【编译原理】手工构造词法分析器

【问题描述】通过设计c语言常见单词的正规文法或正规式，而后得到NFA，再确定化得到DFA，根据DFA的转换矩阵或转换图，用c++语言实现词法分析器。
【输入形式】输入一段完整的c语言程序
【输出形式】各类单词的token字

【样例输入】int main(){

    int a = 10;

    double b = -20.9;

    if(a<=b)        

    a+=b;

    return a;

}



【样例输出】

line1:(type, int)

line1:(keyword, main)

line1:(bracket, ()

line1:(bracket, ))

line1:(bracket, {)

line2:(type, int)

line2:(identify, a)

line2:(OPT, =)

line2:(integer, 10)

line2:(bracket, ;)

line3:(type, double)

line3:(identify, b)

line3:(OPT, =)

line3:(decimal, -20.9)

line3:(bracket, ;)

line4:(keyword, if)

line4:(bracket, ()

line4:(identify, a)

line4:(OPT, <=)

line4:(identify, b)

line4:(bracket, ))

line5:(identify, a)

line5:(OPT, +=)

line5:(identify, b)

line5:(bracket, ;)

line6:(keyword, else)

line6:(identify, a)

line6:(OPT, =)

line6:(integer, 0)

line6:(bracket, ;)

line7:(keyword, return)

line7:(identify, a)

line7:(bracket, ;)

line8:(bracket, })

【样例说明】需要识别的关键字包括void, int, main, double, return, float, if, else, do, while, for, scanf, printf, char, sqrt, abs, 运算符(算术、关系、逻辑、位)；需要识别的其他单词有标识符, 整数（十进制形式、指数形式），实数（十进制形式、指数形式），字符串；过滤注释及空格。

【评分标准】根据设计文档的质量、lex文件的正确性，代码的正确性、代码的时间空间复杂度、识别单词的种类等综合评分

构造DFA:

看起来也不是DFA

直接上代码：

#include <iostream>
#include <string.h>
#include <vector>

using namespace std;

vector<string> keyword = { "scanf","printf","if","else","for","while","return","do","main","abs","sqrt","float" };
vector<string> type = { "int","void","char","double","short" };
vector<char> bracket = { ',', '\\', ';', ':', '(', ')', '[',  ']', '{', '}', '"', '\'' };

struct Store {
    int line = 0;
    string token;
    string word;
};

vector<Store> msg;

class Compile
{
public:
    void scan();
    void print();
    void Pushmsg(string s);
    void JudgeFloat(string s);
private:
    int line = 1;
    char ch = ' ';
    string word;
    Store temp;
};
void Compile::Pushmsg(string s)
{
    temp.line = line;
    temp.token = s;
    temp.word = word;
    msg.push_back(temp);
    word.clear();
}
void Compile::JudgeFloat(string s)
{
    word += ch;
    ch = getchar();
    if (ch == '+' || ch == '-')
    {
        word += ch;
        ch = getchar();
    }

    if (ch < '1' || ch > '9')
    {
        cout << "Error at Line " << line << ": Illegal floating point number \"" << word << "\".\n";
        exit(-1);
    }
    else
    {
        word += ch;
        while ((ch = getchar()) && (ch >= '0' && ch <= '9'))
        {
            word += ch;
        }
        Pushmsg("float");
    }
}
void Compile::scan()
{
    /*读取第一个有效字符*/
    ch = getchar();
    while (ch == ' ' || ch == '\n')
    {
        if (ch == '\n')
            line++;
        ch = getchar();
    }

    while (ch != '\0' && ch != EOF)
    {
        
        while (ch == ' ' || ch == '\n')
        {
            if (ch == '\n')
                line++;
            ch = getchar();
        }
        
        //判断是否整数、小数、浮点数
        if (ch == '+' || ch == '-')
        {
            word += ch;
            ch = getchar();
            if (ch == '=')
            {
                word += ch;
                Pushmsg("OPT");
                ch = getchar();
            }
            else
                if (ch >= '1' && ch <= '9')
                {
                    word += ch;
                    while ((ch = getchar()) && (ch >= '0' && ch <= '9'))
                    {
                        word += ch;
                    }
                    Pushmsg("integer");
                }
        }

        if (ch >= '1' && ch <= '9')
        {
            word += ch;
            ch = getchar();
            while (ch >= '0' && ch <= '9')
            {
                word += ch;
                ch = getchar();
            }
            //判度小数、浮点数
            if (ch == '.')
            {
                word += ch;
                ch = getchar();
                while (ch >= '0' && ch <= '9')
                {
                    word += ch;
                    ch = getchar();
                }
                if (ch == 'e')
                {
                    JudgeFloat(word);
                }
                else
                    Pushmsg("decimal");
            }
            else
                Pushmsg("integer");
        }

        int flag = 0;
        if (ch == '_' || isalpha(ch))
        {
            while (isalnum(ch)||ch == '_')
            {
                word += ch;
                ch = getchar();
            }
            
            for (vector<string>::iterator it = keyword.begin(); it != keyword.end(); it++)
            {
                if (word == (*it))
                {
                    Pushmsg("keyword");
                    flag = 1;
                }
            }
            if (flag == 0)
            {
                for (vector<string>::iterator it = type.begin(); it != type.end(); it++)
                {
                    if (word == (*it))
                    {
                        Pushmsg("type");
                        flag = 1;
                    }
                }
                if (flag == 0)
                {
                    Pushmsg("identify");
                }
            }
            
        }
        
        if (ch == '/')
        {
            char temp = getchar();
            if (temp == '/')
            {
                do {
                    ch = getchar();
                } while (ch != '\n');
            }
            else
                if (temp == '*')
                {
                    do {
                        ch = getchar();
                    } while (ch != '/');
                    ch = getchar();
                }
                else 
                    if (temp == '=')
                    {
                        word += "/=";
                        Pushmsg("OPT");
                        ch = getchar();
                    }
                    else
                    {
                        word += '/';
                        Pushmsg("OPT");
                        ch = getchar();
                    }
        }
        // 0 开头的情况
        if (ch == '0')
        {
            word += ch; ch = getchar();
            Pushmsg("integer");
        }

        //对typeidentify进行判断
        char temp;
        if (ch == '%' || ch == '&')
        {
            temp = getchar();
            if (isalnum(temp))
            {
                word += ch;
                word += temp;
                Pushmsg("typeidentify");
                ch = getchar();
            }
        }

        int flag2 = 0;
        //对bracket进行匹配
        for (vector<char>::iterator it = bracket.begin(); it != bracket.end(); it++)
        {
            if (ch == (*it))
            {
                word += ch; ch = getchar();
                Pushmsg("bracket");
                break;
                flag2 = 1;
            }
        }
        //bracket没有匹配成功，对OPT进行判断
        if (flag2 == 0)
        {
            switch (ch)
            {
            case '*':
            case '=':
            case '<':
            case '>':
            case '!':
                temp = getchar();
                if (temp == '=') {
                    word += ch;
                    word += temp; 
                    ch = getchar();
                    Pushmsg("OPT");
                    break;
                }
                else {
                    word += ch; ch = getchar();
                    Pushmsg("OPT");
                    break;
                }
                break;
            default:
                break;
            }
        }
    }
     
}

void Compile::print()
{
    for (vector<Store>::iterator it = msg.begin(); it != msg.end(); it++)
        cout << "line" << (*it).line << ":(" << (*it).token << ", " << (*it).word << ')' << endl;

}
int main()
{
    Compile cp;
    cp.scan();
    cp.print();
}