Generating Multiple Derived Events from a Single Source Event

The methods that are published to SAS Micro Analytic Service can be used to generate multiple derived events when a single source event is received. This is accomplished by coding an array output argument for each derived event field. The values in the arrays are then used to populate the fields of multiple derived events.

A SAS Micro Analytic Service module method can produce one or more arrays. When such a method is mapped to a SAS Event Stream Processing source window, and when at least one of the output array names matches a derived event field name, multiple derived events can be generated. The number of elements in the matching array or arrays at run time determines the number of events that are generated. As is the case with scalar outputs, any array output that does not match a derived event field name is ignored.

When you are authoring a method that is capable of generating more than one event, the best practice is to produce parallel arrays of the same size. When you follow this practice, each array element contributes one value to each of the derived events, in order. The arrays can contain different numbers of elements across method calls (for example, they might generate one derived event for one call, three derived events for another, and so on). However, for a single method call, the best practice is to produce the same number of elements in each array.

If you ignore these best practices, you can still generate multiple events, but keep the following rules in mind:

Scalar output values, if any, are repeated in the corresponding fields of every derived event.

SAS Event Stream Processing requires every event to have a unique key composed of one or more event fields. Therefore, the source event's key cannot be duplicated in multiple generated derived events. In that case, SAS Event Stream Processing halts processing and returns a duplicate key error. Because of this, the method author must ensure that each derived event has a unique key value. To do that, produce an array containing unique key values and mark the matching event field as a key in the derived event schema. This technique is useful whenever you author your own module.

SAS Micro Analytic Service can execute modules that are generated by SAS analytics products, where the analytic functions do not produce arrays of unique key values. To accommodate such functions, SAS Micro Analytic Service provides a key generation feature. To use this feature, add a key field named _masRowNum of type int32 to the derived event schema, making it part of a composite key. When _masRowNum is present, SAS Micro Analytic Service populates the field with the row number of each derived event, in order, starting with 1. This feature ensures that the composite key value is unique across multiple generated derived events.

Here is an example of a derived event schema:

ID*:int32,_masRowNum*:int32,symbol:int32,
   quantity:int32,price:double,total:double

When a set of derived events is generated from a single source event, you can explicitly set opcodes or flags in the derived events either individually or as a group. To set individual opcodes, use the _outOpcodeArray meta argument. Similar to _outOpcode, _outOpcodeArray can also be used to suppress the generation of any of the individual derived events.

To set flags in the derived events individually, use the _outFlagsArray meta argument.

To set the same opcode in all the derived events that are generated from a single source event, use the _outOpcode meta argument and omit the _outOpcodeArray meta argument. Similarly, to set the same flags in all the derived events that are generated from a single source event, use the _outFlags meta argument and omit the _outFlagsArray meta argument.

For information about the meta arguments _outOpcodeArray, _outFlagsArray, _outOpcode, and _outFlags, see Operation Codes and Flags.