事情不會總是那么理想的，如果你要在C/C++程序里處理XML怎么辦呢？

選擇一：市面上的XML lib還是有幾個的，最有名的當(dāng)然是libxml。我一年前用過，很不錯，我還特意寫了一份簡明教程，后來不知擱哪兒了。

選擇二：MS的MSXML，我要介紹的就是這個。

先說一下在MSDN哪里找文檔吧，往下看的時候也好有個參考：在Index里打：Windows Media Services 9 Series SDK=>Programming Reference=>Programming Reference (C++)=>XML DOM Interfaces (C++)。什么？Windows Media？呵呵，不錯，我覺得這個guide反而是最清楚的，你直接找MSXML，得到的結(jié)果，我覺得還沒這個好。

在C程序里調(diào)用MSXML基本就是一堆COM接口，不過在Visual Studio里操作先要做點簡單的設(shè)置：

在你的Project里Add References=>COM標(biāo)簽=>Microsoft XML v4.0，5.0其實也有了，但因為是和Office一起發(fā)布的，覺得有點怪，不想用，反正也未必用什么很怪異的功能，4.0可以了。

然后在加入這兩行：

#include <msxml2.h>
#import <msxml4.dll>

頭文件和dll庫。什么？在哪里加？頭文件或者c/cpp文件啊，哪里合適放哪兒。

然后就開始編程了，先定義兩個必用的變量：

IXMLDOMDocumentPtr xmlFile = NULL;
IXMLDOMElement* xmlRoot = NULL;

為什么是必用的？　　汗．．．

第一步當(dāng)然是初始化COM：

if(FAILED(CoInitialize(NULL))) ....

接下來初始化xmlFile對象：

if(FAILED(xmlFile.CreateInstance("Msxml2.DOMDocument.4.0"))) ...

然后就可以加載xml文件了：

_variant_t varXml(L"C:\\test.xml"); //L for unicode
VARIANT_BOOL varOut;
xmlFile->load(varXml, &varOut);

取得root element：

xmlFile->get_documentElement(&xmlRoot))

取得第一級element：

IXMLDOMNodeList* xmlChildNodes = NULL;
xmlRoot->get_childNodes(&xmlChildNodes);

遍歷所有第一級element:

IXMLDOMNode* currentNode = NULL;
while(!FAILED(xmlChildNodes->nextNode(&currentNode)) && currentNode != NULL)
{
//do something
}

取得當(dāng)前element的名稱：

BSTR nodeName;
currentNode->get_nodeName(&nodeName);

取得當(dāng)前element的一個attribute（假設(shè)這個attribute叫type）的值：

IXMLDOMNamedNodeMap* attributes = NULL;
IXMLDOMNode* attributeName = NULL;
_bstr_t bstrAttributeName = "type";
BSTR nameVal;
currentNode->get_attributes(&attributes);
attributes->getNamedItem(bstrAttributeName, &attributeName);
attributeName->get_text(&nameVal);

需要注意的是，你要記住釋放所有的借口，IXMLDOM***->Release()，這可不是.NET，有人幫你GC，你得自己調(diào)用Release()來減reference count，it‘s COM, remember?

好了，大致就這樣，順便提一下XPath：

_bstr_t bstrXmlQuery = L"/books/book[@type=scifi and @author=fox]";
IXMLDOMNodeList* nodes = NULL;
if(FAILED(xmlRoot->selectNodes(bstrXmlQuery, &nodes)) || FAILED(nodes->get_length(&length)) || length == 0)
//no match found or something went wrong
else
//match found

上面是找這樣的node:

<books>
<book type="scifi" author="fox">....
</book>
....
</books>

具體的XPath語法就查手冊吧，到處都有。

哦，對了，忘了說：如果你全部用ATL的類的話，借口的調(diào)用會簡單一點，不過很容易轉(zhuǎn)換的，比如：

IXMLDOMDocument* 對應(yīng) IXMLDOMDocumentPtr（我這里用了），其他基本也是加個Ptr，我不廢話了。

最后提供一個sample，我臨時攢的。工作的時候?qū)懙某绦虍?dāng)然不能拿來貼的，呵呵。這個sample基本就是遍歷整個xml，然后報告一遍文件的結(jié)構(gòu)，對每個node，如果它有一個叫id的attribute，就同時打印id的值。If you want the complete VS project, shoot me an email. But I guess no one really needs it anyway, right, : )

