You are viewing the documentation for Blueriq 14. Documentation for other versions is available in our documentation directory.
A function is a special relationship between inputs and outputs. The function transforms the input into the desired output. In Blueriq, functions do exactly that, and they can make use of the rule engine and flows. As the inputs and outputs of a function are clearly defined, it is clear what information is expected by the function, and which attributes have an updated value after the function call. The internal workings of a function are hidden, to reduce the long lists in studio and the mental load of the Business Engineer. Both these points lead to a high maintainability of your model.
You may think of a function as a computation or a service which is provided to you as a caller of the function. Examples of functions are:
- Calculating the interest that you need to pay for a loan over the next years
- Inputs: Loan amount, Number of years, Interest rate, Monthly payment
- Output: Total amount of money owed after given number of years
- Deciding if you are eligible for a subsidy for solar panels
- Inputs: Type of roof, Size of roof in square meters, Number of solar panels, Costs of solar panels, Number of inhabitants
- Outputs: Decision, Subsidy amount, Reason for denial
Intended use
Blueriq functions are intended for decoupling smaller calculations. This is what they are good at and should be used for. There are other scenarios, in which you might consider functions, for which a function may not be the best choice. Below there are some scenarios discussed
- A function which has side effects. For example when a function stores data in an aggregate or writes data to an external system. This would mean that the outcome of a function can differ after calling it multiple times. This may be confusing to users which expect the outcome to be identical.
- A function which interacts with the process engine. In basis this is the same as above, as the state of the process engine would change, and the result of the function may be different on the second call. In fact using the process engine in a function currently does not work at all.
- A function with complex parameters. When you need to send a complex structure with instances which have relations to other instances and so on, then a web service is a much better option. Functions intentionally have simple parameters.
- A recursive function. You could model a function which calls another function. There is however no mechanism in place to prevent infinite looping and you have to be really carefull on what you do. Therefore this is discouraged.
Modeling
A function call can be compared to a service call in your flow. It behaves more or less the same, except that a different part of the model is used for execution. To use a given function, you create a function call by filling in the input parameters, and mapping the output parameters. Here is an example:
Now, you can place this function call in a flow:
You can use the exception exit in case the function fails to execute.
Functions in library
Of course functions can be modeled in a project structure with multiple stacked modules, when doing this you have to be aware of this fact: when calling the function, by default it will select the function definition from the lowest implementation module possible. It is possible to call the function if specified at a higher (implementation) module, but you have to explicitly point to that definition.
Input validation
When modeling a function it is possible to check the validity of your input parameters. The input type is always checked, but other validations can be added to the parameters. Please note that when a multi-valued attribute as target attribute, your parameter also becomes multivalued. The validations can be added by adding a validation rule or validation type to the input attribute. This validation will then be triggered when calling the function. When a validation is triggered there will be an error message, on screen for the user when the function is called from within Blueriq or in the JSON result when exposed and called as a webservice.
When an input parameter is added to the function flow which has a validation, this will show in the parameters tab of the flow as a validation icon. 
Exposing the function as a REST webservice
We offer the option to expose a function as a REST webservice. This is done by selecting the Exposed as webservice checkbox in the general tab.
When you reload projects in the runtime after saving the flow with this option checked, the function can be approached as a webservice.
The URL of the webservice will be:
http://<environment>:<port>/Runtime/server/api/v2/function?project=<project>&function=<flowname>&version=<branch>&module=<module>
In which <environment>, <port>, <project>, <flowname> & <branch> can be found by selecting the flow and right clicking on the start button on the runtime overview and then copying the shortcut.
The call to the webservice is a JSON message containing the parameters defined in the function. The HTTP method being called is POST.
When calling the webservice, the HTTP headers Content-Type and Accept have to be set with application/json
For the example used to calculate the loan remainder this would look like:
{
"LoanAmount":100000,
"NumberOfYears":5,
"InterestRate":3.5,
"MonthlyPayment":500
}
The webservice will also return a JSON message containing the output parameters of the function.
For the example it would look like:
{
"MoneyOwed":25000
}
Changing contract
When a function flow is exposed as a web service, changing the parameters may break contract with the using party. Be careful when changing the parameters while the exposed web service is already in use.
Development Dashboard widget
When looking for the endpoint of a function which is exposed as a webservice you can make use of the Development dashboard widget "Blueriq Exposed functions"
This will show a widget in which you can select the exact repository, branch, project, module and function of which you would want to see the url. When a function is selected the input and output parameters and corresponding parameter types are shown (the contract).
- Unknown values will be mapped to null in JSON
- Interpret null values are interpret as UNKNOWN when a primitive value is expected
Shortcut to exposed function flow
It is also possible to make a shortcut to an exposed function flow.
To do this you must add a function shortcut in the application.properties with the following content:
|
The url to call the shortcut will be: /Runtime/server/api/v2/function/shortcut/{shortcutname}/
If the property blueriq.production.shortcutsOnly is set and the runtime is running in production mode, only function shortcuts can be used to start a function.
Support for Test Paths was added in Blueriq 14.11.3.
Swagger feed for exposed function flows
A Swagger feed of Function Flows exposed as a webservice is provided by the Runtime, see Swagger for more information.
Securing an exposed function flow
By default, an exposed function flow can be called without authentication. It is necessary to add the 'securityEnabled' property to the shortcut, this will make sure that the endpoint is secured. After securing the endpoint, a role can be added to the flow to make sure the user is authorized. If you add one or more roles to the function flow, the runtime will require credentials via basic authentication.
Further Reading
Overview
Content Tools