bpmn-to-code

ADR 004: Strategy Pattern for Multi-Engine Support ​

Status ​

Accepted

Context ​

Different BPMN process engines (Camunda 7, Zeebe) use different XML namespaces, element attributes, and extension mechanisms. Extracting process information requires engine-specific parsing logic.

Decision ​

Use Strategy pattern with a registry of engine-specific extractors:

kotlin
Map<ProcessEngine, EngineSpecificExtractor>

Each extractor implements EngineSpecificExtractor interface and handles its engine's specifics. The ExtractBpmnAdapter selects the appropriate extractor based on the target engine.

Consequences ​

Positive ​

  • Extensibility: New engines added by implementing EngineSpecificExtractor
  • Separation: Engine-specific logic isolated in dedicated classes
  • Maintainability: Changes to one engine don't affect others
  • Clear contract: Interface defines what extraction must provide

Negative ​

  • Cannot share parsing logic between similar engines (code duplication)
  • Adding common functionality requires updating all extractors

Implementation ​

kotlin
// Interface
interface EngineSpecificExtractor {
    fun extract(file: File): BpmnModel
}

// Registry
val extractors = mapOf(
    ProcessEngine.ZEEBE to ZeebeModelExtractor(),
    ProcessEngine.CAMUNDA_7 to Camunda7ModelExtractor()
)

Future engines (e.g., Flowable, jBPM) can be added by creating new extractor implementations.