【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&＃xff0c;而后得到NFA&＃xff0c;再确定化得到DFA&＃xff0c;根据DFA的转换矩阵或转换图&＃xff0c;用c&＃43;&＃43;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 &＃61; 10;double b &＃61; -20.9;if(a<&＃61;b)a&＃43;&＃61;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, &＃61;)line2:(integer, 10)line2:(bracket, ;)line3:(type, double)line3:(identify, b)line3:(OPT, &＃61;)line3:(decimal, -20.9)line3:(bracket, ;)line4:(keyword, if)line4:(bracket, ()line4:(identify, a)line4:(OPT, <&＃61;)line4:(identify, b)line4:(bracket, ))line5:(identify, a)line5:(OPT, &＃43;&＃61;)line5:(identify, b)line5:(bracket, ;)line6:(identify, a)line6:(OPT, &＃61;)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main, return, if, else, do, while, for, scanf, printf, sqrt, abs&＃xff1b;type类型包括void, int, double, float, char&＃xff1b;运算符(算术、关系、逻辑、位)&＃xff1b;需要识别的其他单词有标识符, 整数&＃xff08;十进制形式、指数形式&＃xff09;&＃xff0c;实数&＃xff08;十进制形式、指数形式&＃xff09;&＃xff0c;字符串&＃xff08;输出类型名为string&＃xff09;&＃xff1b;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&＃xff0c;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

实现代码&＃xff1a;

#include
#include
#include
#include using namespace std;vector<string> keyword &＃61; {"scanf", "printf", "if", "else", "for", "while", "return", "do", "main", "abs", "sqrt"};
vector<string> type &＃61; {"int", "void", "char", "double", "short", "float"};
vector<char> bracket &＃61; {&＃39;,&＃39;, &＃39;\\&＃39;, &＃39;;&＃39;, &＃39;:&＃39;, &＃39;(&＃39;, &＃39;)&＃39;, &＃39;[&＃39;, &＃39;]&＃39;, &＃39;{&＃39;, &＃39;}&＃39;, &＃39;"&＃39;, &＃39;\&＃39;&＃39;};struct Node
{int line &＃61; 0;string type;string word;
};vector<Node> stack;
int line &＃61; 1;
char text[int(1e5)] &＃61; "";
char ch &＃61; &＃39; &＃39;;
int len &＃61; 0;
int i &＃61; 0;
string word;
Node temp;void makeword(string s)
{temp.line &＃61; line;temp.type &＃61; s;temp.word &＃61; word;stack.push_back(temp);word.clear();
}void JudgeE()
{word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];if (ch &＃61;&＃61; &＃39;&＃43;&＃39; || ch &＃61;&＃61; &＃39;-&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];}if (ch >&＃61; &＃39;1&＃39; && ch <&＃61; &＃39;9&＃39;){word &＃43;&＃61; ch;while ((ch &＃61; text[&＃43;&＃43;i]) && (ch >&＃61; &＃39;0&＃39; && ch <&＃61; &＃39;9&＃39;))word &＃43;&＃61; ch;makeword("float");}else{cout << "Error at Line " << line << ": Illegal floating point number \"" << word << "\".\n";exit(-1);}
}void jump()
{while (ch &＃61;&＃61; &＃39; &＃39; || ch &＃61;&＃61; &＃39;\n&＃39;){if (ch &＃61;&＃61; &＃39;\n&＃39;)line&＃43;&＃43;;ch &＃61; text[&＃43;&＃43;i];}
}void makenumber()
{while (ch >&＃61; &＃39;0&＃39; && ch <&＃61; &＃39;9&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];}if (ch &＃61;&＃61; &＃39;.&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];while (ch >&＃61; &＃39;0&＃39; && ch <&＃61; &＃39;9&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];}if (ch &＃61;&＃61; &＃39;e&＃39;)JudgeE();elsemakeword("decimal");}else if (ch &＃61;&＃61; &＃39;e&＃39;)JudgeE();elsemakeword("integer");
}void fun()
{ch &＃61; text[i];jump();while (ch !&＃61; &＃39;\0&＃39; && ch !&＃61; EOF){jump();if (ch &＃61;&＃61; &＃39;&＃43;&＃39; || ch &＃61;&＃61; &＃39;-&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];if (ch &＃61;&＃61; &＃39;&＃61;&＃39;){word &＃43;&＃61; ch;makeword("OPT");ch &＃61; text[&＃43;&＃43;i];}elsemakenumber();}if (ch >&＃61; &＃39;0&＃39; && ch <&＃61; &＃39;9&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];makenumber();}if (isalnum(ch) || ch &＃61;&＃61; &＃39;_&＃39;){while (isalnum(ch) || ch &＃61;&＃61; &＃39;_&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];}if (find(keyword.begin(), keyword.end(), word) !&＃61; keyword.end())makeword("keyword");else if (find(type.begin(), type.end(), word) !&＃61; type.end())makeword("type");elsemakeword("identify");}if (ch &＃61;&＃61; &＃39;/&＃39;){char temp &＃61; text[&＃43;&＃43;i];if (temp &＃61;&＃61; &＃39;/&＃39;)do{ch &＃61; text[&＃43;&＃43;i];} while (ch !&＃61; &＃39;\n&＃39;);else if (temp &＃61;&＃61; &＃39;*&＃39;){do{ch &＃61; text[&＃43;&＃43;i];} while (ch !&＃61; &＃39;/&＃39;);ch &＃61; text[&＃43;&＃43;i];}else if (temp &＃61;&＃61; &＃39;&＃61;&＃39;){word &＃43;&＃61; "/&＃61;";makeword("OPT");ch &＃61; text[&＃43;&＃43;i];}else{word &＃43;&＃61; &＃39;/&＃39;;makeword("OPT");ch &＃61; text[&＃43;&＃43;i];}}if (ch &＃61;&＃61; &＃39;0&＃39;){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];makeword("integer");}if (ch &＃61;&＃61; &＃39;%&＃39; || ch &＃61;&＃61; &＃39;&&＃39;){if (isalnum(text[i&＃43;1])){word &＃43;&＃61; ch;word &＃43;&＃61; text[&＃43;&＃43;i];ch &＃61; text[&＃43;&＃43;i];makeword("typeidentify");}}if (find(bracket.begin(), bracket.end(), ch) !&＃61; bracket.end()){word &＃43;&＃61; ch;ch &＃61; text[&＃43;&＃43;i];makeword("bracket");}if (ch &＃61;&＃61; &＃39;*&＃39; || ch &＃61;&＃61; &＃39;&＃61;&＃39; || ch &＃61;&＃61; &＃39;<&＃39; || ch &＃61;&＃61; &＃39;>&＃39; || ch &＃61;&＃61; &＃39;!&＃39;){word &＃43;&＃61; ch;if (text[i &＃43; 1] &＃61;&＃61; &＃39;&＃61;&＃39;)word &＃43;&＃61; text[&＃43;&＃43;i];ch &＃61; text[&＃43;&＃43;i];makeword("OPT");}}
}int main()
{text[len] &＃61; getchar();while (text[len] !&＃61; &＃39;\0&＃39; && text[len] !&＃61; EOF){text[&＃43;&＃43;len] &＃61; getchar();}fun();for (vector<Node>::iterator iter &＃61; stack.begin(); iter !&＃61; stack.end(); iter&＃43;&＃43;)cout << "line" << (*iter).line << ":(" << (*iter).type << ", " << (*iter).word << &＃39;)&＃39; << endl;
}