Guided Rules with the JBoss BRMS (Guvnor)

Exclusive offer: get 50% off this eBook here
JBoss Drools Business Rules

JBoss Drools Business Rules — Save 50%

Capture, automate, and reuse your business processes in a clear English language that your computer can understand.

$29.99    $15.00
by Paul Browne | March 2009 | JBoss Java Open Source

In this article by Paul Browne, we're going to read about the Guvnor rule editor, and use it to write some more sophisticated rules. In particular, we're going to:

  • Show how to put information into and out of our rules
  • Build a fact model to hold this information
  • Import our newly built model into Guvnor
  • Create guided rules using this fact model
  • Run and test our new fact-based rules

Passing information in and out

The main reason for the simplicity of our Hello World example was that it neither took in any information, nor passed any information out—the rule always fired, and said the same thing. In real life, we need to pass information between our rules and the rest of the system. You may remember that in our tour of the Guvnor, we came across models that solved this problem of 'How do we get information into and out of the rules?'.

Here is the spreadsheet that we will use as an example in this article:

If we want to duplicate this in our model/JavaBean, we would need places to hold four key bits of sales-related information.

  • Customer Name: String (that is, a bit of text)
  • Sales: Number
  • Date of Sale: Date
  • Chocolate Only Customer: Boolean (that is, a Y/N type field)

We also need a description for this group of information that is useful when we have many spreadsheets/models in our system (similar to the way this spreadsheet tab is called Sales).

Note that one JavaBean (model) is equal to one line in the spreadsheet. Because we can have multiple copies of JavaBeans in memory, we are able to represent the many lines of information that we have in a spreadsheet. Later, we'll loop and add 10, 100, or 1000 lines (that is, JavaBeans) of information into Drools (for as many lines as we need). As we loop, adding them one at a time, the various rules will fire as a match is made.

Building the fact model

We will now build this model in Java using the Eclipse editor. We're going to do it step-by-step.

  1. Open the Eclipse/JBoss IDE editor that you installed earlier. If prompted, use the default workspace. (Unless you've a good reason to put it somewhere else.)
  2. From the menu bar at the top the screen, select File | New Project. Then choose Java Project from the dialog box that appears. You can either select this by starting to type "Java Project" into the wizard, or by finding it by expanding the various menus.
  3. In the Create a new Java Project dialog that appears, give the project a name in the upper box. For our example, we'll call it SalesModel (one word, no spaces).
  4. Accept the other defaults (unless you have any other reason to change them). Our screen will now look something like this:

When you've finished entering the details, click on Finish. You will be redirected to the main screen, with a new project (SalesModel) created. If you can't see the project, try opening either the Package or the Navigator tab.

When you can see the project name, right-click on it. From the menu, choose New | Package. The New Java Package dialog will be displayed, as shown below. Enter the details as per the screenshot to create a new package called org.sample, and then click on Finish.

If you are doing this via the navigator (or you can take a peek via Windows Explorer), you'll see that this creates a new folder org, and within it a subfolder called sample. Now that we've created a set of folders to organize our JavaBeans, let's create the JavaBean itself by creating a class.

To create a new Java class, expand/select the org.sample package (folder) that we created in the previous step. Right-click on it and select New Class. Fill in the dialog as shown in the following screenshot, and then click on Finish:

We will now be back in the main editor, with a newly created class called Sales.java (below). For the moment, there isn't much there—it's akin to two nested folders (a sample folder within one called org) and a new (but almost empty) file / spreadsheet called Sales.

package org.sample;
public class Sales {
}

By itself, this is not of much use. We need to tell Java about the information that we want our class (and hence the beans that it creates) to hold. This is similar to adding new columns to a spreadsheet.

Edit the Java class until it looks something like the code that follows (and take a quick look of the notes information box further down the page if you want to save a bit of typing). If you do it correctly, you should have no red marks on the editor (the red marks look a little like the spell checking in Microsoft Word).

package org.sample;
import java.util.Date;
public class Sales {
private String name;
private long sales;
private Date dateOfSale;
private boolean chocolateOnlyCustomer;
public String getName() {
return name;
}
public void setName(String name) {
this.name = name;
}
public long getSales() {
return sales;
}
public void setSales(long sales) {
this.sales = sales;
}
public Date getDateOfSale() {
return dateOfSale;
}
public void setDateOfSale(Date dateOfSale) {
this.dateOfSale = dateOfSale;
}
public boolean isChocolateOnlyCustomer() {
return chocolateOnlyCustomer;
}
public void setChocolateOnlyCustomer(boolean choclateOnlyCustomer) {
this.chocolateOnlyCustomer = chocolateOnlyCustomer;
}
}

Believe it or not, this piece of Java code is almost the same as the Excel Spreadsheet we saw at the beginning of the article. If you want the exact details, let's go through what it means line by line.

