Улучшения работы с текстовыми координатами (#110)

* #107 - оптимизация ITextCoordinateSystemComputer.GetLines * #107 - правка тестов * #109 - Вынос логики построения координат в сервис * #109 - тесты
Stepami · Feb 2, 2025 · 524fb4c · 524fb4c
1 parent 29e047c
commit 524fb4c
Show file tree

Hide file tree

Showing 5 changed files with 64 additions and 27 deletions.
diff --git a/src/Domain/HydraScript.Domain.FrontEnd/Lexer/ITextCoordinateSystemComputer.cs b/src/Domain/HydraScript.Domain.FrontEnd/Lexer/ITextCoordinateSystemComputer.cs
@@ -2,5 +2,19 @@ namespace HydraScript.Domain.FrontEnd.Lexer;
 
 public interface ITextCoordinateSystemComputer
 {
+    /// <summary>
+    /// Возвращает список индексов переноса строки внутри фрагмента исходного кода<br/>
+    /// Всегда начинается с -1
+    /// </summary>
     public IReadOnlyList<int> GetLines(string text);
+
+    /// <summary>
+    /// Строит координату в формате (Строка, Столбец)
+    /// </summary>
+    /// <param name="absoluteIndex">Индекс символа от начала строки, в диапазоне [0, ДлинаСтроки)</param>
+    /// <param name="newLineList">Список индексов переноса строки <see cref="GetLines"/></param>
+    /// <returns></returns>
+    public Coordinates GetCoordinates(
+        int absoluteIndex,
+        IReadOnlyList<int> newLineList);
 }
diff --git a/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Impl/RegexLexer.cs b/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Impl/RegexLexer.cs
@@ -21,8 +21,7 @@ public List<Token> GetTokens(string text)
 
     public IEnumerator<Token> GetEnumerator()
     {
-        var matches = Structure.Regex.Matches(_text);
-        foreach (Match match in matches)
+        foreach (Match match in Structure.Regex.Matches(_text))
         {
             foreach (var type in Structure)
             {
@@ -32,9 +31,8 @@ public IEnumerator<Token> GetEnumerator()
 
                 var value = group.Value;
                 var segment = new Segment(
-                    new Coordinates(group.Index, _lines),
-                    new Coordinates(group.Index + group.Length, _lines)
-                );
+                    computer.GetCoordinates(group.Index, _lines),
+                    computer.GetCoordinates(absoluteIndex: group.Index + group.Length, _lines));
                 var token = new Token(type, segment, value);
 
                 if (type.Error()) throw new LexerException(token);

diff --git a/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Impl/TextCoordinateSystemComputer.cs b/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Impl/TextCoordinateSystemComputer.cs
@@ -6,26 +6,40 @@ public class TextCoordinateSystemComputer : ITextCoordinateSystemComputer
 {
     private readonly SearchValues<char> _sv = SearchValues.Create(['\n']);
 
+    /// <inheritdoc/>
     public IReadOnlyList<int> GetLines(string text)
     {
         var newText = text.EndsWith(Environment.NewLine)
             ? text
             : text + Environment.NewLine;
         var textLength = newText.Length;
 
-        LinkedList<int> indices = [];
+        var indices = new List<int>(capacity: 128) { -1 };
+        var textAsSpan = newText.AsSpan();
         while (true)
         {
-            var start = indices.Last != null ? indices.Last.Value + 1 : 0;
+            var start = indices[^1] + 1;
             if (start == textLength)
                 break;
-            var textAsSpan = newText.AsSpan(
+            var index = textAsSpan.Slice(
                 start,
-                length: textLength - start);
-            var index = textAsSpan.IndexOfAny(_sv);
-            indices.AddLast(start + index);
+                length: textLength - start).IndexOfAny(_sv);
+            indices.Add(start + index);
         }
 
         return indices.ToList();
     }
+
+    /// <inheritdoc/>
+    public Coordinates GetCoordinates(int absoluteIndex, IReadOnlyList<int> newLineList)
+    {
+        for (var i = 1; i < newLineList.Count; i++)
+            if (absoluteIndex <= newLineList[i])
+            {
+                var offset = newLineList[i - 1];
+                return new Coordinates(Line: i, Column: absoluteIndex - offset);
+            }
+
+        return new Coordinates();
+    }
 }
diff --git a/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Token.cs b/src/Domain/HydraScript.Domain.FrontEnd/Lexer/Token.cs
@@ -39,21 +39,8 @@ public static implicit operator string(Segment segment) =>
 [ExcludeFromCodeCoverage]
 public record Coordinates(int Line, int Column)
 {
-    public Coordinates(int absolutePos, IReadOnlyList<int> system) :
-        this(0, 0)
+    public Coordinates() : this(Line: 1, Column: 1)
     {
-        for (var i = 0; i < system.Count; i++)
-            if (absolutePos <= system[i])
-            {
-                var offset = i == 0 ? -1 : system[i - 1];
-                (Line, Column) = (i + 1, absolutePos - offset);
-                break;
-            }
-
-        if (Line == 0)
-        {
-            (Line, Column) = (system.Count + 1, 1);
-        }
     }
 
     public override string ToString() => $"({Line}, {Column})";

diff --git a/tests/HydraScript.Tests/Unit/FrontEnd/TextCoordinateSystemComputerTests.cs b/tests/HydraScript.Tests/Unit/FrontEnd/TextCoordinateSystemComputerTests.cs
@@ -13,15 +13,15 @@ public void GetLines_NoNewLine_SingleIndexResult()
     {
         const string text = "let x = 0";
         var result = _sut.GetLines(text);
-        result.Should().BeEquivalentTo([text.Length + Environment.NewLine.Length - 1]);
+        result.Should().BeEquivalentTo([-1, text.Length + Environment.NewLine.Length - 1]);
     }
 
     [Fact]
     public void GetLines_HasNewLine_SingleIndexResult()
     {
         var text = "let x = 0" + Environment.NewLine;
         var result = _sut.GetLines(text);
-        result.Should().BeEquivalentTo([text.Length - 1]);
+        result.Should().BeEquivalentTo([-1, text.Length - 1]);
     }
 
     [Fact]
@@ -39,9 +39,33 @@ public void GetLines_HasNewLines_MultipleIndicesResult()
         var result = _sut.GetLines(text);
         result.Should().BeEquivalentTo(
         [
+            -1,
             stmt1.Length + Environment.NewLine.Length - 1,
             stmt1.Length + stmt2.Length + Environment.NewLine.Length * 2 - 1,
             stmt1.Length + stmt2.Length + stmt3.Length + Environment.NewLine.Length * 3 - 1
         ]);
     }
+
+    [Theory, MemberData(nameof(EmptySystems))]
+    public void GetCoordinates_EmptySystem_StartCoordinatesReturned(IReadOnlyList<int> newLineList)
+    {
+        var coord = _sut.GetCoordinates(
+            absoluteIndex: Random.Shared.Next(0, int.MaxValue),
+            newLineList);
+        coord.Should().Be(new Coordinates());
+    }
+
+    public static TheoryData<IReadOnlyList<int>> EmptySystems =>
+        new([[], [-1]]);
+
+    [Fact]
+    public void GetCoordinates_NotEmptySystem_CorrectCoordinatesReturned()
+    {
+        // abcd\nabc\naaaa\n
+        //     4    8     13
+        IReadOnlyList<int> newLineList = [-1, 4, 8, 13];
+        // координата символа b
+        var coord = _sut.GetCoordinates(absoluteIndex: 6, newLineList);
+        coord.Should().Be(new Coordinates(Line: 2, Column: 2));
+    }
 }