实验：实现C语言小子集程序的词法分析

实现C语言小子集程序的词法分析

（1）以表1的小语言（C语言小子集）为例实现词法分析

（2）设计单词属性值；设计表格（表示标识符表），单词符号及机内表示
（3）编码实现词法分析程序
（4）要求：
- 输入方式：文件（源文件）
- 输出方式：词法分析的结果（单词串）输出到屏幕上同时保存文件上（目标文件）。
（5）要求实现：

• 对正确源程序的识别；
• 对包含有注释//和/* */的源程序的识别；
• 对包含错误标识符和错误数的源程序的识别并作相应的错误处理。
  （6）自己设计3-5个测试实例，要求覆盖上述功能。测试实例可参看(附录A)

附录A

源代码

//词法生成器
#include<iostream>
#include<vector>
using namespace std;
#include<fstream>
#include<iomanip>
#include<map>
#include<string>
map<string, int>SYMBOL;
vector<string>VARIABLE;//保存标识符
vector<int>v;//处理行
const int num = 25500;//文本最多25500个字符
bool Flag = true;//注释
string FileName = "3.txt";
FILE *f_out = fopen("符号表.txt", "w");
FILE *f = fopen("标识符表.txt", "w");
void LoadSymbol()
{
    int sum = 0, i = 0;
    string s = "";
    char temp[num];
    ifstream f_in("C语言小子集定义表.txt");
    f_in.getline(temp, num, 0);//一次性读入所有符号，保存到temp中
    //cout << temp << endl;
    while (temp[i] != '#')
    {
        while (temp[i] != ':')
            s += temp[i], i++;
        i++;
        while (temp[i] != '\n')
            sum = sum * 10 + (temp[i] - '0'), i++;
        SYMBOL[s] = sum;
        s = "", sum = 0;
        i++;
    }
    /*map<string, int>::iterator it;
    for (it = SYMBOL.begin();it != SYMBOL.end();it++)
    {
        cout << (*it).first << ":" << (*it).second << endl;
    }*/
}
int Find(string s)//查询函数，如果查到存在此关键词便返回值，否则返回零，不包括常数以及标识符
{
    map<string, int>::iterator it;
    it = SYMBOL.find(s);
    if (it != SYMBOL.end())
    {
        return SYMBOL[s];
    }
    else
        return 0;
}
//预处理（去空格以及注释）- Tab无法处理
void PreDeal(string &s)
{
    //处理注释
    int index = 0;
    if (s.find("//") != -1)
    {
        index = s.find("//");
        s = s.substr(0, index);
        cout << s << endl;
    }
    else if (s.find("/*") != -1)
    {
        index = s.find("/*");
        if (s.find("*/") == -1)
        {
            Flag = false;
            s = "";
        }
        else
        {
            index = s.find("*/");
            s = s.substr(index + 2, s.size() - 1);
        }
    }
    else if (s.find("*/") != -1 && Flag == false)
    {
        index = s.find("*/");
        s = s.substr(index + 2, s.size() - 1);
        Flag = true;
    }

