This is a solution for a coding challenge.
Due to limited time, the solution is only tested using assert statements. The conditions are the based on the boolean statements and sample data suggested in the instruction.
I utilized a Python dictionary instead of a multidimensional numpy (compiled in C) array to sacrifice super fast access time for decreased memory waste when the number of rules is small. The implementation for file parsing and packet rule matching is designed in a very functional manner, so that they can easily be called in parallel or concurrently if desired.
More thorough testing and actual concurrency/parallelism would be nice. Rules for frequently used ports or protocols can be cached in memory while others can be stored on disk, further reducing resource strain. The whole algorithm can be implemented in and compiled from C for further speedup.
This is one of the more interesting coding challenges I have had. Check out my other projects, one of which also has to do with high performance packet processing and cybersecurity. For Illumio, I am mostly interested in the data team. You can see my projects on github which extensively works with all kind of data management, processing, analysis, and visualization, using technologies such as MySQL, Spark, R, and mathematical models such as deep learning, time-series, and hypothesis testing.