词法分析器就是通过扫描一段程序判断是否是关键字、标识符、常数、分界符、运算符。一般分为一符一种和经典五中;
这里我用的是经典五中,此词法分析器是用c++编写的;
/*
保留字|关键字:1
操作符|运算符:2
分界符:3
标识符:4
常数:5
无识别:6
*/
主要代码为:
#include<iostream> using namespace std; #define MAX 10 /* 保留字|关键字:1 操作符|运算符:2 分界符:3 标识符:4 常数:5 无识别:6 */ char ch = ' '; char* keyWord[10] = {"void","main","break","include","begin","end","if","else","while","switch"}; char token[20];//定义获取的字符 //判断是否是关键字 bool isKey(char * token) { for(int i = 0;i < MAX;i++) { if(strcmp(token,keyWord[i]) == 0) return true; } return false; } //判断是否是字母 bool isLetter(char letter) { if((letter >= 'a' && letter <= 'z')||(letter >= 'A' && letter <= 'Z')) return true; else return false; } //判断是否是数字 bool isDigit(char digit) { if(digit >= '0' && digit <= '9') return true; else return false; } //词法分析 void analyze(FILE *fpin) { while((ch = fgetc(fpin)) != EOF){ if(ch == ' '||ch == '\t'||ch == '\n'){} else if(isLetter(ch)){ char token[20]={'\0'}; int i=0; while(isLetter(ch)||isDigit(ch)){ token[i] = ch; i++; ch = fgetc(fpin); } //回退一个指针 fseek(fpin,-1L,SEEK_CUR); if(isKey(token)){ //关键字 cout<<token<<"\t1"<<"\t关键字"<<endl; } else{ //标识符 cout<<token<<"\t4"<<"\t标识符"<<endl; } } else if(isDigit(ch)||(ch == '.')) { int i=0; char token[20]={'\0'}; while(isDigit(ch)||(ch == '.'&&isDigit(fgetc(fpin)))) { if(ch == '.')fseek(fpin,-1L,SEEK_CUR); token[i] = ch; i++; ch = fgetc(fpin); } fseek(fpin,-1L,SEEK_CUR); //属于无符号常数 cout<<token<<"\t5"<<"\t常数"<<endl; } else switch(ch){ //运算符 case '+':{ ch = fgetc(fpin); if(ch == '+')cout<<"++"<<"\t2"<<"\t运算符"<<endl; else { cout<<"+"<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; case '-':{ ch = fgetc(fpin); if(ch == '-')cout<<"--"<<"\t2"<<"\t运算符"<<endl; else { cout<<"-"<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; case '*':cout<<ch<<"\t2"<<"\t运算符"<<endl;break; case '/':cout<<ch<<"\t2"<<"\t运算符"<<endl;break; //分界符 case '(':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case ')':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case '[':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case ']':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case ';':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case '{':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; case '}':cout<<ch<<"\t3"<<"\t分界符"<<endl;break; //运算符 case '=':{ ch = fgetc(fpin); if(ch == '=')cout<<"=="<<"\t2"<<"\t运算符"<<endl; else { cout<<"="<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; case ':':{ ch = fgetc(fpin); if(ch == '=')cout<<":="<<"\t2"<<"\t运算符"<<endl; else { cout<<":"<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; case '>':{ ch = fgetc(fpin); if(ch == '=')cout<<">="<<"\t2"<<"\t运算符"<<endl; else { cout<<">"<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; case '<':{ ch = fgetc(fpin); if(ch == '=')cout<<"<="<<"\t2"<<"\t运算符"<<endl; else { cout<<"<"<<"\t2"<<"\t运算符"<<endl; fseek(fpin,-1L,SEEK_CUR); } }break; //无识别 default: cout<<ch<<"\t6"<<"\t无识别符"<<endl; } } } int main(){ char input[30]; FILE *fpin; cout<<"请输入源文件名:\n"<<endl; for(;;){ cin>>input; if((fpin = fopen(input,"r")) != NULL) break; else cout<<"路径输入错误"<<endl; } cout<<"****************词法分析结果********************"<<endl; analyze(fpin); fclose(fpin); }
运行结果:
来源:https://www.cnblogs.com/ya-qiang/p/8987926.html