Implement Remove method for Trie and UkkonenTrie

DemoApp/DemoApp.csproj

@@ -6,6 +6,7 @@
         <Nullable>enable</Nullable>
         <UseWindowsForms>true</UseWindowsForms>
         <ImplicitUsings>enable</ImplicitUsings>
+        <IsPackable>false</IsPackable>
     </PropertyGroup>
 
     <ItemGroup>

README.md

@@ -38,6 +38,10 @@ This small library contains a bunch of trie data structures all having the same
 public interface ITrie {
   IEnumerable Retrieve(string query);
   void Add(string key, TValue value);
+
+  // Note, only Trie and UkkonenTrie are only supported.
+  void Remove(string key, TValue value);
+  void Remove(string key, params TValue[] values);
 }
 ```
 
@@ -61,6 +65,12 @@ public interface IGenericTrie<TKey, TValue> where TKey : IEquatable<TKey> {
 
 At the moment only `UkkonenTrie` implements this interface.
 
+## Removal
+
+The current implementation allows for item removal in two specific tree types: `Trie` and `UkkonenTree`.
+
+Please note that removing an item does **not** optimize or alter the internal tree structure. This limitation exists due to the complexities involved in reducing a trie tree without adversely affecting its other functionalities.
+
 ## Performance
 
 All diagrams are given in logarithmic scale on the x-axis and y-axis.
@@ -74,5 +84,4 @@ All diagrams are given in logarithmic scale on the x-axis and y-axis.
 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.
 
-![](https://raw.githubusercontent.com/OliBomby/trienet/master/img/trie-demo-app.png)
-
+![](https://raw.githubusercontent.com/OliBomby/trienet/master/img/trie-demo-app.png)

SampleConsoleApp/SampleConsoleApp.csproj

@@ -5,6 +5,7 @@
         <TargetFramework>net6.0</TargetFramework>
         <ImplicitUsings>enable</ImplicitUsings>
         <Nullable>enable</Nullable>
+        <IsPackable>false</IsPackable>
     </PropertyGroup>
 
     <ItemGroup>

TrieNet.Test/BaseTrieTest.cs

@@ -13,15 +13,20 @@ namespace TrieNet.Test;
 public abstract class BaseTrieTest {
     [OneTimeSetUp]
     public virtual void Setup() {
-        Trie = CreateTrie();
-        for (var i = 0; i < Words40.Length; i++) Trie.Add(Words40[i], i);
+        Trie = CreateDefaultTrie();
+    }
+
+    public ITrie<int> CreateDefaultTrie() {
+        var trie = CreateTrie();
+        for (var i = 0; i < Words40.Length; i++) trie.Add(Words40[i], i);
+        return trie;
     }
 
     protected ITrie<int> Trie { get; private set; }
 
     protected abstract ITrie<int> CreateTrie();
 
-    public readonly string[] Words40 = {
+    public static readonly string[] Words40 = {
         "daubreelite",
         "daubingly",
         "daubingly",

TrieNet.Test/Performance/FakeTrie.cs

@@ -24,4 +24,12 @@ public void Add(string key, T value) {
         var keyValPair = new KeyValuePair<string, T>(key, value);
         stack.Push(keyValPair);
     }
+
+    public void Remove(string key, T value) {
+        throw new System.NotImplementedException();
+    }
+
+    public void Remove(string key, params T[] values) {
+        throw new System.NotImplementedException();
+    }
 }

TrieNet.Test/SuffixTrieTest.cs

@@ -10,7 +10,7 @@
 namespace TrieNet.Test;
 
 public class SuffixTrieTest {
-    public readonly string[] Words20 = {
+    public static readonly string[] Words20 = {
         "overcontribution",
         "overcontribute",
         "overcontraction",
@@ -45,7 +45,7 @@ public virtual void Setup() {
         Trie2.Add("aabacdefac", 0);
