edadma / b-tree

B+ Tree implementation in Scala

GitHub

b-tree

b-tree is a B+ Tree implementation in Scala. b-tree is designed to be both generic (type parameters for keys and values, and an abstract type for node references) and general (doesn't care how the tree is stored). An extending class needs to implement a number of simple methods and node type that provide storage independence.

Scaladoc library documentation can be found at http://edadma.github.io/b-tree.

As an example, the following code

import xyz.hyperreal.btree.{MemoryBPlusTree, FileBPlusTree}


object Example1 extends App {
  val memoryTree = new MemoryBPlusTree[String, Any]( 5 )
  
  memoryTree.insertKeys( "k", "z", "p", "d", "b", "v", "h", "x", "o", "y", "c", "t", "j", "n", "f", "l", "s", "q", "i", "m", "e", "u", "w", "a", "g", "r" )
  memoryTree.diagram( "memoryTree" )
  
  val fileTree = new FileBPlusTree[String, Any]( "btree", 5 )
  
  fileTree.insertKeys( "k", "z", "p", "d", "b", "v", "h", "x", "o", "y", "c", "t", "j", "n", "f", "l", "s", "q", "i", "m", "e", "u", "w", "a", "g", "r" )
  fileTree.diagram( "fileTree" )
  fileTree.close

  println( memoryTree.boundedKeysIterator(('>, "c"), ('<, "l")).mkString(", ") )
  println( memoryTree.reverseKeysIterator.mkString(", ") )
}

produces identical B+ Tree structures in memory and on disk, and therefore the generated image files memoryTree and fileTree will be identical. The diagram that is produced is

tree

The call to boundedKeysIterator above produces an iterator over the keys

d, e, f, g, h, i, j, k

since these are the ones that are strictly between c and l. The call to reverseKeysIterator produces an iterator over the keys

z, y, x, w, v, u, t, s, r, q, p, o, n, m, l, k, j, i, h, g, f, e, d, c, b, a

Adding b-tree to a project

There are two ways to add b-tree to a project. If you just want either an in-memory B+ tree "map" or file based B+ tree persistant storage for your project, then the concrete classes MemoryBPlusTree and FileBPlusTree work out-of-the-box. On the other hand, if you want to integrate the B+ tree algorithms into your project, then you can extend the abstract class BPlusTree by implementing it's storage abstraction methods to hook it into your project.

Extending BPlusTree

b-tree has one abstract class called BPlusTree which provides for interaction with a B+ tree. The concrete methods operate through a storage abstraction layer composed of 1 abstract type (the node type), 4 variables, and 34 methods. To use b-tree in a project, it is necessary to extend BPlusTree and provide implementations for all the abstract members. 34 methods sounds like a lot but they are all easy to implement. A lot of effort was put into coding BPlusTree so that it's abstraction layer would be dead simple to implement.

There are two example implementations provided: MemoryBPlusTree and FileBPlusTree. MemoryBPlusTree is just a normal in-memory data structure where the node type is a simple class. FileBPlusTree provides for persistant storage of key/value pairs in disk files. To see how easy it can be to implement the abstraction layer just look at the source code for MemoryBPlusTree.

To begin working on adding b-tree to a project, visit the scaladoc library documentation. Next, click on BPlusTree and then select visibility "All" to see all the protected members. Everything under "Type Members" and "Abstract Value Members" must be implemented. All the information needed is in the member descriptions and in the class description at the top.

License

b-tree is distributed under the MIT License, meaning that you are free to use it in your free or proprietary software.

Usage

Use the following elements to use b-tree in your Maven project:

<repository>
	<id>hyperreal</id>
	<url>https://dl.bintray.com/edadma/maven</url>
</repository>

<dependency>
	<groupId>xyz.hyperreal</groupId>
	<artifactId>b-tree</artifactId>
	<version>0.3</version>
</dependency>

Add the following to your build.sbt file to use b-tree in your SBT project:

resolvers += "Hyperreal Repository" at "https://dl.bintray.com/edadma/maven"

libraryDependencies += "xyz.hyperreal" %% "b-tree" % "0.3"

Building

Requirements

  • SBT 0.13.13+
  • Java 8+
  • Scala 2.12.0+
  • graphviz, imageMagick (optionally, for tree diagrams)

Clone

git clone git://github.com/edadma/b-tree.git

Compile and Test

For compiling and testing, it's best to use the following command:

test-only -- -l slow

to exclude the "stress tests" that are tagged as "slow". Some stress testing is still done with this command. If you just type test, you may have to wait many minutes for all the tests to complete. The example above can then be executed with

test:runMain Example1