  • The braces ({ and }) are a bit like tabs. We use them to organize our code.
  • package—This data holder will live in the subdirectory sample within the directory org.
  • import—List of any other data formats that we need (for example, dates). Text and number data formats are automatically imported.
  • Public class Sales—This is the mould that we'll use to create a JavaBean. It's equivalent to a spreadsheet with a Sales tab.
  • Private String name—create a text (string) field and give it a column heading of 'name'. The private bit means 'keep it hidden for the moment'.
  • The next three lines do the same thing, but for sales (as a number/long), dateOfSale (as a date) and chocolateOnlyCustomer (a Boolean or Y/N field).
  • The rest of the lines (for example, getName and setName) are how we control access to our private hidden fields. If you look closely, they follow a similar naming pattern.

The get and set lines (in the previous code) are known as accessor methods. They control access to hidden or private fields. They're more complicated than may seem necessary for our simple example, as Java has a lot more power than we're using at the moment.
Luckily, Eclipse can auto-generate these for us. (Right-click on the word Sales in the editor, then select Source | Generate Getters and Setters from the context menu. You should be prompted for the Accessor methods that you wish to create.)

Once you have the text edited like the sample above, check again that there are no spelling mistakes. A quick way to do this is to check the Problems tab in Eclipse (which is normally at the bottom of the screen).

Now that we've created our model in Java, we need to export it so that we can use in the Guvnor.

  1. In Eclipse, right-click on the project name (SalesModel) and select Export.
  2. From the pop-up menu, select jar (this may be under Java; you might need to type jar to bring it up). Click on Next. The screen shown above will be displayed.
  3. Fill out this screen. Accept the defaults, but give the JAR file a name (SalesModel.jar) and a location (in our case C:tempSalesModel.jar). Remember these settings as we'll need them shortly.
  4. All being well, you should get an 'export successful' message when you click on the Finish button, and you will be able to use Windows Explorer to find the JAR file that you just created.
JBoss Drools Business Rules Capture, automate, and reuse your business processes in a clear English language that your computer can understand.
Published: April 2009
eBook Price: $29.99
Book Price: $49.99
See more
Select your format and quantity:

Importing the fact model into Guvnor

Switching back to the Guvnor, we're now going to create a package to hold our brand new model that we created in the previous steps.

  1. From the lefthand side menu, select the Packages tab.
  2. Go to Package | Create New Package (org.sample) | Create Package.
  3. From just below the Packages tab, highlight Create New and then select the New model (jar) of fact classes.
  4. Fill out the dialog box that is displayed, as follows. The Name (SalesModel) and Package (org.sample) should be the same as the ones we created in Eclipse. Click on the OK button.

  5. Back in the main Guvnor screen, check that everything is in place.

In addition to the upload button (that we're going to use in a minute), this screen also has a Download button (to retrieve JAR files that you may have uploaded earlier). It also has the usual Save, Copy, Archive, and Delete options.

To upload our fact model JAR into Guvnor , follow the steps shown below:

  1. Click on the Browse button, and then navigate to and select the JAR file that we exported earlier. Click on OK.
  2. After returning to the screen above, click on the upload  button (actually, the 'up' arrow icon to the right of upload ).
  3. If everything goes well, you will get the message File was uploaded successfully. Click on OK to return to the SalesModel/Package tab.
  4. Save the updated package (by clicking on the Save Changes button). As with other saves/check-ins you'll be asked for an optional checkin comment.
  5. We can check whether Guvnor has successfully picked up the new package information by expanding the org.package that was created.

You can see from this example that in the Configuration section, under Imported types, our class (org.sample.Sales) is listed. This means that Guvnor has not only uploaded our class file, but will also allow us to write rules that use this class. Now that Guvnor knows the format of the information that we want to pass in and out, we can start writing rules using Guvnor.

Guided rules using the fact model

Back at the chocolate factory, we've decided to implement a customer loyalty scheme. When any customer has sales of greater than 100 dollars, we want to give them a flat rate discount of 10 dollars. To put it in a slightly more 'rules-like' format, our new business rule will look something like this:

when
we have a sale greater than 100
then
Give a discount (by adding a 'negative' sale)

Yes, as a business rule it's slightly clunky, but it keeps things simple. In real life we'd just update our sales object with the new balance after the discount (and keep a note of what discount was given).