         Trie2.Add("aabacdefac", 1);
     }
-
+    
     [TestCase("a", new[] { 0, 1, 3, 8 })]
     [TestCase("b", new[] { 2 })]
     [TestCase("c", new[] { 4, 9 })]

TrieNet.Test/TrieTest.cs

@@ -1,6 +1,7 @@
 // This code is distributed under MIT license. Copyright (c) 2013 George Mamaladze
 // See license.txt or http://opensource.org/licenses/mit-license.php
 
+using System;
 using System.Linq;
 using NUnit.Framework;
 using TrieNet.Trie;
@@ -23,4 +24,47 @@ public void ExhaustiveParallelAddFails() {
                 .ForAll(phrase => trie.Add(phrase, phrase.GetHashCode()));
         }
     }
+
+    [Test]
+    public void RemoveKey() {
+        var trie = CreateDefaultTrie();
+
+        Assert.AreEqual(new [] { 21, 22, 23}, trie.Retrieve("capo"));
+        Assert.AreEqual(new [] { 22, 23}, trie.Retrieve("capoc"));
+
+        trie.Remove("capoc", 22);
+
+        Assert.AreEqual(new [] { 21, 23}, trie.Retrieve("capo"));
+        Assert.AreEqual(new [] { 23}, trie.Retrieve("capoc"));
+    }
+
+    [Test]
+    public void RemoveMultipleKeys() {
+        var trie = CreateDefaultTrie();
+
+        trie.Remove("capoc", 22, 23);
+
+        Assert.AreEqual(new [] { 21}, trie.Retrieve("capo"));
+        Assert.AreEqual(Enumerable.Empty<int>(), trie.Retrieve("capoc"));
+    }
+
+    [Test]
+    public void RemovePartialMatchedKey_RemoveAllEntries() {
+        var trie = CreateDefaultTrie();
+
+        trie.Remove("capo", 22, 23);
+
+        Assert.AreEqual(new [] { 21 }, trie.Retrieve("capo"));
+        Assert.AreEqual(Array.Empty<int>(), trie.Retrieve("capoc"));
+    }
+
+    [Test]
+    public void RemoveLongerKey_HasNoEffect() {
+        var trie = CreateDefaultTrie();
+
+        trie.Remove("capocissss", 22);
+
+        Assert.AreEqual(new [] { 21, 22, 23}, trie.Retrieve("capo"));
+        Assert.AreEqual(new [] { 22, 23}, trie.Retrieve("capoc"));
+    }
 }

TrieNet.Test/UkkonenTreeTest.cs

@@ -11,4 +11,48 @@ public class UkkonenTreeTest : SuffixTrieTest {
     protected override ISuffixTrie<int> CreateTrie() {
         return new CharUkkonenTrie<int>(0);
     }
+
+    [Test]
+    public void RemoveKey() {
+        var trie = CreateTestTrie();
+
+        Assert.AreEqual(new [] { 15, 16, 17 }, trie.Retrieve("archi"));
+
+        trie.Remove("architecturesque", 17);
+
+        Assert.AreEqual(new [] { 15, 16 }, trie.Retrieve("archi"));
+    }
+
+    [Test]
+    public void RemoveMultipleValues() {
+        var trie = CreateTestTrie();
+
+        trie.Remove("architis", 15, 16);
+
+        Assert.AreEqual(new [] { 17 }, trie.Retrieve("archi"));
+    }
+
+    [Test]
+    public void RemovePartialMatchedKey_RemoveAllEntries() {
+        var trie = CreateTestTrie();
+
+        trie.Remove("archi", 15, 16);
+
+        Assert.AreEqual(new [] { 17 }, trie.Retrieve("archi"));
+    }
+
+    [Test]
+    public void RemoveLongerKey_HasNoEffect() {
+        var trie = CreateTestTrie();
+
+        trie.Remove("architissssssss", 15, 16);
+
+        Assert.AreEqual(new [] { 15, 16, 17 }, trie.Retrieve("archi"));
+    }
+
+    static CharUkkonenTrie<int> CreateTestTrie() {
+        var trie = new CharUkkonenTrie<int>();
+        for(var i =0; i < Words20.Length; ++i) trie.Add(Words20[i], i);
+        return trie;
+    }
 }

TrieNet/ITrie.cs

@@ -14,4 +14,6 @@ namespace TrieNet;
 public interface ITrie<TValue> {
     IEnumerable<TValue> Retrieve(string query);
     void Add(string key, TValue value);
