Files
clients/CrawlerList/sdaumcafemanage.cpp

199 lines
6.1 KiB
C++

#include "sdaumcafemanage.h"
#include "widget.h"
#include <QSqlQuery>
#include <QSqlError>
#include <QDebug>
#include <QFile>
#include <QNetworkProxy>
SDaumCafeManage::SDaumCafeManage(QObject *pObject) : SManage(pObject)
{
m_nID = 0;
}
QString SDaumCafeManage::makeGetListQuery(QString _str,QDate _date,int _nPage)
{
//http://search.daum.net/search?w=cafe&m=board&q=%EC%84%B1%ED%98%95&find=off&sort=recency&p=3&sd=20140501000000&ed=20140502235959&period=u&page=1&DA=PGD
//http://search.daum.net/search?w=cafe&DA=PGD&m=board&q=%EC%84%B1%ED%98%95&find=off&sort=recency&p=3&sd=20140501000000&ed=20140501235959&period=u
//http://search.daum.net/search?w=cafe&m=board&q=%EC%84%B1%ED%98%95&find=off&sort=recency&p=3&period=u&sd=20140506000000&ed=20140506235959&page=1&DA=PGD
QString strOut;
strOut = "http://search.daum.net/search?w=cafe&m=board&q=";
strOut += EncodetoUtf8(_str,true);
strOut += "&find=off&sort=recency&p=";
strOut += QString::number(_nPage);
strOut += "&period=u&sd=";
strOut += _date.toString("yyyyMMdd");
strOut += "000000";
strOut += "&ed=";
strOut += _date.toString("yyyyMMdd");
strOut += "235959";
strOut += "&page=1&DA=PGD";
//InsertLog("DaumCafe : URL");
//InsertLog(strOut);
qDebug() << strOut;
return strOut;
}
void SDaumCafeManage::Start()
{
m_nMode = E_PROCESS_LIST_RUN;
m_bFinalLast = false;
}
bool SDaumCafeManage::Update()
{
if (m_bFinalLast) return m_bFinalLast;
switch(m_nMode)
{
case E_PROCESS_LIST_RUN:
if (UseProcess() == false)
{
m_strListQuery = makeGetListQuery(m_strKeyword,m_date,m_ncList);
m_pMain->InsertLog("Start : " + QString::number(m_ncList) + " Date : " + m_date.toString("yyyy-MM-dd"));
{
#if defined(Q_OS_WIN32)
m_pro[0].start("CrawlerProcess",QStringList()<< "daum" << "cafe_list" << m_strListQuery << m_strGroupID << m_strKeywordID);
#else
m_pro[0].start("./CrawlerProcess",QStringList()<< "daum" << "cafe_list" << m_strListQuery << m_strGroupID << m_strKeywordID);
#endif
m_ncList+=1;
}
m_nMode = E_PROCESS_LIST_FINISH_WAIT;
m_nWait = 0;
}
break;
case E_PROCESS_URL_RUN:
if (UseProcess() == false)
{
m_pMain->InsertLog("(" + QString::number(m_ncUrl+1) + "/" + QString::number(m_strListURL.size()) + ")");
{
#if defined(Q_OS_WIN32)
m_pro[0].start("AjaxCrawlerProcess",QStringList() << "daum_cafe" << m_strListURL.at(m_ncUrl) << m_strGroupID << m_strKeywordID );
#else
m_pro[0].start("./AjaxCrawlerProcess",QStringList() << "daum_cafe" << m_strListURL.at(m_ncUrl) << m_strGroupID << m_strKeywordID );
#endif
//m_pro[0].start("CrawlerProcess",QStringList() << "daum" << "cafe_data" << m_strListURL.at(m_ncUrl) << m_strGroupID << m_strListQuery << m_strKeywordID );
m_ncUrl++;
}
m_nMode = E_PROCESS_URL_FINISH_WAIT;
m_nWait = 0;
}
break;
case E_PROCESS_LIST_FINISH_WAIT:
case E_PROCESS_URL_FINISH_WAIT:
m_nWait++;
if (m_nWait >= 180)
{
//for(int i = 0; i < C_PROCESS_MAX ; i++)
{
// if (m_pro[0].state() == QProcess::Running)
{
m_pMain->InsertLog("Kill Process.");
m_pro[0].kill();
}
}
ReLoadList(); //return m_bFinalLast;
}
break;
}
return m_bFinalLast;
}
void SDaumCafeManage::processFinished(QProcess *_pPro,QString _strOut)
{
switch(m_nMode)
{
case E_PROCESS_LIST_FINISH_WAIT:
{
m_bLast = false;
bool reloaded = false;
if (_strOut.right(4) == "last" || m_ncList >= 1000)
m_bLast = true;
if (m_bLast == false && m_nUntilPage > 0)
{
if (m_ncList >= m_nUntilPage)
m_bLast = true;
}
// if (_strOut.right(5) == "block")
// m_bFinalLast = true;
if (_strOut.right(5) == "block")
{
reloaded = true;
ReLoadList();
}
if(_strOut.right(7) == "loading")
{
reloaded = true;
ReLoadList();
}
QStringList strOutList = _strOut.split("\n");
if(strOutList.length() > 2)
{
if(_strOut.split("\n").at(2).trimmed().length() == 0)
{
if(reloaded == false)
{
ReLoadList();
}
}
}
else
{
if(reloaded == false)
{
ReLoadList();
}
}
m_strListURL.clear();
foreach(QString str,_strOut.split("\n"))
{
if (str.isEmpty()) continue;
if (str.at(0) == QChar('o'))
m_strListURL.push_back(str.right(str.length()-2).trimmed());
}
m_ncUrl = 0;
if (m_strListURL.size() == 0)
{
m_nMode = E_PROCESS_LIST_RUN;
CheckLast();
//if(m_bLast == false)
//ReLoadList();
}
else
m_nMode = E_PROCESS_URL_RUN;
break;
}
case E_PROCESS_URL_FINISH_WAIT:
m_nMode = E_PROCESS_LIST_RUN;
if (m_ncUrl >= m_strListURL.size())
{
m_nMode = E_PROCESS_LIST_RUN;
CheckLast();
m_bLast = false;
}
else
m_nMode = E_PROCESS_URL_RUN;
break;
}
}
void SDaumCafeManage::ReLoadList()
{
m_nMode = E_PROCESS_LIST_RUN;
m_ncList -= 1;
if(m_ncList < 1)
m_ncList = 1;
m_pMain->InsertLog("Reload List");
}