TrieNet - The library provides .NET Data Structures for Prefix String Search and Substring (Infix) Search to Implement Auto-completion and Intelli-sense. This is a modern .NET update for the old TrieNet package.

Usage

  nuget install TrieNet2

using TrieNet.Ukkonen;
	
...

var trie = new CharUkkonenTrie<int>(3);
//var trie = new GenerixSuffixTrie<char, int>(3);

trie.Add("hello", 1);
trie.Add("world", 2);
trie.Add("hell", 3);

var result = trie.Retrieve("hel");
// result = { 1, 3 };

var result2 = trie.RetrieveSubstrings("ll");
// result2 = { new WordPosition(2, 1), new WordPosition(2, 3) };

Implementation

This small library contains a bunch of trie data structures all having the same interface:

public interface ITrie {
  IEnumerable Retrieve(string query);
  void Add(string key, TValue value);
}

Class	Description
Trie	The simple trie, allows only prefix search, like .Where(s => s.StartsWith(searchString))
SuffixTrie	Also allows infix search, like .Where(s => s.Contains(searchString))
PatriciaTrie	Compressed trie, more compact, a bit more efficient during look-up, but a quite slower during build-up.
SuffixPatriciaTrie	The same as PatriciaTrie, also enabling infix search.
ParallelTrie	Very primitively implemented parallel data structure which allows adding data and retrieving results from different threads simultaneusly.
CharUkkonenTrie	The same as SuffixPatriciaTrie, but more compact, more efficient during look-up, and quite a lot faster during build-up. This is the best choice.

There is also a generic interface which lets you search for more than just strings:

public interface IGenericTrie<TKey, TValue> where TKey : IEquatable<TKey> {
    IEnumerable<TValue> Retrieve(ReadOnlySpan<TKey> query);
    void Add(ReadOnlyMemory<TKey> key, TValue value);
}

At the moment only UkkonenTrie implements this interface.

Performance

All diagrams are given in logarithmic scale on the x-axis and y-axis.

Demo app

The app demonstrates indexing of large text files and look-up inside them. Indexing usually takes only a few seconds and the look-up delay will be unnoticeable for the user.