5 - Actions

The Action Sequence document is the definition, the Runtime Context provides an execution environment and the Components are the business logic.  A Component performs a single function, a group of related functions or is a bridge between the Pentaho BI Platform and an external application.  The jFree Reports component is an example of a component that interfaces the platform to the jFree Reports reporting engine. 

There are two major functions that a Component gets called to do - validate and execute.  Validate is called to verify that the inputs are valid and all resources required to execute are available.  Execute actually performs the action. 

The action-definition node in the Action Sequence document defines how the component should function.  It is the place to define and map inputs, outputs and any other metadata or settings the component will require when it is executed.  The name of the component that executes the action definition is identified using the component-name node.  The name is the name of a Java class that is dynamically loaded at runtime.  When referring to the built in Pentaho components, the fully qualified name of the component should not be used, just the class name. For example, use EmailComponent instead of org.pentaho.plugin.email.EmailComponent. 

Action-inputs

The action-inputs and action-resources define the parameters that will be passed into the component when it executes.  Some components have required inputs that must be available or runtime error will occur.  Some inputs may be specified but are optional.  For example, the Email Component requires a to but the cc is optional.  There are several ways to satisfy a required input. It can be provided as a passed parameter with the same name to the component. It can be mapped to a different name and passed to the component. It can be hard coded with a constant value or in some cases it could be prompted for.  By default the email component should have an action input that looks like:

<action-inputs>
    <to type="string" />
          . . .
</action-inputs>

Action Input Mapping

What happens if a component requires an input named x and I want to pass it a suitable parameter named y?  You can map any action-input to a different name using the mapping attribute.  Again, using our trusty Email Component example, let's assume we have a JavaScript rule that performs a query and returns the email address for a user in an output-parameter named EMAIL.  Let's further assume that I want to use that parameter as the to parameter in the Email Component.  The XML fragment would look like:

<action-inputs>
    <to type="string" mapping="EMAIL"/>
                   . . .
</action-inputs>

Constant Values as Inputs

If a component requires an input parameter and it is not found in the action-inputs, the component will automatically look for an XML node with the same name, in the component-definition section.  If it finds one, it will use the value from there.  The XML fragment would look like:

<action-inputs>
</action-inputs>

<component-definition>
    <to>joe.pentaho@pentaho.org</to>
</component-definition>

Replaceable Parameters

All components have the ability to use the built in template mechanism to perform text replacement using input parameters.  This means that a constant value can be modified using the value of an action-input. Hmmm that wasn't very clear either.  Let's try with an example:

<action-inputs>
    <EMAIL type="string"/>
         . . .
</action-inputs>

<component-definition>
    <to>{EMAIL}@pentaho.org</to>
</component-definition>

The text within the curly braces {} is replaced by an input parameter with the same name.  In this example, if the value of the action-input EMAIL is joe.pentaho then the value of the to parameter passed to the email component would be joe.pentaho@pentaho.org.

Add more about component-definition and action-outputs