A formula expression analysis implemented by myself

When playing games recently, there is a need to parse a formula into a sequence of pre-order traversal, and then store this sequence string in the database. In the game server, read this sequence and perform evaluation. The work of parsing this formula is placed in the game editor and I am responsible for it.

The requirements of the formula are as follows: the formula only supports + - * /. ( ) and keywords in the game world, such as role.curlv, which represent the current level of the character. These keywords are finally translated into something like this [0.12], where 0 means role and 12 means curlv attribute.

A simple formula looks like this: role.curlv * 10 / 100 + 50.

At first I thought this problem was not difficult, but when I actually started it, I found that it was very difficult. According to the characteristics of the formula, I spent 2 days to barely realize it with the stack. However, due to the length of the code, there are too many judgment conditions. It is regarded as wild by cto. At the same time, he gave me a blog and let me see his professional writing style. After reading that blog, I realized that it is a pity that I did not learn the principles of compilation in college (there is no such course in college). I can't understand the grammar derivation inside. In desperation, I googled it myself. I went to the library to borrow books and spent several days learning the basics of compilation principles. According to the method in the book, I finally wrote a class.

The following is the source code:

 

/** * @file XFormula.h * @auth Jamy * @email suzhijie325@163.com * @date 2010年11月13日 **/ #ifndef __XFORMULA_H__ #define __XFORMULA_H__ #include <string> #include <strstream> #include <vector> #include <memory.h> using std::string; using std::strstream; using std::vector; //语法树节点 struct Node { Node* left; Node* right; string data; Node():left(NULL), right(NULL) {} }; /* 公式的文法规则 exp -> term { addop teram } addop -> "+" | "-" teram -> factor { mulop factor } mulop -> "*" | "/" factor -> num | "(" exp ")" | keyword num -> { [0-9] } keywrod -> letter '.' letter letter -> { [a-z] } */ class XFormula { enum TFormulType { TFT_Keyword = 1, //关键字 TFT_Num = 2, //数值 TFT_Add = 3, // ‘+’ TFT_Sub = 4, // ‘-‘ TFT_Mul = 5, // ‘*’ TFT_Div = 6, // ‘/’ TFT_LParen = 7, // ‘(‘ TFT_RParen = 8, // ‘)’ TFT_Error = 9, // 未预期类型 TFT_Enum_Count }; public: XFormula(void); ~XFormula(void); //获取先序序列 string get_sequence(){ strstream_ << std::ends; return strstream_.str();} //把关键字 skill.curlv转换为[0.17]的形式 string get_format_by_string(string& str); //创建二叉树 void create_tree(string& str); private:///文法解析模块，用的是递归预测分析方法 //exp -> term { addop teram } Node* expression(int& cur_pos, string& str, string& outVal); // teram -> factor { mulop factor } Node* term(int& cur_pos, string& str, string& outVal); // factor -> num | "(" exp ")" | object Node* factor(int& cur_pos, string& str, string& out_val); //从str中cur_pos中获取一个记号，值在outVal中，返回值表示记号的类型。 int get_token(int& cur_pos, string& str, string& out_val); private: //从二叉树转换为先序遍历的序列 void create_sequence(Node* pnf); //获取序列的字符流 strstream strstream_; //存放指针在对象析构的时候自动释放内存。 vector<Node*> node_vec_; }; #endif// __XFORMULA_H__

 

