fjfnaranjo / fmtasks

Flask+MongoDB task storage service

Flask+MongoDB task storage service

This is a simple code test for a selection process.

It is a RESTful API implementing a generic document storage service. It considers the documents "tasks" by convention (to comply with the spec).

It uses Flask for the HTTP service and MongoDB for the storage.

It uses the Flask development server to expose itself in the port 5000 (more notes below about this).

The app uses "tasks" for the MongoDB collection name and "fmtasks" for the database name.

The only configuration implemented is a enviroment variable to change the default MongoDB URI: MONGODB_URI (also explained below).

Design notes

This application development did some concessions in the design scope:

1. Docker layering: The Docker container in this implementation is designed to be the final Docker layer. If I was planning to use this application in a production environment, I would not include the ENTRYPOINT, the run configuration or the EXPOSE. The application itself could be the layer just before the last one. And the last one would be a nginx server or some other wsgi server. The reason to simplify the design was that I know is a code test and it seemed a unnecessary inconvenience to deal with for the test evaluator. Also, I would have implemented the Flask app itself as a Python package.

2. REST semantics: On a real project, a REST library can be used to implement the more diverse semantics in HTTP interaction with the service. I have stick to simple methods to avoid dealing with the header management required for a more strict RESTful API. Also, the HTTP response codes are over simplified sometimes.

3. Avoidance of existing implementations: There are libraries implementing the exact behavior required for this code test, like Eve Framework. They even manage the interaction with the database, listings, responses, etc. In a normal productive environment, Eve by itself can do everything here, much more, and better. It only needs some models definitions.

4. Configuration: The app only needs one configuration parameter, the MongoDB URI, and its optional. I did a quick implementation using environment variables but a proper option parser would be advisable in a real project.

API

• Path: The API exposes the documents and operations under "/task/". When necessary, the MongoDB hexadecimal id is appended to the path to signify a particular document.

  • HTTP Methods:
  • POST for the "create" operation.
  • GET for the "read" operation.
  • PUT for the "update" operation.
  • DELETE for the "delete" operation.

• Data encoding: The API uses application/json for the Content-Type both for sending and for receiving. The content for the documents is specified in a simple json structure. When necesary, the id of the objects is specified in the path.

  {'content': 'Sample content.'}

• Sample requests: I implemented the application using a TDD approach. You can find sample requests in the application test suite.

Docker

The code test specs required the application to be 'dockerized'.

You can find both Docker and Docker Compose definition files. Run the following command in the repository root folder to deploy the application using Docker Compose:

$ docker-compose up

Alternatively, if you do not have Docker Compose installed, you can run the app providing a MongoDB connection string as an environment variable:

$ docker build -t fmtasks .
$ docker run --env MONGODB_URI=mongodb://example.com -p 5000:5000 fmtasks

Change "mongodb://example.com" for the appropriate URI.

I used "fmtasks" for the image name in the command but you can change it (in both lines).