Rule Engine Overview

RobustMQ Rule Engine provides lightweight in-connector data processing. It is not designed to replace stream-compute systems like Flink. The goal is to cover common transformation needs in MQTT and connector pipelines with lower operational complexity.

Processing Model

The runtime pipeline is:

decode -> ops[] -> encode

• decode: normalize source payload into Vec<Map<String, Value>>
• ops[]: run ordered operators on normalized records
• encode: serialize transformed records back to Bytes

Current First Operator

The first implemented operator is Extract.

• Input: field_mapping (source_field -> target_field)
• Supports source path styles:
  • dot path, e.g. session.mqtt.topic
  • JSON Pointer, e.g. /payload/alarms/0/active
• If source field is missing, target value is filled with "-"
• Current Extract semantics are projection output: only mapped target fields are kept

Related Example

See the end-to-end demo here:

• Rule Engine Processing Demo