You need pytest on your system to run the tests. There are no non-standard library dependencies, so you can simply pip3 install pytest or equivalent on your system, or set up a virtual environment and install the requirements.txt via pip3 install -r requirements.txt.

To run tests run pytest.

You can run the sample data by calling the script e.g. python3 main.py sample_data.txt, passing the file you want to process as the first positional parameter.

You can also run sample data from the web with a simple lambda interface at https://f55xysqwn6dfq3ry55d4jk52lu0mzrgr.lambda-url.us-east-1.on.aws/. The interface is all contained in lambda_handler and has a text area you can paste your data to process.

You can run this lamdba yourself by copying the entire main as the lambda function code with lambda_handler (the default name anyway) as the entry point. There are no external dependencies so packaging is not necessary.

The approach taken was to let the world contain the logic to run the simulation, represented by the class World. The robots (class Robot) act as a container for coordinates and direction. The world owns the interface to run the simulation, and tracks where robots fall off the grid. The World class is then wrapped to expose it to different interfaces such as a CLI and a small lambda based web interface.

After reviewing the project statement Python was chosen as the language of choice due to the following properties

• Personal familiarity. Because of the tight timeframe of the project we want to ensure strong familiarity with a language and its standard library. This means Java, Python, C#, and JS/TS were considered.
• In the specifications it can be deduced that the path of one robot can have an effect on subsequent robot if it is "lost" (goes out of bounds). This tells me that parllelism (something other languages might be stronger with due to the lack of the global interpreter lock) is not a priority.
• The project can be modeled in a small number of classes, with most of the code base being in the application logic rather than the class structure.
• There is file IO, and the opportunity for a small CLI. Both of these are easily achievable in Python whereas the overhead for such things in some other languages considered (Java, C#, JS) are more complex.
• We can provide a simple web interface by deploying this as a lambda URL. This is not a priority but a nice quality of life feature that is easily achievable with Python. In an actual client scenario this enables non-technical users easy access without a lot of effort or maintainence.