#include "stdafx.h"
#include <windows.h>
#include <msxml2.h>
#import <msxml4.dll>

HANDLE logFile = NULL;

#define INDENT 4

#define TESTHR(hr) \
{ \
if(FAILED(hr)) goto fail; \
}

void PrintChild(IXMLDOMNodeList* nodeList, int level)
{
if(nodeList == NULL)
return;

IXMLDOMNode* currentNode = NULL;
IXMLDOMNodeList* childNodes = NULL;
IXMLDOMNamedNodeMap* attributes = NULL;
IXMLDOMNode* attributeID = NULL;

while(!FAILED(nodeList->nextNode(&currentNode)) && currentNode != NULL)
{
BSTR nodeName;
TESTHR(currentNode->get_nodeName(&nodeName));
DWORD dwBytesWritten;
for(int i=0; i<level*INDENT; i++)
WriteFile(logFile, L" ", (DWORD)(sizeof(WCHAR)), &dwBytesWritten, NULL);

//WCHAR msg[MAX_SIZE];
//wsprintf(msg, L"%s ", nodeName);
WriteFile(logFile, nodeName, (DWORD)(wcslen(nodeName)*sizeof(WCHAR)), &dwBytesWritten, NULL);

TESTHR(currentNode->get_attributes(&attributes));
if(attributes!=NULL)
{
_bstr_t bstrAttributeName = "id";
BSTR idVal;
TESTHR(attributes->getNamedItem(bstrAttributeName, &attributeID));
if(attributeID != NULL)
{
TESTHR(attributeID->get_text(&idVal));
WriteFile(logFile, L" ", (DWORD)(sizeof(WCHAR)), &dwBytesWritten, NULL);
WriteFile(logFile, idVal, (DWORD)(wcslen(idVal)*sizeof(WCHAR)), &dwBytesWritten, NULL);
WriteFile(logFile, L"\r\n", (DWORD)(2*sizeof(WCHAR)), &dwBytesWritten, NULL);
attributeID->Release(); attributeID = NULL;
}
else
{
WriteFile(logFile, L"\r\n", (DWORD)(2*sizeof(WCHAR)), &dwBytesWritten, NULL);
}
attributes->Release(); attributes = NULL;

}
else
{
WriteFile(logFile, L"\r\n", (DWORD)(2*sizeof(WCHAR)), &dwBytesWritten, NULL);
}

TESTHR(currentNode->get_childNodes(&childNodes));
PrintChild(childNodes, level+1);
currentNode=NULL;
}

fail:
if(childNodes!=NULL)
childNodes->Release();
if(attributeID!=NULL)
attributeID->Release();
if(attributes!=NULL)
attributes->Release();
if(currentNode != NULL)
currentNode->Release();
}

int _tmain(int argc, _TCHAR* argv[])
{

IXMLDOMDocumentPtr xmlFile = NULL;
IXMLDOMElement* xmlRoot = NULL;
_variant_t varXml(L"C:\\demo1.xml");

logFile = CreateFile(L"log.txt", GENERIC_WRITE, 0, NULL, CREATE_ALWAYS, FILE_ATTRIBUTE_NORMAL, NULL);
if(logFile == INVALID_HANDLE_VALUE)
goto fail;

TESTHR(CoInitialize(NULL));

TESTHR(xmlFile.CreateInstance("Msxml2.DOMDocument.4.0"));

VARIANT_BOOL varOut;
TESTHR(xmlFile->load(varXml, &varOut));

TESTHR(xmlFile->get_documentElement(&xmlRoot));

BSTR rootName;
DWORD dwBytesWritten;
TESTHR(xmlRoot->get_nodeName(&rootName));
WriteFile(logFile, rootName, (DWORD)(wcslen(rootName)*sizeof(WCHAR)), &dwBytesWritten, NULL);
WriteFile(logFile, L"\r\n", (DWORD)(2*sizeof(WCHAR)), &dwBytesWritten, NULL);

IXMLDOMNodeList* xmlChildNodes = NULL;
TESTHR(xmlRoot->get_childNodes(&xmlChildNodes));

PrintChild(xmlChildNodes, 2);

fail:
if(logFile != INVALID_HANDLE_VALUE)
CloseHandle(logFile);
if(xmlChildNodes!=NULL)
xmlChildNodes->Release();
if(xmlRoot!=NULL)
xmlRoot->Release();
return 0;
}