+    void Remove(string key, TValue value);
+    void Remove(string key, params TValue[] values);
 }

TrieNet/PatriciaTrie/PatriciaSuffixTrie.cs

@@ -39,6 +39,14 @@ public void Add(string key, TValue value) {
             innerTrie.Add(currentSuffix, new WordPosition<TValue>(position, value));
     }
 
+    public void Remove(string key, TValue value) {
+        throw new NotSupportedException();
+    }
+
+    public void Remove(string key, params TValue[] values) {
+        throw new NotImplementedException();
+    }
+
     private static IEnumerable<Tuple<StringPartition, int>> GetAllSuffixes(int minSuffixLength, string word) {
         for (var i = word.Length - minSuffixLength; i >= 0; i--)
             yield return new Tuple<StringPartition, int>(new StringPartition(word, i), i);

TrieNet/PatriciaTrie/PatriciaTrie.cs

@@ -25,6 +25,14 @@ public virtual void Add(string key, TValue value) {
         Add(new StringPartition(key), value);
     }
 
+    public void Remove(string key, TValue value) {
+        throw new NotSupportedException();
+    }
+
+    public void Remove(string key, params TValue[] values) {
+        throw new NotImplementedException();
+    }
+
     internal override void Add(StringPartition keyRest, TValue value) {
         GetOrCreateChild(keyRest, value);
     }

TrieNet/Trie/ConcurrentTrie.cs

@@ -15,4 +15,12 @@ public IEnumerable<TValue> Retrieve(string query) {
     public void Add(string key, TValue value) {
         Add(key, 0, value);
     }
+
+    public void Remove(string key, TValue value) {
+        throw new NotSupportedException();
+    }
+
+    public void Remove(string key, params TValue[] values) {
+        throw new NotImplementedException();
+    }
 }

TrieNet/Trie/SuffixTrie.cs

@@ -39,6 +39,14 @@ public void Add(string key, T value) {
             innerTrie.Add(suffix, new WordPosition<T>(position, value));
     }
 
+    public void Remove(string key, T value) {
+        throw new NotImplementedException();
+    }
+
+    public void Remove(string key, params T[] values) {
+        throw new NotImplementedException();
+    }
+
     private static IEnumerable<Tuple<string, int>> GetAllSuffixes(int minSuffixLength, string word) {
         for (var i = word.Length - minSuffixLength; i >= 0; i--) {
             var partition = new StringPartition(word, i);

TrieNet/Trie/Trie.cs

@@ -15,4 +15,12 @@ public IEnumerable<TValue> Retrieve(string query) {
     public void Add(string key, TValue value) {
         Add(key, 0, value);
     }
+
+    public void Remove(string key, TValue value) {
+        RemoveFromKey(key, new [] { value });
+    }
+
+    public void Remove(string key, params TValue[] values) {
+        RemoveFromKey(key, values);
+    }
 }

TrieNet/Trie/TrieNode.cs

@@ -3,17 +3,18 @@
 
 using System;
 using System.Collections.Generic;
+using System.Diagnostics.CodeAnalysis;
+using System.Linq;
 
 namespace TrieNet.Trie;
 
 [Serializable]
 public class TrieNode<TValue> : TrieNodeBase<TValue> {
     private readonly Dictionary<char, TrieNode<TValue>> children;
-    private readonly Queue<TValue> values;
+    private readonly List<TValue> values = new();
 
     protected TrieNode() {
         children = new Dictionary<char, TrieNode<TValue>>();
-        values = new Queue<TValue>();
     }
 
     protected override int KeyLength => 1;
@@ -35,7 +36,8 @@ protected override TrieNodeBase<TValue> GetOrCreateChild(char key) {
         return result;
     }
 
