Turn SpanUnit.MaxLength into a dynamic variable. (#17)

Megrez.Tests/Megrez.Tests.csproj

@@ -4,7 +4,7 @@
     <TargetFramework>net6.0</TargetFramework>
     <Nullable>enable</Nullable>
     <IsPackable>false</IsPackable>
-    <ReleaseVersion>2.5.3</ReleaseVersion>
+    <ReleaseVersion>2.5.4</ReleaseVersion>
   </PropertyGroup>
 
   <ItemGroup>

Megrez.Tests/MegrezTests.cs

@@ -20,9 +20,9 @@ public void Test01_SpanUnitInternalAbilities() {
     Compositor.Node n3 = new(keyArray: new() { "gao1", "ke1", "ji4" }, spanLength: 3,
                              unigrams: langModel.UnigramsFor(new() { "gao1ke1ji4" }));
     Assert.AreEqual(actual: span.MaxLength, expected: 0);
-    span.Add(node: n1);
+    span.Append(node: n1);
     Assert.AreEqual(actual: span.MaxLength, expected: 1);
-    span.Add(node: n3);
+    span.Append(node: n3);
     Assert.AreEqual(actual: span.MaxLength, expected: 3);
     Assert.AreEqual(actual: span.NodeOf(length: 1), expected: n1);
     Assert.AreEqual(actual: span.NodeOf(length: 2), expected: null);
@@ -33,8 +33,8 @@ public void Test01_SpanUnitInternalAbilities() {
     Assert.AreEqual(actual: span.NodeOf(length: 2), expected: null);
     Assert.AreEqual(actual: span.NodeOf(length: 3), expected: null);
 
-    span.Add(node: n1);
-    span.Add(node: n3);
+    span.Append(node: n1);
+    span.Append(node: n3);
     span.DropNodesOfOrBeyond(length: 2);
     Assert.AreEqual(actual: span.MaxLength, expected: 1);
     Assert.AreEqual(actual: span.NodeOf(length: 1), expected: n1);
@@ -44,7 +44,7 @@ public void Test01_SpanUnitInternalAbilities() {
     Assert.AreEqual(actual: span.MaxLength, expected: 0);
     Assert.AreEqual(actual: span.NodeOf(length: 1), expected: null);
     Compositor.Node n114514 = new(new(), 114_514, new());
-    Assert.IsFalse(span.Add(n114514));
+    Assert.IsFalse(span.Append(n114514));
     Assert.IsNull(span.NodeOf(length: 0));
     Assert.IsNull(span.NodeOf(length: Compositor.MaxSpanLength + 1));
   }

Megrez.sln

@@ -52,6 +52,6 @@ Global
 		$0.DotNetNamingPolicy = $4
 		$4.DirectoryNamespaceAssociation = PrefixedHierarchical
 		$0.StandardHeader = $5
-		version = 2.5.3
+		version = 2.5.4
 	EndGlobalSection
 EndGlobal

Megrez/Megrez.csproj

@@ -3,16 +3,16 @@
   <PropertyGroup>
     <TargetFramework>net6.0</TargetFramework>
     <Nullable>enable</Nullable>
-    <ReleaseVersion>2.5.3</ReleaseVersion>
+    <ReleaseVersion>2.5.4</ReleaseVersion>
     <PackageId>vChewing.Megrez</PackageId>
     <Authors>Shiki Suen</Authors>
     <Company>Atelier Inmu</Company>
     <Copyright>(c) 2022 and onwards The vChewing Project for Megrez-specific changes; (c) 2022 and onwards Lukhnos Liu for upstream contents.</Copyright>
     <RepositoryUrl>https://github.com/ShikiSuen/MegrezNT</RepositoryUrl>
     <NeutralLanguage>zh-TW</NeutralLanguage>
-    <AssemblyVersion>2.5.3</AssemblyVersion>
-    <FileVersion>2.5.3</FileVersion>
-    <Version>2.5.3</Version>
+    <AssemblyVersion>2.5.4</AssemblyVersion>
+    <FileVersion>2.5.4</FileVersion>
+    <Version>2.5.4</Version>
     <Product>Megrez</Product>
     <GeneratePackageOnBuild>True</GeneratePackageOnBuild>
     <PackageReadmeFile>README.md</PackageReadmeFile>

