Project

rb_heap

0.05
Low commit activity in last 3 years
No release in over a year
Light-weight priority queue implementation using a heap
2005
2006
2007
2008
2009
2010
2011
2012
2013
2014
2015
2016
2017
2018
2019
2020
2021
2022
2023
2024
 Dependencies

Development

~> 1.13
~> 10.0
~> 3.0
 Project Readme

Heap

This is an heap implementation that can be used for priority queues. Internally it represents the heap as an array. It also contains an in-place heapsort implementation.

gem install rb_heap

heap = Heap.new(:<)

# << and .add are aliases
heap << 3 << 1 << 2
heap.add(4)

heap.peak # => 4
heap.pop # => 4
heap.pop # => 3
heap.pop # => 2
heap.pop # => 1

Different kind of heaps

The default heap is a min heap, but you can also specifiy it explicitely.

minHeap = Heap.new
anotherMinHeap = Heap.new(:<)

Creating a max heap is straight-forward:

maxHeap = Heap.new(:>)

You can also pass a custom comparison function using a block:

minAbsHeap = Heap.new{|a, b| a.abs < b.abs}
minAbsHeap << 3 << 1 << -2
minAbsHeap.pop # => 1
minAbsHeap.pop # => -2
minAbsHeap.pop # => 3

You can use this comparison function to compare specific fields:

Person = Struct.new(:name, :age)

ageHeap = Heap.new{|a, b| a.age < b.age}
ageHeap << Person.new("Richard Hendricks", 26)
ageHeap << Person.new("Erlich Bachman", 32)
ageHeap << Person.new("Dinesh Chugtai", 30)

ageHeap.pop.name # => Richard Hendricks
ageHeap.pop.name # => Dinesh Chugtai
ageHeap.pop.name # => Erlich Bachman

pop (O(log n)) and peak (O(1))

pop removes the top element of the heap and returns it. peak returns the top element but doesn't change the heap itself.

When the heap doesn't have any elements, nil is returned. You can also explicitly check if the heap still has elements using the empty? method.

add (O(log n))

You can add elements to the heap using .add. As you already saw << is aliased to .add.

replace (O(log n))

You can replace the top element with a new one using this method. Of course you could also pop and add but that would cause two rebalance operations. .replace is optimised to only take one.

heap = Heap.new
heap << 1 << 2 << 3
heap.replace(4)

heap.pop # => 2
heap.pop # => 3
heap.pop # => 4

offer (O(log n))

A lot of times you only want to replace the top element if it's smaller/larger than the potentially new element.

heap = Heap.new
heap << 1

heap.offer(0) # It's a min heap and 1 > 0, so it's not replaced
heap.peak # => 1

heap.offer(2) # 1 < 2, so it's replaced
heap.peak # => 2

This method assumes that there's already a top element, so you have to check yourself that you don't call it on an empty heap.

A potential use case would be finding the largest k numbers in a stream of incoming numbers, without actually storing all numbers.

Other utilities

size

Returns the amount of elements that are in the heap.

empty?

Returns a boolean that indicates whether the heap has any elements.

to_a

Gives you all the heap contents as an array, but you shouldn't make any assumptions about the order of the elements.

Sorting

Heapsort is pretty light-weight to implement once you have a working heap. For this reason this library also comes with an in-place heapsort function.

Heap.sort([3,1,2]) # => [1, 2, 3]
Heap.sort([3,1,2], :>) # => [3, 2, 1]
Heap.sort([3,1,-2]){|a, b| a.abs < b.abs} # => [1, -2, 3]

Developing

You can run the unit tests using rake.

Future ideas

Ideas for functions that could be added:

  • Create a heap from an array. This is possible in O(n). Right now one could repeatly add all the array elements into a new heap, but that'd be O(n log n)
  • Merging two heaps

I guess it would also be fun to implement different kind of heaps, but maybe that's beyond the scope if this library.