更新v0.0.3

GarthTB · Nov 19, 2024 · 4781e3c · 4781e3c
1 parent 9896f70
commit 4781e3c
Show file tree

Hide file tree

Showing 4 changed files with 20 additions and 11 deletions.
diff --git a/CodeLord/CodeLord.csproj b/CodeLord/CodeLord.csproj
@@ -13,7 +13,7 @@
     <ApplicationIcon>Icon\favicon.ico</ApplicationIcon>
     <Product>Codelord</Product>
     <Title>Codelord</Title>
-    <Version>0.0.2</Version>
+    <Version>0.0.3</Version>
     <Company>Garth</Company>
     <Authors>Garth</Authors>
     <Copyright>© Garth 2024</Copyright>

diff --git a/CodeLord/Components/Concater.cs b/CodeLord/Components/Concater.cs
@@ -25,10 +25,10 @@ private static string JD(string head, string tail)
             var a = "abcdefghijklmnopqrstuvwxyz"; // 所有码元
             var oldEnd = head[^1];
             var newStart = tail[0];
+            if (head.Length < 4 && y.Contains(head[^1])) // 不足4码且以音码结尾，后补空格
+                head = $"{head} ";
             if (!a.Contains(newStart) && oldEnd == ' ') // 标点开头且前为空格，直接替换空格
                 return $"{head[..^1]}{tail}";
-            if (tail.Length < 4 && y.Contains(tail[^1])) // 不足4码且以音码结尾，后补空格
-                tail = $"{tail} ";
             if (x.Contains(newStart) && a.Contains(oldEnd)) // 形码开头且无标点断开，前加空格
                 tail = $" {tail}";
             return $"{head}{tail}";

diff --git a/CodeLord/Components/Encoder.cs b/CodeLord/Components/Encoder.cs
@@ -33,14 +33,18 @@ private static string[] FindShortest(ConcurrentDictionary<string, List<string>>
         {
             Console.WriteLine($"共需遍历{text.Length}字，正在遍历...");
             Dictionary<int, HashSet<string>> tempWays = new(text.Length) { [0] = [""] }; // 键为末尾位置，值为编码集合
+            var maxLen = dict.Keys.Max(x => x.Length); // 词库中的最大字数
+            var slices = Enumerable.Range(0, text.Length)
+                                   .Select(i => new string(text.Skip(i).Take(maxLen).ToArray()))
+                                   .ToArray()
+                                   .AsSpan(); // 预先切片以提升性能
 
             for (int i = 0; i < text.Length; i++)
             {
-                var heads = tempWays[i];
-                var minLen = heads.Min(x => x.Length);
-                var trimHeads = heads.Where(x => x.Length == minLen).Take(limit);
-                var tails = FindBranches(dict, text, i);
-                foreach (var head in trimHeads)
+                var heads = tempWays[i].OrderBy(x => x.Length)
+                                       .Take(limit); // 最短路径
+                var tails = FindBranches(dict, slices[i]);
+                foreach (var head in heads)
                     foreach (var (length, codes) in tails) // code无重复项
                         foreach (var code in codes)
                             RecordWays(i + length, Concater.Join(codeID, head, code), tempWays);
@@ -53,12 +57,12 @@ private static string[] FindShortest(ConcurrentDictionary<string, List<string>>
             Console.WriteLine($"\n遍历完成，共得到{ways.Length}种最短编码。");
             return ways;
 
-            static (int, List<string>)[] FindBranches(ConcurrentDictionary<string, List<string>> dict, string text, int i)
+            static (int, List<string>)[] FindBranches(ConcurrentDictionary<string, List<string>> dict, string text)
             {
-                var branches = dict.Where(x => text[i..].StartsWith(x.Key))
+                var branches = dict.Where(x => text.StartsWith(x.Key))
                                    .Select(x => (x.Key.Length, x.Value)) // Value无重复项
                                    .ToArray();
-                return branches.Length == 0 ? [(1, [text.Substring(i, 1)])] : branches;
+                return branches.Length == 0 ? [(1, [text[0..1]])] : branches;
             }
 
             static void RecordWays(int endIndex, string code, Dictionary<int, HashSet<string>> tempWays)

diff --git a/README.md b/README.md
@@ -29,6 +29,11 @@
 
 ## 更新日志
 
+### [0.0.3] - 20241120
+
+- 改善长文本编码性能
+- 完善键道规则
+
 ### [0.0.2] - 20241119
 
 - 改为流式计算，以避免卡顿并减少内存消耗