httpdownload.cpp

基于Winsock2的支持断点续传和SOCKS代理的HTTP下载类 CHttpDownload

	m_nRetryDelay = nRetryDelay;
	m_nRetryMax	  = nRetryMax;
	
	// 检查一下m_nRetryMax,如果为0，设为缺省值
	if( (RETRY_TYPE_TIMES == m_nRetryType) && (0 == m_nRetryMax) )
		m_nRetryMax = DEFAULT_RETRY_MAX;
}

// 获取下载文件的状态
BOOL CHttpDownload::GetDownloadFileStatus(LPCTSTR lpszDownloadUrl, DWORD &dwFileSize, CTime &FileTime)
{
	// 检验要下载的URL是否为空
	m_strDownloadUrl = lpszDownloadUrl;
	m_strDownloadUrl.TrimLeft();
	m_strDownloadUrl.TrimRight();
	if( m_strDownloadUrl.IsEmpty() )
		return FALSE;

	// 检验要下载的URL是否有效
	if ( !ParseURL(m_strDownloadUrl, m_strServer, m_strObject, m_nPort))
	{
		// 在前面加上"http://"再试
		m_strDownloadUrl = _T("http://") + m_strDownloadUrl;
		if ( !ParseURL(m_strDownloadUrl, m_strServer, m_strObject, m_nPort) )
		{
			TRACE(_T("Failed to parse the URL: %s\n"), m_strDownloadUrl);
			return FALSE;
		}
	}

	m_strTempSavePath	= "|";
	m_bStopDownload		= FALSE;

	if ( SendRequest(TRUE) !=  SENDREQUEST_SUCCESS )
		return FALSE;

	dwFileSize  = m_dwDownloadSize;
	FileTime	= m_TimeLastModified;

	return TRUE;
}


// 从URL里面拆分出Server和Object来
BOOL CHttpDownload::ParseURL(LPCTSTR lpszURL, CString &strServer, CString &strObject,USHORT& nPort)
{
	CString strURL(lpszURL);
	strURL.TrimLeft();
	strURL.TrimRight();
	
	// 清除数据
	strServer = _T("");
	strObject = _T("");
	nPort	  = 0;

	int nPos = strURL.Find("://");
	if( nPos == -1 )
		return FALSE;

	// 进一步验证是否为http://
	CString strTemp = strURL.Left( nPos+lstrlen("://") );
	strTemp.MakeLower();
	if( strTemp.Compare("http://") != 0 )
		return FALSE;

	strURL = strURL.Mid( strTemp.GetLength() );
	nPos = strURL.Find('/');
	if ( nPos == -1 )
		return FALSE;

	strObject = strURL.Mid(nPos);
	strTemp   = strURL.Left(nPos);
	
	///////////////////////////////////////////////////////////////
	/// 注意：并没有考虑URL中有用户名和口令的情形和最后有#的情形
	/// 例如：http://abc@def:www.yahoo.com:81/index.html#link1
	/// 
	//////////////////////////////////////////////////////////////

	// 查找是否有端口号
	nPos = strTemp.Find(":");
	if( nPos == -1 )
	{
		strServer = strTemp;
		nPort	  = DEFAULT_HTTP_PORT;
	}
	else
	{
		strServer = strTemp.Left( nPos );
		strTemp	  = strTemp.Mid( nPos+1 );
		nPort	  = (USHORT)_ttoi((LPCTSTR)strTemp);
	}
	return TRUE;
}

// 从返回到头里获得必要的信息
UINT CHttpDownload::GetInfo(LPCTSTR lpszHeader, DWORD &dwContentLength, DWORD &dwStatusCode, CTime &TimeLastModified)
{
	dwContentLength = 0;
	dwStatusCode	= 0;
	TimeLastModified= CTime::GetCurrentTime();

	CString strHeader = lpszHeader;
	strHeader.MakeLower();

	//拆分出HTTP应答的头信息的第一行
	int nPos = strHeader.Find("\r\n");
	if (nPos == -1)
		return HTTP_FAIL;
	CString strFirstLine = strHeader.Left(nPos);

	// 获得返回码: Status Code
	strFirstLine.TrimLeft();
	strFirstLine.TrimRight();
	nPos = strFirstLine.Find(' ');
	if( nPos == -1 )
		return HTTP_FAIL;
	strFirstLine = strFirstLine.Mid(nPos+1);
	nPos = strFirstLine.Find(' ');
	if( nPos == -1 )
		return HTTP_FAIL;
	strFirstLine = strFirstLine.Left(nPos);
	dwStatusCode = (DWORD)_ttoi((LPCTSTR)strFirstLine);
	
