HttpSendRequest和Http头

转自：http://jetyi.blog.51cto.com/1460128/628676 

Windows中有一组WinINet函数(http://msdn.microsoft.com/en-us/library/aa385473(v=VS.85).aspx),其中关于向Internet发送/接受请求的函数比较奇怪,尤其是HttpSendRequest函数问题更是诡异,下面是代码示例(该代码只是分析出问题,但没找到原因)

这组测试代码假设网络都是正常的.

 

BOOL SendHttpHeaderTest()
{
    BOOL bRet = FALSE;
    if (ERROR_SUCCESS != ::InternetAttemptConnect(0))
        return FALSE;

    if (!::InternetCheckConnection(_T("http://www.baidu.com"), FLAG_ICC_FORCE_CONNECTION, 0))
        return FALSE;

    TCHAR szModuleFile[MAX_PATH] = {0};
    ::GetModuleFileName(::GetInstance(), szModuleFile, MAX_PATH);
    LPCTSTR lpPath = ::PathFindFileName(szModuleFile);
    HINTERNET hOpen = ::InternetOpen(lpPath, INTERNET_OPEN_TYPE_PRECONFIG_WITH_NO_AUTOPROXY, NULL, NULL, 0);
    DWORD dwErr = ::GetLastError();//返回0

    HINTERNET hConnect = ::InternetConnect(hOpen, _T("www.baidu.com"), INTERNET_DEFAULT_HTTP_PORT, NULL, NULL, INTERNET_SERVICE_HTTP, 0, 0);
    dwErr = ::GetLastError();//返回0

#if 1
    DWORD dwFlag = INTERNET_FLAG_KEEP_CONNECTION | INTERNET_FLAG_NO_AUTO_REDIRECT | INTERNET_FLAG_NO_CACHE_WRITE | INTERNET_FLAG_NO_COOKIES;
    HINTERNET hOpenRequest = ::HttpOpenRequest(hConnect, _T("GET"), _T("img/baidu_sylogo1.gif"), _T("HTTP/1.1"), _T("http://www.baidu.com/"), NULL, dwFlag, 0);
    dwErr = ::GetLastError();//返回122


    bRet = ::HttpSendRequest(hOpenRequest, NULL, 0, NULL, 0);
    dwErr = ::GetLastError();//返回0
#endif

    TCHAR szBuff[BUFF_LEN_1024] = {0};
    DWORD dwBuffSize = BUFF_LEN_1024;
    bRet = ::HttpQueryInfo(hOpenRequest, HTTP_QUERY_STATUS_CODE, (LPVOID)szBuff, &dwBuffSize, NULL);
    dwErr = ::GetLastError();//返回0

    //Reference to http://www.w3.org/Protocols/rfc2616/rfc2616-sec10.html
    int nStatusCode = _tstoi(szBuff);// 值为200,证明请求数据baidu_sylogo1.gif是成功的.
    if (nStatusCode<200 || 206<nStatusCode)
        bRet = FALSE;

    bRet = TRUE;
    ::InternetCloseHandle(hConnect);

    ::InternetCloseHandle(hOpen);

    return bRet;
}

 

#if 1 ... #endif HttpOpenRequest说明要向www.baidu.com请求baidu_sylogo1.gif,但这个函数是不会向www.baidu.com发送任何数据的,只有调用HttpSendRequest时才会发送这个请求.

  HttpOpenRequest函数的返回值很有意思,可以看到它的返回值是有效的,证明对该函数的调用是成功的,但问题dwErr返回值是122,它的含义是 ERROR_INSUFFICIENT_BUFFER: The data area passed to a system call is too small,但不知道什么意思,也不知到如何才能使dwErr成为0.

