Sanitizing node-crossed candidates. (#20)

Megrez.Tests/LMDataForTests.cs

@@ -190,6 +190,7 @@ nian2zhong1 年中 -11.373044
 gao1ke1ji4 高科技 -9.842421
 zhe4yang4 這樣 -6.000000 // Non-LibTaBE
 ni3zhe4 你這 -9.000000 // Non-LibTaBE
+ke1ke1 顆顆 -8.000000 // Non-LibTaBE
 jiao4 教 -3.676169
 jiao4 較 -3.24869962
 jiao4yu4 教育 -3.32220565

Megrez.Tests/Megrez.Tests.csproj

@@ -4,7 +4,7 @@
     <TargetFramework>net6.0</TargetFramework>
     <Nullable>enable</Nullable>
     <IsPackable>false</IsPackable>
-    <ReleaseVersion>2.6.0</ReleaseVersion>
+    <ReleaseVersion>2.6.2</ReleaseVersion>
   </PropertyGroup>
 
   <ItemGroup>

Megrez.Tests/MegrezTests.cs

@@ -538,4 +538,42 @@ public void Test21_Compositor_hardCopy() {
     List<Node> resultB = compositorB.Walk().WalkedNodes;
     Assert.True(resultA.SequenceEqual(resultB));
   }
+
+  [Test]
+  public void Test22_Compositor_SanitizingNodeCrossing() {
+    SimpleLM theLM = new(input: StrSampleData);
+    string rawReadings = "ke1 ke1";
+    Compositor compositor = new(langModel: theLM, separator: "");
+    foreach (string key in rawReadings.Split(separator: " ")) {
+      compositor.InsertKey(key);
+    }
+    int a = compositor.FetchCandidatesAt(givenLocation: 1, filter: Compositor.CandidateFetchFilter.BeginAt)
+                .Select(x => x.KeyArray.Count)
+                .Max();
+    int b = compositor.FetchCandidatesAt(givenLocation: 1, filter: Compositor.CandidateFetchFilter.EndAt)
+                .Select(x => x.KeyArray.Count)
+                .Max();
+    int c = compositor.FetchCandidatesAt(givenLocation: 0, filter: Compositor.CandidateFetchFilter.BeginAt)
+                .Select(x => x.KeyArray.Count)
+                .Max();
+    int d = compositor.FetchCandidatesAt(givenLocation: 2, filter: Compositor.CandidateFetchFilter.EndAt)
+                .Select(x => x.KeyArray.Count)
+                .Max();
+    Assert.AreEqual(actual: $"{a} {b} {c} {d}", expected: "1 1 2 2");
+    compositor.Cursor = compositor.Length;
+    compositor.InsertKey("jin1");
+    a = compositor.FetchCandidatesAt(givenLocation: 1, filter: Compositor.CandidateFetchFilter.BeginAt)
+            .Select(x => x.KeyArray.Count)
+            .Max();
+    b = compositor.FetchCandidatesAt(givenLocation: 1, filter: Compositor.CandidateFetchFilter.EndAt)
+            .Select(x => x.KeyArray.Count)
+            .Max();
+    c = compositor.FetchCandidatesAt(givenLocation: 0, filter: Compositor.CandidateFetchFilter.BeginAt)
+            .Select(x => x.KeyArray.Count)
+            .Max();
+    d = compositor.FetchCandidatesAt(givenLocation: 2, filter: Compositor.CandidateFetchFilter.EndAt)
+            .Select(x => x.KeyArray.Count)
+            .Max();
+    Assert.AreEqual(actual: $"{a} {b} {c} {d}", expected: "1 1 2 2");
+  }
 }

Megrez.sln

@@ -52,6 +52,6 @@ Global
 		$0.DotNetNamingPolicy = $4
 		$4.DirectoryNamespaceAssociation = PrefixedHierarchical
 		$0.StandardHeader = $5
-		version = 2.6.0
+		version = 2.6.2
 	EndGlobalSection
 EndGlobal

Megrez/Megrez.csproj

@@ -3,16 +3,16 @@
   <PropertyGroup>
     <TargetFramework>net6.0</TargetFramework>
     <Nullable>enable</Nullable>
-    <ReleaseVersion>2.6.0</ReleaseVersion>
+    <ReleaseVersion>2.6.2</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>2.6.0</AssemblyVersion>
-    <FileVersion>2.6.0</FileVersion>
-    <Version>2.6.0</Version>
+    <AssemblyVersion>2.6.2</AssemblyVersion>
+    <FileVersion>2.6.2</FileVersion>
+    <Version>2.6.2</Version>
     <Product>Megrez</Product>
     <GeneratePackageOnBuild>True</GeneratePackageOnBuild>
     <PackageReadmeFile>README.md</PackageReadmeFile>

