Question: How to deal with models referencing each other? about schematics HOT 5 CLOSED

@mgedmin

Hey man, I resolved it the following way:
https://schematics.readthedocs.io/en/latest/usage/validation.html#model-level-validation

You can write a method and create an instance of the model there and call validate().
For the example above I did it like this:

class Product(Model):
    vendors = ListType(BaseType)

    def validate_vendors(self, data: dict, value: list[VendorItem]):
        if value:
            for vendor_item in value:
                Vendor(vendor_item).validate()

class Vendor(Model):
    products = ListType(BaseType)

    def validate_products(self, data: dict, value: list[ProductItem]):
        if value:
            for product_item in value:
                Product(product_item).validate()

Keep an eye on the naming convention and the attributes. Otherwise the method won't be called correctly.
It is definitely not an ideal solution but it works for my purposes and maybe for yours to.
I use schematics for Item-Validation in scrapy and errors are displayed correctly in the statistics etc.

Hope this helps! 👍

from schematics.

I have the same question.

from schematics.

This seems less fragile than my hack, which was

class TreeNode(Model):
    children = ListType(ModelType(Model))  # should be ListType(ModelType(TreeNode))

# schematics doesn't support recursive models, so we monkey-patch that in 
TreeNode.children.field.model_class = TreeNode
TreeNode.children.field.fields = TreeNode.fields

from schematics.

Unsurprisingly my hack broke somewhere during the upgrade to schematics 2.x (AttributeError: property 'model_class' of 'ModelType' object has no setter).

from schematics.

Actually schematics 2.0 supports self-recursive models (#335), so no workaround should be necessary.

from schematics.