CompilerDesignIFLR1/CompilerDesignIflr1/LexicalAnalysis.cs

using System;
using System.Collections.Generic;
using System.Linq;
using System.Text;
using System.Text.RegularExpressions;
using System.Threading.Tasks;
using CompilerDesignIFlr1;

namespace CompilerDesignIflr1
{
    internal class LexicalAnalysis
    {
        string Text { get; set; } = "";
        Dictionary<string, LR1Unit> Symbol = [];
        Dictionary<string, LR1Unit> Keyword = [];
        Dictionary<string, LR1Unit> Patterns = [];
        internal List<LR1Unit> Result;

        internal LexicalAnalysis(LR1Creator creator, string codeFilePath)
        {
            foreach (var (_, unit) in creator.TokenUnit)
            {
                var val = unit.Grammar[0];
                if (val.Contains('['))
                    Patterns.Add(val, unit);
                else if (val.Select(x => char.IsLetterOrDigit(x)).All(x => x))
                    Keyword.Add(val, unit);
                else
                    Symbol.Add(val, unit);
            }
            PrintStrSplitResult();
            var list = StrSplit(File.ReadAllText(codeFilePath));
            foreach (string s in list)
                Console.WriteLine(s);
            Result = Analyze(list);
            PrintAnalyzeResult();
        }

        internal Stack<LR1Unit> GetStack()
        {
            Stack<LR1Unit> stack = [];
            for (int i = 1; i <= Result.Count; i++)
            {
                stack.Push(Result[^i]);
            }
            return stack;
        }

        internal void PrintStrSplitResult()
        {
            Console.WriteLine("Symbol");
            foreach (var (a, b) in Symbol)
                Console.WriteLine($"{a}: {b}");
            Console.WriteLine("keyword");
            foreach (var (a, b) in Keyword)
                Console.WriteLine($"{a}: {b}");
            Console.WriteLine("Regex");
            foreach (var (a, b) in Patterns)
                Console.WriteLine($"{a}: {b}");
        }

        internal void PrintAnalyzeResult()
        {
            Console.WriteLine("units");
            foreach (var unit in Result)
                Console.WriteLine($"{unit.Name}: {unit.Value}");
        }

        public List<LR1Unit> Analyze(List<string> list)
        {
            List<LR1Unit> ans = [];
            for (int i = 0; i < list.Count; i++)
            {
                string s = list[i];
                if (Keyword.TryGetValue(s, out var unit))
                {
                    var temp = unit.Clone();
                    temp.Value = s;
                    ans.Add(temp);
                }
                else if(i + 1 < list.Count && Symbol.TryGetValue(s + list[i + 1], out var symbol))
                {
                    var temp = symbol.Clone();
                    temp.Value = s + list[i + 1];
                    i++;
                    ans.Add(temp);
                }
                else if (Symbol.TryGetValue(s, out symbol))
                {
                    var temp = symbol.Clone();
                    temp.Value = s;
                    ans.Add(temp);
                }
                else
                {
                    bool noAnswer = true;
                    foreach (var (pattern, ut) in Patterns)
                    {

                        if (Regex.IsMatch(s, '^' + pattern + '$'))
                        {
                            var temp = ut.Clone();
                            temp.Value = s;
                            ans.Add(temp);
                            noAnswer = false;
                            break;
                        }
                    }
                    if (noAnswer)
                    {
                        Console.WriteLine($"无法识别的字符串: {s}");
                    }
                }
            }
            return ans;
        }

        public List<string> StrSplit(string s)
        {
            List<string> lt = new List<string>();
            int l = 0;
            int r = 0;

            while (r < s.Length)
            {
                if (!char.IsLetterOrDigit(s[r]))
                {
                    string k = s.Substring(l, r - l).Trim();
                    if (k.Length != 0)
                    {
                        lt.Add(k);
                    }
                    k = s[r].ToString().Trim();

                    if (k.Length != 0)
                    {
                        if(k is "\'" or "\"")
                        {
                            l = r;
                            int temp = r + 1;
                            while (temp < s.Length && s[temp].ToString().Trim() != k)
                                temp++;
                            lt.Add(s.Substring(l, temp - l + 1));
                            r = temp;
                        }
                        else
                            lt.Add(k);
                    }
                    l = r + 1;
                    r = l;
                }
                else
                {
                    r += 1;
                }
            }

            if (l < s.Length)
            {
                lt.Add(s.Substring(l, r - l));
            }
            lt.Add("#");
            return lt;
        }
    }
}