Megrez/src/3_KeyValuePaired.cs

@@ -165,32 +165,39 @@ public enum CandidateFetchFilter {
   /// <summary>
   /// 返回在當前位置的所有候選字詞（以詞音配對的形式）。<para/>如果組字器內有幅位、且游標
   /// 位於組字器的（文字輸入順序的）最前方（也就是游標位置的數值是最大合規數值）的
-  /// 話，那麼這裡會用到 location - 1、以免去在呼叫該函式後再處理的麻煩。
+  /// 話，那麼這裡會對 location 的位置自動減去 1、以免去在呼叫該函式後再處理的麻煩。
   /// </summary>
-  /// <param name="location">游標位置。</param>
+  /// <param name="givenLocation">游標位置，必須是顯示的游標位置、不得做任何事先糾偏處理。</param>
   /// <param name="filter">候選字音配對陣列。</param>
   /// <returns></returns>
-  public List<KeyValuePaired> FetchCandidatesAt(int location, CandidateFetchFilter filter = CandidateFetchFilter.All) {
+  public List<KeyValuePaired> FetchCandidatesAt(int? givenLocation = null,
+                                                CandidateFetchFilter filter = CandidateFetchFilter.All) {
     List<KeyValuePaired> result = new();
     if (Keys.IsEmpty()) return result;
+    int location = Math.Max(0, Math.Min(givenLocation ?? Cursor, Keys.Count));
+    if (filter == CandidateFetchFilter.EndAt) {
+      if (location == Keys.Count) filter = CandidateFetchFilter.All;
+      location -= 1;
+    }
     location = Math.Max(0, Math.Min(location, Keys.Count - 1));
-
     // 按照讀音的長度（幅位長度）來給節點排序。
     List<NodeAnchor> anchors =
         FetchOverlappingNodesAt(location).StableSorted((x, y) => x.SpanLength.CompareTo(y.SpanLength));
     string keyAtCursor = Keys[location];
-    foreach (Node theNode in anchors.Select(x => x.Node).Where(x => !x.KeyArray.IsEmpty())) {
+    foreach (NodeAnchor theAnchor in anchors) {
+      Node theNode = theAnchor.Node;
       foreach (Unigram gram in theNode.Unigrams) {
         switch (filter) {
           case CandidateFetchFilter.All:
             // 得加上這道篩選，所以會出現很多無效結果。
             if (!theNode.KeyArray.Contains(keyAtCursor)) continue;
             break;
           case CandidateFetchFilter.BeginAt:
-            if (theNode.KeyArray.First() != keyAtCursor) continue;
+            if (theAnchor.SpanIndex != location) continue;
             break;
           case CandidateFetchFilter.EndAt:
             if (theNode.KeyArray.Last() != keyAtCursor) continue;
+            if (theNode.SpanLength >= 2 && theAnchor.SpanIndex + theAnchor.SpanLength - 1 != location) continue;
             break;
         }
         result.Add(new(theNode.KeyArray, gram.Value));

Megrez/src/4_SpanUnit.cs

@@ -112,11 +112,13 @@ public bool DropNodesOfOrBeyond(int length) {
   /// <returns>一個包含所有與該位置重疊的節點的陣列。</returns>
   internal List<NodeAnchor> FetchOverlappingNodesAt(int givenLocation) {
     List<NodeAnchor> results = new();
-    if (Spans.IsEmpty() || givenLocation >= Spans.Count) return results;
+    givenLocation = Math.Max(0, Math.Min(givenLocation, Keys.Count - 1));
+    if (Spans.IsEmpty()) return results;
 
     // 先獲取該位置的所有單字節點。
     foreach (int spanLength in new BRange(1, Spans[givenLocation].MaxLength + 1)) {
       if (Spans[givenLocation].NodeOf(spanLength) is not {} node) continue;
+      if (string.IsNullOrEmpty(node.KeyArray.Joined())) continue;
       results.Add(new(node, givenLocation));
     }
 
@@ -127,6 +129,7 @@ internal List<NodeAnchor> FetchOverlappingNodesAt(int givenLocation) {
       if (alpha > bravo) continue;
       foreach (int theLength in new BRange(alpha, bravo + 1)) {
         if (Spans[theLocation].NodeOf(theLength) is not {} node) continue;
+        if (string.IsNullOrEmpty(node.KeyArray.Joined())) continue;
         results.Add(new(node, theLocation));
       }
     }