Skip to content
Adaptive Radix Trees implemented in Go
Go Makefile
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
examples
test/assets
.gitignore
.travis.yml
LICENSE
README.md
api.go
consts.go
doc.go
factory.go
go.mod
go.sum
makefile
node.go
node_test.go
tree.go
tree_dump.go
tree_dump_test.go
tree_test.go
tree_traversal.go
utils.go

README.md

An Adaptive Radix Tree Implementation in Go

Build Status Coverage Status Go Report Card GoDoc

This library provides a Go implementation of the Adaptive Radix Tree (ART).

Features:

  • Lookup performance surpasses highly tuned alternatives
  • Support for highly efficient insertions and deletions
  • Space efficient
  • Performance is comparable to hash tables
  • Maintains the data in sorted order, which enables additional operations like range scan and prefix lookup
  • O(k) search/insert/delete operations, where k is the length of the key
  • Minimum / Maximum value lookups
  • Ordered iteration
  • Prefix-based iteration
  • Support for keys with null bytes, any byte array could be a key

Usage

package main

import (
    "fmt"
    "github.com/plar/go-adaptive-radix-tree"
)

func main() {

    tree := art.New()

    tree.Insert(art.Key("Hi, I'm Key"), "Nice to meet you, I'm Value")
    value, found := tree.Search(art.Key("Hi, I'm Key"))
    if found {
        fmt.Printf("Search value=%v\n", value)
    }

    tree.ForEach(func(node art.Node) bool {
        fmt.Printf("Callback value=%v\n", node.Value())
        return true
    })

    for it := tree.Iterator(); it.HasNext(); {
        value, _ := it.Next()
        fmt.Printf("Iterator value=%v\n", value.Value())
    }
}

// Output:
// Search value=Nice to meet you, I'm Value
// Callback value=Nice to meet you, I'm Value
// Iterator value=Nice to meet you, I'm Value

Documentation

Check out the documentation on godoc.org

Performance

plar/go-adaptive-radix-tree outperforms kellydunn/go-art by avoiding memory allocations during search operations. It also provides prefix based iteration over the tree.

Benchmarks were performed on datasets extracted from different projects:

  • The "Words" dataset contains a list of 235,886 english words. [2]
  • The "UUIDs" dataset contains 100,000 uuids. [2]
  • The "HSK Words" dataset contains 4,995 words. [4]
go-adaptive-radix-tree # Average time Bytes per operation Allocs per operation
Tree Insert Words 10 146,326,506 ns/op 41,299,883 B/op 1,326,167 allocs/op
Tree Search Words 30 44,075,933 ns/op 0 B/op 0 allocs/op
Tree Insert UUIDs 20 86,062,471 ns/op 19,638,904 B/op 547,648 allocs/op
Tree Search UUIDs 50 35,808,749 ns/op 0 B/op 0 allocs/op
go-art
Tree Insert Words 5 272,047,975 ns/op 81,628,987 B/op 2,547,316 allocs/op
Tree Search Words 10 129,011,177 ns/op 13,272,278 B/op 1,659,033 allocs/op
Tree Insert UUIDs 10 140,309,246 ns/op 33,678,160 B/op 874,561 allocs/op
Tree Search UUIDs 20 82,120,943 ns/op 3,883,131 B/op 485,391 allocs/op

To see more benchmarks just run

$ make benchmark

References

[1] The Adaptive Radix Tree: ARTful Indexing for Main-Memory Databases (Specification)

[2] C99 implementation of the Adaptive Radix Tree

[3] Another Adaptive Radix Tree implementation in Go

[4] HSK Words. HSK(Hanyu Shuiping Kaoshi) - Standardized test of Standard Mandarin Chinese proficiency.

You can’t perform that action at this time.