
      # ...
    end
  end
end

It is possible to lock individual checkbox items. In the following example, Item 1 is not locked, so the user can check and uncheck it freely. Item 2 is checked because it is included in the default, and the user can not uncheck it. Item 3 is unchecked because it is not included in the default, and the user can not check it.

input :locked_checkbox_example,
      title: 'Locked Checkbox Input Example',
      type: 'checkbox',
      default: ['value2'],
      options: {
        list_options: [
          {
            label: 'Item 1',
            value: 'value1'
          },
          {
            label: 'Item 2 - Locked checked',
            value: 'value2',
            locked: true
          },
          {
            label: 'Item 3 - Locked unchecked',
            value: 'value3',
            locked: true
          }
        ]
      }

Ordering Inputs

When a group or suite displays all of its descendants’ inputs, they may be in an unintuitive order. They can be reordered using input_order.

group do
  input_order :input_2, :input_1
  
  test do
    input :input_1
  end
  
  test do
    input :input_2
  end
end

Additional Input Instructions

If a developer wants to include additional input instructions, they can define input_instructions which will be displayed above the inputs.

group do
  input_instructions %(
    Register Inferno as a standalone application using the following information:

    * Redirect URI: `#{SMARTAppLaunch::AppRedirectTest.config.options[:redirect_uri]}`

    Enter in the appropriate scope to enable patient-level access to all
    relevant resources. If using SMART v2, v2-style scopes must be used. In
    addition, support for the OpenID Connect (openid fhirUser), refresh tokens
    (offline_access), and patient context (launch/patient) are required.
  )
end

Outputs

Defining Outputs

The output method defines an output. It is used in a test’s definition block to define which outputs a test provides, and within a test’s run block to assign a value to an output. Multiple outputs can be defined and assigned at once.

test do
  output :value1
  output :value2, :value3
  
  run do
    output value1: 'ABC'
    output value2: 'DEF',
           value3: 'GHI'
  end
end

test do
  # These inputs will automatically get their values from the previous test's
  # outputs.
  input :value1, :value2, :value3
  ...
end

output for defining outputs in the API docs

output for assigning values to outputs in the API docs

Handling Complex Objects

Since inputs and outputs are all stored as strings, special handling is needed if passing complex objects between tests. This can generally be handled using JSON serialization. Ruby hashes and arrays, as well as FHIR model classes, support the to_json method, which turns an object into a JSON string.

test do
  output :complex_object_json
  
  run do
    ...
    output complex_object_json: hash_or_array_or_fhir_resource.to_json
  end
end

test do
  input :complex_object_json
  
  run do
    assert_valid_json(complex_object_json) # For safety

    complex_object = JSON.parse(complex_object_json)
    ...
  end
end

Behind the Scenes

Inputs and outputs work as a single key-value store scoped to a test session. The main differences between them are:

• An input’s value can not be changed during a test
• Inputs support additional metadata for display in the UI (title, description, etc.)

Since inputs and outputs form a single key-value store, a value will be overwritten if multiple tests write to the same output. However, each test result stores the input and output values that were present for that particular test.