    //处理空格
    string temp = "";
    for (int i = 0;i < s.size();)
    {
        while (s[i] == ' '&&s[i + 1] == ' ')
        {
            i++;
        }
        temp += s[i];
        i++;
    }
    s = temp;
}
//处理单行语句
void Deal(string &s, int line)
{
    //fputs("adsdsa", f_out);
    PreDeal(s);
    string temp = "";
    for (int i = 0;i < s.size();)
    {
        if (s[i] == '>')
        {
            if (s[i + 1] == '=')
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL[">="]);
                fputs("->,", f_out);
                i = i + 2;
            }
            else
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL[">"]);
                fputs("->,", f_out);
                i++;
            }
        }
        else if (s[i] == '<')
        {
            if (s[i + 1] == '=')
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL["<="]);
                fputs("->,", f_out);
                i = i + 2;
            }
            else
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL["<"]);
                fputs("->,", f_out);
                i++;
            }
        }
        else if (s[i] == '=')
        {
            if (s[i + 1] == '=')
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL["=="]);
                fputs("->,", f_out);
                i = i + 2;
            }
            else
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL["="]);
                fputs("->,", f_out);
                i++;
            }
        }
        else if (s[i] == '!'&&s[i + 1] == '=')
        {
            fputs("<", f_out);
            fprintf(f_out, "%d,", SYMBOL["!="]);
            fputs("->,", f_out);
            i = i + 2;
        }
        else if (s[i] == '&'&&s[i + 1] == '&')
        {
            fputs("<", f_out);
            fprintf(f_out, "%d,", SYMBOL["&&"]);
            fputs("->,", f_out);
            i = i + 2;
        }
        else if (s[i] == '|'&&s[i + 1] == '|')
        {
            fputs("<", f_out);
            fprintf(f_out, "%d,", SYMBOL["||"]);
            fputs("->,", f_out);
            i = i + 2;
        }
        else if (Find(temp = s[i]))
        {
            fputs("<", f_out);
            fprintf(f_out, "%d,", SYMBOL[temp]);
            fputs("->,", f_out);
            i++;
        }
        else if (true)
        {
            temp = "";
            while (s[i] != ' '&&s[i] != '\n'&&s[i] != ')'&&s[i] != ';'&&s[i] != '('&&s[i] != '{'&&s[i] != '}'&&s[i] != '['&&s[i] != ']'&&s[i] != ','&&s[i] != '+'&&s[i] != '-'&&s[i] != '*'&&s[i] != '/'&&s[i] != '%'&&s[i] != '>'&&s[i] != '<'&&s[i] != '='&&s[i] != '!'&&s[i] != '&')
            {
                temp += s[i];
                i++;
                if (i == s.size())break;
            }
            if (Find(temp))
            {
                fputs("<", f_out);
                fprintf(f_out, "%d,", SYMBOL[temp]);
                fputs("->,", f_out);
            }
            else if (temp != "")
            {
                if (temp[0] >= '0'&&temp[0] <= '9' && (temp[temp.size() - 1] >= 'a'&&temp[temp.size() - 1] <= 'z' || temp[temp.size() - 1] >= 'A'&&temp[temp.size() - 1] <= 'Z'))
                {
                    fputs("LexicalError,", f_out);
                    v.push_back(line);
                }
                else if ((temp[0] >= 'a'&&temp[0] <= 'z' || temp[0] >= 'A'&&temp[0] <= 'Z'))
                {
                    fputs("<", f_out);
                    fprintf(f_out, "%d,", 1);
                    fprintf(f_out, "%s>,", temp.c_str());
                    VARIABLE.push_back(temp.c_str());
                }
                else 
                {
                    int sum = 0;
                    while (s[i] != ' '&&s[i] != ';'&&s[i] != ')'&&s[i] != ';')
                        sum = sum * 10 + (s[i] - '0'), i++;
                    fputs("<", f_out);
                    fprintf(f_out, "%d,", 2);
                    fprintf(f_out, "%s>,", temp.c_str());
                }
            }
        }
        if (s[i] == ' ' || s[i] == '\n')
            i++;
    }
}
void LoadFile()//读入文本
{
    string s = "";
    int i = 0, line = 1;
    char temp[num];
    ifstream f_in(FileName);
    f_in.getline(temp, num, 0);
    while (temp[i] != '#')
    {
        while (temp[i] != '\n')
        {
            s += temp[i];
            i++;
        }
        if (Flag == false)
        {
            if (s.find("*/") != -1)
                Flag = true;
            fputs("\n", f_out);
        }
        else
            Deal(s, line);
        fputs("\n", f_out);
        s = "", i++, line++;
    }
    if (v.size() == 0)
        fputs("LexicalError(s) 0 errors\n", f_out);
    else
    {
        fputs("LexicalError(s) on line(s) ", f_out);
        fprintf(f_out, "%d", v[0]);
        for (int j = 1;j < v.size();j++)
            fprintf(f_out, ",%d", v[j]);
        fputs("\n", f_out);
    }
    fputs("#", f_out);
    v.clear();
    fclose(f_out);
}
void Print_symbol()
{
    cout << "打印符号表:" << endl;
    char temp[num];
    ifstream f_in("符号表.txt");
    f_in.getline(temp, num, 0);//一次性读入所有符号，保存到temp中
    cout << temp << endl;
}
void Print_variable()
{
    FILE*f = fopen("标识符表.txt", "w");
    cout << endl << "打印标识符表:" << endl;
    fputs("标识符表:\n", f);
    for (int i = 0;i < VARIABLE.size();i++)
    {
        cout << fixed << setw(10) << i + 1 << ":" << VARIABLE[i] << " ";
        fprintf(f, "%d:%s\n", i + 1, VARIABLE[i].c_str());
        if ((i + 1) % 5 == 0)cout << endl;
    }
    cout << endl;
}
int main()
{
    cout << "输入文本名称：" << endl;
    cin >> FileName;
    LoadSymbol();//加载C语言小子集定义表至map中
    LoadFile();
    Print_symbol();//打印符号表
    Print_variable();//打印标识符表
    cout << endl;
    return 0;
}