refact: 将编译中各个阶段接口化 (#35)

Reviewed-on: PostGuard/Canon#35

Canon.Core/Abstractions/ICCodeGenerator.cs

@@ -0,0 +1,11 @@
+using Canon.Core.CodeGenerators;
+
+namespace Canon.Core.Abstractions;
+
+/// <summary>
+/// 支持生成C语言代码的接口
+/// </summary>
+public interface ICCodeGenerator
+{
+    public void GenerateCCode(CCodeBuilder builder);
+}

Canon.Core/Abstractions/IGrammarParser.cs

@@ -8,13 +8,19 @@ namespace Canon.Core.Abstractions;
 /// <summary>
 /// 语法分析器接口
 /// </summary>
-public abstract class GrammarParserBase
+public interface IGrammarParser
 {
-    public abstract ITransformer BeginTransformer { get; }
+    public ITransformer BeginTransformer { get; }
 
-    public abstract NonTerminator Begin { get; }
+    public NonTerminator Begin { get; }
 
-    public SyntaxNodeBase Analyse(IEnumerable<SemanticToken> tokens)
+    /// <summary>
+    /// 分析指定的词法记号流并构建对应的语法树
+    /// </summary>
+    /// <param name="tokens">输入的词法记号流</param>
+    /// <returns>语法树的根节点</returns>
+    /// <exception cref="InvalidOperationException">语法分析错误</exception>
+    public ProgramStruct Analyse(IEnumerable<SemanticToken> tokens)
     {
         Stack<AnalyseState> stack = [];
         stack.Push(new AnalyseState(BeginTransformer, SyntaxNodeBase.Create(SemanticToken.End)));
@@ -36,7 +42,7 @@ public abstract class GrammarParserBase
                 {
                     // 如果是归约到起始符
                     // 那么就直接返回不继续进行归约
-                    return top.Node;
+                    return top.Node.Convert<ProgramStruct>();
                 }
 
                 List<SyntaxNodeBase> children = [];

Canon.Core/Abstractions/ILexer.cs

@@ -0,0 +1,11 @@
+using Canon.Core.LexicalParser;
+
+namespace Canon.Core.Abstractions;
+
+/// <summary>
+/// 词法分析器接口
+/// </summary>
+public interface ILexer
+{
+    public IEnumerable<SemanticToken> Tokenize(ISourceReader reader);
+}

Canon.Core/Abstractions/ISourceReader.cs

@@ -0,0 +1,31 @@
+using System.Diagnostics.CodeAnalysis;
+
+namespace Canon.Core.Abstractions;
+
+/// <summary>
+/// 读取源代码的接口
+/// </summary>
+public interface ISourceReader
+{
+    /// <summary>
+    /// 尝试读取下一个字符
+    /// </summary>
+    /// <param name="c">读取到的字符</param>
+    /// <returns>是否成功读取</returns>
+    public bool TryReadChar([NotNullWhen(true)] out char? c);
+
+    /// <summary>
+    /// 源文件名称
+    /// </summary>
+    public string FileName { get; }
+
+    /// <summary>
+    /// 当前读取字符的行号
+    /// </summary>
+    public uint Line { get; }
+
+    /// <summary>
+    /// 当前读取字符的列号
+    /// </summary>
+    public uint Pos { get; }
+}

Canon.Core/CodeGenerators/CCodeBuilder.cs

@@ -0,0 +1,22 @@
+using System.Text;
+
+namespace Canon.Core.CodeGenerators;
+
+/// <summary>
+/// 构建C语言代码
+/// </summary>
+public class CCodeBuilder
+{
+    private readonly StringBuilder _builder = new();
+
+    public void AddString(string code)
+    {
+        _builder.Append(code);
+    }
+
+
+    public string Build()
+    {
+        return _builder.ToString();
+    }
+}

Canon.Core/GrammarParser/Grammar.cs

@@ -22,7 +22,7 @@ public class Grammar
     /// </summary>
     public required LrState BeginState { get; init; }
 
-    public GrammarParserBase ToGrammarParser()
+    public IGrammarParser ToGrammarParser()
     {
         Dictionary<LrState, Transformer> transformers = [];
 
@@ -71,10 +71,10 @@ public class Grammar
         return new GrammarParser(transformers[BeginState], Begin);
     }
 
