layer-interfaces

Interfaces Layer

Purpose: Expose Nomarr to the outside world via HTTP (FastAPI), CLI (Typer), and web handlers.

Interfaces are thin adapters. They do three things:

1. Validate inputs
2. Call one service method
3. Serialize outputs

No business logic lives here.

---

Allowed Imports

# ✅ Allowed
from nomarr.services import LibraryService, TaggingService
from nomarr.helpers.dto import LibraryDict, FileDict
from nomarr.interfaces.api.types import LibraryResponse  # Pydantic models

Forbidden Imports

# ❌ NEVER import these in interfaces
from nomarr.workflows import ...      # Call services, not workflows
from nomarr.components import ...     # Call services, not components
from nomarr.persistence import ...    # No direct DB access

---

The One Service Call Rule

Each route handler should call exactly one service method.

# ✅ Good - one service call
@router.get("/​library/​{library_id}")
def get_library(library_id: str, library_service: LibraryService = Depends(...)) -> LibraryResponse:
    library = library_service.get_library(library_id)
    return LibraryResponse.from_dto(library)

# ❌ Bad - multiple service calls
@router.post("/​process/​{library_id}")
def process(library_id: str, library_service: LibraryService = Depends(...), tagging_service: TaggingService = Depends(...)):
    library = library_service.get_library(library_id)
    tagging_service.tag_library(library.key)  # ← Extract to service method
    return {"status": "ok"}

If you need multiple service calls: Extract a service method that orchestrates them.

---

Data Flow

Request Flow

JSON → Pydantic Request Model → .to_dto() → Service (DTO)

Response Flow

Service (DTO) → .from_dto() → Pydantic Response Model → JSON

Pydantic models live only in interfaces. Never let them leak into services.

---

Authentication Rules

MANDATORY: All API endpoints require authentication except login.

Web API (/​api/​web/​*)

• Uses verify_session for session token authentication
• All routes MUST include dependencies=[Depends(verify_session)] or _session: dict = Depends(verify_session) as a parameter
• Exception: /​api/​web/​auth/​login is the only unauthenticated endpoint

v1 API (/​api/​v1/​*)

• Uses verify_key for API key authentication
• All routes MUST include dependencies=[Depends(verify_key)]
• Exception: /​api/​v1/​public/​* is intentionally public (version info)

API Consumer Separation — DO NOT MIX

The web frontend MUST ONLY call /​api/​web/​* endpoints.

The v1 API uses API key authentication. The web frontend uses session authentication. These are incompatible — the frontend cannot authenticate to v1 endpoints.

If functionality exists in v1 but the frontend needs it, create a parallel route under web API.

---

Error Handling

• HTTP routes: Raise HTTPException
• CLI commands: Raise typer.Exit(1)
• Let services/​workflows raise domain exceptions, catch them here

@router.get("/​file/​{file_key}")
def get_file(file_key: str, library_service: LibraryService = Depends(...)) -> FileResponse:
    file = library_service.get_file(file_key)
    if not file:
        raise HTTPException(status_code=404, detail="File not found")
    return FileResponse.from_dto(file)

---

Validation Checklist

Before committing interface code, verify:

• Does this file import from workflows, components, or persistence? → Violation
• Does this route call more than one service method? → Extract to service
• Does this route contain business logic (loops, branching, computation)? → Move to service
• Are Pydantic models staying in this layer only? → Services return DTOs
• Is the DTO-to-Pydantic conversion explicit? → Use .from_dto()
• Does this route have authentication? → Add verify_session (web) or verify_key (v1)
• Is the frontend calling /​api/​v1/​*? → Create web API route instead

---

Layer Scripts

This skill includes validation scripts in .github/​skills/​layer-interfaces/​scripts/:

lint.py

Runs all linters on the interfaces layer:

python .github/​skills/​layer-interfaces/​scripts/​lint.py

Executes: ruff, mypy, vulture, bandit, radon, lint-imports

check_naming.py

Validates interfaces naming conventions:

python .github/​skills/​layer-interfaces/​scripts/​check_naming.py

Checks:

• Files must end in _if.py
• Route handlers should be thin (validate, call service, serialize)