VC++抓取网页解析数据

////////////////////////////////////////////////////////////////////////// // 链接并获取网页数据 ////////////////////////////////////////////////////////////////////////// BOOL GetData(CString URL,CString filename) { HINTERNET hInternet ; HINTERNET hOpenUrl ; hInternet = InternetOpen("Application",INTERNET_OPEN_TYPE_PRECONFIG,NULL, NULL,INTERNET_FLAG_NO_CACHE_WRITE) ; if (hInternet != NULL) { hOpenUrl = InternetOpenUrl(hInternet,URL,NULL,0,INTERNET_FLAG_RELOAD,0) ; if (hOpenUrl != NULL)// If Open URL OK { fstream file(filename,ios::out); BOOL bLoop ; DWORD lpdwNumberOfBytesRead, dwSize ; bLoop = true; dwSize = 512 ; LPSTR lpBuffer ; for (int row=1;bLoop;row++) { lpBuffer = new char[dwSize+1] ; InternetReadFile(hOpenUrl,(LPVOID)lpBuffer,dwSize,&lpdwNumberOfBytesRead) ; if (lpdwNumberOfBytesRead == 0 || row == 200){ bLoop = false ; } else if (row < 86) { continue; } else{ lpBuffer[lpdwNumberOfBytesRead] = '/0'; //--------------------------------------------- int len=MultiByteToWideChar(CP_UTF8, 0, (LPCTSTR)lpBuffer, -1, NULL,0); unsigned short* wszGBK = new unsigned short[len+1]; memset(wszGBK, 0, len * 2 + 2); MultiByteToWideChar(CP_UTF8, 0, (LPCTSTR)lpBuffer, -1, (LPWSTR)wszGBK, len); len = WideCharToMultiByte(CP_ACP, 0, (LPCWSTR)wszGBK, -1, NULL, 0, NULL, NULL); char *szGBK=new char[len + 1]; memset(szGBK, 0, len + 1); WideCharToMultiByte (CP_ACP, 0, (LPCWSTR)wszGBK, -1, (LPSTR)szGBK, len, NULL,NULL); file<<szGBK; delete szGBK; delete wszGBK; //--------------------------------------------- } delete lpBuffer; } file.close(); } else{ AfxMessageBox("Error: Can not open URL."); return FALSE; } if (hOpenUrl != NULL){ InternetCloseHandle(hOpenUrl); } } else AfxMessageBox("Error: Can not create Internet Session."); if (hInternet != NULL) InternetCloseHandle(hInternet) ; filecount ++ ; //又下载完毕了一个文件 if (filecount == 38) { hThreadPD1 = CreateThread(//创建线程1 NULL,0, (LPTHREAD_START_ROUTINE)ThreadFuncPD1, NULL, 0,&ThreadIDPD1 ); WaitForSingleObject(hThreadPD1,0); //WaitForSingleObject(hThread,INFINITE); hThreadPD2 = CreateThread(//创建线程1 NULL,0, (LPTHREAD_START_ROUTINE)ThreadFuncPD2, NULL, 0,&ThreadIDPD2 ); WaitForSingleObject(hThreadPD2,INFINITE); } return TRUE; } ////////////////////////////////////////////////////////////////////////// // 处理获取到的网页数据 ////////////////////////////////////////////////////////////////////////// BOOL ProcessData(const char* infile,const char* outfile) { fstream file1,file2; file1.open(infile,ios::in);//读入数据文件 if(!file1) { AfxMessageBox("抓取的网页TXT数据文件丢失,请先下载完毕!"); return FALSE; } if (strcmp(infile,name101) == 0 || strcmp(infile,name201) == 0) {//第一次写文件 加标题 file2.open(outfile,ios::out); file2<<"股票代码/t股票名称/t涨跌幅/t最新价/t涨跌额 "<<endl;/// file2.clear(); } else{ file2.open(outfile,ios::out | ios::app);//写入数据文件 } if(!file2) { AfxMessageBox("定义的目录在本机不存在!"); return FALSE; } file2.clear(); char buffer[110]; char code[7]; char name[10]; char old[10]; char now[10]; char start[10]; const int _MAXROW = 3000; bool codeok=false; bool nameok=false; bool oldok =false; bool nowok =false; bool startok =false; for(int row=0;row<_MAXROW && !file1.eof() ; row++) { file1.clear(); file1.getline(buffer,110,'/n');//从文件获取一行 try{ for(int i=1;i<110 && buffer[i] ;i++){//分析那一行数据 //---------------------------------------------------------------- if (!codeok && buffer[i-1] == 'n' && buffer[i] == 'k' && buffer[i+1] == '"' && buffer[i+2] == '>' ) {//找到了股票代码 for(int j=0;j<6;j++,i++) code[j] = buffer[i+3];//入数组 code[6]=NULL; codeok = true; break;//分析一行数据结束 } //---------------------------------------------------------------- if (codeok && !nameok && buffer[i] == 'k' && buffer[i+1] == '"' && buffer[i+2] == '>' ) {//找到了股票的名称 for(int j=0;j<8;j++,i++){ name[j] = buffer[i+3];//入数组 if (buffer[i+4]=='<') break; } name[j+1]=NULL; nameok = true; break;//分析一行数据结束 } //---------------------------------------------------------------- if (codeok && !nowok && buffer[i-1] == '"' && buffer[i] == '>' ) {//找到了股票的当前价格 for(int j=0;buffer[i]!='<';j++,i++) now[j] = buffer[i+1];//入数组 now[j-1]=NULL; nowok = true; break;//分析一行数据结束 } //---------------------------------------------------------------- if (codeok && !oldok && buffer[i-1] == '"' && buffer[i] == '>' ) {//找到了股票的涨跌幅 for(int j=0;buffer[i]!='<';j++,i++) old[j] = buffer[i+1];//入数组 old[j-1]=NULL; oldok = true; break;//分析一行数据结束 } //---------------------------------------------------------------- if (codeok && !startok && buffer[i-1] == '"' && buffer[i] == '>' ) {//找到了股票的涨跌额 for(int j=0;buffer[i]!='<';j++,i++) start[j] = buffer[i+1];//入数组 start[j-1]=NULL; startok = true;//最后一个不要break } //---------------------------------------------------------------- if (codeok && nameok && nowok && oldok && startok) {//所有数据获取完毕 codeok = false; nameok = false; nowok = false; oldok = false; startok =false; file2<<code<<"/t/t"<<name<<"/t"<<old<<"/t/t"<<now<<"/t/t"<<start<<"/t"<<endl; break;//一行结束，也是一只股票读取结束 } }//分析一行 throw i; }//try catch (int) { } }//分析文件 file1.close(); file2.close(); return TRUE; }