Messenger API

A simple messenger API written in Python with Flask.

Getting Started

Dependencies

• Linux or Mac OS
• Python 3.9.X
  • See requirements.txt for python dependencies
• SQLite 3.36.X
• jq (optional)
  • Makes it easy to pretty print when testing with cURL
• Docker (optional)
  • Only necessary if the are issues with any of the above dependencies

Running

As Script

First install the dependencies. Note that the following commands assume Arch Linux.

$ pacman -S sqlite3 python3 python-pip jq
$ pip install -r requirements.txt

$ ./messenger_api.py
Launching Messenger API
 * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

Within Docker

I have included a simple Dockerfile which can be be used to make it easier to run the project without installing the dependencies.

$ docker run --network host -e 127.0.0.1 -it $(docker build -q .)
Launching Messenger API
 * Running on http://127.0.0.1:5000/ (Press CTRL+C to quit)

Testing

As Script

If you have all of the dependencies installed, tests are run as a script.

$ ./tests.py
.............
----------------------------------------------------------------------
Ran 13 tests in 0.142s

OK

Within Docker

Alternatively, the tests can also be run within a docker container to manage the dependencies.

$ docker run -it $(docker build -q .) ./tests.py
.............
----------------------------------------------------------------------
Ran 13 tests in 3.431s

OK

Notes

Interpretation of Prompt

...with a limit of 100 messages or all messages in last 30 days.

I took this very literally to say that there should be two filtering options that the user may choose between (all and limit) when requesting messages from the server. Messages older than 30 days are not returned regardless of filter.

The default limit is 100 if no filter is specified in the request.

Data Older Than 30 Days

While the API limits the retrieved messages to the past 30 days, there is not yet a background process which purges older messages from the system. Currently, this data is simply inaccessible.

Group Messaging

I have gone slightly beyond the minimum requirements of the project. Instead of simply supporting messaging between two distinct users, I have added support for group messaging. Groups are defined a list of the usernames within the group. It would be the responsibility of the frontend to recognize that a message is a group message and display it appropriately. To make it easier for the frontend to determine which group a message should fall into, I have guaranteed that usernames are always reported in alphabetical order when returned in the recipients array.

In retrospect, the addition of multiple recipients made it harder to filter the messages when retrieving them from the database. I was able to use the LIKE command to search for a specific user, but this command is slow, especially on large data sets. A more optimal SQL query would have been possible if I had used named groups instead.

Message IDs

I exposed message IDs in the API so that future APIs could be added which allow messages to be edited or deleted.

Next Steps

• Deploy with WSGI server such as Gunicorn instead of the Flask dev server.
• Migrate database from SQLite to PostgreSQL for more scalability.
• Integrate websockets so that messages are pushed to recipients immediately instead of requiring the frontend to poll for new messages.

Acknowledgments

Invaluable Guide for using Flask with SQLite

• Flask Sqlite3 Pattern

Helpful Blog Post on Unit Testing Flask with SQLite

• In-memory SQLite database and Flask: a threading trap

Code Snippet to Hide Flask Development Server Warning

• Prevent Flask Production Env Warning

API Design and Documentation Inspiration

• Lithic

Example Data Inspiration

• Teenage Mutant Ninja Turtles Quotes