MDSL (Micro-)Service Contracts Generator

Introduction and Motivation

The Microservices Domain Specific Language (MDSL) is a Domain-Specific Language (DSL) to specify (micro-)service contracts and data representations, jointly realizing the technical part of the API Description pattern from Microservice API Patterns (MAP).

Our MDSL generator automatically produces (micro-)service contracts out of strategic DDD context maps written in CML. The generator creates the contracts according to the following mapping, which reflects our proposal how we would derive (micro-)services from models based on strategic DDD . The generator aims at providing assistance regarding how your system can be implemented as a (micro-)service-oriented architecture.

Language Mapping

CML Input MDSL Output Description
Upstream Bounded Contexts from upstream-downstream relationships Service Specification (API description) We create one service specification for each upstream Bounded Context of your Context Map.
Exposed Aggregates Endpoint Every exposed Aggregate of your upstream Bounded Context results in one endpoint.
Public methods/operations of the aggregate root entity or of Services. Operation Your exposed Aggregates should contain methods/operations, either on the aggregate root entity or in Services. For every method/operation in those objects we generate an operation in MDSL.
Parameters & return values of methods/operations Base types or data type specifications if possible If you use primitive data types in CML, they are mapped to the base types of MDSL. If you refer to objects (such as entities) in CML, we produce a corresponding parameter tree. Types which are not further declared are mapped to abstract, unspecified elements (P).
Upstream Bounded Contexts from upstream-downstream relationships API provider For the upstream Bounded Context we also generate an API provider.
Downstream Bounded Contexts from upstream-downstream relationships API client Downstream Bounded Contexts are mapped to corresponding API clients.

Data Type Mapping

The base/primitive types are mapped to Atomic Parameters as this:

CML type MDSL type
String D<string>
int or Integer D<int>
long or Long D<long>
double or Double D<double>
boolean D<bool>
Blob D<raw>
Date D<string> (no date available in MDSL)
Note: Types in CML are case-sensitive. For example: If you write "string" instead of "String", you create a new abstract data type instead of using the primitive type "String".

If you declare a method with multiple parameters or refer to an object (such as Entity or Value Object) in CML, we generate a corresponding [Parameter Tree(https://microservice-api-patterns.org/patterns/structure/representationElements/ParameterTree). For example the following entity would be mapped to the (rather flat) parameter tree below:

CML input:

Entity Address {
  String street
  String lockbox nullable
  int postalCode
  String city
}

MDSL data type result:

data type Address { "street":D<string>, "lockbox":D<string>?, "postalCode":D<int>, "city":D<string> }

All abstract data types that are not base types and not specified in CML (no references to objects) will produce an abstract, unspecified placeholder element P in MDSL, as the following example illustrates:

data type JustAnUnspecifiedParameterType P

Example

An exemplary API description in MDSL, generated by Context Mapper, is:

// Generated from DDD Context Map 'Insurance-Example_Context-Map.cml' at 21.10.2019 17:48:52 CEST.
API description CustomerManagementContextAPI
usage context PUBLIC_API for BACKEND_INTEGRATION

data type Address { "street":D<string>, "postalCode":D<int>, "city":D<string> }
data type AddressId P
data type changeCustomerParameter { "firstname":D<string>, "lastname":D<string> }

endpoint type CustomersAggregate
  serves as INFORMATION_HOLDER_RESOURCE
  exposes
    operation createAddress
      with responsibility "Creates new address for customer"
      expecting
        payload Address
      delivering
        payload AddressId
    operation changeCustomer
      with responsibility "Changes existing customer address"
      expecting
        payload changeCustomerParameter
      delivering
        payload D<bool>

// Generated from DDD upstream Bounded Context 'CustomerManagementContext' implementing OPEN_HOST_SERVICE (OHS) and PUBLISHED_LANGUAGE (PL).
API provider CustomerManagementContextProvider
  // The customer management context is responsible for managing all the data of the insurance companies customers.
  offers CustomersAggregate
  at endpoint location "http://localhost:8001"
    via protocol "RESTfulHTTP"

// Generated from DDD upstream Bounded Context 'CustomerManagementContext' implementing OPEN_HOST_SERVICE (OHS) and PUBLISHED_LANGUAGE (PL).
API client PolicyManagementContextClient
  // This bounded context manages the contracts and policies of the customers.
  consumes CustomersAggregate
API client CustomerSelfServiceContextClient
  // This context represents a web application which allows the customer to login and change basic data records like the address.
  consumes CustomersAggregate

IPA

Note: This example has been generated from our insurance example, which you can find in our examples repository.