Of course, we're going to write this rule using the Guvnor. I'll show you the full step-by-step answer soon, but the end result will look something similar to the following:

Some key notes and buttons to use in the Guvnor are:

  1. You're going to create a new rule using the guided editor in Guvnor.
  2. Click the '+' next to WHEN and THEN to add new conditions/consequences (such as greater than 100 and Insert Sales -10).
  3. The 'green arrow' icon allows you to refine these further.
  4. Guvnor will pick up the sales model that we imported earlier and offer it as choice to you on a menu.
  5. If you make a mistake, the '-' icon allows you to delete a line.

The step-by-step answer

Before we write a rule we must make sure that we have a category assigned to it. We can use any existing category or we can create a new one (under the Admin tab, expand Categories | New Category). For this step-by-step example, we've created a new SalesCategory. But categories are just descriptive tags, so it will work with pretty much any name.

After you've chosen a category, follow these steps:

  1. Create a new business rule by selecting menu option rules | create new | New business rule (guided editor).
  2. Enter the following values in the screen that is displayed. We will give the new rule a name (SalesDiscount—although anything descriptive is OK). We will assign a category (the SalesCategory that we created earlier). Then we will pick the package (org.sample) from the drop-down list. After entering a description (optional) we need to click on OK.

  3. We'll then be taken to the guided business rule editor that we saw earlier on our quick tour. In the main section, click on the '+' sign next to the WHEN label, to add a condition (that is, to restrict the circumstances under which our rule will fire).

  4. The Add a condition to the rule dialog will be displayed. We're going to choose a fact type of Sales. (Actually, this will be the only fact type in the drop-down list.) This means that our rule will fire only when a sales fact is present. To put it another way, our rule only applies to the rules spreadsheet.

  5. After choosing this, we'll automatically be taken back to the guided rule-editing screen.
  6. Back on the main screen, we'll see that Sales has been added as a condition. Currently, this rule will fire for all sales, so we want to restrict it to only those sales of more than 100 dollars.

  7. Because we want to elaborate on the WHEN condition, we need to click on the green 'arrow' icon immediately next to the Sales condition. We'll be shown the Modify constraints for Sales dialog box, shown as follows:

  8. The first dropdown field contains a list of all of the fields (columns) available for the Sales object. We'll choose sales (that is, the dollar value or amount) from the drop-down list.
  9. Back in the guided editor, another line will have been added. The default value is please choose. Change this to greater than or equal to, as shown in the following diagram.

  10. Now we need a value to compare this field to (as part of the filter). Click on the pencil icon to set this. In the Field value dialog box that is displayed, click on Literal value. Literal values are numbers we can enter directly.

  11. Back in the guided editor, a new text box will have appeared. Enter 100 (the value we want to use in our rule) in this text box.

That's it—the WHEN part of the rule is done and should look like the objective picture that we saw at the very start of this sample). Now would be a good time to save the rule (and enter a comment if you see fit).

The THEN part is somewhat easier, in that there are fewer steps to create it.

  1. Click on the green plus sign next to the Then section. The Add a new action dialog box will be displayed:

  2. We choose to insert a new fact (in the first dropdown). Sales will be the only option in this menu. Inserting a new fact is like adding a new row to the Excel Spreadsheet (that is another line of information into the memory).
  3. Back in the guided editor, the text Insert Sales will now be displayed. Click on the green arrow next to this to begin setting to values for our new sales object (this is similar to saying: once we insert a new line into the Excel Spreadsheet, here are the values that I want to use in the newly created cells).

  4. First we're going to add a name field. This will appear in the main guided editor. Click on the pencil next to the name field to add a value. Enter the value Discount (so that the purpose of the new line/new fact is clear).
  5. Now that we've created our first field, we can repeat the process for the second one—Sales. Add a field (click the green icon next to Insert sales) and choose sales. Back in the guided editor, click on the pencil next to the new line, and enter a value of -10.
  6. By now, the screen will look like the screenshot back at the start of this section. Now is a good time to save your rule. Validate the rule by clicking on the Validate button. All being well, you will see a dialog box similar to the following example. To close the dialog box, click on the 'x' in the upper-right corner of the dialog box.

Remember that validation is only a check to catch the most obvious errors. It's still possible to get warning messages when we run and test our rule in the next section.

To confirm that the guided editor has written the rule for us, click on the View source button. The meaning (give a discount for sales of 100) is pretty much as we'd expect. We've converted the rule into plain English.

Rule "SalesDiscount"
//name of the rule
// use the slightly more readable mvel
when
we can find a fact / line of more than 100 sales
then
create a new line / fact
set the name to "discount"
set the sales to -10
add the line back to the model
end
// end of rule

You'll notice that this text is read-only. How do we try this rule out?

