Sportradar Code Challenge

A demonstration of an ETL pipeline based on NHL APIs.

Purpose

As a Data Analyst, I want to be able to quickly generate CSV files for a given hockey team or player. The data for these CSV files should come from the API at https://statsapi.web.nhl.com/api/v1/.

The documentation for this API can be found at https://github.com/sportradarus/sportradar-api-challenge.

Expected Behavior

1. Begin a console instance in Visual Studio.
2. User is prompted to select Team or Player.
3. User is prompted for an ID.
4. User is prompted for a year.
5. The application sends a request to the appropriate API endpoint.
6. The API returns a JSON response with the requested data.
7. The application parses the JSON and generates a CSV file with the data.
8. User is prompted for a directory path and a file path.
9. CSV file is saved to that path, if it is valid.

Instructions to run

1. Checkout Git repo in Visual Studio 2019.
2. Right-click the NHL_API project.
3. Choose Debug >> Start New Instance.

Highlights

• Resources/Enums/PipelineType.cs is an enum of pipeline types, and the code is set up to use switch statements in most places. This would make it easy to expand this application to support other pipeline types, such as a Tournament pipeline.
• Resources/JsonConverters provides a library of custom JSON converters. These are used in this project to serialize the API JSON responses into usable objects. This provides flexibility to handle complex JSON from multiple responses, including things like nested objects as in the case of SeasonJsonConverter, which has a nested serializer for Game entities.
• All required functions are neatly organized into their own Service classes. Not only does this make it easier to find a logical area of the code, it also keeps the main program loop clean and free of minutia. This helps future developers, who are maintaining the code, to quickly and intuitively understand what's happening in the code. It also has the benefit of eliminating code duplication, as these service functions can be re-used in many places.
• Comments. Comments are very near and dear to me, as I feel they break up the monotany of the code and help give a more human-readable flow to the logic. You'll find both function summaries and common comments everywhere in this project.
• Error handling. Try/Catch blocks to print helpful messages and allow the program to gracefully continue in the event of an error.

Areas to Improve

• If this were a real project for production, I would spend more time on the JSON Converters and their associated Models. They could be fully fleshed out to handle every facet of the API. But for the sake of time and sanity, I just went with the key entities.
• Testing. If this were going to production, I would absolutely write a test for every logical part of this project. But a lot of that is just tedious, and for the sake of this demo, the skills can be demonstrated with just a few tests.

Test Strategy

Two tests around the Team pipeline; one to hit the API and compare against some expected values, and another to test a non-existent Team ID to verify the proper exception is returned.

As mentioned above, I would do similar tests for the Player pipeline, but those would just be repitions of the same idea as the Team pipeline tests.

I could also write some tests for other components, such as the JSON converters and the CSV service.