  根据msdn中的描述,这段代码可以用下面一段代码替换:

 

#if 2
    HINTERNET hOpenRequest = ::HttpOpenRequest(hConnect, NULL, NULL, NULL, NULL, NULL, dwFlag, 0);
    dwErr = ::GetLastError();//仍然返回122

    LPCTSTR lpHeader =
    _T("GET /img/baidu_sylogo1.gif /HTTP/1.1")//A line
    _T("Referer: http://www.baidu.com/\r\n")  //B line
    _T("Host: www.baidu.com")                 //C line
    _T("\r\n\r\n");                           //D line

    DWORD dwHeaderLen = _tcslen(lpHeader);
    bRet = ::HttpAddRequestHeaders(hOpenRequest, lpHeader, dwHeaderLen, HTTP_ADDREQ_FLAG_ADD|HTTP_ADDREQ_FLAG_REPLACE);

    bRet = ::HttpSendRequest(hOpenRequest, NULL, 0, NULL, 0);
    dwErr = ::GetLastError();//返回0
#endif
// 注意观察lpHeaer的值,有如下几个特点:1)它的结尾D行有两个"\r\n\r\n";2)A行结尾没有\r\n;3)B行结尾有\r\n;4)C行结尾没有\r\n.
// 下面测试一下这4个特点究竟哪个会对HttpSendRequest的调用产生影响.
Test1: // A行结尾有\r\n
    LPCTSTR lpHeader = _T("GET /img/baidu_sylogo1.gif /HTTP/1.1\r\n")_T("Referer: http://www.baidu.com/")_T("Host: www.baidu.com")_T("\r\n\r\n");
    // HttpSendRequest返回0, dwErr = 12150, 含义ERROR_HTTP_HEADER_NOT_FOUND:The requested header could not be located.
Test2: // A/B/C行结尾没有\r\n
    LPCTSTR lpHeader = _T("GET /img/baidu_sylogo1.gif /HTTP/1.1")_T("Referer: http://www.baidu.com/")_T("Host: www.baidu.com")_T("\r\n\r\n");
    // HttpSendRequest返回1, dwErr = 0
Test3: // D行结尾一个\r\n
    LPCTSTR lpHeader = _T("GET /img/baidu_sylogo1.gif /HTTP/1.1")_T("Referer: http://www.baidu.com/")_T("Host: www.baidu.com")_T("\r\n");
    // HttpSendRequest返回1, dwErr = 0
Test4: // D行结尾没有\r\n
    LPCTSTR lpHeader = _T("GET /img/baidu_sylogo1.gif /HTTP/1.1")_T("Referer: http://www.baidu.com/")_T("Host: www.baidu.com");
    // HttpSendRequest返回1, dwErr = 0

 从以上4个测试可以看出,HttpSendRequest的第二个参数header是一个字符串,A行必须不能有\r\n,其它行可以有,也可以没有.

 在测试中还发现,仅仅只有A行会导致HttpSendRequest调用失败,必须得有B行.

 其实lpHeader的值A行刚好对应HttpOpenRequest的参数lpszVerb,lpszObjectName和lpszVersion;B行对应参数lpszReferer.所以,#if 2...endif 还可以用下面的代码代替:

 

#if 3
    hOpenRequestHandle = ::HttpOpenRequest(hConnectHandle, NULL, NULL, NULL, NULL, NULL, dwFlag, dwContext);
    LPCTSTR lpHeader =  _T("GET /img/baidu_sylogo1.gif /HTTP/1.1")_T("Referer: http://www.baidu.com/");
    DWORD dwLen = _tcslen(lpHeader);
    bRet = ::HttpAddRequestHeaders(hOpenRequestHandle, lpHeader, nLen, HTTP_ADDREQ_FLAG_REPLACE|HTTP_ADDREQ_FLAG_ADD);//返回1
    bRet = ::HttpSendRequest(hOpenRequestHandle, NULL, 0, NULL, 0);//返回1
#endif

以上这些测试都是在VS2008系统中测试出来的,如果有谁发现HttpSendRequest发送的HTTP头结构文档及其它特点,还请告诉我,不胜感激.