Megrez/src/1_Compositor.cs

@@ -430,7 +430,7 @@ public int Update(bool updateExisting = false) {
         }
         List<Unigram> unigramsB = TheLangModel.UnigramsFor(joinedKeyArray);
         if (unigramsB.IsEmpty()) continue;
-        Spans[position].Add(new(joinedKeyArray, theLength, unigramsB));
+        Spans[position].Append(new(joinedKeyArray, theLength, unigramsB));
         nodesChanged += 1;
       }
     }

Megrez/src/2_Walker.cs

@@ -26,7 +26,7 @@ public partial struct Compositor {
       List<List<Vertex>> vertexSpans = new();
       foreach (SpanUnit _ in Spans) vertexSpans.Add(new());
       foreach ((int i, SpanUnit span) in Spans.Enumerated()) {
-        foreach (int j in new BRange(1, span.MaxLength + 1)) {
+        foreach (int j in new BRange(1, Math.Max(1, span.MaxLength) + 1)) {
           if (span.NodeOf(j) is not {} theNode) continue;
           vertexSpans[i].Add(new(theNode));
         }

Megrez/src/4_SpanUnit.cs

@@ -16,12 +16,12 @@ public class SpanUnit {
     /// <summary>
     /// 節點陣列。每個位置上的節點可能是 null。
     /// </summary>
-    public List<Node?> Nodes = new();
+    public Dictionary<int, Node> Nodes = new();
     /// <summary>
     /// 該幅位單元內的所有節點當中持有最長幅位的節點長度。
     /// 該變數受該幅位的自身操作函式而被動更新。
     /// </summary>
-    public int MaxLength { get; private set; }
+    public int MaxLength => Nodes.Keys.Count > 0? Nodes.Keys.Max() : 0;
     /// <summary>
     /// 該幅位單元內的節點的幅位長度上限。
     /// </summary>
@@ -30,17 +30,22 @@ public class SpanUnit {
     /// <summary>
     /// 幅位乃指一組共享起點的節點。
     /// </summary>
-    public SpanUnit() {
-      Clear();  // 該函式會自動給 MaxLength 賦值。
-    }
+    public SpanUnit() => Clear();  // 該函式會自動給 MaxLength 賦值。
 
     /// <summary>
     /// 清除該幅位單元內的全部的節點，且重設最長節點長度為 0，然後再在節點陣列內預留空位。
     /// </summary>
-    public void Clear() {
-      Nodes.Clear();
-      foreach (int _ in new BRange(0, MaxSpanLength)) Nodes.Add(null);
-      MaxLength = 0;
+    public void Clear() => Nodes.Clear();
+
+    /// <summary>
+    /// 往該幅位塞入一個節點。
+    /// </summary>
+    /// <param name="node">要塞入的節點。</param>
+    /// <returns>該操作是否成功執行。</returns>
+    public bool Append(Node node) {
+      if (!AllowedLengths.Contains(node.SpanLength)) return false;
+      Nodes[node.SpanLength] = node;
+      return true;
     }
 
     /// <summary>
@@ -52,26 +57,7 @@ public void Clear() {
     /// 於是就提供了這個專門的工具函式。
     /// </remarks>
     /// <param name="givenNode">要參照的節點。</param>
-    public void Nullify(Node givenNode) {
-      BRange theRange = new(lowerbound: 0, upperbound: Nodes.Count);
-      foreach (int theIndex in theRange) {
-        Node? currentNode = Nodes[theIndex];
-        if (!Equals(currentNode, givenNode)) continue;
-        Nodes[theIndex] = null;
-      }
-    }
-
-    /// <summary>
-    /// 往該幅位塞入一個節點。
-    /// </summary>
-    /// <param name="node">要塞入的節點。</param>
-    /// <returns>該操作是否成功執行。</returns>
-    public bool Add(Node node) {
-      if (!AllowedLengths.Contains(node.SpanLength)) return false;
-      Nodes[node.SpanLength - 1] = node;
-      MaxLength = Math.Max(MaxLength, node.SpanLength);
-      return true;
-    }
+    public void Nullify(Node givenNode) => Nodes.Remove(givenNode.SpanLength);
 
