refeat: ILexer接口适配 (#38)

Co-authored-by: Huaps <1183155719@qq.com>
Co-authored-by: duqoo <92306417+duqoo@users.noreply.github.com>
Reviewed-on: PostGuard/Canon#38

Canon.Core/LexicalParser/LexRules.cs

@@ -0,0 +1,83 @@
+namespace Canon.Core.LexicalParser;
+
+public static class LexRules
+{
+    // 保留关键字
+    private static readonly string[] _keywords =
+    [
+        "Program", "Const", "Var", "Procedure",
+        "Function", "Begin", "End", "Array",
+        "Of", "If", "Then", "Else",
+        "For", "To", "Do", "Integer",
+        "Real", "Boolean", "Character", "Divide",
+        "Not", "Mod", "And", "Or"
+    ];
+
+    private static readonly string[] _delimiter = [";", ",", ":", ".", "(", ")", "[", "]", "'", "\"", ".."];
+
+    private static readonly string[] _operator = ["=", "<>", "<", "<=", ">", ">=", "+", "-", "*", "/", ":="];
+
+    // 判断字符
+    public static bool IsDigit(char _ch) {
+        if (_ch >= '0' && _ch <= '9') return true;
+        return false;
+    }
+
+    public static bool IsHexDigit(char _ch)
+    {
+        if ((_ch >= '0' && _ch <= '9') || (_ch<= 'F' && _ch >= 'A')) return true;
+        return false;
+    }
+
+    public static bool IsLetter(char _ch) {
+        if ((_ch >= 'A' && _ch <= 'Z') || (_ch >= 'a' && _ch <= 'z' || _ch == '_')) {
+            return true;
+        }
+        return false;
+    }
+
+    public static bool IsKeyword(string tokenString)
+    {
+
+        foreach (var t in _keywords)
+        {
+            if (string.Equals(tokenString, t, StringComparison.OrdinalIgnoreCase)) return true;
+        }
+        return false;
+    }
+
+
+    public static bool IsDelimiter(char ch)
+    {
+        foreach (var delimiter in _delimiter)
+        {
+            if (delimiter.Contains(ch))
+            {
+                return true;
+            }
+        }
+        return false;
+    }
+
+    public static bool IsOperator(char ch)
+    {
+        foreach (var o in _operator)
+        {
+            if (o.Contains(ch))
+            {
+                return true;
+            }
+        }
+        return false;
+    }
+
+    public static bool IsBreakPoint(char ch)
+    {
+        if (ch == ' ' || ch == '\n' || ch == '\t' || ch == '\r')
+        {
+            return true;
+        }
+
+        return false;
+    }
+}

Canon.Core/LexicalParser/LexemeFactory.cs

@@ -0,0 +1,95 @@
+using Canon.Core.Enums;
+
+namespace Canon.Core.LexicalParser;
+
+public static class LexemeFactory
+{
+
+    public static SemanticToken MakeToken(SemanticTokenType tokenType,string literal,uint _line,uint _chPos)
+    {
+        SemanticToken? token;
+        switch (tokenType)
+        {
+            case SemanticTokenType.Character:
+                CharacterSemanticToken characterSemanticToken = new CharacterSemanticToken()
+                {
+                    LinePos = _line, CharacterPos = _chPos, LiteralValue = literal,
+                };
+                token = characterSemanticToken;
+                break;
+            case SemanticTokenType.Identifier:
+                IdentifierSemanticToken identifierSemanticToken = new IdentifierSemanticToken()
+                {
+                    LinePos = _line, CharacterPos = _chPos, LiteralValue = literal,
+                };
+                token = identifierSemanticToken;
+                break;
+            default:
+                throw new ArgumentOutOfRangeException(nameof(tokenType), tokenType, null);
+        }
+
+        return token;
+
+
+    }
+
+    public static KeywordSemanticToken MakeToken(KeywordType keywordType,string literal,uint _line,uint _chPos)
+    {
+        KeywordSemanticToken keywordSemanticToken = new KeywordSemanticToken
+        {
+            LinePos = _line,
+            CharacterPos = _chPos,
+            LiteralValue = literal,
+            KeywordType = keywordType
+        };
+        return keywordSemanticToken;
+    }
+
+    public static DelimiterSemanticToken MakeToken(DelimiterType delimiterType,string literal,uint _line,uint _chPos)
+    {
+        DelimiterSemanticToken delimiterSemanticToken = new DelimiterSemanticToken()
+        {
+            LinePos = _line,
+            CharacterPos = _chPos,
+            LiteralValue = literal,
+            DelimiterType = delimiterType
+        };
+        return delimiterSemanticToken;
+    }
+
+    public static NumberSemanticToken MakeToken(NumberType numberType,string literal,uint _line,uint _chPos)
+    {
+        string temp = literal;
+        string result;
+        if (numberType == NumberType.Hex)
+        {
+            result = string.Concat("0x", temp.AsSpan(1, temp.Length - 1));
+        }
+        else
+        {
+            result = temp;
+        }
+
+        NumberSemanticToken numberSemanticToken = new NumberSemanticToken()
+        {
+            LinePos = _line,
+            CharacterPos = _chPos,
+            LiteralValue = result,
+            NumberType = numberType
+        };
+        return numberSemanticToken;
+
+    }
+
+    public static OperatorSemanticToken MakeToken(OperatorType operatorType,string literal,uint _line,uint _chPos)
+    {
+        OperatorSemanticToken operatorSemanticToken = new OperatorSemanticToken()
+        {
+            LinePos = _line,
+            CharacterPos = _chPos,
+            LiteralValue = literal,
+            OperatorType = operatorType
+        };
+        return operatorSemanticToken;
+    }
+}

Canon.Core/LexicalParser/SemanticToken.cs

@@ -339,18 +339,4 @@ public class EndSemanticToken : SemanticToken
     public override SemanticTokenType TokenType => SemanticTokenType.End;
 }
 
-/// <summary>
-/// 错误类型记号
-/// </summary>
-public class ErrorSemanticToken : SemanticToken
-{
-    public override SemanticTokenType TokenType => SemanticTokenType.Error;
-
-    public static bool TryParse(uint linePos, uint characterPos, LinkedListNode<char> now,
-        out IdentifierSemanticToken? token)
-    {
-        token = null;
-        return false;
-    }
-}