This is a micro framework for building HTTP APIs on a high level of abstraction on top of aiohttp. It allows to create jsonapi-like HTTP interface to models in declarative way and leaves it possible to fine tune at any level.

Install the package

pip install aiohttp_baseapi

Creates project directory structure in current directory (it also contains two sample apps)

baseapi-start-project

Install the dependencies

cd src/
pip install -r .meta/packages

Configure Postgres DB connection

echo "DATABASE = {'host':'localhost','port':5432,'database':'test','user':'test','password':'test','minsize':1,'maxsize':10}" > ./settings_local.py

Create migrations

alembic revision --autogenerate -m "init"

Run migrations

alembic upgrade head

Run the application

python ./main.py --port=9000

That's it! You can try the API. It contains "default" app - it just prints all existing API methods under "/" location. And "demo" app - it contains two sample models.

http GET :9000
http POST :9000/authors <<< '{"data":{"name":"John", "surname": "Smith"}}'
http POST :9000/books <<< '{"data":{"category": "Fiction","name":"Birthday","is_available":true, "author_id": 1}}'
http GET ":9000/books?filter[name]=Birthday&include=authors"
http GET :9000/books/1
http PUT :9000/books/1 <<< '{"data":{"is_available":false}}'
http DELETE :9000/books/1

There are some built-in features you can use:

• filtration
• sorting
• fields selection
• pagination
• inclusion
• validation

Request and response formats are inspired by jsonapi.org, but have some differences.

Retrieved information can be filtered using filter GET-parameter. You can filter by multiple fields: filter[fieldname1]=foo&filter[fieldname2]=bar. In this case in response there are values for which the fieldname1 equals to foo and fieldname2 equals to bar. Also you can filter enumerating desired values: filter[fieldname]=foo,bar. In this case in response there are values for which the fieldname equals to foo or bar.

Retrieved information can be sorted using sort GET-parameter. Also you can sort enumerating multiple values: sort=fieldname1,-fieldname2. In this case values in response will be sorted by fieldname1 ascending, and then by fieldname2 descending.

Using parameter fields you can retrieve only those fields which you need. For example: fields=fieldname1,fieldname2. In this case values in response will have only fieldname1 and fieldname2 fields.

Pagination (limit and offset) can be performed using page parameter. Usage: page[limit]=10&page[offset]=20 - standard pagination (20 items skipped, maximum 10 returned).

Also there is possibility to attach related entities using parameter include. One can apply described above features (filtration, sorting, etc.) to included entities. It will affect only included entities. Examples: filter[entity.fieldname]=foo - filtration; sort[entity]=fieldname - sorting; fields[entity]=fieldname - choosing fields; page[entity.limit]=10 - pagination.

The data passed in modifying requests (POST, PUT, etc.) can be validated using json-schema (which can be auto-generated from model description) or manually. Default data provider is database, but you can use anything you wish.

Run:

$ pip install -r .meta/packages_unit
$ cd src
$ make unit-test