About the Collector Node sample

The Collector Node sample contains a message flow with two Collector nodes.

The sample is based on the following scenario:

The flow creates an invoice record from three separate parts of a specific order (customer order details, warehouse information, and shipping costs).
The flow groups all the invoices into batches, which are sent at 90 second intervals.

The invoice for an order is generated by taking input from three sources:

Input from a WebSphere MQ message, which contains details of the original customer order: item ordered, quantity, and cost.
Input from an external file, which is named to match the original customer number and other details from the original request. The file also contains information from the warehouse about how many days the delivery takes.
Input from a WebSphere MQ message, which contains the postage cost for the customer order that is identified by the CustomerNumber field value.

The following diagram shows the CollectorNodeSampleFlow message flow:

Screen capture of the CollectorNodeSampleFlow message flow

The first Collector node in the message flow demonstrates:

Collecting input from different input sources, two from WebSphere MQ messages and one from an external file.
Using pattern matching on the message content, and the file name to correlate input for each collection that is created.
Completing collections by using the Quantity property that is set on each input terminal.

The second Collector node in the message flow demonstrates:

Setting no pattern matching on message content.
Completing collections by using the Timeout property on the input terminal.
Batching message collections.
Using the TimeoutNotification node to control when complete collections are propagated.

The warning symbol that is displayed on the Collector nodes in the diagram indicates that the Catch terminals are not connected.

The CollectorNodeSampleFlow message flow

Message flow input

The message flow takes input from the following three sources:

Input to the MQInput node, MQ COLL_IN1, are XML messages containing details of the item ordered. The format of the input messages is:

<Invoice> 
     <CustomerNumber>111</CustomerNumber> 
     <Product>Toaster</Product> 
     <Quantity>1</Quantity> 
     <UnitPrice>4.99</UnitPrice> 
</Invoice>

Input to the FileInput node, fileinput, are files containing the customer order number as part of the file name, for example 111.order. These files contain delivery information from the warehouse. The format of the input message is:
```
<Order><OrderNumber>111</OrderNumber><Product>Toaster</Product><Quantity>1</Quantity><Delivery>2
days</Delivery></Order> 
```
Input to the MQInput node, MQ COLL_IN3, is the current postage rate that must be added to the price of the ordered items. The format of the input message is:
```
<Packaging>
  <CustomerNumber>111</CustomerNumber>
  <Postage>1.95</Postage>
</Packaging>
```

The input from the three input nodes is sent to the first Collector node in the flow, CollectByCustNum, which is configured to complete a collection when one message from each input source matches the criteria configured in the Collector node.

Pattern matching can be used to correlate messages into a message collection based on the content of incoming data. Pattern matching can be defined on a given input terminal. The pattern matching that is configured on each input terminal can use two fields to set up the pattern matching:

A correlation path to set the location in the incoming data of the value that is matched by using an XPath expression.
A correlation pattern that can extract the whole value, or use wildcards to extract part of the incoming data.

In the Collector node, pattern matching is set in the Collection definition table. You can find this table on the Basic page of the Properties view.

The CollectByCustNum Collector node has three input terminals to collect input from the three input nodes:

Input terminal IN1 receives input from MQInput node MQ COLL_IN1
Input terminal IN2 receives input from FileInput node fileinput
Input terminal IN3 receives input from MQInput node MQ COLL_IN3

The following diagram shows the three input terminals:

Screen capture of the CollectByCustNum Collector node

On CollectByCustNum the input from MQInput node MQ COLL_IN1 and FileInput node fileinput are correlated on CustomerNumber. No pattern matching is required on input terminal IN3, instead the input from MQInput node MQ COLL_IN3 contains a postage price that is added to each collection.

In the following example, the CustomerNumber is 111. The input to the three input terminals is:

Input terminal IN1: the messages arriving on input terminal IN1 from MQ COLL_IN1 have the format:
```
 <Invoice> 
     <CustomerNumber>111</CustomerNumber>  
     <Product>Toaster</Product> 
     <Quantity>1</Quantity> 
     <UnitPrice>4.99</UnitPrice> 
</Invoice> 
```
The Correlation Path set on input terminal IN1 is set to $Body/Invoice/CustomerNumber so that the correlating value in the preceding input data is the customer number 111.
Input terminal IN2: the file that is read in by the FileInput node has a file name in the format <CustomerNumber>.order, for example 111.order.
The Correlation Path set on input terminal IN2 is set to $LocalEnvironment/File/Name. The Correlation Pattern set on input terminal IN2 is set to *.order.
The FileInput node sets $LocalEnvironment/File/Name to the name of the file that is read. In the example this value is set to 111.order. By using this setting for the Correlation Path, together with the wildcard pattern *.order, for the Correlation Pattern, you can correlate the input on input terminal IN2 with the customer number 111.
Input terminal IN3: the postage price for the order must be added to the order. The Correlation Path that is set on input terminal I31 is set to $Body/Packaging/CustomerNumber so that the correlating value is the customer number 111.

Message collections are created with subtrees containing the original order, delivery information from the warehouse, and a postage price.

The completed message collection is propagated to the Compute node, Compute1, which combines the input to create a new invoice message. This Compute node performs the following tasks:

Multiplies the quantity of items with the price of the items ordered, and then adds the postage price from input terminal IN3 to create the final price.
Adds a delivery element from the XML data from input terminal IN2.

The new message that is created by the Compute node, has the following structure:

<Invoice> 
    <CustomerNumber>111</CustomerNumber> 
    <Product>Toaster</Product> 
    <Quantity>1</Quantity>  
    <Price>8.94</Price> 
    <Delivery>2 days</Delivery> 
</Invoice>

The invoice message that is created in Compute1 is propagated to the second Collector node in the message flow, CollectUntilTrigger. This Collector node has one input terminal that is called IN4. In this Collector node the Timeout property is used to complete message collections, rather than the Quantity property that is used by the first Collector node. The Timeout property is set to 30 seconds so that all messages received within the 30 second timeout period form a completed collection.

Event coordination is also enabled in this node. A Timeout Notification node is connected to the Control terminal of the Collector node. Therefore, completed collections are not propagated to the next node in the flow until the control terminal is triggered by the TimeoutNotification node. The TimeoutNotification node is set to send a message every 90 seconds.

The message collections are sent to a Compute node, Compute2. The Compute node takes the message data from each collection and adds time and date information to form the final output message. Each collection might contain more than one invoice depending on the number of invoice messages that are added to each collection in Collector node CollectUntilTrigger.

The final output message is in the form:

<Invoices> 
  <Invoice> 
     <InvoiceNumber>111</InvoiceNumber> 
      <Product>Toaster</Product> 
     
<NumberOrdered>1</NumberOrdered> 
      <TotalCost>9.94</TotalCost> 
      <Delivery>2 days</Delivery>       
 </Invoice> 
  <Invoice> 
     <InvoiceNumber>222</InvoiceNumber> 
      <Product>Food
Processor</Product> 
     
<NumberOrdered>2</NumberOrdered> 
      <TotalCost>75.93</TotalCost> 
      <Delivery>4
days</Deliver>       
 </Invoice>  
 <Date>2007-08-10</Date> 
 <Time>17:18:22</Time> 
</Invoices>

Notice in the message flow that a Compute node follows each Collector node. You must have either a Compute node or a JavaCompute node following a Collector node, because only these nodes can parse the message tree that is propagated by a Collector node. You can use a Compute node or JavaCompute node to take message parts from the message collection and to generate new messages that can be processed by other nodes in your message flow.

Back to sample home