Launch the app

1. Download .Net Core runtime and SDK here
2. Add your tsv file and specify the path in the appsettings.json
3. Build and publish the solution: dotnet publish -r [Runtime] -c Release and target the runtime you want (list here).
4. Launch the solution: ./bin/Release/netcoreapp2.2/[Runtime]/Algolia.exe.
5. Make queries on your browser or preferred tools on http://localhost:5000 and/or https://localhost:5001

Design

The algorithm is simple. We first read the tsv file and create a list, sorted by datetime and stored in memory. When we receive a call on the count or the popular API, we just look for the first date that is inside the datePrefix and iterate over the list from here until we leave the datePrefix.

For count, we add elements to a structure that guarantees uniqueness. We returns the size of this structure at the end of the iteration.

For popular, we use a dictionary to count the number of each queries and return the top size at the end of the iteration.

Both of those API run in O(n), n being the number of queries done during the specified date range.

Improvement

This works well because the input data fits in memory. For bigger input, we could consider some kind of MapReduce algorithm, or probabilistic algorithm if an approximation is good enough (like HyperLogLog for example).