Skip to main content
Version: 1.14

Basic Components

Nodes work with a data records. They can produce, fetch, send, collect data or organize data flow. Each node has at least two parameters: Name and Description. Name has to be unique in a scenario. Description is a narrative of your choice.

Most of the nodes, with source and sink nodes being notable exceptions, have both input and at least one output flow.

Sinks and filters can be disabled by selecting Disable checkbox.

 

Variable​

A Variable component is used to declare a new variable; in the simplest form a variable declaration looks like in the example below. As the event was read from the Kafka topic, the #input variable holds its content and its value is assigned to a newly declared myFirstVariable variable.

alt_text

As you can see in the variable configuration form below, Nussknacker inferred the data type of the #input variable from the information already available to Nussknacker.

alt_text

In the next example #input variable is used to create an expression returning a boolean value. If the input Kafka topic contains JSON objects and they contain operation field, the value of this field can be obtained in the following way:

#input.operation

Note that internally Nussknacker converts JSON’s object into SpEL’s map.

alt_text

 

RecordVariable​

The specialized record-variable component can be used to declare a record variable (object in JSON)

alt_text

The same can be achieved using a plain Variable component, just make sure to write a valid SpEL expression.

alt_text

 

Filter​

Filter passes records which satisfy the filtering condition to true sink.

filter graph single

You can additionally define false sink. Records from the source which meet the filter's condition go to the true sink, and others go to the false sink.

filter graph

The Expression field should contain the SpEL expression for the filtering conditions and should produce a boolean value.

filter window

 

Choice​

Choice is more advanced variant of filter component - instead of one filtering condition, you can define multiple conditions in some defined order. It distributes incoming records among output branches in accordance with the filtering conditions configured for those branches.

choice graph

Each record from the source is tested against the condition defined for outgoing node. If #input.color is blue record goes to the blue sink.
If #input.color is green record goes to the green sink. For every other value record goes to the sink for others because condition true is always true. Order of evaluation of conditions is the same as visible in form. You can modify the order using drag & drop functionality. Order is also visible on graph in edges description as a number. Be aware that layout button can change displayed order of nodes, but it has no influence on order of evaluation.

choice window

 

Split​

Split node logically splits processing into two or more parallel branches. Each branch receives all data records and processes them independently and in parallel.

In the Request - Response processing mode you can use this feature to paralellize and hence speed up the processing. You must use a sequence of Union and Collect nodes to merge parallelly executed branches and collect the results from these branches. A discussion of Request - Response scenario with multiple branches can be found here. In the Streaming processing mode the most typical reason for using a Split node is to define dedicated logic and dedicated sink for each of the branches.

split graph

Example: (Streaming processing mode) - every record from the source goes to sink 1 and sink 2.

Split node doesn't have additional parameters.

 

ForEach​

for_each

for-each transforms the stream so that subsequent nodes are executed once for every value (possibly multiple times). This node has two parameters:

  • Elements - list of values over which to loop. It can contain both fixed values and expressions evaluated during execution.
  • Output Variable Name - the name of the variable to which element value will be assigned.

For example, when

  • Elements is {#input.value1, #input.value2}
  • Output Variable Name is outputVar

then nodes that follow for-each will be executed twice and the value of current element can be referenced as #outputVar.

 

Union​

union_window

Union merges multiple branches into one branch.

In the Streaming processing mode events from the incoming branches are passed to the output branch without an attempt to combine or match them.  

In the Request - Response processing mode only one response sink can return value. If you have parallel branches of processing the Union node is used to merge them and then Collect node to collect results of processing in each of the merged branches. Check Into doc for details on how to interpret scenario graph in different processing modes.

The #input variable will be no longer available downstream the union node; a new variable will be available instead, which is defined in the union node.

Branch names visible in the node configuration form are derived from node names preceding the union node.

Example: union_example

Entry fields:

  • Output Variable Name - the name of the variable containing results of the merge (replacing previously defined variables, in particular #input).
  • Output Expression - there is one expression for each of the input branches. When there is an incoming event from a particular input branch, the expression defined for that branch is evaluated and passed to the output branch. The expressions defined for respective branches need to be of identical data type. In the example above it is always a record containing fields branchName and value.

Note, that the #input variable used in the Output expression field refers to the content of the respective incoming branch.