自己实现的一个公式表达式解析

最近在做游戏的时候，有个需求，需把一个公式解析为先序遍历的序列，再把这个序列字符串存放到数据库中，在游戏服务器中，读取这个序列，并进行求值而。解析这个公式的工作就放在游戏编辑器的中，并由本人负责。

公式的需求如下：公式只支持+ - * /. ( ) 还有游戏世界中的关键字，比如role.curlv，表示角色当前的等级。这些关键字最后被翻译为像这样[0.12]的形式，其中0表示role， 12表示curlv属性。

简单的公式像这个样子:role.curlv * 10 / 100 + 50。

开始以为这个问题不难，然而在真正动手的时候才发现困难重重，自己根据公式的的特征，花2天的时间用堆栈勉强的实现了。但由于代码冗长，判断条件过多。被cto视为野路派的。同时他给了我一篇博客，让我看看人家专业的写法。看完了那篇博客之后，我才发现，大学时候没有学过编译原理（大学就没有这门课程）是一件很遗憾的事儿。看里面的文法推导完全看不懂。无奈之下，自己google之。去图书馆借书，花了好几天在学习编译原理的基础。照着书上的方法终于写出了一个类。

以下是源码：

/** * @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; }

使用方法:

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; }

在vs2005下编译通过。在vs2010中由于vs2010中对sting的升级，会导致越界。

运行结果: