Primer: pydantic

Most classes in flodym are based on the pydantic library, which provides safety in data handling. More specifically, most classes in flodym are pydanctic.BaseModels.

Their main property is that the arguments passed to the class initialization are replaced by so-called fields. While a normal class would be defined and initialized like this:

class Person:
    def __init__(self, name):
        self.name = name


person_a = Person("Adam")
print(person_a.name)

Adam

this would look slightly different for pydantic, where the __init__ function is pre-implemented:

from pydantic import BaseModel


class Person(BaseModel):
    name: str


person_a = Person(name="Adam")
print(person_a.name)

Adam

There are two major differences to normal classes:

• The input is type-checked. In the example, name has to be a string, and e.g. initializing Person like this: Person(name=2) will lead to an error.

• Normal classes do not guarantee that parameters passed to init end up as class attributes. So the following will lead to an error:

class Person:
    def __init__(self, first_name: str, last_name: str):
        self.initials = f"{first_name[0]}.{last_name[0]}."


person_a = Person(first_name="Adam", last_name="Smith")

try:
    print(person_a.first_name)
except AttributeError as e:
    print(f"AttributeError: {e}")

AttributeError: 'Person' object has no attribute 'first_name'

In pydantic, what you put in at initialization is always what you get as attributes.

This creates transparency.

But on the other hand, it can be more convenient to have some operations done on initialization.

Therefore, flodym sometimes offers class methods to create objects from different arguments, like for the example above:

class Person(BaseModel):

    initials: str

    @classmethod
    def from_name(cls, first_name: str, last_name: str):
        initials = f"{first_name[0]}.{last_name[0]}."
        return cls(initials=initials)


person_a = Person.from_name(first_name="Adam", last_name="Smith")
print(person_a.initials)

A.S.

For further information on pydantic, please refer to the pydantic documentation.