XML在web的應用是很廣泛的，但對於普通c++程序員恐怕用得不多，xml靈活的格式使得一些設置文件描述變得很容易，但是應用他總是困難的，網絡上XML解析器龐大的嚇人，如果爲了解析很簡單的XML也要學習一大堆庫有點恐怖，反正我是沒興趣，用MSXML版本的不同也很煩人，COM的應用也很囉嗦，所以我寫了一個簡單的解析器，可以供簡單應用，高手就不要看了，這個解析器只支持XML的子集如下：（用於設置文件是足夠了，還可以擴展）
<?xml?>
<根元素>
<元素 屬性列/>
<元素 屬性列/>
<元素 屬性列/>文本</元素>
<元素><子元素>文本</子元素><子元素>文本</子元素></元素>
</根元素>
可以解析成
<SXmlUnknow>
<SXMLElement>
<SXMLElement 屬性列/>
<SXMLElement 屬性列/>
<SXMLElement 屬性列/>SXMLText</元素>
<SXMLElement><SXMLElement>SXMLText</SXMLElement><SXMLElement>SXMLText</SXMLElement></SXMLElement>
</SXMLElement>
例：
#include <iostream>#include "sxml.h"
/*<?xml?>
<書店> <書本 書名="VC++" 價格="50" 折扣="1"/> <書本 書名="VB" 價格="50" 折扣="0.8"/> <書本 書名="C#" 價格="50" 折扣="0.7">有源程序光盤</書本> <掛曆><年份>2006</年份><價格>50</價格></掛曆>
</書店>*/int main(int argc, char *argv[]){ try
{
SXmlDOM dom; //dom.parse("<?xml?><書店><書本 書名="VC++" 價格="50" 折扣="1"/><書本 書名="VB" 價格="50" 折扣="0.8"/><書本 書名="C#" 價格="50" 折扣="0.7">有源程序光盤</書本><掛曆><年份>2006</年份><價格>50</價格></掛曆></書店>") dom.parseFile("test.xml"); //查詢 SXmlElement& root = dom.root(); cout << "有" << root.count("書本") << "本書!"<< endl; cout << "VB的價格:" << (int)root.item("書本", 1)["價格"] * (float)root.item("書本", 1)["折扣"] << endl; cout << root.at(0).xml(); SXmlElement& gl = root.item("掛曆"); cout << gl.item("年份").text() << endl; //遍歷 SXmlNodeList::iterator iter; for(iter = root.begin(); iter!=root.end(); iter++) cout << (*iter)->xml(); //賦值 root.item("書本", 1)["價格"] = 60; root.item("書本", 1)["折扣"] = 0.5; cout << "VB的價格:" << (int)root.item("書本", 1)["價格"] * (float)root.item("書本", 1)["折扣"] << endl; root.item("書本").text() = "有軟盤"; //保存 dom.saveFile("else.xml");//屬性列的順序將自動排列, 採用了C++ map<>
} catch(exception &e) { cout << e.what() << endl; } return 0;} 解析器源文件:SXML.h
/*
支持XML的子集只支持元素、屬性和文本，可用於簡單的XML設置文件的解析，採用標準C++不依賴平臺及編譯器
支持類似文檔：
<SXMLElement>
<SXMLElement 屬性列/>
<SXMLElement 屬性列/>
<SXMLElement 屬性列/>SXMLText</元素>
<SXMLElement><SXMLElement>SXMLText</SXMLElement><SXMLElement>SXMLText</SXMLElement></SXMLElement>
</SXMLElement>
提供兩個函數用於實體引用,
SXmlEntityXml 用於 "</">&/'" ======> {"<", """, ">", "&", "'"}
SXmlEntityParse 用於 {"<", """, ">", "&", "'"} =====> "</">&/'"
測試版 llbird [email protected] http://blog.csdn.net/wujian53
*/
#ifndef _SXML_H
#define _SXML_H
#pragma warning(disable: 4530)
#pragma warning(disable: 4786)
#include <map>
#include <iterator>
#include <fstream>
#include <vector>
#include <string>
#include <algorithm>
#include <sstream>
#include <exception>
#include <list>
using namespace std;
struct AttributeType
{
string _Text;
template<typename T>
T operator=(T val){
stringstream ss;
ss << val;
_Text = ss.str();
return val;
}
AttributeType& operator=(AttributeType& val){
_Text = val._Text;
return val;
}
operator const char*() { return _Text.c_str(); }
operator string() { return _Text; }
operator int() { return atoi(_Text.c_str()); }
operator long() { return atol(_Text.c_str()); }
operator float() { return atof(_Text.c_str()); }
operator double() { return atof(_Text.c_str()); }
};
inline ostream& operator<<(ostream &out, AttributeType x){ return out << x._Text;}
inline istream& operator>>(istream &in, AttributeType x) { return in >> x._Text; }
#ifdef _USE_STRING
typedef string ValueType;
#else
typedef AttributeType ValueType;
#endif
enum SXML_Type{ XML_UNKNOW, XML_ELEMENT, XML_PI, XML_COMMENT, XML_Text, XML_CHAR_DATA, XML_DOC_Type, XML_DOCUMENT};
///異常
class SXmlExecption : public exception
{
string _Src;
public:
SXmlExecption(string s) : _Src(s) { }
~SXmlExecption()throw() { }
const char* what() const throw() { return _Src.c_str(); }
};
//除去頭尾空格
inline void trim_c_str(const char *&fptr, const char *&lptr)
{
while(fptr<lptr && isspace(*fptr)) fptr++;
while(fptr<lptr && isspace(*(lptr-1))) lptr--;
}
//除去頭尾空格
inline string& trim_string(string& str)
{
string::size_type pos;
for(pos=0; pos<str.length() && isspace(str.at(pos)); pos++);
str.erase(0, pos);
for(pos=str.length()-1; pos>=0 && isspace(str.at(pos)); pos--);
str.erase(pos + 1);
return str;
}
///將字符轉換爲實體引用 '<' --> "<"
inline string SXmlEntityXml(string str)
{
char *entity[] = {"lt", "quot", "gt", "amp", "apos"};
char *ref = "</">&/'", *f;
for(string::size_type pos=0; pos < str.length(); pos++)
if((f = strchr(ref, str[pos])))
str.replace(pos, 1, string("&") + entity[f-ref] + ";");
return str;
}
///將實體引用轉換爲字符 "<" --> '<'
inline string SXmlEntityParse(string str)
{
char *entity[] = {"lt", "quot", "gt", "amp", "apos"};
char *ref = "</">&/'", i;
string r;
for(string::size_type pos = 0, bpos, epos, rlen = 1;
(bpos=str.find('&', pos))!=string::npos; pos = bpos + rlen)
{
epos = str.find(';', bpos);
if(epos == string::npos)
throw SXmlExecption("找不到實體引用的右邊界';'");
r.assign(str.begin()+bpos+1, str.begin()+epos);
for(i = 0; i < 5; i++)
if(r == entity[i])
{
str.replace(bpos, epos - bpos + 1, 1, ref[i]);
break;
}
if(i == 5)
throw SXmlExecption("不支持的實體引用!");
}
return str;
}
///屬性列
struct SXmlAttribute
{
map<string, ValueType> _Map;
void clear() { _Map.clear(); }
map<string, ValueType>::size_type size() { return _Map.size(); }
ValueType& value(string key) { return _Map[key]; }
string xml()
{
string str;
for(map<string, ValueType>::iterator iter=_Map.begin();
iter!=_Map.end(); iter++)
str += " " + iter->first + "=/"" + (string)(iter->second) + "/"";
return str;
}
///屬性列處理: key1="val_1" key2="val_2" key3="&#XXX;"
void parse(const char *first, const char *last)
{
trim_c_str(first, last);
for(const char *ptr=first, *fptr, *L, *R; ptr < last; ptr=R+1)
{
if((fptr = find(ptr, last, '='))==last)
break;
trim_c_str(ptr, fptr);
if((L = find(fptr+1, last, '/"')) != last)
R = find(L+1, last, '/"');
if(L==last || R==last)
break;
_Map[string(ptr, fptr)] = string(L+1, R);
}
}
};
///節點基類
struct SXmlNode
{
SXML_Type _Type;
SXmlNode* _pParent;
string _Text;
SXmlNode(SXmlNode *parent=NULL) : _pParent(parent) { }
virtual ~SXmlNode(){}
SXmlNode* get_parent() { return _pParent; }
SXML_Type& type() { return _Type; }
string& operator=(const char *s) { return _Text=s, _Text; }
string& operator=(string& s) { return _Text=s, _Text; }
virtual string xml() { return _Text; } //>內存數據轉換爲XML
operator string&() { return _Text; }
void parse(const char *p) { parse(p, p+strlen(p)); } //>解析XML
void parse(const char *f, const char *l) { _Text.assign(f, l); }
};
///字符塊節點
struct SXmlText : SXmlNode
{
SXmlText(SXmlNode *parent=NULL) : SXmlNode(parent) { _Type = XML_Text; }
};
///其他節點
struct SXmlUnknow : SXmlNode
{
SXmlUnknow(SXmlNode *parent=NULL) : SXmlNode(parent) { _Type = XML_UNKNOW; }
};
#ifdef _USE_LIST
typedef list<SXmlNode*> SXmlNodeList ;
#else
typedef vector<SXmlNode*> SXmlNodeList ;
#endif
///元素節點
struct SXmlElement : public SXmlNode
{
SXmlNodeList _pNodes;
SXmlAttribute _Attributes;
typedef SXmlNodeList::iterator iterator;
SXmlElement(SXmlNode *parent=NULL) : SXmlNode(parent) { _Type = XML_ELEMENT; }
virtual ~SXmlElement() { clear(); }
string& tag() { return _Text; }
SXmlNodeList::size_type size() { return _pNodes.size(); }
SXmlNodeList& nodes() { return _pNodes; }
iterator begin() { return _pNodes.begin(); }
iterator end() { return _pNodes.end(); }
ValueType& value(string key) { return _Attributes.value(key); }
ValueType& operator[](string key) { return _Attributes.value(key); }
SXmlElement& operator =(const SXmlElement& x)
{
_Text = x._Text;
_pNodes = x._pNodes;
return *this;
}
SXmlNode& at(SXmlNodeList::size_type n = 0)
{
if(n<0 || n>=size())
throw SXmlExecption("子節點編號超出範圍!");
SXmlNodeList::iterator iter = _pNodes.begin();
advance(iter, n);
return *(*iter);
}
SXmlNodeList::size_type count()
{
return size();
}
SXmlNodeList::size_type count(string tag_name)
{
SXmlNodeList::iterator iter;
SXmlNodeList::size_type i;
for(iter = _pNodes.begin(), i = 0; iter != _pNodes.end(); iter++)
if((*iter)->_Type==XML_ELEMENT&& (*iter)->_Text==tag_name )
i++;
return i;
}
SXmlElement& newItem(string tag_name)
{
string &str = trim_string(tag_name);
if(!str.length())
throw SXmlExecption("元素節點標籤不應爲空字符串!");
SXmlElement* p = new SXmlElement(this);
p->tag() = str;
_pNodes.push_back((SXmlNode*)p);
return (*p);
}
SXmlElement& item(string name, SXmlNodeList::size_type n = 0)
{
SXmlNodeList::iterator iter;
SXmlNodeList::size_type i;
for(iter = _pNodes.begin(), i = 0; iter != _pNodes.end(); iter++)
if((*iter)->_Type==XML_ELEMENT&& (*iter)->_Text==name && i++==n)
return *((SXmlElement*)(*iter));
throw SXmlExecption("找不到/"" + name +"/"子節點!");
}
string& text()
{
for(SXmlNodeList::iterator iter = _pNodes.begin(); iter != _pNodes.end(); iter++)
{
if((*iter)->_Type == XML_Text)
return (string&)(*((SXmlText*)(*iter)));
}
SXmlNode* p = (SXmlNode*)new SXmlText(this);
_pNodes.insert(_pNodes.begin(), 1, p);
return (string&)(*p);
}
///清除所有節點
void clear()
{
_Attributes.clear();
for(SXmlNodeList::iterator iter=_pNodes.begin(); iter!=_pNodes.end(); iter++)
delete (*iter);
_pNodes.clear();
}
virtual string xml()
{
string str;
if(type() == XML_ELEMENT)
{
str = string("<") + _Text;
if(_Attributes.size())
str += _Attributes.xml();
if(_pNodes.size())
{
SXmlNodeList::iterator iter;
for(iter=_pNodes.begin(), str += '>'; iter!=_pNodes.end(); iter++)
str += (*iter)->xml();
str += "</" + _Text + ">";
}
else
str += "/>";
}
else
if(_pNodes.size())
{
