编译原理小型C语言词法分析器

#include<iostream>

#include<string>

#include<map>

using namespace std;

map<int, string>Key_words, Symbols;//存储关键字，存储算符和界符

string S[1000], Ans[1000];//存储输入，输出

int cnt = 0, wrong_cnt = 0, wrong_line[1000];//存储有效行数量，错误行数量，错误行记录

bool f = 1;//注释标记

void Initialize_Key_words() {//初始化关键字

 Key_words[3] = "void", Key_words[4] = "int", Key_words[5] = "float", Key_words[6] = "char", Key_words[7] = "if", Key_words[8] = "else",

  Key_words[9] = "while", Key_words[10] = "do", Key_words[11] = "for", Key_words[12] = "return", Key_words[13] = "main";

}

void Initialize_Symbols() {//初始化算符和界符

 Symbols[14] = "+", Symbols[15] = "-", Symbols[16] = "*", Symbols[17] = "/", Symbols[18] = "%", Symbols[19] = "<",

  Symbols[20] = "<=", Symbols[21] = ">", Symbols[22] = ">=", Symbols[23] = "==", Symbols[24] = "!=", Symbols[25] = "&&",

  Symbols[26] = "||", Symbols[27] = "=", Symbols[28] = "(", Symbols[29] = ")", Symbols[30] = "[", Symbols[31] = "]",

  Symbols[32] = "{", Symbols[33] = "}", Symbols[34] = ";", Symbols[35] = ",";

}

int isKey_words(string s) {//判断是否为关键字

 for (int i = 3; i <= 13; i++)if (Key_words[i] == s)return i;

 return -1;

}

int isSymbol(string s) {//判断是否为算符和界符

 for (int i = 14; i <= 35; i++)if (Symbols[i] == s)return i;

 return -1;

}

string isNum(string s) {//判断是否为数字

 for (int i = 0; i < s.size(); i++)

  if (s[i] >= '0' && s[i] <= '9');

  else return "-1";

 return s;

}

string isLetter(string s) {//判断是否为字母

 for (int i = 0; i < s.size(); i++)

  if ((s[i] >= 'a' && s[i] <= 'z') || (s[i] >= 'A' && s[i] <= 'Z'));

  else return "-1";

 return s;

}

string isIdentifier(string s) {//判断是否是标识符

 if ((s[0] >= 'a' && s[0] <= 'z') || (s[0] >= 'A' && s[0] <= 'Z') || s[0] == '_')return s;

 else return "-1";

}

void check(string temp, int cnt) {//检测每个单词并存储答案

 string print;

 if (isKey_words(temp) != -1)print += "<", print += to_string(isKey_words(temp)), print += ",->,";

 else if (isSymbol(temp) != -1)print += "<", print += to_string(isSymbol(temp)), print += ",->,";

 else if (isNum(temp) != "-1")print += "<2,", print += temp, print += ">,";

 else if (isLetter(temp) != "-1")print += "<1,", print += temp, print += ">,";

 else {

  if (isIdentifier(temp) != "-1")print += "<1,", print += temp, print += ">,";

  else print += "LexicalError,", wrong_line[wrong_cnt++] = cnt + 1;

 } 

 Ans[cnt] += print;

}

bool fun(int cnt) {//词法分析处理

 //预处理单词

 string temp;

 for (int i = 0; i < S[cnt].size(); i++) {

  string ch;

  ch += S[cnt][i];//提取每个字符

  if (isSymbol(ch) != -1) {//将界符和算符分成单独的单词

   if (ch == "<" || ch == ">" || ch == "=") {//特判字符长度为二的界符和算符

    if (i + 1 < S[cnt].size() && S[cnt][i + 1] == '=' && f)temp += ' ', temp += ch, temp += '=', temp += ' ', i++;

    else if (f)temp += ' ', temp += ch, temp += ' ';

   }

   else if (ch == "/") {//特判注释

    if (i + 1 < S[cnt].size() && S[cnt][i + 1] == '/')goto loop;//遇到"//"就跳过后面单词的识别

    else if (i + 1 < S[cnt].size() && S[cnt][i + 1] == '*')f = 0;//遇到"/*"就封锁输入

   }

   else if (ch == "*" && i + 1 < S[cnt].size() && S[cnt][i + 1] == '/')f = 1;//直到遇到"*/"正常输入

   else if (f)temp += ' ', temp += ch, temp += ' ';

  }

  else if (ch == "!" && i + 1 < S[cnt].size() && S[cnt][i + 1] == '=' && f)temp += ' ', temp += ch, temp += '=', temp += ' ', i++;//特判!=

  else if (ch == "&" && i + 1 < S[cnt].size() && S[cnt][i + 1] == '&' && f)temp += ' ', temp += ch, temp += '&', temp += ' ', i++;//特判&&

  else if (ch == "|" && i + 1 < S[cnt].size() && S[cnt][i + 1] == '|' && f)temp += ' ', temp += ch, temp += '|', temp += ' ', i++;//特判||

  else if (f)temp += S[cnt][i];

 }

loop:

 temp += ' '; if (temp == " ")return 0;

 //拆单词

 string slice;

 for (int i = 0; i < temp.size(); i++) {

  if (temp[i] != ' ')slice += temp[i];//遇到连续非空字符就提取成一个单词

  else if (slice.size() != 0) {

   check(slice, cnt);//检测该单词

   slice = "";

  }

 }

 return 1;

}

int main() {

 Initialize_Key_words(), Initialize_Symbols();//初始化

 while (1) {

  getline(cin, S[cnt]);//按行输入

  if (S[cnt] == "#")break;

  if (S[cnt].size() != 0 && fun(cnt))cnt++;//去掉注释后，如果该行字符长度不为零，则行数加以

 }

 for (int i = 0; i < cnt; i++)cout << Ans[i] << endl;//输出答案

 if (wrong_cnt != 0) {//如果有错误行，输出错误行序号

  cout << "LexicalError(s) on line(s) ";

  for (int i = 0; i < wrong_cnt; i++)cout << wrong_line[i] << ",";

 }

 return 0;

}