-    private class GrammarParser(ITransformer beginTransformer, NonTerminator begin) : GrammarParserBase
+    private class GrammarParser(ITransformer beginTransformer, NonTerminator begin) : IGrammarParser
     {
-        public override ITransformer BeginTransformer { get; } = beginTransformer;
+        public ITransformer BeginTransformer { get; } = beginTransformer;
-        public override NonTerminator Begin { get; } = begin;
+        public NonTerminator Begin { get; } = begin;
     }
 
     private class Transformer : ITransformer

Canon.Core/SyntaxNodes/NonTerminatedSyntaxNode.cs

@@ -21,12 +21,14 @@ public abstract class NonTerminatedSyntaxNode : SyntaxNodeBase, IEnumerable<Synt
             {
                 yield return child;
             }
-
+            else
-            NonTerminatedSyntaxNode nonTerminatedNode = child.Convert<NonTerminatedSyntaxNode>();
-
-            foreach (SyntaxNodeBase node in nonTerminatedNode)
             {
-                yield return node;
+                NonTerminatedSyntaxNode nonTerminatedNode = child.Convert<NonTerminatedSyntaxNode>();
+
+                foreach (SyntaxNodeBase node in nonTerminatedNode)
+                {
+                    yield return node;
+                }
             }
         }
     }

Canon.Core/SyntaxNodes/ProgramStruct.cs

@@ -1,4 +1,5 @@
-using Canon.Core.Enums;
+using Canon.Core.CodeGenerators;
+using Canon.Core.Enums;
 
 namespace Canon.Core.SyntaxNodes;
 
@@ -20,4 +21,9 @@ public class ProgramStruct : NonTerminatedSyntaxNode
     {
         return new ProgramStruct { Children = children };
     }
+
+    public override void GenerateCCode(CCodeBuilder builder)
+    {
+        builder.AddString("#include <PascalCoreLib.h>");
+    }
 }

Canon.Core/SyntaxNodes/SyntaxNodeBase.cs

@@ -1,9 +1,11 @@
-using Canon.Core.Enums;
+using Canon.Core.Abstractions;
+using Canon.Core.CodeGenerators;
+using Canon.Core.Enums;
 using Canon.Core.LexicalParser;
 
 namespace Canon.Core.SyntaxNodes;
 
-public abstract class SyntaxNodeBase
+public abstract class SyntaxNodeBase : ICCodeGenerator
 {
     public abstract bool IsTerminated { get; }
 
@@ -19,6 +21,14 @@ public abstract class SyntaxNodeBase
         return result;
     }
 
+    /// <summary>
+    /// 语法树节点基类对于生成C代码的默认实现
+    /// </summary>
+    public virtual void GenerateCCode(CCodeBuilder builder)
+    {
+
+    }
+
     public static SyntaxNodeBase Create(SemanticToken token)
     {
         return new TerminatedSyntaxNode { Token = token };

Canon.Generator/GrammarGenerator/GeneratedGrammarParser.cs

@@ -7,11 +7,11 @@ namespace Canon.Generator.GrammarGenerator;
 public class GeneratedGrammarParser(
     Dictionary<string, GeneratedTransformer> transformers,
     string beginState,
-    NonTerminator begin) : GrammarParserBase
+    NonTerminator begin) : IGrammarParser
 {
-    public override ITransformer BeginTransformer => transformers[beginState];
+    public ITransformer BeginTransformer => transformers[beginState];
 
-    public override NonTerminator Begin => begin;
+    public NonTerminator Begin => begin;
 
     public string GenerateCode(string namespaceValue)
     {
@@ -74,7 +74,7 @@ public class GeneratedGrammarParser(
                        """);
         builder.Append('\n');
 
-        builder.Append("public class GeneratedGrammarParser : GrammarParserBase\n")
+        builder.Append("public class GeneratedGrammarParser : IGrammarParser\n")
             .Append("{\n");
 
         builder.Append("\tprivate static readonly Dictionary<string, GeneratedTransformer> s_transformers = new()\n")
@@ -104,10 +104,10 @@ public class GeneratedGrammarParser(
                        """);
         builder.Append("\n");
 
-        builder.Append('\t').Append("public override ITransformer BeginTransformer => ")
+        builder.Append('\t').Append("public ITransformer BeginTransformer => ")
             .Append($"s_transformers[\"{beginState}\"];\n");
 
