C語言解譯器-16 文法分析（1）

Parser類用於對C代碼進行文法分析並構造文法樹。

/// <summary>    /// Parser class parses the simple C source code to build    /// the syntax serial.    /// </summary>    public class Parser    {        public enum ParsingErrorType        {            Warning,            UnknownError,            UndefinedVariable,            UnexpectedKeyword,            TokenExpected,            SyntaxError,            TypeError,            FormatError,            FunctionBodyUndefined        };        public class ParsingEventArg        {            public Context Context;        }        public class ParsingWarningEventArg        {            public Context Context;            public Context.LocationInfo Location;            public string Description;        }        public class ParsingErrorEventArg        {            public Context Context;            public Context.LocationInfo Location;            public ParsingErrorType Error;            public string Description;            public bool Continue;        }        public event EventHandler<ParsingEventArg>          OnParsing;        public event EventHandler<ParsingWarningEventArg>   OnParsingWarning;        public event EventHandler<ParsingErrorEventArg>    OnParsingError;                ///////////////////////////////        // Private member variables        ///////////////////////////////        private SourceCode m_sourceCode = null;        private Word m_lastWord = null;        private Expression.ExpressionNode m_lastExpNode = null;        private Dictionary<char, char> m_escapeCharDict = new Dictionary<char, char>();        private int m_errorCount = 0;        private int m_warningCount = 0;        public String Source        {            get { return m_sourceCode.Text; }        }        public bool HasError        {            get { return m_errorCount > 0; }        }        public bool HasWarning        {            get { return m_warningCount > 0; }        }        public int ErrorCount        {            get { return m_errorCount; }        }        public int WarningCount        {            get { return m_warningCount; }        }        public int MaxError = 0;        public int MaxWarning = 65535;        public Context Parse(SourceCode src)        {            m_sourceCode = src;            Context ctx = new Context();            if (Parse(ctx, src))            {                //文法分析完成後，校正是否有未使用的變數或函數，以及對申明而未定義函數的使用                if (ValidateContextReference(ctx))                    return ctx;            }            return null;        }        ...

文法分析的主要方法定義如下：

private bool Parse(Context ctx, SourceCode src)        {            bool res = true;            ctx.Location.FirstLine = src.AbsoluteLine;            ctx.Location.FirstPos = src.AbsolutePos;            foreach (SourceCode stmt in src.SplitStatement()) // 逐語句進行處理            {                try                {                    // 檢測do...while的while部分                    if (ctx.Children.Count > 0 && ctx.Children.Last() is ControlFlow.DoWhileLoop)                    {                        if ((ctx.Children.Last() as ControlFlow.DoWhileLoop).Condition == null)                        {                            Word wordOfWhile = GetWord(stmt);                            if (wordOfWhile.Text != "while")                            {                                if (!NotifyError(ctx, wordOfWhile.Location, ParsingErrorType.SyntaxError, "\"while\" is expected."))                                    return false;                            }                            else                            {                                res = ParseControl_While(ctx, stmt, new Context.LocationInfo()                                {                                    FirstLine = wordOfWhile.AbsoluteLine,                                    FirstPos = wordOfWhile.AbsoluteStartPos                                });                                if (!res)                                    return false;                                else                                    continue;                            }                        }                    }                    if (stmt.Text.EndsWith(";")) // 語句                    {                        res = ParseStatement(ctx, stmt);                    }                    else                    {                        if (stmt.Text.EndsWith("}")) // 函數體或塊                        {                            if (stmt.Text.StartsWith("{")) // 塊                            {                                SourceCode blockSrc = new SourceCode()                                {                                    LineOffset = stmt.AbsoluteLine,                                    PosOffset = stmt.AbsolutePos,                                    Text = stmt.Text.Substring(1, stmt.Text.Length - 2)                                };                                Block block = new Block()                                {                                    Name = Context.GetAnonymousName("block"),                                    Location = new Context.LocationInfo()                                    {                                        FirstLine = stmt.AbsoluteLine,                                        FirstPos = stmt.AbsolutePos                                    }                                };                                ctx.AddChild(block);                                res = Parse(block, blockSrc);                                block.Location.LastLine = stmt.AbsoluteLine;                                block.Location.LastPos = stmt.AbsolutePos;                            }                            else                            {   // 函數                                // 過濾控制結構                                Word wordOfControlFlow = GetWord(stmt);                                if (Context.IsControlFlow(wordOfControlFlow.Text))                                {                                    res = ParseControlFlow(ctx, stmt, wordOfControlFlow);                                }                                else                                {                                    stmt.ResetPos();                                    res = ParseFunction(ctx, stmt, wordOfControlFlow.Location);                                }                            }                        }                    }                }                catch (ParseException pe)                {                    if (!NotifyError(ctx, ctx.Location, ParsingErrorType.SyntaxError, pe.Message))                        return false;                }                if (!res)                    return false;            } // for            ctx.Location.LastLine = src.AbsoluteLine;            ctx.Location.LastPos = src.AbsolutePos;            return true;        }

語句處理，分三種情況：申明、控制結構及運算式。

private bool ParseStatement(Context ctx, SourceCode src)        {            Word firstWord = GetWord(src);            if (Context.IsDataType(firstWord.Text)) // 以類型打頭            {                //變數或函數申明                return ParseDeclare(ctx, src, firstWord);            }            else                if (Context.IsControlFlow(firstWord.Text)) // 控制結構                {                    //Control                     return ParseControlFlow(ctx, src, firstWord);                }                else                {                    // 運算式                    src.ResetPos();                    return ParseExpression(ctx, src, firstWord.Location);                }        }

函數解析的實現如下：

private bool ParseFunction(Context ctx, SourceCode src, Context.LocationInfo loc)        {            // 定位函數體            while (!src.Eof && src.CurrentChar != '{')                src.NextChar();            // 頭部位置資訊            Context.LocationInfo headerLoc = loc;            headerLoc.LastPos = src.AbsolutePos - 1;            // 函數頭部            SourceCode funcHeader = new SourceCode()            {                PosOffset = loc.FirstPos,                LineOffset = loc.FirstLine,                Text = src.Text.Substring(0, src.Column)            };            // 解析頭部            // 如成功, 一個FunctionDefine將被添加到當前Context的尾            if (!ParseStatement(ctx, funcHeader))                 return false;            src.NextChar(); // skip '{'            // 函數體            SourceCode bodyStmt = new SourceCode()            {                PosOffset = src.AbsolutePos,                LineOffset = src.AbsoluteLine,                Text = src.Text.Substring(src.Column, src.Text.Length - src.Column - 1)            };            // 函數對象            Function.FunctionDefine funcDef = ctx.Children.Last() as Function.FunctionDefine;            funcDef.AddChild(new Block()            {                Name = Context.GetAnonymousName("block")            });            // 遞迴解析函數體            if (Parse(funcDef.Body, bodyStmt))            {                funcDef.Location = headerLoc;                return true;            }            return false;        }

控制結構的解析如下：

private bool ParseControlFlow(Context ctx, SourceCode src, Word wordOfControlFlow)        {            bool res = false;            switch (wordOfControlFlow.Text)            {                case "if": res = ParseControl_If(ctx, src, wordOfControlFlow.Location); break;                case "else": res = ParseControl_Else(ctx, src, wordOfControlFlow.Location); break;                case "for": res = ParseControl_For(ctx, src, wordOfControlFlow.Location); break;                case "do": res = ParseControl_DoWhile(ctx, src, wordOfControlFlow.Location); break;                case "while": res = ParseControl_While(ctx, src, wordOfControlFlow.Location); break;                case "switch": res = ParseControl_Switch(ctx, src, wordOfControlFlow.Location); break;                case "continue": res = ParseControl_Continue(ctx, src, wordOfControlFlow.Location); break;                case "break": res = ParseControl_Break(ctx, src, wordOfControlFlow.Location); break;                case "return": res = ParseControl_Return(ctx, src, wordOfControlFlow.Location); break;                default:                    {                        // Unsupported control flow.                        if (!NotifyError(ctx, wordOfControlFlow.Location, ParsingErrorType.SyntaxError, "Unsupported keyword."))                            return false;                    }                    break;            } // switch            if (res)                NotifyParsing(ctx.Children.Last());            return res;        }