	// 检查返回码
	if( dwStatusCode >= 300 && dwStatusCode < 400 ) //首先检测一下服务器的应答是否为重定向
	{
		nPos = strHeader.Find("location:");
		if (nPos == -1)
			return HTTP_FAIL;

		CString strRedirectFileName = strHeader.Mid(nPos + strlen("location:"));
		nPos = strRedirectFileName.Find("\r\n");
		if (nPos == -1)
			return HTTP_FAIL;

		strRedirectFileName = strRedirectFileName.Left(nPos);
		strRedirectFileName.TrimLeft();
		strRedirectFileName.TrimRight();
		
		// 设置Referer
		m_strReferer = m_strDownloadUrl;

		// 判断是否重定向到其他的服务器
		nPos = strRedirectFileName.Find("http://");
		if( nPos != -1 )
		{
			m_strDownloadUrl = strRedirectFileName;
			// 检验要下载的URL是否有效
			if ( !ParseURL(m_strDownloadUrl, m_strServer, m_strObject, m_nPort))
				return HTTP_FAIL;
			return HTTP_REDIRECT;
		}

		// 重定向到本服务器的其他地方
		strRedirectFileName.Replace("\\","/");
		
		// 是相对于根目录
		if( strRedirectFileName[0] == '/' )
		{
			m_strObject = strRedirectFileName;
			return HTTP_REDIRECT;
		}
		
		// 是相对当前目录
		int nParentDirCount = 0;
		nPos = strRedirectFileName.Find("../");
		while (nPos != -1)
		{
			strRedirectFileName = strRedirectFileName.Mid(nPos+3);
			nParentDirCount++;
			nPos = strRedirectFileName.Find("../");
		}
		for (int i=0; i<=nParentDirCount; i++)
		{
			nPos = m_strDownloadUrl.ReverseFind('/');
			if (nPos != -1)
				m_strDownloadUrl = m_strDownloadUrl.Left(nPos);
		}
		m_strDownloadUrl = m_strDownloadUrl+"/"+strRedirectFileName;

		if ( !ParseURL(m_strDownloadUrl, m_strServer, m_strObject, m_nPort))
			return HTTP_FAIL;
		return HTTP_REDIRECT;
	}

	// 服务器错误，可以重试
	if( dwStatusCode >=500 )
		return HTTP_ERROR;

	// 客户端错误，重试无用
	if( dwStatusCode >=400 && dwStatusCode <500 )
		return HTTP_FAIL;
		
	// 获取ContentLength
	nPos = strHeader.Find("content-length:");
	if (nPos == -1)
		return HTTP_FAIL;

	CString strDownFileLen = strHeader.Mid(nPos + strlen("content-length:"));	
	nPos = strDownFileLen.Find("\r\n");
	if (nPos == -1)
		return HTTP_FAIL;

	strDownFileLen = strDownFileLen.Left(nPos);	
	strDownFileLen.TrimLeft();
	strDownFileLen.TrimRight();

	// Content-Length:
	dwContentLength = (DWORD) _ttoi( (LPCTSTR)strDownFileLen );

	// 获取Last-Modified:
	nPos = strHeader.Find("last-modified:");
	if (nPos != -1)
	{
		CString strTime = strHeader.Mid(nPos + strlen("last-modified:"));
		nPos = strTime.Find("\r\n");
		if (nPos != -1)
		{
			strTime = strTime.Left(nPos);
			strTime.TrimLeft();
			strTime.TrimRight();
			TimeLastModified = GetTime(strTime);
		}
	}
	return HTTP_OK;
}

// 将字符串转化成时间
CTime CHttpDownload::GetTime(LPCTSTR lpszTime)
{
	int nDay,nMonth,nYear,nHour,nMinute,nSecond;

	CString strTime = lpszTime;
	int nPos = strTime.Find(',');
	if (nPos != -1)
	{
		strTime = strTime.Mid(nPos+1);
		strTime.TrimLeft();

