Skip to content

Multi-model pattern

Documents with different schemas could be stored in a single collection and managed correctly. UnionDoc class is used for this.

It supports find and aggregate methods. For find, it will fetch all the found documents into the respective Document classes.

Documents with union_doc in their settings can still be used in find and other queries. Queries of one such class will not see the data of others.

Example

Create documents:

from beanie import Document, UnionDoc


class Parent(UnionDoc):  # Union
    class Settings:
        name = "union_doc_collection"  # Collection name
        class_id = "_class_id"  # _class_id is the default beanie internal field used to filter children Documents


class One(Document):
    int_field: int = 0
    shared: int = 0        

    class Settings:
        name = "One" # Name used to filer union document 'One', default to class name
        union_doc = Parent


class Two(Document):
    str_field: str = "test"
    shared: int = 0

    class Settings:
        union_doc = Parent

The schemas could be incompatible.

Insert a document

await One().insert()
await One().insert()
await One().insert()

await Two().insert()

Find all the documents of the first type:

docs = await One.all().to_list()
print(len(docs))

>> 3 # It found only documents of class One

Of the second type:

docs = await Two.all().to_list()
print(len(docs))

>> 1 # It found only documents of class Two

Of both:

docs = await Parent.all().to_list()
print(len(docs))

>> 4 # instances of both classes will be in the output here

Aggregations will work separately for these two document classes too.