#include "StdAfx.h" #include "XFormula.h" #include <iostream> #include <stack> #include <string> #include <fstream> using namespace std; //校验是不是数字 bool isNumber(char c) { return c >= '0' && c < '9'; } //只支持小写 bool isLetter(char c) { return c >= 'a' && c <= 'z'; } XFormula::XFormula(void) { } XFormula::~XFormula(void) { for (int i = 0; i < (int) node_vec_.size(); ++i) { delete node_vec_[i]; node_vec_[i] = NULL; } } void XFormula::create_sequence(Node* pnf) { strstream sstream; if (pnf) { char c = pnf->data[0]; if (isLetter(c)) { string s = pnf->data; string str = get_format_by_string(s); if (str.empty()) { sstream << "请检查：" << s << ends; throw sstream.str(); } strstream_ << "[" << str << "]" << ","; } else { strstream_ << pnf->data << ","; } create_sequence(pnf->left); create_sequence(pnf->right); } } //校验给定的字符串是不是合法。通过文件校验 string XFormula::get_format_by_string(string& str) { strstream sstream; string::size_type pos = str.find('.'); if (pos == str.npos) { sstream << "无效的字符串" << str << ends; throw sstream.str(); } string s1 = str.substr(0, pos); string s2 = str.substr(pos + 1); ifstream f("c:/data/object.txt"); if (f.fail()) { sstream << "打开文件c://data//object.txt失败" << ends; throw sstream.str(); } string nPos = ""; while (!f.eof()) { char temp[256] = { 0 }; f.getline(temp, 256); string l(temp); string::size_type lpos = l.find('='); if (lpos != l.npos) { string lname = l.substr(0, lpos); if (lname == s1) { nPos = l.substr(lpos + 1); break; } } } f.close(); string pPos = ""; if (!nPos.empty()) { string path = string("c:/data/").append(s1).append(".txt"); ifstream ff(path.c_str()); if (ff.fail()) { sstream << "打开文件" << path << "失败" << endl; throw sstream.str(); } while (!ff.eof()) { char temp[256] = { 0 }; ff.getline(temp, 256); string l(temp); string::size_type lpos = l.find('='); if (lpos != l.npos) { string lname = l.substr(0, lpos); if (lname == s2) { pPos = l.substr(lpos + 1); break; } } } ff.close(); } if (!pPos.empty()) { return nPos + string(".") + pPos; } else { sstream << s2 << "处错误" << endl; throw sstream.str(); } return ""; } void XFormula::create_tree(string& str) { int cur_pos = 0; string err; Node* pNode = expression(cur_pos, str, err); if (cur_pos < str.size()) { strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } create_sequence(pNode); } int XFormula::get_token(int& cur_pos, string& str, string& out_val) { out_val.clear(); while (cur_pos < (int) str.size()) { char nextChar = str[cur_pos]; if (nextChar == '/n' || nextChar == ' ' || nextChar == '/t') { cur_pos++; continue; } switch (nextChar) { case '+': cur_pos++; out_val.push_back(nextChar); return TFT_Add; case '-': cur_pos++; out_val.push_back(nextChar); return TFT_Sub; case '*': cur_pos++; out_val.push_back(nextChar); return TFT_Mul; case '/': cur_pos++; out_val.push_back(nextChar); return TFT_Div; case '(': cur_pos++; out_val.push_back(nextChar); return TFT_LParen; case ')': cur_pos++; out_val.push_back(nextChar); return TFT_RParen; default: break; } if (isLetter(nextChar) || nextChar == '.') { while ((isLetter(nextChar) || nextChar == '.') && (cur_pos < (int) str.size())) { out_val.push_back(nextChar); cur_pos++; nextChar = str[cur_pos]; } return TFT_Keyword; } if (isNumber(nextChar)) { while (isNumber(nextChar) && (cur_pos < (int) str.size())) { out_val.push_back(nextChar); cur_pos++; nextChar = str[cur_pos]; } return TFT_Num; } strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } } // exp -> term { addop teram } Node* XFormula::expression(int& cur_pos, string& str, string& out_val) { Node* pNode = term(cur_pos, str, out_val); int tokenType = TFT_Error; while (cur_pos < (int) str.size()) { if ((str[cur_pos] == '+' || str[cur_pos] == '-')) { tokenType = get_token(cur_pos, str, out_val); switch (tokenType) { case TFT_Add: case TFT_Sub: { Node* pAddOp = new Node(); node_vec_.push_back(pAddOp); pAddOp->data = outVal; pAddOp->left = pNode; pAddOp->right = term(cur_pos, str, out_val); pNode = pAddOp; } break;; default: { strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } } } else { break; } } return pNode; } // teram -> factor { mulop factor } Node* XFormula::term(int& cur_pos, string& str, string& out_val) { int tokenType = TFT_Error; Node* pNode = factor(cur_pos, str, out_val); while (cur_pos < (int) str.size()) { if ((str[cur_pos] == '*' || str[cur_pos] == '/')) { tokenType = get_token(cur_pos, str, out_val); switch (tokenType) { case TFT_Mul: case TFT_Div: { Node* pMulNode = new Node(); node_vec_.push_back(pMulNode); pMulNode->data = out_val; pMulNode->left = pNode; pMulNode->right = factor(cur_pos, str, out_val); pNode = pMulNode; } break; default: { strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } } } else { break; } } return pNode; } // factor -> num | "(" exp ")" | keyword Node* XFormula::factor(int& cur_pos, string& str, string& out_val) { Node* pNode = NULL; switch (get_token(cur_pos, str, out_val)) { case TFT_Keyword: case TFT_Num: { pNode = new Node(); node_vec_.push_back(pNode); pNode->data = out_val; } break; case TFT_LParen: { pNode = expression(cur_pos, str, out_val); if (get_token(cur_pos, str, out_val) != TFT_RParen) { strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } break; } default: { strstream sstream; sstream << "位置:" << cur_pos << ",字符:/"" << str[cur_pos - 1] << "/"附近有错" << ends; throw sstream.str(); } } return pNode; }

 

Instructions:

int _tmain(int argc, _TCHAR* argv[]) { string str = "1+skill.curlv+1+2*2+3"; cout << "表达式:" << str << endl; XFormula f; int i = 0; try { f.create_tree(str); string ret = f.get_sequence(); cout << ret << endl; } catch (char* s) { cout << s << endl; } return 0; }

Compile and pass under vs2005. In vs2010, due to the upgrade of sting in vs2010, it will lead to out of bounds.

operation result: