ルビ置換を簡素化

OUCC · Apr 4, 2024 · acdfb43 · acdfb43
1 parent c82b434
commit acdfb43
Show file tree

Hide file tree

Showing 4 changed files with 45 additions and 42 deletions.
diff --git a/Epub/KoeBook.Epub/Models/Paragraph.cs b/Epub/KoeBook.Epub/Models/Paragraph.cs
@@ -6,5 +6,5 @@ public sealed class Paragraph : Element
 {
     public ScriptLine? ScriptLine { get; set; }
     public Audio? Audio => ScriptLine?.Audio;
-    public string? Text { get; set; }
+    public string Text { get; set; } = "";
 }
diff --git a/Epub/KoeBook.Epub/Services/AnalyzerService.cs b/Epub/KoeBook.Epub/Services/AnalyzerService.cs
@@ -39,31 +39,16 @@ public async ValueTask<BookScripts> AnalyzeAsync(BookProperties bookProperties,
         }
         _epubDocumentStoreService.Register(document, cancellationToken);
 
-        var scriptLines = new List<ScriptLine>();
-        foreach (var chapter in document.Chapters)
-        {
-            foreach (var section in chapter.Sections)
+        var scriptLines = document.Chapters.SelectMany(c => c.Sections)
+            .SelectMany(s => s.Elements)
+            .OfType<Paragraph>()
+            .Select(p =>
             {
-                foreach (var element in section.Elements)
-                {
-                    if (element is Paragraph paragraph)
-                    {
-                        var line = paragraph.Text;
-                        // rubyタグがあればルビのdictionaryに登録
-                        var rubyDict = ExtractRuby(line).ToDictionary();
+                // ルビを置換
+                var line = ReplaceBaseTextWithRuby(p.Text);
 
-                        foreach (var ruby in rubyDict)
-                            _rubyReplacements.TryAdd(ruby.Key, ruby.Value);
-                        // ルビを置換
-                        line = ReplaceBaseTextWithRuby(line, rubyDict);
-
-                        var scriptLine = new ScriptLine(line, "", "");
-                        paragraph.ScriptLine = scriptLine;
-                        scriptLines.Add(scriptLine);
-                    }
-                }
-            }
-        }
+                return p.ScriptLine = new ScriptLine(line, "", "");
+            }).ToList();
 
         // 800文字以上になったら１チャンクに分ける
         var chunks = new List<string>();
@@ -85,24 +70,10 @@ public async ValueTask<BookScripts> AnalyzeAsync(BookProperties bookProperties,
         return bookScripts;
     }
 
-    private static IEnumerable<KeyValuePair<string, string>> ExtractRuby(string text)
-    {
-        return RubyRegex()
-            .Matches(text)
-            .Select(m => KeyValuePair.Create(m.Groups[1].Value, m.Groups[2].Value));
-    }
-
-    private static string ReplaceBaseTextWithRuby(string text, Dictionary<string, string> rubyDict)
+    private static string ReplaceBaseTextWithRuby(string text)
     {
         // 元のテキストからルビタグをすべてルビテキストに置き換える
-        var resultText = text;
-        foreach (var pair in rubyDict)
-        {
-            var rubyTag = $"<ruby><rb>{pair.Key}</rb><rp>（</rp><rt>{pair.Value}</rt><rp>）</rp></ruby>";
-            resultText = resultText.Replace(rubyTag, pair.Value);
-        }
-
-        return resultText;
+        return RubyRegex().Replace(text, m => m.Groups[2].Value);
     }
 
     [GeneratedRegex("<ruby><rb>(.*?)</rb><rp>（</rp><rt>(.*?)</rt><rp>）</rp></ruby>")]

diff --git a/KoeBook.Test/Epub/AnalyzerServiceTest.cs b/KoeBook.Test/Epub/AnalyzerServiceTest.cs
@@ -0,0 +1,32 @@
+using System.Runtime.CompilerServices;
+using KoeBook.Epub.Services;
+
+namespace KoeBook.Test.Epub;
+
+public class AnalyzerServiceTest
+{
+    [Theory]
+    [InlineData("aa", "aa")]
+    [InlineData("<ruby><rb>漢字</rb><rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>", "かんじ")]
+    [InlineData("ああ<ruby><rb>漢字</rb><rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>あああ", "ああかんじあああ")]
+    [InlineData("""
+        ああ<ruby><rb>漢字</rb><rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>あああ
+        ああ<ruby><rb>漢字</rb><rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>あああ
+        ああ<ruby><rb>漢字1</rb><rp>（</rp><rt>かんじ1</rt><rp>）</rp></ruby>あああ
+        """, "ああかんじあああ\nああかんじあああ\nああかんじ1あああ")]
+    [InlineData("<ruby><rb>佐久平</rb><rp>《</rp><rt>さくだいら</rt><rp>》</rp></ruby>　<ruby><rb>啓介</rb><rp>《</rp><rt>けいすけ</rt><rp>》</rp></ruby>",
+        "<ruby><rb>佐久平</rb><rp>《</rp><rt>さくだいら</rt><rp>》</rp></ruby>　<ruby><rb>啓介</rb><rp>《</rp><rt>けいすけ</rt><rp>》</rp></ruby>")]
+    [InlineData("<ruby><rb>漢字</rb>\n<rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>", "<ruby><rb>漢字</rb>\n<rp>（</rp><rt>かんじ</rt><rp>）</rp></ruby>")]
+    public void ReplaceBaseTextWithRuby(string input, string expected)
+    {
+        var result = AnalyzerServiceProxy.ReplaceBaseTextWithRuby(null, input);
+
+        Assert.Equal(expected, result);
+    }
+}
+
+file static class AnalyzerServiceProxy
+{
+    [UnsafeAccessor(UnsafeAccessorKind.StaticMethod)]
+    public static extern string ReplaceBaseTextWithRuby(AnalyzerService? _, string text);
+}
diff --git a/KoeBook.Test/Epub/EpubDocumentTest.cs b/KoeBook.Test/Epub/EpubDocumentTest.cs
@@ -89,7 +89,7 @@ public void EnsureParagraph()
         var element = Assert.Single(section.Elements);
         var paragraph = Assert.IsType<Paragraph>(element);
         Assert.Null(paragraph.Audio);
-        Assert.Null(paragraph.Text);
+        Assert.Empty(paragraph.Text);
         Assert.Null(paragraph.ClassName);
 
         // 空でないときは無視
@@ -129,7 +129,7 @@ public void EnsureParagraph()
         element = Assert.Single(document.Chapters[0].Sections[1].Elements);
         paragraph = Assert.IsType<Paragraph>(element);
         Assert.Null(paragraph.Audio);
-        Assert.Null(paragraph.Text);
+        Assert.Empty(paragraph.Text);
         Assert.Null(paragraph.ClassName);
 
         // インデックスは正しく指定する必要がある