boost::xml——基本操作以及中文乱码解决方案

下面是本人使用boost库的xml部分的基础操作，并且解决对于大家使用boost库读写中文xml内容出现的乱码问题。

1.实现boost库xml基本操作
2.解决boost对xml中中文乱码问题
3.实现普通字符串和宽字符串的傻瓜切换（模仿tchar.h）
4.代码运行环境为VS2010，需要导入boost库才能正常运行
5.VS2010运行时可能会发生错误。例如：cl.exe 或者 cvtres.exe 报错。 解决办法就是重新打开项目或者切换其它正常项目运行一下（反正我是这么解决的）

下面是源码部分。

注：为了说明问题使用一个cpp文件完成所有操作。

/*实现单字节和宽字节的自动转换*/

//note_1: G prefix is global meaning
//note_2: template<typename type> ,type only support tstring, can't support int and so on.
//note_3: In VS2010 update1, it maybe encounter bug for VS2010 self. example: "error MSB6006: CL.exe" or "LINK : fatal error LNK1158: cvtres.exe", solution is reopen project
//note_4: switch std::string and std::wstring in use macro CHINESE_CHARSET. The major solution is chinese unreadable code with xml file

//Windows API
//#include <tchar.h>
//#include <wchar.h>

//C++ standard template library
#include <iostream>
#include <string>
#include <exception>
#include <map>

//third library head files
#include <boost/property_tree/ptree.hpp>
#include <boost/property_tree/xml_parser.hpp>
#include <boost/program_options/detail/convert.hpp>
#include <boost/program_options/detail/utf8_codecvt_facet.hpp>
#include <boost/foreach.hpp>

using namespace std;
using namespace boost;
using namespace boost::property_tree;


#define CHINESE_CHARSET //控制处理宽字符还是单字节字符，其中宽字符可以对中文进行支持

#ifdef CHINESE_CHARSET
#define tptree boost::property_tree::wptree
#define tstring std::wstring
#define _USERT(x) L ## x
#else
#define tptree boost::property_tree::ptree
#define tstring std::string
#define _USERT(x) x
#endif

bool LoadXML(tptree &o_pt, const string i_strFileName)
{
    //set locale
#ifdef CHINESE_CHARSET
    std::locale current_locale(locale(""), new boost::program_options::detail::utf8_codecvt_facet());
#else
    std::locale current_locale;
#endif

    //read xml
    try
    {
        boost::property_tree::read_xml(i_strFileName, o_pt, boost::property_tree::xml_parser::trim_whitespace, current_locale);
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false;
    }
    
    return true;
}

bool SaveXML(const tptree &i_pt, const string i_strFileName)
{
    //set locale
#ifdef CHINESE_CHARSET
    std::locale current_locale(locale(""), new boost::program_options::detail::utf8_codecvt_facet());
    boost::property_tree::xml_parser::xml_writer_settings<wchar_t> settings(L'	', 1, L"utf-8");
#else
    std::locale current_locale;
    boost::property_tree::xml_parser::xml_writer_settings<char> settings('	', 1, "utf-8");
#endif

    try
    {
        boost::property_tree::write_xml(i_strFileName, i_pt, current_locale, settings);
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false;
    }

    return true;
}

//是否存在直接子节点或者属性，返回-1：出现错误；返回0：没有子节点或者属性；返回1：有属性或者有子节点
int IsExistChildNode(const tptree &i_pt, const tstring i_strNodePath)
{
    try
    {
        const tptree pt = i_pt.get_child(i_strNodePath);
        if (pt.empty())
        {
            return 0; //no child
        }
        else
        {
            return 1; //have child or attribute
        }
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return -1; //error: exception
    }

    return -1;
}

//查看是否有子节点或者属性，如果没有则返回false，如果有则返回true
//通过o_nChildNodeCount和o_nChildNodeAttributeCount返回数目
//通过o_bChildNodeOrAttribute判断是直接子节点还是属性：true直接子节点；false属性
bool IsChildNodeOrAttr(const tptree &i_pt
    , const tstring i_strNodePath
    , bool &o_bChildNodeOrAttribute
    , int &o_nChildNodeCount //直接子节点数目
    , int &o_nChildNodeAttributeCount) //直接子节点属性数目
{
    o_bChildNodeOrAttribute = false; //initialize
    o_nChildNodeCount = 0; //initialize child node count
    o_nChildNodeAttributeCount = 0; //initialize attribute count

    tstring l_strChildNode; //local object
    tstring l_strXmlAttr(_USERT("<xmlattr>")); //with l_strChildNode
    tstring l_strAttrPath(_USERT(".<xmlattr>")); //mark <xmlattr>
    bool l_bflag = true; //valve

