Backend API Hiring Challenge

Introduction

Datapane is an API-driven product that provides client libraries / commandline applications that talk to an API server to handle and process datasets.

This challenge involves building a small API server and corresponding command-line client, both in Python, that allow uploading and processing CSV files, such as those included in the repo.

Task

For this task we'll be building a very simple Python 3-based REST API server and command-line client.

You may use any Python libraries and technologies of your choice, for instance FastAPI, etc, to build the API server.

API Server

The API server should have a single root endpoint, /dataset/, that allows list and CRUD operations over a dataset object via the following API / HTTP verbs:

• GET /datasets/ - list the uploaded datasets
• POST /datasets/ - creates a dataset. This endpoint takes a CSV file as input, and stores it somewhere/how on the server as a pandas dataframe. A reference id to this created object is returned by the endpoint.
• GET /datasets/<id>/ - return the file name, and size of the dataset object
• DELETE /datasets/<id>/ - delete the dataset object
• GET /datasets/<id>/excel/ - export the dataset as an excel file
• GET /datasets/<id>/stats/ - return the the stats generated by running df.describe() on the pandas dataframe as a json object
• GET /datasets/<id>/plot/ - generate and return a PDF containing a list of histograms of all the numerical columns in the dataset

Client Library

The client app should be a fully standalone command-line python application, that is easily installable and runnable. The app should provide command-line arguments that correspond and support each of the API actions above - how you structure the command line arguments and what you call them is left up to yourselves.

• list - list the uploaded datasets
• clear - delete all the datasets
• create name-of-dataset.csv - creates a dataset. This endpoint takes a CSV file as input, and stores it somewhere/how on the server as a pandas dataframe. A reference id to this created object is returned by the endpoint.
• delete name-of-dataset.csv - delete the dataset object
• info name-of-dataset.csv - return the file name, and size of the dataset object

Technologies

• A Python 3 framework that supports generating JSON APIs (FastAPI)
• Build systems, tools, and scripts of your choice, e.g. poetry, setup.py, docker, etc.
• Any libraries you may find useful to help your task, we prioritise using existing libraries to accomplish tasks rather than building in-house

Requirements

• Multi-user support and log-in is NOT required for this project
• Instructions should be provided on how to build / bundle / start the system
• You should aim to use the latest Python language features, ecosystem, tooling, and libraries where possible

Optional Requirements

• As CSVs can be untrusted, you should consider running the CSV importing within a container / sandbox

Instructions

• Start the server

  • cd name-of-server-dir
  • uvicorn server:app

• Start the client

  • cd name-of-client-dir
  • python client.py

• Docs

  • 'localhost:8000/docs' for the server