Please write a program stream-sampler that receives and processes an input stream consisting of
single characters. The stream is of unknown and possibly very large length and the program should
work regardless of the size of the input. The program should take the sample size as a parameter
and generate a random representative sample using that many characters.
As for receiving the data the tool should work with two kinds of inputs:
- Values piped directly into the process using cat input.txt | stream-sampler 5
- values generated by using a random source from within the language
Clean and maintainable code, efficiency, memory consumption, and good testing are all important
criteria we would like you to keep in mind when completing the challenge. Try to keep the maximum
allocated memory as low as feasible.
Should something about the challenge be unclear please make an assumption and document it.
Example
Given a sample size of 5 and the following stream of characters as values: THEQUICKBROWNFOXJUMPSOVERTHELAZYDOG
A possible random sample could be: EMETN
Example execution and output for piped processing:
$ dd if=/dev/urandom count=100 bs=1MB | base64 | stream-sampler 5
Random Sample: EMETN
Two solutions have been implemented using Java programming language.
1. RESERVOIR SAMPLING
The random sample is generated from the stream using Reservoir Sampling algorithm designed by Knuth in 1997.
The sampling is done using Java 8 Collector implementation.
2. MINHEAP SAMPLING
The random sample is generated from the stream using MinHeap datastructure.