Loan Broker Example

As the main flow in the Loan Broker application, this flow coordinates data collection among flows to produce an end user response.

The request-response HTTP Inbound Endpoint in this flow receives an end user request. Because it has a two-way message exchange pattern, this HTTP endpoint is responsible for both receiving and returning messages.

Next, the Body to Parameter Map Transformer converts the data format of the message payload from HTTP body data to a Java map. The Loan Broker application only processes Java message payloads.

Then, Mule employs a content-based router to direct the message for further processing. The Choice Flow Control routes each message to one of two processing pathways according to its payload contents.

The create customer request component uses expressions to extract data from the message payload. It uses the data to create a new Java object with three values:

With a new CustomerQuoteRequest object in its payload, the message encounters its first Message Enricher. Throughout this flow, Mule enriches messages with data rather than changing the payload contents. By enriching a message, Mule preserves the payload content so that other elements in the application can access the original data.

The Enrich with creditProfile enricher contains only a Flow Reference Component. This type of component invokes other flows, or subflows, in the application to acquire, then add data to the message. In this case, the lookupCustomerCreditProfile component demands that the LookupCustomerCreditProfile subflow access an external Web service to acquire the borrower’s credit score. Mule enriches the message with the credit score, then passes the message to the next enricher in the flow.

As with its predecessor, the Enrich with Banks enricher uses a flow reference component to invoke a subflow and acquire data. In this case, instead of adding a credit score, Mule uses the result of the LookupBanks subflow to add a list of banks to the message payload.

Mule then uses a Variable Transformer to create an empty list variable. Mule will fill this empty quotes list variable with the quotes it fetches from banks. With an empty list to fill, the message next encounters a Foreach scope. One by one, this iterative processor fetches data to populate each item on the list.

To fetch these data, the flow reference component first invokes the LookupLoanQuote subflow to acquire a quote from a bank. Then, the message enricher adds the quote to the list variable. Foreach continues to invoke, then enrich, until it has acquired a quote from each bank on the list of banks. Foreach then passes the message to the next message processor in the flow.

To illustrate foreach’s behavior with an example, imagine a message payload with the following contents:

Foreach processes the message payload as follows:

The penultimate message processor in this flow references yet another subflow in the application. The findLowestLoanQuote subflow determines which quote in the list is the lowest, then logs the result in the message payload.

Finally, the Object to String Transformer converts the message payload’s data format from Java to a string. The HTTP endpoint sends the response to the end user.

Notice that the Loan-broker-sync flow also contains a Catch Exception Strategy. Rather than use Mule’s default exception strategy this flow uses a customized exception strategy to handle errors. If an error occurs in the flow, the exception strategy’s Set Payload Transformer sets an error message on the payload. The application sends this error message, which reads, Error processing loan request, as a response to the end user.

Invoked upon demand by the Loan-broker-sync flow, this subflow acquires and logs the borrower’s credit score on the message payload.

To acquire the credit score, the customer transformer sets the payload to Customer, as defined by the create customer request expression transformer. (Recall that the Customer variable contains the borrower’s name and SSN.) Mule sends a request to the getCreditProfile SOAP Web service. The HTTP endpoint inserts the Web service’s response into the subflow.

Mule leverages Apache’s CXF framework to build Web services. The Processor Chain that wraps the SOAP Component and HTTP outbound endpoint is a CXF requirement. It ensures that Mule completes all processing activities prior to logging the processing result.

Last in this flow, the Logger Component logs the payload of the Web service’s response on the message payload as the Credit Profile.

The Loan Broker application prevents exposing all banks to all loan quote requests. A bank that caters to premiere clients, for example, would be irked to receive a request for a quote for a small loan from a borrower with poor credit. To prevent such irksome calls to banks’ Web services, the Loan Broker application employs the LookupBanks subflow.

Mule first uses a choice flow control to examine the amount in the payload, then routes the message according to the size of the loan.

Note that the choice flow control directs the message to the first expression that evaluates to true. For example, it directs a quote request for a loan of $30,000 only to the Bank 1, Bank 2 component.

Each expression component in this subflow contains the URIs of the banks willing to provide an interest rate quote. For example, messages that pass into the Bank 3, Bank 4 component earn, as a payload addition, the URIs for Banks 3 and 4. The banks Logger component records the list of appropriate banks to which to send a request.

This sends a quote request to banks' Web services.

First, the variable transformer stores the Mule message payload — the bank’s URI — as a variable named bankUri. (Recall that this subflow receives requests one at a time from foreach in the Loan-broker-sync flow. Each request’s payload a the URI of a bank.)

The create LoanBrokerLoanRequest component uses expressions to extract the borrower’s credit profile (logged by the creditProfile logger in the LookupCustomerCreditProfile flow) from the message payload. It uses the data to create a request to send to the getLoanQuote Web service.
Mule uses a SOAP component — configured as a JAXWS-client — to send the request to a bank’s Web service. The HTTP outbound endpoint dynamically determines where to send the request based on the bank’s URI in the message payload. It receives the response from the banks’ Web service and pushes the response payload to the quote logger to record.

This simple subflow uses an expression component to determine which item, in the list of quotes, offers the lowest interest rate. The Logger records the result.

The expression in the component compares the getInterestRate of items in the list to each other to determine which one is the lowest (see image below).

The remaining six flows in the Loan Broker application are “mock flows”. They act as external Web services to which the five legitimate flows and subflows call to request data.

Each flow contains:

You do not need to include these flows your customized application; they exist in the Loan Broker example only to support a functional example.