github.com/alediaferia/triego

Triego

Despite the name the internal of this data structure is implemented using a Radix Tree.

This project is one of the funding blocks of typeflow-webservice but, although mostly designed around it, it features a rather general purpose API around a Radix Tree.

Implementation considerations

The whole implementation has been designed with prefix iteration efficiency in mind. This implies that feeding the radix tree is way slower than a simple trie, but looking for words or prefixes is way faster due to the reduced amount of nodes to traverse. So, although insertion should be still MAX(O(N*Log(N))), the algorithm requires some additional time mostly due to memory allocations and/or resizing of existing nodes.

Basic usage

package main

import (
  "github.com/alediaferia/triego"
  "fmt"
)

func countTrieNodes(trie *triego.Trie, i *int) {
	if len(trie.Children) == 0 {
		*i = *i + 1
		return
	}
	for _, v := range trie.Children {
		countTrieNodes(v, i)
	}

	*i = *i + 1
}

func main() {
    trie := triego.NewTrie()
    
    trie.AppendWord("trial")
    trie.AppendWord("trie")
    
    fmt.Println("Stored words:")
    for _, w := range trie.Words() {
        fmt.Println(w)
    }
    
    fmt.Println("")
    var nodes int = 0
    countTrieNodes(trie, &nodes)
    fmt.Printf("Number of allocated nodes: %d\n", nodes)
}

Output:

Stored words:
trie
trial

Number of allocated nodes: 4

That is, 3 actual nodes plus the root node. This is because 1 node is required for the "tri" prefix and just 2 additional nodes for "e" and "al":

• tri
  • e
  • al

Advanced concepts

Prefixes iteration

The provided API can be used to iterate over all the prefixes in the radix.

You need to provide the EachPrefix API with a callback having this signature:

type PrefixIteratorCallback   func(PrefixInfo) (skip_subtree, halt bool)

PrefixInfo holds information about the current prefix. You can find out here. The callback can also be used to skip entire subtrees making the search even faster if it meets certain conditions of your choice. To do so, simply return true as the first return argument.

radix.EachPrefix(func(info PrefixInfo) (skip_subtree, halt bool) {
    value := some_computation(info.Prefix)
    
    if value > MAX_THRESHOLD {
        // this value is too high
        // and we are not interested in
        // further prefixes containing
        // the current one so we skip
        // the current subtree altogether
        return true, false
    }
    
    ...
    
    return false, false
})

License

The code in this repository is released under the terms of the MIT license. Copyright (c) Alessandro Diaferia [e-mail]