Code Generator for Solace Event Portal Applications

The Code Generator uses Solace Event Portal applications represented as AsyncApi and uses them to create scaffolded code for use in integration platforms (iPaaS). The purpose is to facilitate development and to provide an integral component in an efficient Development Lifecylce procedure.

Currently, the Code Generator is capable of:

• Mule - Generating Mule flow projects (jar files) suitable for import into Anypoint Studio
• SAP IFlow - Generating SAP IFlow XML documents. The ability to create packaged IFlow projects is delegated to a separate process

Design

Process Description

The Code Generator uses a two-step translation process:

1. Convert the source format into an internal format.
   • This is a direct mapping of elements into the internal format. Little to no knowledge of the final format is required.
2. Convert the internal format into the iPaaS-specific output structure
   • In this step, most of the business logic is applied. Logic such as flow names, references, etc. are created.

Data Models

• AsyncApiAccessor - Uses GSON to parse AsyncApi input document. AsyncApiAccessor provides convenience methods for extracting data elements.
• MapMuleDoc - A simple internal format to facilitate mapping from AsyncApi. The internal format insulates source mapping from the business logic required to produce generated code. Generated code can be in distinct formats as required by the target iPaaS.
• MuleDoc - A model created for the XML representation of Mule Flows. Unlike the actual mule flow docs used by studio (which uses jaxb), the mapper uses Faster Jackson parsers to serialize the XML data.
• BPMN2/SAP IFlow - Imported data format that provides representation for SAP IFlows. The format is XML wireline.

Build

Build for local use

mvn clean package

Build for use as Lib

mvn clean install

List out dependencies

mvn dependency:copy-dependencies -DincludeScope=runtime -DoutputDirectory=./mydeps

Implement

Mule

The Mule mapper is intended to create complete Mule projects from an Eclipse plugin to Anypoint Studio. (need reference)

The mapper procedure can be called statically using convenience methods in com.solace.ep.codegen.mule.eclipse.EclipseProjectGenerator class. The resulting output is a packaged jar file containing a complete MuleFlow project that can be imported into Eclipse.

EclipseProjectGenerator:

• Method: generateEclipseArchiveForMuleFlowFromAsyncApi
• Paramters:
  • groupId - Maven Group ID of the new Mule Project
  • flowNameArtifactId - Artifact ID
  • version - Version
  • asyncApiAsString - AsyncApi string
  • projectOutputPathAsString - Target path of the generated project
  • cloudHubProject - (boolean, optional) Include CloudHub configuration

SAP IFlow

SAP IFlow generator creastes the IFlow BPMN2/XML document from an AsyncApi. Convenience methods are provided in class com.solace.ep.codegen.sap.iflow.SapIFlowGenerator.

Method	Input	Output
writeSapIflowFileFromAsyncApiFile	AsyncApi File Path	IFlow File
writeSapIflowFileFromAsyncApiString	AsyncApi String	IFlow File
writeSapIflowFileFromMapDoc	MapMuleDoc Object	IFlow File
getSapIflowFromAsyncApiString	AsyncApi String	Return IFlow XML String
getSapIflowFromAsyncApiFile	AsyncApi File Path	Return IFlow XML String
getSapIflowFromMapDoc	MapMuleDoc Object	Return IFlow XML String

Project TO-DO

• Change static maven pom.xml to be generated dynamically
  • Add file comments indicating that the project was generated by Solace tooling
• Solution for circular flow-ref in business logic flows
  • There isn't a great solution for how to handle this. Possibly could create a "logger" sub-flow and link all Business logic subflows to it. To prevent accidental infitite loop, which can be annoying even in dev. Would need instructions to customer for how to handle this.
• Update README instructions for how to install and integrate the project

Completed

• ✅ Add ability to identify and write schema files in resources directory
  • Collect schemas from components.schemas
  • Collect schemas components.messages (Not handled as $ref elts) Currently out of EP:
    • json schemas are handled in the schemas blocks
    • avro schemas are handled in-line in messages block
  • Ensure schema files (by content) are written exactly once to the resources directory
  • Schema files should be appropriately named; there will be difference between avro/json; json schema name can be null
    • Establish order of precedence for names
  • Update ingress and egress flows to point to the correct schema files under resources folder
• ✅ Move Solace connector config to global.xml file and include in the project, according to Mule best practices
• ✅ Construct topic string in Dataweave accounting for parameters