Writing scripted rules

Model Expert provides lots of validation rules, but it can not implement every kind of rule you might ever need.

What validations are possible?

You can create scripts which can run for either

1. all meta-model types (not recommended),
2. all instances of a specific element type, or
3. instances of a specific meta-model type and stereotype.
4. or for whole diagrams: see Validating Diagrams

These can check, for example:

• for specific values in a variable,
• whether users have obeyed a naming standard,
• whether some kinds of connector exist.
• any combination of the above.
• how many elements in a diagram
• whether the notes for a diagram are suitable
• …and anything else you can write code for.

Notes for writing Model Expert validation scripts

We suggest keeping the validation scripts simple. It’s quite possible to accidentally replicate lots of the built-in functions of Model Expert using scripts, so be sure to check that what you are coding can’t be done using a Reference Model.

Also, writing scripts is quite difficult: it has taken us quite a lot of time to write and debug our example scripts, so think about this before you start writing lots of scripts.

Creating an EA Script

To add scripting rules to Model Expert, you will need knowledge of the JavaScript – the only scripting language which we support- as well as some knowledge of the EA programming model. We have provided some example scripts, so you may choose to copy and modify these.

For details on creating an EA Script, see the EA help here.

Also see Testing your scripts.

Writing a Model Expert Validation Script

Validation scripts have just four requirements:

1. They must declare themselves to be a Model Expert validation script, by having a special comment at the start of the script. This lets Model Expert quickly find all the Model Expert validation scripts.
2. They must declare what kind of element type and stereotype they should be used for, or whether they are for validating a diagram.
3. Some code which does the checking.
   Your script will be passed a reference to the current Repository, and the ElementGuid of the instance of the EA element which your code will examine, or the guid of the diagram.
4. They must output an XML string of the form:

   <ElementViolation name='Serious problem' severity='7' />
   <ElementViolation name='Not so serious problem' severity='3'  subject='This really needs fixing'   violatorID='12345'

Script type declaration

At the start of the script, you must have a comment like:

//<Validation>Element</Validation>

– this says the script is for validating individual elements

or

//<Validation>Diagram</Validation>

– this script is for validating a whole diagram, optionally with all its contents

Type / Stereotype Declaration

This must be in the form of a code comment. 

//<Type>All</Type>

//<Type>UseCase</Type>

//<Type>UseCase</Type>

//<Stereotype>Special</Stereotype>

Validation Code

The code to do the validation can be quite simple, because the script has been passed the EA Object_ID of the element being validated, as well as a reference to the current EA Repository.

 let e = Repository.GetElementByGuid(Guid.toString());

let outputHTML = "";

 if (e.Scenarios.Count === 1) {
let aScenario = e.Scenarios.GetAt(0);

if (aScenario.Type === "Basic Path") {
outputHTML += "<ElementViolation 
name='Only basic path scenarios for use case' 
severity='3' violatorID = '" + e.ElementID + "' 
subject='" + e.Name + "'/>";
}
}

 if (e.Requirements.Count > 0) {
outputHTML += "<ElementViolation 
name='Use Case has Internal Requirements' 
severity='3' violatorID = '" + e.ElementID + 
"' subject='" + e.Name + "'/>";
}

Notes

You will notice that there are lots of formatting marks in this output – lots of (‘) and (“) and <>  characters. This is just what the script needs to have in order to work.

Output

Each violation instance that your script returns can have the following attributes: