Files
clients/CrawlerList/snavercafemanage.cpp
admin d21d98ac45 뉴스 크롤러 추가
git-svn-id: svn://192.168.0.12/source@149 8346c931-da38-4b9b-9d4c-e48b93cbd075
2015-06-18 08:46:08 +00:00

183 lines
5.4 KiB
C++

#include "snavercafemanage.h"
#include "widget.h"
#include <QSqlQuery>
#include <QSqlError>
#include <QDebug>
#include <QFile>
#include <QNetworkProxy>
SNaverCafeManage::SNaverCafeManage(QObject *pObject) : SManage(pObject)
{
m_nID = 0;
}
QString SNaverCafeManage::makeGetListQuery(QString _str,QDate _date,int _nPage)
{
QString strOut;
strOut = "http://cafeblog.search.naver.com/search.naver?where=article&ie=utf8&query=";
strOut += EncodetoUtf8(_str,true);
strOut += "&t=0&st=date&date_option=6&date_from=";
strOut += _date.toString("yyyy.MM.dd");
strOut += "&date_to=";
strOut += _date.toString("yyyy.MM.dd");
//strOut += "&srchby=text&dup_remove=1&cafe_url=&without_cafe_url=&board=&sm=tab_pge&nso=so%3Add%2Cp%3Afrom";
strOut += "&srchby=text&dup_remove=1&";
strOut += "cafe_url=";
strOut += m_strAuthorship;
strOut += "&without_cafe_url=&board=&sm=tab_pge&nso=so:dd,p:from";
strOut += _date.toString("yyyyMMdd");
strOut += "to";
strOut += _date.toString("yyyyMMdd");
//strOut += "%2Ca%3Aall&start=" + QString::number(_nPage);
strOut += ",a:all&start=" + QString::number(_nPage);
//qDebug() << strOut;
return strOut;
}
void SNaverCafeManage::Start()
{
m_nMode = E_PROCESS_LIST_RUN;
m_bFinalLast = false;
}
bool SNaverCafeManage::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"));
{
m_pro[0].start("CrawlerProcess",QStringList()<< "naver" << "cafe_list" << m_strListQuery << m_strGroupID << m_strKeywordID);
m_ncList+=10;
}
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()) + ")");
{
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "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 >= 300)
{
//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();
}
break;
}
return m_bFinalLast;
}
void SNaverCafeManage::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/10) >= m_nUntilPage)
m_bLast = 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 SNaverCafeManage::ReLoadList()
{
m_nMode = E_PROCESS_LIST_RUN;
m_ncList -= 10;
if(m_ncList < 0)
m_ncList = 1;
m_pMain->InsertLog("Reload List");
}