词法分析器

  1 // lex_analyzer.cpp : 定义控制台应用程序的入口点。
  2 //词法分析器
  3 /*
  4 sys:
  5 1-10:关键字
  6 10:标识符
  7 11:数字
  8 //比较区
  9 20:<
 10 21:<=
 11 22:>
 12 23:>=
 13 24:=
 14 25:==
 15 //运算区
 16 26:*
 17 27:/
 18 28:+
 19 29:-
 20 //分隔符号区
 21 30:;
 22 31:(
 23 32:)
 24 33:{
 25 34:}
 26 35:\n
 27 -1
 28 1000:#
 29 */
 30 #include "stdafx.h"
 31 #include <iostream>
 32 #include <sstream>
 33 #include <string>
 34 using namespace std;
 35 string ss;
 36 int scanner();
 37 const string key[] = {"int","main","return","if","for"};//定义关键字
 38 stringstream prog;
 39 int _tmain(int argc, _TCHAR* argv[])
 40 {
 41     char ch = NULL;
 42     stringstream s;
 43     int row=0;//行号
 44     int id_code;//识别码
 45     cout << "please input some C/C++ code" << endl;
 46     while (ch != ‘#‘)
 47     {
 48         cin.get(ch);
 49         s << ch;
 50     }
 51     ss = s.str();
 52     do{
 53          id_code= scanner();
 54          switch (id_code)
 55          {
 56          case -1: cout << "Error in row " << row << "!" << endl; break;
 57          case 35: row = row++; break;
 58          case 1000:break;
 59          default: cout << "(" << id_code << "," << prog.str() << ")" << endl; break;
 60              break;
 61          }
 62     } while (id_code !=1000);
 63     return 0;
 64 }
 65 int scanner()
 66 {
 67     static int  point=0;
 68     prog.str("");//清空
 69     char cha = NULL;
 70     int syn;
 71     //过滤掉空格
 72     do
 73     {
 74         cha = ss[point++];
 75     } while (cha == ‘ ‘);
 76     //识别标识符或关键字，标识符以_或字母开头
 77     if (cha >= ‘a‘&&cha <= ‘z‘ || cha >= ‘A‘&&cha <= ‘Z‘ || cha == ‘_‘)
 78     {
 79         while (cha >= ‘a‘&&cha <= ‘z‘ || cha >= ‘A‘&&cha <= ‘Z‘ || cha == ‘_‘ || cha >= ‘0‘&&cha <= ‘9‘)
 80         {
 81             prog << cha;
 82             cha = ss[point++];
 83         }
 84         --point;
 85         syn = 10;
 86             //识别关键字
 87         for (int i = 0; i < sizeof(key) / sizeof(key[0]); i++)
 88         {
 89             if (prog.str() == key[i])
 90                 syn = i;
 91         }
 92     }
 93     //识别数字
 94     else if (cha >= ‘0‘&&cha <= ‘9‘)
 95     {
 96         while (cha >= ‘0‘&&cha <= ‘9‘)
 97         {
 98             prog << cha;
 99             cha = ss[point++];
100         }
101         --point;
102         syn = 11;
103     }
104     else switch (cha)
105     {
106     case ‘<‘:
107         prog << cha;
108         cha = ss[point++];
109         if (cha == ‘=‘)
110         {
111             syn = 21;
112             prog << cha;
113         }else {
114             syn = 20;
115             --point;
116         }
117         break;
118     case ‘>‘:
119         prog << cha;
120         cha = ss[point++];
121         if (cha == ‘=‘)
122         {
123             syn = 23;
124             prog << cha;
125         }
126         else {
127             syn = 22;
128             --point;
129         }
130         break;
131     case ‘=‘:
132         prog << cha;
133         cha = ss[point++];
134         if (cha == ‘=‘)
135         {
136             syn = 25;
137             prog << cha;
138         }
139         else {
140             syn = 24;
141             --point;
142         }
143         break;
144     case‘*‘:syn = 26; prog << cha; break;
145     case‘/‘:syn = 27; prog << cha; break;
146     case‘+‘:syn = 28; prog << cha; break;
147     case‘-‘:syn = 29; prog << cha; break;
148     case‘;‘:syn = 30; prog << cha; break;
149     case‘(‘:syn = 31; prog << cha; break;
150     case‘)‘:syn = 32; prog << cha; break;
151     case‘{‘:syn = 33; prog << cha; break;
152     case‘}‘:syn = 34; prog << cha; break;
153     case‘\n‘:syn = 35; break;
154     case ‘#‘:syn = 1000; prog << cha; break;
155     default: syn = -1; break;
156     }
157     return syn;
158 }