     /// <summary>
     /// 丟掉任何不小於給定幅位長度的節點。
@@ -80,18 +66,9 @@ public bool Add(Node node) {
     /// <returns>該操作是否成功執行。</returns>
     public bool DropNodesOfOrBeyond(int length) {
       if (!AllowedLengths.Contains(length)) return false;
+      length = Math.Min(length, MaxSpanLength);
       foreach (int i in new BRange(length, MaxSpanLength + 1)) {
-        if (i > Nodes.Count) continue;  // 防呆
-        Nodes[i - 1] = null;
-      }
-      MaxLength = 0;
-      if (length <= 1) return false;
-      int maxR = length - 2;
-      foreach (int i in new BRange(0, maxR + 1)) {
-        BRange countRange = new(0, maxR - i + 1);
-        if (!countRange.Contains(maxR - i)) continue;  // 防呆
-        MaxLength = maxR - i + 1;
-        break;
+        Nodes.Remove(i);
       }
       return true;
     }
@@ -101,7 +78,7 @@ public bool DropNodesOfOrBeyond(int length) {
     /// </summary>
     /// <param name="length">給定的幅位長度。</param>
     /// <returns>查詢結果。</returns>
-    public Node? NodeOf(int length) => AllowedLengths.Contains(length) ? Nodes[length - 1] : null;
+    public Node? NodeOf(int length) => Nodes.ContainsKey(length) ? Nodes[length] : null;
   }
 
   // MARK: - Internal Implementations.

Megrez/src/5_Vertex.cs

@@ -41,7 +41,7 @@ internal class Vertex {
     /// 這是一個可變的數據結構，用於有向無環圖的構建和單源最短路徑的計算。
     /// </summary>
     /// <param name="node">字詞節點。</param>
-    public Vertex(Node node) { Node = node; }
+    public Vertex(Node node) => Node = node;
 
     /// <summary>
     /// 讓一個 Vertex 順藤摸瓜地將自己的所有的連帶的 Vertex 都摧毀，再摧毀自己。
@@ -62,7 +62,7 @@ public void Destroy() {
   /// </summary>
   /// <param name="u">參照頂點，會在必要時成為 v 的前述頂點。</param>
   /// <param name="v">要影響的頂點。</param>
-  internal void Relax(Vertex u, ref Vertex v) {
+  private void Relax(Vertex u, ref Vertex v) {
     // 從 u 到 w 的距離，也就是 v 的權重。
     double w = v.Node.Score;
     // 這裡計算最大權重：
@@ -76,7 +76,7 @@ internal void Relax(Vertex u, ref Vertex v) {
   /// <summary>
   /// 位相幾何排序處理時的處理狀態。
   /// </summary>
-  internal class TopoSortState {
+  private class TopoSortState {
     public int IterIndex { get; set; }
     public Vertex Vertex { get; }
     public TopoSortState(Vertex vertex, int iterIndex = 0) {
@@ -105,7 +105,7 @@ public TopoSortState(Vertex vertex, int iterIndex = 0) {
   /// </summary>
   /// <param name="root">根頂點。</param>
   /// <returns>排序結果（頂點陣列）。</returns>
-  internal List<Vertex> TopoSort(ref Vertex root) {
+  private List<Vertex> TopoSort(ref Vertex root) {
     List<Vertex> result = new();
     List<TopoSortState> stack = new();
     stack.Add(new(root));