 

以if/else為例，說明控制結構的解析過程。

先看if部分：

private bool ParseControl_If(Context ctx, SourceCode src, Context.LocationInfo loc)        {            src.SkipSpace(); // get '('            if (src.CurrentChar != '(')                if (!NotifyError(ctx, GetLocation(loc.FirstLine, loc.FirstPos, src.AbsoluteLine, src.AbsolutePos), ParsingErrorType.SyntaxError, "'(' is expected."))                    return false;            // 擷取判斷運算式            SourceCode condition = GetParenthesisCode(src);            ControlFlow.IfThen stxIf = new ControlFlow.IfThen()            {                Location = new Context.LocationInfo()                {                    FirstLine = loc.FirstLine,                    FirstPos = loc.FirstPos                }            };            ctx.AddChild(stxIf);            //解析判斷運算式            if (!ParseExpression(stxIf, condition, ref stxIf.Condition))                return false;                        // 嘗試解析then部分代碼            src.SkipSpace();            bool res = false;            Block ThenBlock = new Block();            stxIf.AddChild(ThenBlock);            if (src.CurrentChar == '{') //塊？            {                SourceCode code = GetBlockCode(src);                res = Parse(ThenBlock, code);                stxIf.Location.LastLine = src.AbsoluteLine;                stxIf.Location.LastPos = src.AbsolutePos;            }            else            {   // 單個語句？                SourceCode stmt = new SourceCode()                {                    PosOffset = src.AbsolutePos,                    LineOffset = src.AbsoluteLine,                    Text = src.Text.Substring(src.Column)                };                res = Parse(ThenBlock, stmt);                stxIf.Location.LastLine = stmt.AbsoluteLine;                stxIf.Location.LastPos = stmt.AbsolutePos;            }            // else另案處理            return res;        }

 

再看else部分：

private bool ParseControl_Else(Context ctx, SourceCode src, Context.LocationInfo loc)        {            // else 不能單獨出現，前面必須有個if。這種處理方式也解決了else與if的就近匹配問題。            if (!(ctx.Children.Last() is ControlFlow.IfThen))                if (!NotifyError(ctx, loc, ParsingErrorType.SyntaxError, "\"else\" should not appear here."))                    return false;            // 上一個文法可能是多重if/then            Context lastStx = ctx.Children.Last();            while (lastStx.Children.Count > 2) // Children數大於2，表示具有else部分，則嘗試取得最後一個if/then            {                lastStx = lastStx.Children.Last();            }            // 再次檢測if/then            if (!(lastStx is ControlFlow.IfThen))                if (!NotifyError(ctx, loc, ParsingErrorType.SyntaxError, "Can't find matched \"if\"."))                    return false;            ControlFlow.IfThen stxIf = lastStx as ControlFlow.IfThen;            src.SkipSpace();            bool res = false;            Block elseBlock = new Block();            stxIf.AddChild(elseBlock);            // Block            if (src.CurrentChar == '{')            {                SourceCode code = GetBlockCode(src);                res = Parse(elseBlock, code);                lastStx.Location.LastLine = src.AbsoluteLine;                lastStx.Location.LastPos = src.AbsolutePos;            }            else            {   // Statement                SourceCode stmt = new SourceCode()                {                    PosOffset = src.AbsolutePos,                    LineOffset = src.AbsoluteLine,                    Text = src.Text.Substring(src.Column)                };                res = Parse(elseBlock, stmt);                lastStx.Location.LastLine = stmt.AbsoluteLine;                lastStx.Location.LastPos = stmt.AbsolutePos;            }            return res;        }