Skip to main content
Version: 0.18.21

Action

An Action is a Python class with a run() method that takes a Validation ResultGenerated when data is Validated against an Expectation or Expectation Suite. and does something with it.

Actions are customizable. Great Expectations comes with common Actions for such things as sending email or Slack notifications, updating Data DocsHuman readable documentation generated from Great Expectations metadata detailing Expectations, Validation Results, etc., and storing Validation ResultsGenerated when data is Validated against an Expectation or Expectation Suite. out of the box. However, it is easy to create custom Actions by creating a subclass of Great Expectations' ValidationAction class and overwriting its _run() method. This means that you can configure an Action to do literally anything you are capable of programming in Python in response to a CheckpointThe primary means for validating data in a production deployment of Great Expectations. ValidationThe act of applying an Expectation Suite to a Batch. completing.

Relationship to other objects

Actions are configured inside the action_list parameter of a Checkpoint's configuration, and execute every time the Checkpoint finishes running a Validation. When an Action is run, it will have access to the Validation Results and configured MetricsA computed attribute of data such as the mean of a column. that were generated by the Checkpoint.

Use cases

Configuring, implementing, and executing an Action (custom or otherwise) takes place in the Validate Data step. Creating custom Actions is a process that falls outside the workflow of using Great Expectations.

Actions are configured when Checkpoints are created in the Validate Data step of working with Great Expectations. After that, Checkpoints that have a populated action_list in their configuration will execute the indicated Actions every time they finish running a Validation.

Versatility

The features of a specific Action depend on what that Action is designed to do. An Action can perform anything that can be done with Python code, making them phenomenally versatile.

Convenience

Since some Actions are common, Great Expectations implements them out of the box: You don't have to write every Action as a custom one. These Actions are subclasses of the ValidationAction class, and you can view them in the great_expectations.checkpoint.actions module.

The following are some of the available pre-built Actions:

  • EmailAction: sends an email to a given list of email addresses.
  • MicrosoftTeamsNotificationAction: sends a Microsoft Teams notification to a given webhook.
  • SlackNotificationAction: sends a Slack notification to a given webhook.
  • StoreEvaluationParametersAction: extracts Evaluation Parameters from a Validation Result and stores them in the Store configured for this Action.
  • StoreMetricsAction: extracts Metrics from a Validation Result and stores them in a Metrics Store.
  • StoreValidationResultAction: stores a Validation Result in the ValidationsStore.
  • UpdateDataDocsAction: notifies the site builders of all the data docs sites of the Data Context that a validation result should be added to the data docs.

Access

Actions are not intended to be manually instantiated or accessed. Instead, they are included in the action_list parameter of a Checkpoint's configuration, and when the Checkpoint finishes running a Validation it will then run the Actions in its action_list in order of appearance.

Classes that implement Actions can be found in the great_expectations.checkpoint.actions module, which you can view on GitHub:

Create

Custom actions need to be created in Python code, and can be implemented as Plugins. In order for a Python class to be a valid Action, it must conform to the Action API. Specifically, it must implement a run() method which accepts three required parameters, two named optional parameters, and any number of **kwargs.

Required parameters

  • validation_result_suite: an instance of the ExpectationSuiteValidationResult class.
  • validation_result_suite_identifier: an instance of either the ValidationResultIdentifier class (for open source Great Expectations) or the GeCloudIdentifier (from Great Expectations Cloud).
  • data_asset: an instance of the Validator class.

Optional parameters

  • expectation_suite_identifier
  • checkpoint_identifier

Additional parameters

  • **kwargs: named parameters that are specific to a given Action, and need to be assigned a value in the Action's configuration in a Checkpoint's action_list.

The required and optional named parameters will be automatically passed to the Action from the Checkpoint that the Action is included with, after the Checkpoint completes Validation. This means you can configure your custom Actions to behave conditionally based on the Validation Results your Checkpoint generates, or the values passed along with any of those named parameters. Additional **kwargs parameters can be included, but they cannot be passed automatically by the Checkpoint. Therefore, you will have to specify the name and value for these parameters in the configuration for their Action, in the Checkpoint's action_list.

The best practice when creating a custom Action is to create a subclass of the ValidationAction class found in the great_expectations.checkpoint.actions module. Leave the inherited run() method as its default, and overwrite the _run() method to contain your functionality. There are numerous examples of this practice in the subclasses of ValidationAction located in the great_expectations.checkpoint.actions module, which you can view on GitHub:

If you develop a custom Action, consider making it a contribution in the Great Expectations open source GitHub project. You can also reach out to us on Slack if you need additional guidance in your efforts.

Configure

Actions are configured inside of the action_list parameter for Checkpoints. In general, you will need to provide at least a name (which is user defined and does not need to correspond to anything specific) and, in the in the parameters under action a class_name (which should correspond to the name of the Action's Python class). If you are implementing a custom Action in a Plugin, you will also need to include a module_name in the parameters under action which references where your Plugin is located. Any other keys placed under action will be passed to the Action's class as additional key word arguments.

The following is an example of an action_list configuration that performs some common Actions that are built in to the Great Expectations code base:

YAML
action_list:
- name: store_validation_result
action:
class_name: StoreValidationResultAction
- name: store_evaluation_params
action:
class_name: StoreEvaluationParametersAction
- name: update_data_docs
action:
class_name: UpdateDataDocsAction

To configure Actions in a Checkpoint, see Configure Actions.