using-archipy-logics

Using ArchiPy Logics

Logic classes are pure business rules and the unit-of-work boundary.

Live: https://syntaxarc.github.io/ArchiPy/getting-started/concepts/

Responsibilities

• Accept domain DTOs as input; return domain DTOs as output.
• Framework-agnostic — no FastAPI / gRPC imports.
• Depend on repositories / ports via DI — do not construct adapters.

Unit of Work

When using Postgres SQLAlchemy, decorate public methods:

from archipy.helpers.decorators.sqlalchemy_atomic import postgres_sqlalchemy_atomic_decorator

class UserRegistrationLogic:
    @postgres_sqlalchemy_atomic_decorator
    def register_user(self, input_dto: UserRegistrationInputDTO) -> UserRegistrationOutputDTO:
        ...

• Async stack: async_postgres_sqlalchemy_atomic_decorator.
• Nested atomic-decorator calls reuse the open session — logics may call other logics.

Domain Isolation

✅  OrderCreationLogic → UserQueryLogic → UserRepository
❌  OrderCreationLogic → UserRepository   (bypasses domain boundary)

• May call other domain logics.
• Never import or call a repository from another domain.

Errors

• Raise domain errors for rule violations (BusinessRuleViolationError, AlreadyExistsError, …).
• Infrastructure failures should already be domain errors from the adapter boundary; re-raise with raise ... from e when wrapping.

Do Not

• Put HTTP status mapping or router code here (that belongs in services/).
• Skip *_sqlalchemy_atomic_decorator on write/read paths that need a SQLAlchemy session when SQLAlchemy is in use.
• Construct Redis/DB clients inside logic classes.