.. image:: https://img.shields.io/pypi/v/flask-aggregator.svg
Batch the GET requests to your API into a single POST. Save requests latency and reduce REST chatiness.
I was inspired by this article <http://tech.3scale.net/2013/04/18/accelerate-your-mobile-api-with-nginx-and-lua/>_
from 3scale, and by their NGINX aggregator <https://github.com/solso/api-aggregator>_ - but I wanted something simpler.
Flask-Aggregator adds an endpoint to your Flask application that handles multiple GET requests in a single POST, and returns the response of each GET request in a JSON stream.
It means that instead of sending multiple GET requests:
.. code-block:: sh
-> GET /route1 <- answer1 -> GET /route2 <- answer2 -> GET /route3 <- answer3
You can now just send a single POST that aggregates them all:
.. code-block:: sh
-> POST /aggregate ["/route1", "/route2", "/route3"] <- { "/route1": answer1, <- "/route2": answer2, <- "/route3": answer3 }
Mobile networks.
.. code-block:: sh
$ pip install flask-aggregator
.. code-block:: python
from flask import Flask from flask_aggregator import Aggregator
app = Flask(name) Aggregator(app=app, endpoint="/batch")
.. code-block:: sh
$ python example.py
[go to another shell]
$ curl -H "Content-type: application/json" -X POST 127.0.0.1:5000/batch
--data-raw '["/hello/world", "/hello/ramnes?question=Sup?"]'
{
"/hello/world": "Hello, world!",
"/hello/ramnes?question=Sup?": "Hello, ramnes! Sup?"
}
Not really.
As of today, Flask-Aggregator executes the aggregated requests in a synchronous manner, which makes it only useful if latency is a real issue and response time is not, and that more than N requests are sent at the same time, where N is maximum number of concurrent requests on user's client.
Also, it has limitations such has:
- no automatic caching mechanism browser-side, since it uses a POST request
- no header support at all for now, which means no cookie, etag, or whatever
- no other HTTP verb than GET is supported for now
Last but not least, chances are high that a lot of corner cases are not handled.
MIT