Running this scenario

To start, expand the Package tab and then create a new test scenario. Give it a name (for example TestSales), and select the same package for our rule (org.sample). You will be presented (again) with the blank scenario test screen. The scenario that we're going to create is similar to the following screenshot:

The steps for building this screen are similar to those we used before. We use the '+' sign to insert a new GIVEN/EXPECT, the small green arrow to refine the scenario, and the '–' sign to remove.

  1. Click on the plus sign next to GIVEN and choose to insert a new sales fact (under any name). Click on Add.
  2. Click on the Add a field button that appears. In the dialog box, select the name field.
  3. Click on the green arrow next to sales to add another field (column in our new row). In this case, use Sales and give it a value of 200 (that is, greater than 100!).
  4. Click on the green '+' next to EXPECT. In the New Expectation dialog box that is displayed, click on show list and then choose the Sales Discount rule.
  5. Change the default (that we expect this rule to fire at least once) to Expect Rules, to fire this many times, and then enter '1' in the new text box that appears.
  6. Save this test scenario using the button at the top of the screen.

All being well, if you now click the Run Scenario button, you will get a green bar at the top of the screen saying Results 100%, along with some additional text: Rule [SalesDiscount] was activated 1 times, which indicates that our rule is running as expected.

What just happened?

The test scenario that we created was equivalent to passing in a spreadsheet with one row (that is, one Sales Java object with sales of 200 and a name of Acme Corp). We'd expect our sales discount rule to fire under these circumstances and we tell our test scenario to look out for this. When we run the scenario, our rule behaves as expected and fires, giving Acme Corp a discount of –10 Sales for their order.

Summary

In this article we have covered five main areas. We saw how to get information in and out of our rules, and created the fact model in Java needed to do this. We imported our new fact model into the Guvnor and then built a guided rule around it. Finally, we tested our rule to make sure that it ran correctly.


If you have read this article you may be interested to view :


 

JBoss Drools Business Rules Capture, automate, and reuse your business processes in a clear English language that your computer can understand.
Published: April 2009
eBook Price: $29.99
Book Price: $49.99
See more
Select your format and quantity:

About the Author :


Paul Browne

Paul Browne's first job was selling computers in France and things went steadily downhill from there. He spent millons on behalf of a UK telephone company's procurement department and implemented direct marketing for a well-known Texan computer maker before joining the IT department of a company that builds bright red tractors and other seriously cool machines.

Paul then embraced his techie side (he was writing games in machine code from the age of 11) and started a consultancy that used IT to solve business problems for companies in the financial and public sectors in Ireland, UK , Belgium, and New Zealand. Eight years later, he now works with an Irish government agency that helps similar software companies to grow past their initial teething pains.

More formally, Paul has a bachelor's degree in Business and French from the University of Ulster, a master's degree in Advanced Software from UCD Dublin, a post-grad qualification in Procurement from the Chartered Institute of Procurement and Supply (UK), and will someday complete his ACCA financial exams.

Paul can be found on LinkedIn at http://www.linkedin.com/in/paulbrowne , and via the Red Piranha (Business knowledge) project at http://code.google.com/p/red-piranha/ .

Books From Packt


JBoss Portal Server Development
JBoss Portal Server Development

Learning jQuery 1.3
Learning jQuery 1.3

DWR Java AJAX Applications
DWR Java AJAX Applications

JBoss Tools 3 Developers Guide
JBoss Tools 3 Developers Guide

Java EE 5 Development with NetBeans 6
Java EE 5 Development with NetBeans 6

Spring 2.5 Aspect Oriented Programming
Spring 2.5 Aspect Oriented Programming

Drupal 6 Social Networking
Drupal 6 Social Networking

WordPress Plugin Development: Beginner's Guide
WordPress Plugin Development: Beginner's Guide


Your rating: None Average: 4.5 (2 votes)
Excellent by
Thank you for the wonderful tutorial

Post new comment

CAPTCHA
This question is for testing whether you are a human visitor and to prevent automated spam submissions.
c
n
7
p
c
D
Enter the code without spaces and pay attention to upper/lower case.
Code Download and Errata
Packt Anytime, Anywhere
Register Books
Print Upgrades
eBook Downloads
Video Support
Contact Us
Awards Voting Nominations Previous Winners
Judges Open Source CMS Hall Of Fame CMS Most Promising Open Source Project Open Source E-Commerce Applications Open Source JavaScript Library Open Source Graphics Software
Resources
Open Source CMS Hall Of Fame CMS Most Promising Open Source Project Open Source E-Commerce Applications Open Source JavaScript Library Open Source Graphics Software