pydantic models for input and output?

[hklarner]

Is there a place somewhere to keep pydantic models of vroom input and output? I need them anyways so I thought I could share them. Looks something like this:

from pydantic import BaseModel


class Vehicle(BaseModel):
    id: int

    profile: Optional[str] = None
    description: Optional[str] = None
    start_index: Optional[int] = None
    end_index: Optional[int] = None
    capacity: Optional[Tuple[int]] = None
    skills: Optional[Tuple[int]] = None
    time_window: Optional[Tuple[int, int]] = None  # timestamp or relative seconds
    max_tasks: Optional[int] = None


class Matrix(BaseModel):
    durations: List[List[int]]


class Job(BaseModel):
    id: int
    location_index: int

    description: Optional[str] = None
    service: Optional[int] = None


class VroomRequest(BaseModel):
    vehicles: List[Vehicle]
    matrices: Dict[str, Matrix]
    jobs: List[Job]

[jcoupey]

This does not really belong in the core C++ repo, but we have a WIP repo with python bindings. Best would be to sync with @jonathf over there.

[jonathf]

Hello @hklarner,

I'm going to assume that the idea here is to do validation against the JSON input and output.

Specially for input, that would be quite useful, as it can give much better client side validation even if the execution is server side.

Not that it is a huge jub, but I haven't gotten around to porting JSON I/O to the Python interface. But when I do, adding something like this is a good idea.

I'll keep this issue open for when I get to to that task.

Related to: #5.

[hklarner]

Hi @jonathf
yes, I use the input model for validation but I also use them for code completion in the IDE and as a schema. I am testing VROOM as a service inside an existing project so I have to generate the vroom input and also convert the vroom output into my own output.

[hklarner]

Eventually, of course, I can use pyvroom instead of having vroom-express running ..

[jonathf]

Yeah, sounds like you would benefit a lot from the python interface.