void processPage(int sockfd, t_Buffer* document) {
char *pos = document->ptr;
char query_get[1024] = {0};
char qid[22] = {0};
t_Buffer *header_q = t_Buffer_new(1024);
t_Buffer *document_q = t_Buffer_new(1024);
srand(1200);
// Busca URL de una pregunta de la página
while ( (pos = strstr(pos, SEARCH_QUESTION_TOKEN )) )
{
// Identifica el qid
pos = strstr(pos, SEARCH_QID_TOKEN );
memcpy(qid, pos+5, 21);
sprintf( query_get,\
"GET %s%s%s HTTP/%s\r\nHost: %s\r\nUser-Agent: Mozilla Firefox 11.0 \r\n\r\n",\
SEARCH_QUESTION_TOKEN, SEARCH_QID_TOKEN, qid,\
"1.1", HOST);
// Enviar
if ( send(sockfd, query_get, strlen(query_get), 0) < 0 )
{
perror("send()");
exit(EXIT_FAILURE);
}
receiveHeader(sockfd, header_q);
receiveContent(sockfd, document_q, TE_CHUNKED);
if ( strstr(document_q->ptr, SEARCH_USER) )
printf("http://%s%s%s%s\n", HOST,SEARCH_QUESTION_TOKEN, SEARCH_QID_TOKEN,qid);
sleep(rand() % 3);
}
printf("\n");
// Liberar recursos
t_Buffer_delete(header_q);
t_Buffer_delete(document_q);
}
int HttpClient::requestContent(Url &url, string &contentStr, int expectContentLength) {
// construct a request
string requestStr;
string path = url.getPath();
if (path.empty())
path = "/";
requestStr = "GET " + path + " HTTP/1.0\r\nHost: " + url.getHost()
+ "\r\nUser-Agent: openSE/1.0 (Ubuntu11.04)\r\nAccept-Language: zh,en-us\r\nAccept-Charset: gb2312,utf-8\r\nConnection: Keep-Alive\r\n\r\n";
cout << "requestStr:\n" << requestStr << endl;
// send request:
if (url.getHost() != _preHost) {
if (_preSockFd != -1) {
closesocket(_preSockFd);
_preSockFd = -1;
}
}
int sockFd;
bool sendSuccess = false;
// try to use previous connection
if (_preSockFd != -1) {
sockFd = _preSockFd;
if (rio_writen(sockFd, requestStr.c_str(), requestStr.size()) == -1) {
cerr << "use previous connection:rio_writen error !" << endl;
closesocket(_preSockFd);
_preSockFd = -1;
} else
sendSuccess = true;
}
if (!sendSuccess) {
// try to creat a new connection
sockFd = tcpConnect(url.getIp(), url.getPort());
if (sockFd == -1) {
cerr << "tcpConnect error" << endl;
return -1;
}
// try to use new connection
if (rio_writen(sockFd, requestStr.c_str(), requestStr.size()) == -1) {
cerr << "rio_writen error for requestStr:" << requestStr << endl;
closesocket(sockFd);
return -1;
}
}
//// receive http header
//string headerStr;
//headerStr.reserve(1024);
//if(receiveHeader(sockFd, headerStr, DEFAULT_TIMEOUT_SECONDS) <= 0){
// cerr << "receiveHeader error" << endl;
// closesocket(sockFd);
// _preSockFd = -1;
// return -1;
// }
//cout << "headerStr:\n" << headerStr << endl;
//// parser http header
//httpHeader.setHeaderStr(headerStr);
//// check StatusCode
//int stausCode = httpHeader.getStatusCode();
//if(stausCode == -1){
// cerr << "not find status code in httpHeader: " << httpHeader.getHeaderStr() << endl;
// }
//if(stausCode == 301 || stausCode == 302){
// closesocket(sockFd);
// _preSockFd = -1;
// string locationUrlStr = httpHeader.getLocation();
// if(locationUrlStr.empty()){
// cerr << "error location in httpHeader: " << httpHeader.getHeaderStr() << endl;
// }
// //locationStr = location;
// Url locationUrl(locationUrlStr);
// return requestWebPage(locationUrl, httpHeader, httpContent);
// }
//if(stausCode < 200 || stausCode > 299){
// closesocket(sockFd);
// _preSockFd = -1;
// cerr << "status code beyond [200-300) in httpHeader: " << httpHeader.getHeaderStr() << endl;
// return -1;
// }
//// check content type
//string contentType = httpHeader.getContentType();
//if(contentType.find("image") != string::npos){
// closesocket(sockFd);
// _preSockFd = -1;
// cerr << "contentType is image in httpHeader: " << httpHeader.getHeaderStr() << endl;
// return -1;
// }
//// check ContentLength
//int contentLength = httpHeader.getContentLength();
//if(contentLength == -1){
// //cerr << "contentLength is not finded in httpHeader: " << httpHeader.getHeaderStr() << endl;
// contentLength = MAX_HTTPCONTENT_SIZE/10;
// }
//if(contentLength == 0){
// closesocket(sockFd);
// _preSockFd = -1;
// cerr << "contentLength is 0 in httpHeader: " << httpHeader.getHeaderStr() << endl;
// return -1;
// }
//if(contentLength > MAX_HTTPCONTENT_SIZE){
// closesocket(sockFd);
// _preSockFd = -1;
// cerr << "contentLength > MAX_HTTPCONTENT_SIZE in httpHeader: "
// << httpHeader.getHeaderStr() << endl;
// return -1;
// }
// receive content
if (receiveContent(sockFd, expectContentLength, contentStr, DEFAULT_TIMEOUT_SECONDS)
== -1) {
closesocket(sockFd);
_preSockFd = -1;
cerr << "receiveContent error for url: " << url.getUrlStr() << endl;
cout << contentStr << endl;
return -1;
} else
_preSockFd = sockFd;
cout << "contentStr:\n" << contentStr << endl;
// cout << "content finished,url is:"<<url.getUrlStr()<<endl;
// set http content
return 0;
}