-    protected override TrieNodeBase<TValue> GetChildOrNull(string query, int position) {
+    [return: MaybeNull]
+    protected override TrieNodeBase<TValue> GetChildOrNull([NotNull] string query, int position) {
         if (query == null) throw new ArgumentNullException(nameof(query));
         return
             children.TryGetValue(query[position], out var childNode)
@@ -44,6 +46,10 @@ protected override TrieNodeBase<TValue> GetChildOrNull(string query, int positio
     }
 
     protected override void AddValue(TValue value) {
-        values.Enqueue(value);
+        values.Add(value);
+    }
+
+    protected override void RemoveAll(TValue[] nodeValues) {
+        values.RemoveAll(v => nodeValues.Any(nv => v is not null && v.Equals(nv) || (v is null && nv is null)));
     }
 }

TrieNet/Trie/TrieNodeBase.cs

@@ -4,6 +4,7 @@
 using System;
 using System.Collections.Generic;
 using System.Linq;
+using System.Runtime.CompilerServices;
 
 namespace TrieNet.Trie;
 
@@ -32,6 +33,18 @@ public void Add(string key, int position, TValue value) {
 
     protected abstract void AddValue(TValue value);
 
+    protected virtual void RemoveAll(TValue[] values) {
+        throw new NotSupportedException();
+    }
+
+    protected void RemoveFromKey(string key, TValue[] values) {
+        if (key == null) throw new ArgumentNullException(nameof(key));
+        var child = Enumerable.Range(0, key.Length).Aggregate(this, (node, position) => node?.GetChildOrNull(key, position));
+        if (child is not null)
+            foreach(var node in child.Subtree())
+                node.RemoveAll(values);
+    }
+
     protected abstract TrieNodeBase<TValue> GetOrCreateChild(char key);
 
     protected virtual IEnumerable<TValue> Retrieve(string query, int position) {
@@ -50,6 +63,7 @@ protected virtual IEnumerable<TValue> SearchDeep(string query, int position) {
 
     protected abstract TrieNodeBase<TValue> GetChildOrNull(string query, int position);
 
+    [MethodImpl(MethodImplOptions.AggressiveInlining)]
     private static bool EndOfString(int position, string text) {
         return position >= text.Length;
     }

TrieNet/TrieNet.csproj

@@ -9,12 +9,9 @@
         <Description>.NET Implementations of Trie Data Structures for Substring Search, Auto-completion and Intelli-sense. Includes: patricia trie, suffix trie and a trie implementation using Ukkonen's algorithm. This is a modern .NET update for the old TrieNet package.</Description>
         <GeneratePackageOnBuild>true</GeneratePackageOnBuild>
         <Copyright>Copyright OliBomby 2022</Copyright>
-        <AssemblyVersion>2.0.0</AssemblyVersion>
-        <FileVersion>2.0.0</FileVersion>
         <PackageLicenseUrl></PackageLicenseUrl>
         <PackageProjectUrl>https://github.com/OliBomby/trienet</PackageProjectUrl>
         <PackageTags>ata-structures dotnet algorithms string search ukkonen trie</PackageTags>
-        <Version>2.0.1</Version>
         <RepositoryUrl>https://github.com/OliBomby/trienet</RepositoryUrl>
         <PackageIcon>trienet.png</PackageIcon>
         <Title>TrieNet 2</Title>
@@ -32,5 +29,12 @@
 	<ItemGroup>
 		<None Include="..\license.txt" Pack="true" PackagePath="" />
 	</ItemGroup>
+
+	<ItemGroup>
+	  <PackageReference Include="MinVer" Version="5.0.0">
+	    <PrivateAssets>all</PrivateAssets>
+	    <IncludeAssets>runtime; build; native; contentfiles; analyzers; buildtransitive</IncludeAssets>
+	  </PackageReference>
+	</ItemGroup>
 
 </Project>