When returning a list of objects, it is generally good practice to paginate the resource. This is where Blueprint.paginate steps in.

Pagination is more or less transparent to view function depending on the source of the data that is returned. Two modes are supported: pagination in view function and post-pagination.

Pagination in View Function

In this mode, Blueprint.paginate injects the pagination parameters into the view function as a PaginationParameters object passed as pagination_parameters keyword argument.

It is the responsability of the view function to return only selected elements.

The view function must also specify the total number of elements by setting it as item_count attribute of the PaginationParameters object.

class Pets(MethodView):
    @blp.response(200, PetSchema(many=True))
    def get(self, pagination_parameters):
        pagination_parameters.item_count = Pet.size
        return Pet.get_elements(


This is the mode to use when the data is returned as a lazy database cursor. The view function does not need to know the pagination parameters. It just returns the cursor.

This mode is also used if the view function returns the complete list at no extra cost and there is no interest in specifying the pagination parameters to avoid fetching unneeded data. For instance, if the whole list is already in memory.

This mode makes the view look nicer because everything happens in the decorator and the lazy cursor.

Cursor Pager

In this case, Blueprint.paginate must be passed a pager class to take care of the pagination. flask-smorest provides a pager for list-like objects: Page. For other types, a custom pager may have to be defined.

For instance, the following custom pager works with cursor classes that support slicing and provide a count method returning the total number of elements, such as SQLAlchemy’s Query.

from flask_smorest import Page

class CursorPage(Page):
    def item_count(self):
        return self.collection.count()

class Pets(MethodView):
    @blp.response(200, PetSchema(many=True))
    def get(self):
        return db.session.query(Pet)

Pagination Parameters

Once a view function is decorated with Blueprint.paginate, the client can request a specific range of data by passing query arguments:

GET /pets/?page=2&page_size=10

The view function gets default values for the pagination parameters, as well as a maximum value for page_size.

Those default values are defined as

DEFAULT_PAGINATION_PARAMETERS = {"page": 1, "page_size": 10, "max_page_size": 100}

They can be modified globally by overriding DEFAULT_PAGINATION_PARAMETERS class attribute of the Blueprint class or overridden in a specific view function by passing them as keyword arguments to Blueprint.paginate.

Pagination Header

When pagination is used, a 'X-Pagination' header is added to the response. It contains the pagination information.

# {
#     'total': 1000, 'total_pages': 200,
#     'page': 2, 'first_page': 1, 'last_page': 200,
#     'previous_page': 1, 'next_page': 3,
# }

The name of the header can be changed by overriding PAGINATION_HEADER_NAME class attribute of the Blueprint class. When setting this attribute to None, no pagination header is added to the response.