Microservice API Patterns (MAP) Decorators

The MDSL language allows modelers to specify the roles of endpoints and operations according to the endpoint- and operation-level responsibility patterns in MAP. Our generators match the corresponding pattern names in comments on Aggregates and methods. The following CML code illustrates how the MAP patterns can be added in CML. In this case we use the Information Holder Resource pattern on the Aggregate level and the Retrieval Operation pattern on the method level:

BoundedContext CustomerManagementContext {

  "INFORMATION_HOLDER_RESOURCE"
  Aggregate Customers {

    Entity Customer {
      aggregateRoot

      - SocialInsuranceNumber sin
      String firstname
      String lastname
      - List<Address> addresses

      "RETRIEVAL_OPERATION"
      def @Address getAddress(AddressId addressId);
    }

    Entity Address
  }
}

The patterns are detected by our generator and mapped to the corresponding language features of MDSL. The following MDSL code illustrates the resulting resource for the Aggregate specified above:

data type Address { "street":D<string>, "postalCode":D<int>, "city":D<string> }
data type AddressId P

endpoint type CustomersAggregate
  serves as INFORMATION_HOLDER_RESOURCE
  exposes
    operation getAddress
      with responsibility RETRIEVAL_OPERATION
      expecting
        payload AddressId
      delivering
        payload Address

As illustrated above, the patterns on the endpoint/resource level are added with the serves as keyword and on the operation level with the with responsibility keyword. In the following we list the supported patterns:

Endpoint Role Patterns

Operation Responsibility Patterns

Please refer to the pattern texts on the MAP website for explanations of these decorators.

User Guide

You can generate MDSL (micro-)service contracts from your CML model as follows.

When you a right-click on a CML file in Eclipse, you a Context Mapper context menu will pop up. The action MDSL: Generate Service Contracts allows you to generate the contracts for all upstreams in your Context Map:

MDSL Generator

Note that the Context Mapper menu entry is also available within the context menu uf the CML editor. (right-click anywhere in the editor).

All MDSL files will be generated into the src-gen folder of the project:

MDSL Generator Result

Note: The MDSL Eclipse plugin is not yet available for download on an Eclipse update site. At the moment you can open the `*.mdsl` files with a text editor only (with no syntax highlighting and editor support). You can also request access to the MDSL repository on GitHub by [contacting the MAP team](https://microservice-api-patterns.org/about).

Protected Regions

After you have generated an MDSL contract, you can add protected regions for data types, endpoint types, API providers, and API clients if you want to modify parts of the contract and protect them from being overwritten. The following example shows the corresponding comments that are required to begin and end the four different protected regions:

// Generated from DDD Context Map 'Insurance-Example_Context-Map.cml' at 21.10.2019 17:48:52 CEST.
API description CustomerManagementContextAPI
usage context PUBLIC_API for BACKEND_INTEGRATION

// ** BEGIN PROTECTED REGION for data types

// ** END PROTECTED REGION for data types

data type Address { "street":D<string>, "postalCode":D<int>, "city":D<string> }
data type AddressId P
data type changeCustomerParameter { "firstname":D<string>, "lastname":D<string> }

// ** BEGIN PROTECTED REGION for endpoint types

// ** END PROTECTED REGION for endpoint types

endpoint type CustomersAggregate
  serves as INFORMATION_HOLDER_RESOURCE
  exposes
    operation createAddress
      with responsibility "Creates new address for customer"
      expecting
        payload Address
      delivering
        payload AddressId
    operation changeCustomer
      with responsibility "Changes existing customer address"
      expecting
        payload changeCustomerParameter
      delivering
        payload D<bool>

// ** BEGIN PROTECTED REGION for API providers

// ** END PROTECTED REGION for API providers

// Generated from DDD upstream Bounded Context 'CustomerManagementContext' implementing OPEN_HOST_SERVICE (OHS) and PUBLISHED_LANGUAGE (PL).
API provider CustomerManagementContextProvider
  // The customer management context is responsible for managing all the data of the insurance companies customers.
  offers CustomersAggregate
  at endpoint location "http://localhost:8001"
    via protocol "RESTfulHTTP"

// ** BEGIN PROTECTED REGION for API clients

// ** END PROTECTED REGION for API clients

// Generated from DDD downstream Bounded Context 'PolicyManagementContext' implementing CONFORMIST (CF).
API client PolicyManagementContextClient
  // This bounded context manages the contracts and policies of the customers.
  consumes CustomersAggregate
API client CustomerSelfServiceContextClient
  // This context represents a web application which allows the customer to login and change basic data records like the address.
  consumes CustomersAggregate

IPA

The protected regions allow you to guard data types, endpoints, API providers, and API clients into so that they are not overwritten at re-generation. Thus, you can call the MDSL generator on the same file again and all objects within a protected region will still be there and remain unchanged.

For example, you can move a set of data types into the corresponding protected region if you changed the data types manually after generation and want to protect them:

// Generated from DDD Context Map 'Insurance-Example_Context-Map.cml' at 21.10.2019 17:48:52 CEST.
API description CustomerManagementContextAPI
usage context PUBLIC_API for BACKEND_INTEGRATION

// ** BEGIN PROTECTED REGION for data types

data type Address { "street":D<string>, "postalCode":D<int>, "city":D<string>, "manuallyChangedThisDataType":D<string> }

// ** END PROTECTED REGION for data types

data type AddressId P
data type changeCustomerParameter { "firstname":D<string>, "lastname":D<string> }

// removed the rest here to save space ...

IPA

MDSL Support

The current version of our MDSL generator is compatible with the MDSL version 1.2.0. For further questions regarding MDSL, please visit its website https://socadk.github.io/MDSL.