Ignore empty string ("") filters in Django Ninja FilterSchema like django-filter

[blasferna]

With django-filter, filters with empty strings (e.g. ?name=&code=) are ignored and all records are returned.

In Django Ninja, using FilterSchema like:

from ninja import FilterSchema, Field
from typing import Optional

class ProductListFilters(FilterSchema):
    category_id: int
    q: Optional[str] = Field(
        None,
        q=["name__icontains", "code__icontains"]
    )
    name: Optional[str] = Field(
        None,
        q="name__icontains"
    )
    status: Optional[str] = None

If I query:

?q=&category_id=1&name=&status=

it does not ignore empty string filters (name= or status=), and the results are not as expected.
I have to omit these params to get all records.

Is there a way to make Django Ninja treat empty strings as None and ignore those filters automatically, like django-filter?
Any built-in option or best practice for this?

[vitalik]

@blasferna you can use model_validator:

from ninja import FilterSchema, Field, Query
from typing import Optional
from pydantic import model_validator


class ProductListFilters(FilterSchema):
    category_id: int
    q: Optional[str] = Field(None, q=["name__icontains", "code__icontains"])
    name: Optional[str] = Field(None, q="name__icontains")
    status: Optional[str] = None

    @model_validator(mode="before")
    def remove_empty_strings(cls, values):
        # changing empty strings to None
        for f in cls.__pydantic_fields__:
            if getattr(values, f, None) == "":
                setattr(values, f, None)
        return values


@api.get("/filters")
def test_filters(request, filters: Query[ProductListFilters]):
    return filters.dict(exclude_none=True)

[blasferna]

Thanks @vitalik, that solution works perfectly!

To make this reusable across different filters, I created a custom base schema. This way, all filter schemas can inherit this behavior:

from ninja import FilterSchema as BaseFilterSchema
from pydantic import model_validator

class FilterSchema(BaseFilterSchema):
    @model_validator(mode="before")
    def remove_empty_strings(cls, values):
        # Convert empty string fields to None
        for f in cls.__pydantic_fields__:
            if getattr(values, f, None) == "":
                setattr(values, f, None)
        return values

class ProductListFilters(FilterSchema):
    category_id: int
    q: Optional[str] = Field(None, q=["name__icontains", "code__icontains"])
    name: Optional[str] = Field(None, q="name__icontains")
    status: Optional[str] = None