Minor

rampaa · Aug 23, 2024 · 92c80a8 · 92c80a8
1 parent 812c857
commit 92c80a8
Show file tree

Hide file tree

Showing 3 changed files with 24 additions and 17 deletions.
diff --git a/JL.Core/Utilities/ExtensionMethods.cs b/JL.Core/Utilities/ExtensionMethods.cs
@@ -39,22 +39,29 @@ public static T GetEnum<T>(this string description) where T : struct, Enum
         // return default;
     }
 
-    internal static List<string> ListUnicodeCharacters(this string s)
+    internal static List<string> ListUnicodeCharacters(this ReadOnlySpan<char> text)
     {
-        List<string> textBlocks = new(s.Length);
-        for (int i = 0; i < s.Length; i++)
+        List<string> textBlocks = new(text.Length);
+        for (int i = 0; i < text.Length; i++)
         {
-            if (char.IsHighSurrogate(s, i)
-                && s.Length > i + 1
-                && char.IsLowSurrogate(s, i + 1))
+            char highSurrogateCandidate = text[i];
+            if (char.IsHighSurrogate(highSurrogateCandidate)
+                && text.Length > i + 1)
             {
-                textBlocks.Add(char.ConvertFromUtf32(char.ConvertToUtf32(s, i)));
-
-                ++i;
+                char lowSurragateCandidate = text[i + 1];
+                if (char.IsLowSurrogate(lowSurragateCandidate))
+                {
+                    textBlocks.Add(char.ConvertFromUtf32(char.ConvertToUtf32(highSurrogateCandidate, lowSurragateCandidate)));
+                    ++i;
+                }
+                else
+                {
+                    textBlocks.Add(highSurrogateCandidate.ToString());
+                }
             }
             else
             {
-                textBlocks.Add(s[i].ToString());
+                textBlocks.Add(highSurrogateCandidate.ToString());
             }
         }
 

diff --git a/JL.Core/Utilities/JapaneseUtils.cs b/JL.Core/Utilities/JapaneseUtils.cs
@@ -174,7 +174,7 @@ public static partial class JapaneseUtils
 
     private static readonly FrozenSet<char> s_expressionTerminatingCharacters = s_leftToRightBracketDict.Keys.Union(s_leftToRightBracketDict.Values).Union(s_sentenceTerminatingCharacters).ToFrozenSet();
 
-    private static int FirstKatakanaIndex(string text)
+    private static int FirstKatakanaIndex(ReadOnlySpan<char> text)
     {
         int textLength = text.Length;
         for (int i = 0; i < textLength; i++)
@@ -219,7 +219,7 @@ public static string KatakanaToHiragana(string text)
         return textInHiragana.ToString();
     }
 
-    internal static List<string> LongVowelMarkToKana(string text)
+    internal static List<string> LongVowelMarkToKana(ReadOnlySpan<char> text)
     {
         List<string> unicodeTextList = text.ListUnicodeCharacters();
 
@@ -278,7 +278,7 @@ internal static List<string> LongVowelMarkToKana(string text)
         return stringBuilders.ConvertAll(static sb => sb.ToString());
     }
 
-    public static List<string> CreateCombinedForm(string text)
+    public static List<string> CreateCombinedForm(ReadOnlySpan<char> text)
     {
         List<string> combinedForm = new(text.Length);
 
@@ -300,12 +300,12 @@ public static List<string> CreateCombinedForm(string text)
         return combinedForm;
     }
 
-    internal static bool IsKatakana(string text)
+    internal static bool IsKatakana(ReadOnlySpan<char> text)
     {
         return s_katakanaToHiraganaDict.ContainsKey(text[0]);
     }
 
-    public static int FindExpressionBoundary(string text, int position)
+    public static int FindExpressionBoundary(ReadOnlySpan<char> text, int position)
     {
         int endPosition = text.Length;
         for (int i = position; i < text.Length; i++)
@@ -412,7 +412,7 @@ internal static string FindSentence(string text, int position)
         return sentence;
     }
 
-    private static int FirstPunctuationIndex(string text)
+    private static int FirstPunctuationIndex(ReadOnlySpan<char> text)
     {
         int charIndex = 0;
         foreach (Rune rune in text.EnumerateRunes())

diff --git a/JL.Core/Utilities/TextUtils.cs b/JL.Core/Utilities/TextUtils.cs
@@ -6,7 +6,7 @@ namespace JL.Core.Utilities;
 
 public static class TextUtils
 {
-    private static int FirstInvalidUnicodeSequenceIndex(string text)
+    private static int FirstInvalidUnicodeSequenceIndex(ReadOnlySpan<char> text)
     {
         for (int i = 0; i < text.Length; i++)
         {