		CString strDay,strMonth,strYear,strHour,strMinute,strSecond;
		CString strAllMonth = "jan,feb,mar,apr,may,jan,jul,aug,sep,oct,nov,dec";
		strDay = strTime.Left(2);
		nDay = atoi(strDay);
		strMonth = strTime.Mid(3,3);
		strMonth.MakeLower();
		nPos = strAllMonth.Find(strMonth);
		if (nPos != -1)
		{
			strMonth.Format("%d",((nPos/4)+1));
			nMonth = atoi(strMonth);
		}
		else
			nMonth = 1;
		strTime = strTime.Mid(6);
		strTime.TrimLeft();
		nPos = strTime.FindOneOf(" \t");
		if (nPos != -1)
		{
			strYear = strTime.Left(nPos);
			nYear = atoi(strYear);
		}
		else
			nYear = 2000;
		strTime = strTime.Mid(nPos+1);
		strHour = strTime.Left(2);
		nHour = atoi(strHour);
		strMinute = strTime.Mid(3,2);
		nMinute = atoi(strMinute);
		strSecond = strTime.Mid(6,2);
		nSecond = atoi(strSecond);
	}
	
	CTime time(nYear,nMonth,nDay,nHour,nMinute,nSecond);
	return time;
}


// 停止下载
void CHttpDownload::StopDownload()
{
	m_bStopDownload = TRUE;
}


// STATIC BASE64解码
int CHttpDownload::Base64Decode(LPCTSTR lpszDecoding, CString &strDecoded)
{
	int nIndex =0;
	int nDigit;
    int nDecode[ 256 ];
	int nSize;
	int nNumBits = 6;

	if( lpszDecoding == NULL )
		return 0;
	
	if( ( nSize = lstrlen(lpszDecoding) ) == 0 )
		return 0;

	// Build Decode Table
	for( int i = 0; i < 256; i++ ) 
		nDecode[i] = -2; // Illegal digit
	for( i=0; i < 64; i++ )
	{
		nDecode[ m_strBase64TAB[ i ] ] = i;
		nDecode[ '=' ] = -1; 
    }

	// Clear the output buffer
	strDecoded = _T("");
	long lBitsStorage  =0;
	int nBitsRemaining = 0;
	int nScratch = 0;
	UCHAR c;
	
	// Decode the Input
	for( nIndex = 0, i = 0; nIndex < nSize; nIndex++ )
	{
		c = lpszDecoding[ nIndex ];

		// 忽略所有不合法的字符
		if( c> 0x7F)
			continue;

		nDigit = nDecode[c];
		if( nDigit >= 0 ) 
		{
			lBitsStorage = (lBitsStorage << nNumBits) | (nDigit & 0x3F);
			nBitsRemaining += nNumBits;
			while( nBitsRemaining > 7 ) 
			{
				nScratch = lBitsStorage >> (nBitsRemaining - 8);
				strDecoded += (nScratch & 0xFF);
				i++;
				nBitsRemaining -= 8;
			}
		}
    }	

	return strDecoded.GetLength();


}

// STATIC BASE64编码
int CHttpDownload::Base64Encode(LPCTSTR lpszEncoding, CString &strEncoded)
{
	int nDigit;
	int nNumBits = 6;
	int nIndex = 0;
	int nInputSize;

	strEncoded = _T( "" );
	if( lpszEncoding == NULL )
		return 0;

	if( ( nInputSize = lstrlen(lpszEncoding) ) == 0 )
		return 0;

	int nBitsRemaining = 0;
	long lBitsStorage	=0;
	long lScratch		=0;
	int nBits;
	UCHAR c;

	while( nNumBits > 0 )
	{
		while( ( nBitsRemaining < nNumBits ) &&  ( nIndex < nInputSize ) ) 
		{
			c = lpszEncoding[ nIndex++ ];
			lBitsStorage <<= 8;
			lBitsStorage |= (c & 0xff);
			nBitsRemaining += 8;
		}
		if( nBitsRemaining < nNumBits ) 
		{
			lScratch = lBitsStorage << ( nNumBits - nBitsRemaining );
			nBits    = nBitsRemaining;
			nBitsRemaining = 0;
		}	 
		else 
		{
			lScratch = lBitsStorage >> ( nBitsRemaining - nNumBits );
			nBits	 = nNumBits;
			nBitsRemaining -= nNumBits;
		}
		nDigit = (int)(lScratch & m_nBase64Mask[nNumBits]);
		nNumBits = nBits;
		if( nNumBits <=0 )
			break;
		
		strEncoded += m_strBase64TAB[ nDigit ];
	}
	// Pad with '=' as per RFC 1521
	while( strEncoded.GetLength() % 4 != 0 )
		strEncoded += '=';

	return strEncoded.GetLength();

}