    try
    {
        int l_bExistChildNode = IsExistChildNode(i_pt, i_strNodePath);
        if (0==l_bExistChildNode)
        {
            return false; //no child node or attribute
        }
        BOOST_FOREACH (const tptree::value_type &v, i_pt.get_child(i_strNodePath))
        {
            l_strChildNode = v.first.data();
            if (l_strChildNode==l_strXmlAttr && true==l_bflag)
            {
                BOOST_FOREACH (const tptree::value_type &v, i_pt.get_child(i_strNodePath+l_strAttrPath))
                {
                    ++o_nChildNodeAttributeCount;
                    o_bChildNodeOrAttribute = false; //is attribute
                }
                l_bflag = false;
            }
            else
            {
                ++o_nChildNodeCount;
                o_bChildNodeOrAttribute = true; //is child node
            }
        }
        return true;
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//返回获取直接子节点是否完成：false：获取失败；true：获取成功
//获取的直接子节点保存在o_mapChildNode容器中
//注意：使用前请先确认是否有直接子节点，可以调用IsChildNodeOrAttr查看o_bChildNodeOrAttribute
template<typename type> bool GetChildNodes(const tptree &i_pt, const tstring i_strNodePath, multimap<tstring, type> &o_mmapChildNodes)
{
    tstring l_strChildNode;
    tstring l_strXmlAttr(_USERT("<xmlattr>")); //with l_strChildNode
    type l_tpChildNodeValue;

    try
    {
        int l_bExistChildNode = IsExistChildNode(i_pt, i_strNodePath);
        if (0==l_bExistChildNode)
        {
            return false; //no child node or attribute
        }
        BOOST_FOREACH (const tptree::value_type &v, i_pt.get_child(i_strNodePath))
        {
            l_strChildNode = v.first.data();
            l_tpChildNodeValue = v.second.get_value<type>();
            if (0!=l_strChildNode.compare(l_strXmlAttr))
            {
                o_mmapChildNodes.insert(pair<tstring, type>(l_strChildNode, l_tpChildNodeValue));
            }
            else
            {
                throw runtime_error("This node include with attribute.
Please call IsChildNodeOrAttr function before.");
            }
        }
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//返回获取当前节点属性是否完成：false：获取失败；true：获取成功
//获取的当前节点属性保存在o_mapCurrentNodeAttrs容器中
//注意：使用前请先确认当前节点是否有属性，可以调用IsChildNodeOrAttr查看o_bChildNodeOrAttribute
template<typename type> bool GetCurrentNodeAttrs(const tptree &i_pt, const tstring i_strNodePath, map<tstring, type> &o_mapCurrentNodeAttrs)
{
    tstring l_strChildNodeAttr;
    tstring l_strXmlAttr(_USERT("<xmlattr>")); //with l_strChildNode
    tstring l_strAttrPath(_USERT(".<xmlattr>")); //mark <xmlattr>
    type l_tpChildNodeAttrValue;

    try
    {
        int l_bExistChildNode = IsExistChildNode(i_pt, i_strNodePath);
        if (0==l_bExistChildNode)
        {
            return false; //no child node or attribute
        }
        if (l_strXmlAttr==i_pt.get_child(i_strNodePath).begin()->first.data())
        {
            BOOST_FOREACH (const tptree::value_type &v, i_pt.get_child(i_strNodePath+l_strAttrPath))
            {
                l_strChildNodeAttr = v.first.data();
                l_tpChildNodeAttrValue = v.second.get_value<type>();
                o_mapCurrentNodeAttrs.insert(pair<tstring, type>(l_strChildNodeAttr, l_tpChildNodeAttrValue));
            }
            return true;
        }
        else
        {
            throw runtime_error("This node exclude with attribute.
Please call IsChildNodeOrAttr function before.");
        }
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//返回获取当前节点的直接子节点属性是否完成：false：获取失败；true：获取成功
//获取的当前节点的直接子节点属性保存在o_mapChildNodesAttr容器中
//注意：使用前请先确认当前节点的直接子节点是否有属性，可以调用IsChildNodeOrAttr查看o_bChildNodeOrAttribute
//注意：本功能只应用在当前节点的多个直接子节点具有相同节点名的情况，对于多个直接子节点具有不同节点名情况不适用
template<typename type> bool GetChildNodesAttrs(const tptree &i_pt, const tstring i_strNodePath, multimap<tstring, map<tstring, type>> &o_mmapChildNodesAttrs)
{
    tstring l_strChildNode;
    tstring l_strChildNodeAttr;
    type l_tpChildNodeAttrValue;
    tstring l_strXmlAttr(_USERT("<xmlattr>")); //with l_strChildNode
    tstring l_strAttrPath(_USERT(".<xmlattr>")); //mark <xmlattr>

    try
    {
        int l_bExistChildNode = IsExistChildNode(i_pt, i_strNodePath); //check child node
        if (0==l_bExistChildNode)
        {
            return false; //no child node
        }

        //tstring l_strChildNode = i_pt.get_child(i_strNodePath).begin()->first.data();
        tstring l_strChildNode = i_pt.get_child(i_strNodePath).front().first.data();
        tstring l_strChildNodeAttrPath = i_strNodePath+_USERT(".")+l_strChildNode+l_strAttrPath;
        int l_bExistChildNodeAttr = IsExistChildNode(i_pt, l_strChildNodeAttrPath); //check child node attribute
        if (0==l_bExistChildNodeAttr)
        {
            return false; //no child node attribute
        }

        BOOST_FOREACH (const tptree::value_type &v, i_pt.get_child(i_strNodePath))
        {
            map<tstring, type> l_mapChildNodeAttrs;
            l_strChildNode = v.first.data();
            BOOST_FOREACH (const tptree::value_type &subv, v.second.get_child(l_strXmlAttr))
            {
                l_strChildNodeAttr = subv.first.data();
                l_tpChildNodeAttrValue = subv.second.get_value<type>();
                l_mapChildNodeAttrs.insert(pair<tstring, type>(l_strChildNodeAttr, l_tpChildNodeAttrValue));
            }
            o_mmapChildNodesAttrs.insert(pair<tstring, map<tstring, type>>(l_strChildNode, l_mapChildNodeAttrs));
        }
        return true;
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//提供指向指定节点的可读写迭代器io_iter，使用i_strNodeValue修改指定节点的值
bool SetCurrentNodeValue(tptree::iterator io_iter, const tstring i_strNodeValue)
{
    try
    {
        io_iter->second.put_value<tstring>(i_strNodeValue);
        return true;
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//提供指向指定节点属性的可读写迭代器io_iter，使用i_strNodeAttrValue修改指定节点的值
bool SetCurrentNodeAttrValue(tptree::iterator io_iter, const tstring i_strNodeAttrValue)
{
    try
    {
        io_iter->second.put_value<tstring>(i_strNodeAttrValue);
        return true;
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//提供指向指定节点的迭代器io_iter，将该节点删除
//节点io_pt就是迭代器io_iter指定的节点本身
//注：也可以删除节点属性，删除属性请使用<xmlattr>
bool DelCurrentNode(tptree &io_pt, tptree::iterator i_iter)
{
    try
    {
        io_pt.erase(i_iter);
        return true;
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return false; //error: exception
    }

    return false;
}

//当前节点io_pt
//将该节点以及所有具有相同节点名字节点删除
//注：也可以删除节点属性，删除属性请使用<xmlattr>
tptree::size_type DelCurrentNodes(tptree &io_pt, const tstring &i_strChildNode)
{
    try
    {
        return io_pt.erase(i_strChildNode);
    }
    catch (const std::exception &e)
    {
        cout << "Error:" << typeid(e).name() << ": ";
        cout << e.what() << endl;
        return -1; //error: exception
    }

    return -1;
}


void main()
{
    //test code
    string l_strFileName1("F:\Desktop\20140411\测试\testxml.xml");
    string l_strFileName2("F:\Desktop\20140411\测试\testxml2.xml");
    tptree pt;
    bool f = LoadXML(pt, l_strFileName1);

    //测试IsExistChildNode函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person.<xmlattr>.id"));
    //int l_nState1 = IsExistChildNode(pt, l_strNodePath);

    //测试IsExistChildNode函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person")); //有直接子节点或者属性
    //int l_nState2 = IsExistChildNode(pt, l_strNodePath);

    //测试IsChildNodeOrAttr函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f3 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);

    //测试IsChildNodeOrAttr函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f3 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);

    //测试GetChildNodes函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f4 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);
    //bool f5;
    //multimap<tstring, tstring> l_mapChildNodes;
    //if (l_bChildNodeOrAttribute)
    //{
    //    f5 = GetChildNodes<tstring>(pt, l_strNodePath, l_mapChildNodes);
    //}

    //测试GetChildNodes函数
    //typ==int error
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f4 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);
    //bool f5;
    //multimap<tstring, int> l_mapChildNodes;
    //if (l_bChildNodeOrAttribute)
    //{
    //    f5 = GetChildNodes<int>(pt, l_strNodePath, l_mapChildNodes);
    //}

    //测试GetChildNodes函数
    //tstring l_strNodePath(_USERT("config.departments.department"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f4 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);
    //bool f5;
    //multimap<tstring, tstring> l_mapChildNodes;
    //if (l_bChildNodeOrAttribute)
    //{
    //    f5 = GetChildNodes<tstring>(pt, l_strNodePath, l_mapChildNodes);
    //}

    //测试GetCurrentNodeAttrs函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person"));
    //multimap<tstring, tstring> l_mapChildNodes;
    //bool f5 = GetChildNodes<tstring>(pt, l_strNodePath, l_mapChildNodes);

    //测试GetCurrentNodeAttrs函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person"));
    //bool l_bChildNodeOrAttribute;
    //int l_nChildNodeCount;
    //int l_nChildNodeAttributeCount;
    //bool f4 = IsChildNodeOrAttr(pt, l_strNodePath, l_bChildNodeOrAttribute, l_nChildNodeCount, l_nChildNodeAttributeCount);
    //bool f5;
    //map<tstring, tstring> l_mapChildNodeAttrs;
    //if (!l_bChildNodeOrAttribute)
    //{
    //    f5 = GetCurrentNodeAttrs<tstring>(pt, l_strNodePath, l_mapChildNodeAttrs);
    //}

    //测试GetChildNodesAttrs函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //bool f5;
    ////map<tstring, tstring> l_mapChildNodeAttrs;
    //multimap<tstring, map<tstring, tstring>> l_mmapChildNodesAttrs;
    //f5 = GetChildNodesAttrs<tstring>(pt, l_strNodePath, l_mmapChildNodesAttrs);

    //测试SetCurrentNodeValue函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //tptree::iterator iter= pt.get_child(l_strNodePath).begin();
    //tstring l_strNodeValue = _USERT("testvalue");
    //bool f6 = SetCurrentNodeValue(iter, l_strNodeValue);

    //测试SetCurrentNodeAttrValue函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person.<xmlattr>"));
    //tptree::iterator iter= pt.get_child(l_strNodePath).begin();
    //tstring l_strNodeValue = _USERT("testvalue");
    //bool f6 = SetCurrentNodeAttrValue(iter, l_strNodeValue);

    //测试DelCurrentNode函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //tptree &childpt = pt.get_child(l_strNodePath);
    //tptree::iterator iter = childpt.begin();
    //bool f6 = DelCurrentNode(childpt, ++iter);
    
    //测试DelCurrentNode函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person.<xmlattr>"));
    //tptree &childpt = pt.get_child(l_strNodePath);
    //tptree::iterator iter = childpt.begin();
    //bool f6 = DelCurrentNode(childpt, ++iter);

    //测试DelCurrentNodes函数
    //tstring l_strNodePath(_USERT("config.departments.department.info"));
    //tptree &childpt = pt.get_child(l_strNodePath);
    //tptree::key_type kt = childpt.begin()->first;
    //tptree::size_type st6 = DelCurrentNodes(childpt, kt);

    //测试DelCurrentNodes函数
    //tstring l_strNodePath(_USERT("config.departments.department.info.person.<xmlattr>"));
    //tptree &childpt = pt.get_child(l_strNodePath);
    //tptree::key_type kt = childpt.begin()->first;
    //tptree::size_type st6 = DelCurrentNodes(childpt, kt);

    //bool f2 = SaveXML(pt, l_strFileName2);

    cout << endl;
}

 testxml.xml测试文件：

<?xml version="1.0" encoding="utf-8"?>
<config>
  <mainformsize height="600" width="330" top="50" left="500"/>
  <applicationtitle>GoCom统一通讯应用平台</applicationtitle>
  <loginmark>
      <fontsize>14</fontsize>
      <fontcolor>$0000ff</fontcolor>
      <lantern>true</lantern>
      <clientversion>5.3.5.17</clientversion>
      <serverip>imbs1.smartdot.com.cn</serverip>
  </loginmark>
  <departments>
      <department>
          <name>donghe</name>
          <level>2</level>
          <personcount>3</personcount>
          <info>
              <person id="1_1_1" name="王慧敏"/>
              <person id="1_1_2" name="刘东升"/>
              <person id="1_1_3" name="张智卓"/>
          </info>
      </department>
      <department>
          <name>boyun</name>
          <level>2</level>
          <personcount>3</personcount>
          <info>
              <person id="1_2_1" name="朗朗"/>
              <person id="1_2_2" name="隋国龙"/>
          </info>
      </department>
  </departments>
</config>

注：代码可以正常运行在VS2010环境下，先下载安装boost库并添加到项目中，代码才能正常运行。​