C语言编写网络爬虫.docxVIP

#include stdio.h #include winsock.h #includeiostream #include string #pragma comment(lib, ws2_32.lib)//加载网络支持的库#define _M_MPPC using namespace std; /************************************************************************/ /* 数据获取类 /************************************************************************/ class Net { public: int UTF8ToGBK(unsigned char * lpUTF8Str,unsigned char * lpGBKStr,int nGBKStrLen);//UTF8 转 GBK char *UTOG(char *lpUTF8Str); string geturl(char *url); }; int Net::UTF8ToGBK(unsigned char * lpUTF8Str,unsigned char * lpGBKStr,int nGBKStrLen) { wchar_t * lpUnicodeStr = NULL; int nRetLen = 0; if(!lpUTF8Str) return 0; nRetLen = ::MultiByteToWideChar(CP_UTF8,0,(char *)lpUTF8Str,-1,NULL,NULL); lpUnicodeStr = new WCHAR[nRetLen + 1]; nRetLen = ::MultiByteToWideChar(CP_UTF8,0,(char *)lpUTF8Str,-1,lpUnicodeStr,nRetLen); if(!nRetLen) return 0; nRetLen = ::WideCharToMultiByte(CP_ACP,0,lpUnicodeStr,-1,NULL,NULL,NULL,NULL); if(!lpGBKStr) { if(lpUnicodeStr) delete []lpUnicodeStr; return nRetLen; } if(nGBKStrLen nRetLen) { if(lpUnicodeStr) delete []lpUnicodeStr; return 0; } nRetLen = ::WideCharToMultiByte(CP_ACP,0,lpUnicodeStr,-1,(char *)lpGBKStr,nRetLen,NULL,NULL); if(lpUnicodeStr) delete []lpUnicodeStr; return nRetLen; } char*Net::UTOG(char *lpUTF8Str) { char * lpGBKStr = NULL; int nRetLen = 0; nRetLen = UTF8ToGBK((unsigned char *)lpUTF8Str,NULL,NULL); lpGBKStr = new char[nRetLen + 1]; nRetLen = UTF8ToGBK((unsigned char *)lpUTF8Str,(unsigned char *)lpGBKStr,nRetLen); return lpGBKStr; } string Net::geturl(char *url) { WSADATA WSAData={0}; SOCKET sockfd; struct sockaddr_in addr; struct hostent *pURL; char myurl[BUFSIZ]; char *pHost = 0, *pGET = 0; char host[BUFSIZ], GET[BUFSIZ]; char header[BUFSIZ] = ; static char text[BUFSIZ]; string Data; int i; if(WSAStartup(MAKEWORD(2,2), WSAData)) { return 0; } strcpy(myurl, url); for (pHost = myurl; *pHost != / *pHost != \0; ++pHost); if ( (int)(pHost - myurl) == strlen(myurl) ) strcpy(GET, /); else strcpy(GET, pHost); *p