tmtaybah / compression_service

Compression Service

A multi-threaded client-server implementation that provides a simple API to compress and decompress a string using the Huffman Coding algorithm.

Description (High Level)

• server creates a socket and a thread for each client that connects to it, it then launches a thread which effectively receives the data from the client and hands it off to service_request to deal with appropriately.
• service_request creates a header object that contains magic value, payload length, and request/status code. Header can also serialize and deserialize its contents (this streamlines receiving and responding to requests).
• based on the request code in the header object service_request calls the relevant api function.
• finally service_request keeps track of some statistics by incrementing bytes received when it's first called with data and bytes sent before sending the response that results from calling the api functions, all variables that keep track of statistics are shared variables across multiple threads and so are locked before being read or written to.
• service_request's response is sent to the client over the TCP socket

Target Platform & Language

• MacOS 10.13
• also tested on Ubuntu 16.04
• python3

Assumptions

• assumes compress takes a lowercase string with no whitespace
• stats is across multiple clients (e.g. for the entire service)
• get stats request returns service statistics at the time of request and does not reflect the bytes used to send get stats response but it does reflect the bytes received in with get stats request
• any prints are mainly for logging purposes
• header values are interpreted as signed integers and statistical values are interpreted as unsigned

Third Party Libraries

• pytest : simplifies testing

Instructions

Building

./build.sh

Running

./run.sh

Testing

Unit test (pytest)

make test

Interactive testing

on two terminal sessions:

1. ./run.sh
2. make run_client

things to do:

• fix max_payload issue

• remove all magic value

• add decompress request

• fix performance measurements

• add logs instead of prints