-        builder.Append('\t').Append("public override NonTerminator Begin => ")
+        builder.Append('\t').Append("public NonTerminator Begin => ")
             .Append(begin.GenerateCode()).Append(";\n");
 
         builder.Append("}\n");

Canon.Tests/CCodeGeneratorTests/BasicTests.cs

@@ -0,0 +1,33 @@
+using Canon.Core.Abstractions;
+using Canon.Core.CodeGenerators;
+using Canon.Core.LexicalParser;
+using Canon.Core.SyntaxNodes;
+using Canon.Tests.GeneratedParserTests;
+
+namespace Canon.Tests.CCodeGeneratorTests;
+
+public class BasicTests
+{
+    private readonly IGrammarParser _parser = GeneratedGrammarParser.Instance;
+
+    [Fact]
+    public void ProgramStructTest()
+    {
+        CCodeBuilder builder = new();
+
+        const string program = """
+                               program DoNothing;
+                               begin
+                               end.
+                               """;
+
+        Lexer lexer = new(program);
+        List<SemanticToken> tokens = lexer.Tokenize();
+        tokens.Add(SemanticToken.End);
+
+        ProgramStruct root = _parser.Analyse(tokens);
+        root.GenerateCCode(builder);
+
+        Assert.Equal("#include <PascalCoreLib.h>", builder.Build());
+    }
+}

Canon.Tests/GeneratedParserTests/GenerateParserTests.cs

@@ -11,7 +11,7 @@ public class GenerateParserTests
         Generators = PascalGrammar.Grammar, Begin = new NonTerminator(NonTerminatorType.StartNonTerminator)
     };
 
-    private readonly GrammarParserBase _parser;
+    private readonly IGrammarParser _parser;
 
     public GenerateParserTests()
     {

Canon.Tests/GrammarParserTests/PascalGrammarTests.cs

@@ -9,7 +9,7 @@ namespace Canon.Tests.GrammarParserTests;
 
 public class PascalGrammarTests
 {
-    private readonly GrammarParserBase _parser = GeneratedGrammarParser.Instance;
+    private readonly IGrammarParser _parser = GeneratedGrammarParser.Instance;
 
     [Fact]
     public void DoNothingTest()
@@ -24,8 +24,9 @@ public class PascalGrammarTests
         List<SemanticToken> tokens = lexer.Tokenize();
         tokens.Add(SemanticToken.End);
 
-        ProgramStruct root = _parser.Analyse(tokens).Convert<ProgramStruct>();
+        ProgramStruct root = _parser.Analyse(tokens);
         Assert.Equal("DoNothing", root.Head.ProgramName.LiteralValue);
+        Assert.Equal(15, root.Count());
     }
 
     [Fact]
@@ -43,7 +44,7 @@ public class PascalGrammarTests
         List<SemanticToken> tokens = lexer.Tokenize();
         tokens.Add(SemanticToken.End);
 
-        ProgramStruct root = _parser.Analyse(tokens).Convert<ProgramStruct>();
+        ProgramStruct root = _parser.Analyse(tokens);
         Assert.Equal("Add", root.Head.ProgramName.LiteralValue);
     }
 
@@ -63,11 +64,11 @@ public class PascalGrammarTests
         List<SemanticToken> tokens = lexer.Tokenize();
         tokens.Add(SemanticToken.End);
 
-        ProgramStruct root = _parser.Analyse(tokens).Convert<ProgramStruct>();
+        ProgramStruct root = _parser.Analyse(tokens);
         Assert.Equal("exFunction", root.Head.ProgramName.LiteralValue);
     }
 
-    private static GrammarParserBase GenerateGrammarParser()
+    private static IGrammarParser GenerateGrammarParser()
     {
         GrammarBuilder builder = new()
         {

Canon.Tests/GrammarParserTests/SimpleGrammarWithEmptyTests.cs

@@ -159,7 +159,7 @@ public class SimpleGrammarWithEmptyTests(ITestOutputHelper testOutputHelper)
         };
 
         Grammar grammar = builder.Build();
-        GrammarParserBase parser = grammar.ToGrammarParser();
+        IGrammarParser parser = grammar.ToGrammarParser();
 
         ITransformer transformer1 = parser.BeginTransformer;
         Assert.Equal(3, transformer1.ShiftTable.Count);