Untyped messages and Business Rules Engine

This blogpost has been superseded by a new version. See: http://biztalkmessages.vansplunteren.net/2010/03/04/untyped-messages-and-business-rules-engine-part-2/

The other day one of my BizTalk buddies asked me if I knew a way to process different messages by the same policy in the Business Rules Engine (BRE). In other words is it possible to create some sort of generic policy that can process different types of messages?

In his scenario the schemas had some fields in common and he needed those fields to be evaluated and optionally set in a single BRE policy.

Of course you can make a policy that depends on different schemas and add rules for each schema. The bad thing about that is that you will get multiple copies of almost the same rule which might lead to a maintenance nightmare.

I created a small sample to show how this can be achieved in a single policy and single rule. Below are three very basic schemas I used in this sample:

<?xml version="1.0" encoding="utf-16"?>
<xs:schema xmlns:b="<a href="http://schemas.microsoft.com/BizTalk/2003">http://schemas.microsoft.com/BizTalk/2003</a>" xmlns="<a href="http://UntypedBRE.FirstSchema">http://UntypedBRE.FirstSchema</a>" targetNamespace="<a href="http://UntypedBRE.FirstSchema">http://UntypedBRE.FirstSchema</a>" xmlns:xs="<a href="http://www.w3.org/2001/XMLSchema">http://www.w3.org/2001/XMLSchema</a>">
  <xs:element name="FirstSchema">
    <xs:complexType>
      <xs:sequence>
        <xs:element name="FirstName" type="xs:string" />
        <xs:element name="LastName" type="xs:string" />
        <xs:element name="IsJohn" type="xs:string" />
      </xs:sequence>
    </xs:complexType>
  </xs:element>
</xs:schema>

<?xml version="1.0" encoding="utf-16"?>
<xs:schema xmlns:b="<a href="http://schemas.microsoft.com/BizTalk/2003">http://schemas.microsoft.com/BizTalk/2003</a>" xmlns="<a href="http://UntypedBRE.SecondSchema">http://UntypedBRE.SecondSchema</a>" targetNamespace="<a href="http://UntypedBRE.SecondSchema">http://UntypedBRE.SecondSchema</a>" xmlns:xs="<a href="http://www.w3.org/2001/XMLSchema">http://www.w3.org/2001/XMLSchema</a>">
  <xs:element name="SecondSchema">
    <xs:complexType>
      <xs:sequence>
        <xs:element name="FirstName" type="xs:string" />
        <xs:element name="LastName" type="xs:string" />
        <xs:element name="IsJohn" type="xs:string" />
      </xs:sequence>
    </xs:complexType>
  </xs:element>
</xs:schema>

<?xml version="1.0" encoding="utf-16"?>
<xs:schema xmlns:b="<a href="http://schemas.microsoft.com/BizTalk/2003">http://schemas.microsoft.com/BizTalk/2003</a>" xmlns="<a href="http://UntypedBRE.ThirdSchema">http://UntypedBRE.ThirdSchema</a>" targetNamespace="<a href="http://UntypedBRE.ThirdSchema">http://UntypedBRE.ThirdSchema</a>" xmlns:xs="<a href="http://www.w3.org/2001/XMLSchema">http://www.w3.org/2001/XMLSchema</a>">
  <xs:element name="ThirdSchema">
    <xs:complexType>
      <xs:sequence>
        <xs:element name="FirstName" type="xs:string" />
        <xs:element name="LastName" type="xs:string" />
        <xs:element name="IsJohn" type="xs:string" />
      </xs:sequence>
    </xs:complexType>
  </xs:element>
</xs:schema>

I my sample policy I want to check the ‘FirstName’ element. If the value is equal to ‘John’ I want to fill the ‘IsJohn’ element with value ‘yes’. The policy (and single rule) should work for all of the above schemas.

The problem is that mentioned schemas belong to a different namespace and have a different root node, hence in BizTalk terms have a different message type. Because XML schema facts in the BRE are by default tightly coupled to a specific schema the consequence is that those facts can only operate on single type of message.

In order to make this generic you have to do the following:

1. Add one of the schemas to the Facts Explorer in the BRE.

2. Make the schema ‘general’

As you can see the Document Type resembles the type of the schema I added. To make this generic I change this value to ‘System.Xml.XmlDocument’. This will make sure that if I use a fact from this schema in a rule it will not be typed to this schema but will be generic.

3. Create the rule

In this step I create rule. The facts will be filled in later.

4. Modify the XML facts.

In the rule I need to evaluate the ‘FirstName’ fact and optionally set the ‘IsJohn’ fact. Because I want this to work on all schemas I need to define the facts in a generic way. If I click on the ‘FirstName’ fact I can see the xpath statements that point to this fact in the property pane:

The ‘Xpath Field’ and ‘Xpath Selector’ properties are directly referring to ‘FirstSchema’ root node and namespace. I change the values to make them generic also:

Note that I’m using ‘self::node()’ here. I described this trick before here.

Now the XML fact is no longer pointing to a specific namespace or root node. It just points to a ‘FirstName’ node somewhere in a Xml document.

There are of course other possible values for ‘XPath Selector’ and ‘XPath Field’ to solve this. It all depends on the schemas. If for example the facts you need all have the same parent node you can make the ‘XPath selector’ select the parent node and the ‘XPath Field’ select the ‘FirstName’ element.

I do the same for the fact I want to update in the action of the rule:

5. Complete the rule by adding the facts.

Finally I can drag the XML facts from the Facts Explorer to my rule to complete the condition and create a new action. Like this:

You can see that the both the condition and the action are not referring (anymore) to any specific schema  but instead to any schema that has ‘FirstName’  and ‘IsJohn’  elements.

Testing the rule with instances from two different schemas shows that this works:

One thing to note about this is that the way I changed the xpath statements for the Xml facts comes with a performance penalty. Using things like ‘//*…….’.  will make the engine go through the whole xml tree which is less efficient then using the original full xpath statement. So if performance is a strict requirement be careful using techniques like these.

Another thing is that I do not check for the existence of the nodes first. The policy will crash when an message is processed that does not contain on of the nodes used in the rule.