Outside of the readme, there are two main files: 1. NAT.h 2. NAT.cpp (main function is located here as well). -------------------------------------------------------------------------------------- PREREQUISITES FOR RUN INSTRUCTIONS: 1. Visual Studio with support for C++ (can be downloaded here https://www.visualstudio.com/vs/cplusplus/). I tested it with Visual Studio 2017. -------------------------------------------------------------------------------------- HOW TO RUN (ON VISUAL STUDIO 2017): 1. Once you've downloaded the files, simply click on NAT.sln. 2. Turn off precomplied headers (you can find instructions here: https://msdn.microsoft.com/en-us/library/1hy7a92h.aspx You need to make sure that you select the correct configuration [release versus debug]). You can view the screenshot 'Project Properties' at the root directory on the master branch of the git repo. Note that if you are using a 32 bit machine, select win32. 2. I've included all the required Visual studio files in the git repo. You can simply make the project by going to Build --> Build Solution 3. The main function is located under NAT.cpp. Once you run the program , you will be prompted to enter the NAT file and then the FLOW file. I have set two sample files in the repo (the ones in the email, NAT.txt and FLOW.txt). The outputs will be written to output.txt under ~/NAT/Output.txt. TO run, click on the green arrow button on the top panel. If you need me to give you a linux instructions/version, please let me know. -------------------------------------------------------------------------------------- IMPORTANT NOTE ON TIME: I did not include reading about NAT, writing the Readme and adding files to the git repo in my time slot. -------------------------------------------------------------------------------------- A: ASSUMPTIONS 1. ONLY NAT file's first pairs can have a *. E.g. Valid Entry: *:8080, 192.168.0.1:80 Invalid Entry: 10.0.1.1:8080: *:80 2. There is only one mapping from Flow to NAT. In other words, a FLOW only has one match in the NAT map. 3. In the NAT flow, if an entry has an asteric, for example *:21,1.2.3.4:12, then there will be no other entry with *:21 as the first IP:PORT to another IP:PORT (e.g. *:21,1.2.4.5:15 will be invalid 4. The NAT and FLOW file are all proper and well-formed, with no duplicates (as in step 2). All IPs are of the form W.X.Y.Z where W, X, Y and Z are integers(e.g. 192) This assumption was set for simplicitiy. -------------------------------------------------------------------------------------- B: HOW I TESTED MY CODE To be honest tested for NAT pair with *:port--> pair , ip:*-->pair and pair-->pair. I didn't have enough time to test for edge cases as I would have ran over 90 minutes. I also tested against the sample provided. -------------------------------------------------------------------------------------- C: ANY INTERESTING CODING, DESIGN or ALGORITHMIC CHOICES ALGORITHMIC CHOICES: The first thing that came to my mind when I saw the * and pairs was a hashmap (unordered_map). However the challange was to deal with the wildcard *. To achieve the needed goal, I placed three maps. M1: One from a non-* pair to a pair M2: One from an ip- * (any port) to pair M3: One from a * (any ip) - port to pair For every line in the NAT file (if the line is not empty) If in the NAT flow, I encountered a non-* pair, I a map from the first pair to the second pair to M1. If in the NAT flow, the port was an asteric, then I added the IP , pair to M2. e.g. 10.11.12.13:*,40.100.20.7:3389 --> The I would map 10.11.12.13 to 40.100.20.7:3389 If in the NAT flow, the ip was an asteric, then I added the Port , pair to M3. e.g. *:21,1.2.3.4:12, then I would map 21 to 1.2.3.4:12. To write to the output, the FLOW file (given that there is only one match) was as follows: For every line (pair) in the FLOW file If there exists an entry of the pair in M1, then write to output Pair in Flow --> What it maps to in M1. If there exists an entry of the pair's ip in M2, then write to outpout Pair in flow --> what the IP maps to M2. If there exists an entry of the pair's port in M3, then write to outpout Pair in flow --> what the port maps to M2. The main benefit of using an unordered_map is that querying can be done in O(1). SOFTWARE ENGINEERING: Standard class/function definitions. Nothing too innovative here. Just separation of functions. -------------------------------------------------------------------------------------- D: KNOWN LIMITATIONS DUE TO TIME CONSTRAINT (a.k.a what I would've done better), HOW TO MAKE IT BETTER I tried to solve my limitations through simplyfing the problem (via assumptions). However, give my assumptions, I ran out of time to try the following: 1. Remove an entry for the maps once it has been written to the output (this needs to be properly tested and looking at edge cases with be a must). This approach would allow freeing up space. However, on the flip side, if there is more than one file that needs to be tested against the same NAT file, then this approach will not work. 2. scaling the problem to very large tables would most likely require some sort of a distributed unordered_map which I am hoping is out of the scope of this challenge. 3. I would say my main technical bottle neck was parsing the pairs (IPs and Ports). I took code from stackoverflow and incorporated into my code. I didn't look at how efficient they are, hence I would spend more time on this to make it more efficient.