SXmlNodeList::iterator iter;
for(iter=_pNodes.begin(); iter!=_pNodes.end(); iter++)
str += (*iter)->xml();
}
return str;
}
void parse(const char *str)
{
parse(str, str + strlen(str));
}
void parse(const char *first, const char *last)
{
const char *ptr = first;//ptr:base pointer
const char *fptr, *nptr, *tptr; //fptr:find pointer nptr:next pointer tptr : temp pointer
SXmlNode *newptr;//create SXmlNode
while(ptr < last)
{
fptr = find(ptr, last, '<');
if(ptr != fptr)
{
newptr = new SXmlText(this);
newptr->_Text.assign(ptr, fptr);
_pNodes.push_back(newptr);
}
if(fptr == last)
break;
nptr = find(fptr, last, '>');
if(nptr == last)
throw SXmlExecption(string(fptr, nptr+1) + "找不到標籤的右邊界'>'");
switch(*(fptr + 1))
{
case '?':
newptr = new SXmlUnknow(this);
newptr->parse(fptr, nptr + 1);
_pNodes.push_back(newptr);
ptr = nptr + 1;
break;
case '!':
if(!(*(fptr + 2)=='-' && *(fptr + 3)=='-'))//不是尋常的註釋
{
const char *cdata_L = "CDATA", *cdata_R = "]]>",*doc_Type_L = "DOCTYPE", *doc_Type_R = "]>";
if((tptr = search(fptr, nptr, cdata_L, cdata_L+4)) != nptr)
if((tptr = search(tptr, last, cdata_R, cdata_R+3)) != last)
nptr = tptr + 2;
else
throw SXmlExecption("CDataSection can not find /"]]>/"");
else
if((tptr = search(fptr, nptr, doc_Type_L, doc_Type_L+7)) != nptr)
{
if((tptr = search(tptr, last, doc_Type_R, doc_Type_R+2)) != last)
nptr = tptr + 1;
else
throw SXmlExecption("DOCTYPE can not find /"]>/"");
}
}
newptr = new SXmlUnknow(this);
newptr->parse(fptr, nptr + 1);
_pNodes.push_back(newptr);
ptr = nptr + 1;
break;
default:
char find_str[] = "/n/r/t/x20/>";
SXmlElement *new_elem = new SXmlElement(this);
_pNodes.push_back(new_elem);
tptr = find_first_of(fptr, nptr, find_str, find_str + 6);
new_elem->_Text = string(fptr+1, tptr-fptr-1);
if(*(nptr-1) == '/')//<elemet .../>
{
new_elem->_Attributes.parse(tptr, nptr-1);
ptr = nptr + 1;
}
else //</elemet>
{
new_elem->_Attributes.parse(tptr, nptr);
string str = string("</") + new_elem->_Text;
tptr = search(nptr+1, last, str.begin(), str.end());
if(tptr == last)
throw SXmlExecption(string(fptr, nptr+1) + "找不到結束標籤</element>");
new_elem->parse(nptr+1, tptr);
nptr = find(tptr, last, '>');
if(nptr == last)
throw SXmlExecption(string(fptr, nptr+1) + "找不到標籤的右邊界'>'");
ptr = nptr + 1;
}
} //switch(*(fptr + 1)) mean: <X
}//while(ptr < last)
}
};
///DOM解析
class SXmlDOM : public SXmlElement
{
public:
SXmlDOM()
{
_Type = XML_DOCUMENT;
}
void parseFile(string file_name)
{
vector<char> buf;
ifstream rf(file_name.c_str());
if(rf)
{
rf >> noskipws;
copy(istream_iterator<char>(rf), istream_iterator<char>(), back_inserter(buf));
clear();
parse(&*buf.begin(), &*buf.end());
}
else
throw SXmlExecption("無法打開讀入文件!");
}
void saveFile(string file_name)
{
ofstream rf(file_name.c_str());
if(rf)
rf << xml();
else
throw SXmlExecption("無法打開寫入文件!");
}
SXmlElement& root()
{
for(SXmlNodeList::iterator iter = _pNodes.begin(); iter != _pNodes.end(); iter++)
if((*iter)->_Type == XML_ELEMENT)
return *((SXmlElement*)(*iter));
throw SXmlExecption("找不到根元素!");
}
};
#endif //_SXML_H