SRI - Fix mapping from Prisma models to domain objects

[ognjenkurtic]

Overview

Currently automapping from Prisma models to domain objects is broken, as automapper complains about using type (Prisma model) when invoking the mapping so we use domain objects as source which causes various issues and worksarounds (look for TODOs with the id of this issue).

example of a wrong mapping:

return this.mapper.map(bpiAccountModel, BpiAccount, BpiAccount);

should be something like:

return this.mapper.map(bpiAccountModel, BpiAccountPrismaClientModel, BpiAccount);

Reference

https://github.com/nartc/mapper/discussions/280

Questions

• Do we to introduce a new type of class (Prisma Model DTO) that only holds the properties of the PrismaModel thus making automapper happy (look at MerkleTreeDto in merkle.profile.ts)? What are the drawbacks if we use this approach?
• Can we stick with PrismaModels but somehow configure Automapper to like them? i.e. pojo strategy

Acceptance

• mapping is performed by giving automapper either a prisma model (type) or a new object which represenets the model (class).

• All mappings in the app are fixed according to the new approach

• New approach is documented in the readme

• All tests pass

• Tasks

• [x] Research and decide on the approach

• [x] Implement POC

• [x] Apply everyhwere

• [x] Update readme

[ognjenkurtic]

This branch has a POC of using the pojos strategy of automapper to achieve mapping from prisma model to a domain object, on the example of reading a bpi subject via an id:

https://github.com/eea-oasis/baseline/tree/feature/740-fix-mappings-from-prisma-models-to-domain-objects

Obviously some more polishing is needed but the overall approach works and can be utilized to solve the issue we currently have. Please have a look.

@biscuitdey @Kasshern @Therecanbeonlyone1969 @Ybittan