뉴스 크롤러 추가

git-svn-id: svn://192.168.0.12/source@149 8346c931-da38-4b9b-9d4c-e48b93cbd075
This commit is contained in:
admin
2015-06-18 08:46:08 +00:00
parent f1629c85f3
commit d21d98ac45
8 changed files with 240 additions and 5 deletions

View File

@@ -17,11 +17,13 @@ SOURCES += main.cpp\
smanage.cpp \
snavercafemanage.cpp \
snaverblogmanage.cpp \
sdaumcafemanage.cpp
sdaumcafemanage.cpp \
snavernewsmanage.cpp
HEADERS += widget.h \
smanage.h \
snavercafemanage.h \
snaverblogmanage.h \
sdaumcafemanage.h
sdaumcafemanage.h \
snavernewsmanage.h

View File

@@ -65,6 +65,38 @@ QString SManage::EncodetoUtf8(QString _str,bool _bExt)
return strOut;
}
QString SManage::EncodetoEucKr(QString _str,bool _bExt)
{
QString strOut;
for (int i = 0; i < _str.length(); i++)
{
QString strTran = _str.at(i);
QByteArray byte = strTran.toLocal8Bit();
if (byte.length() == 2)
{
QByteArray byteHan = byte.toHex().toUpper();
for (int j = 0; j < 2 ; j++)
{
strOut += "%";
for (int k = 0; k < 2; k++)
strOut += byteHan.at(j*2+k);
}
}
else if (_str.at(i).isDigit() || _str.at(i).isLower() || _str.at(i).isUpper())
strOut += _str.at(i);
else if (_bExt && _str.at(i) == '@' || _str.at(i) == '*' || _str.at(i) == '_' || _str.at(i) == '.')
strOut += _str.at(i);
else if (_bExt && _str.at(i) == ' ')
strOut += '+';
else
{
strOut += "%";
strOut += byte.toHex().toUpper();
}
}
return strOut;
}
bool SManage::UseProcess()
{
for(int i = 0; i < C_PROCESS_MAX ; i++)

View File

@@ -28,6 +28,7 @@ public:
protected:
virtual void Start() = 0;
QString EncodetoUtf8(QString _str,bool _bExt=false);
QString EncodetoEucKr(QString _str,bool _bExt=false);
virtual void processFinished(QProcess *pPro,QString _strOut) = 0;
bool UseProcess();
void CheckLast();

View File

@@ -8,7 +8,7 @@
SNaverCafeManage::SNaverCafeManage(QObject *pObject) : SManage(pObject)
{
m_nID = 0;
m_nID = 0;
}
QString SNaverCafeManage::makeGetListQuery(QString _str,QDate _date,int _nPage)

View File

@@ -0,0 +1,163 @@
#include "snavernewsmanage.h"
#include "widget.h"
#include <QDebug>
SNaverNewsManage::SNaverNewsManage(QObject *pObject) : SManage(pObject)
{
m_nID = 0;
}
QString SNaverNewsManage::makeGetListQuery(QString _str,QDate _date,int _nPage)
{
QString strOut = "http://news.naver.com/main/search/search.nhn?query=";
strOut += EncodetoEucKr(_str,true);
strOut += "&st=news.all&q_enc=EUC-KR&r_enc=UTF-8&r_format=xml&rp=none&sm=all.basic&ic=all&ie=MS949&so=datetime.dsc&stDate=range:";
strOut += _date.toString("yyyyMMdd:yyyyMMdd");
strOut += "&detail=1&pd=4&start=1&display=25&startDate=";
strOut += _date.toString("yyyy-MM-dd");
strOut += "&endDate=";
strOut += _date.toString("yyyy-MM-dd");
strOut += "&page=" + QString::number(_nPage);
return strOut;
}
void SNaverNewsManage::Start()
{
m_nMode = E_PROCESS_LIST_RUN;
m_bFinalLast = false;
}
bool SNaverNewsManage::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" << "news_list" << m_strListQuery);
m_ncList++;
}
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("AjaxCrawlerProcess",QStringList() << m_strListURL.at(m_ncUrl) << m_strGroupID << 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)
{
m_pMain->InsertLog("Kill Process.");
m_pro[0].kill();
ReLoadList();
}
break;
}
return m_bFinalLast;
}
void SNaverNewsManage::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 >= 160)
m_bLast = true;
if (m_bLast == false && m_nUntilPage > 0)
{
if (m_ncList >= 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 SNaverNewsManage::ReLoadList()
{
m_nMode = E_PROCESS_LIST_RUN;
m_ncList--;
if(m_ncList < 0)
m_ncList = 1;
m_pMain->InsertLog("Reload List");
}

View File

@@ -0,0 +1,29 @@
#ifndef SNAVERNEWSMANAGE_H
#define SNAVERNEWSMANAGE_H
#include "smanage.h"
class SNaverNewsManage : public SManage
{
public:
SNaverNewsManage(QObject *pObject);
enum E_PROCESS_STATE
{
E_PROCESS_LIST_RUN = 0,
E_PROCESS_LIST_FINISH_WAIT,
E_PROCESS_URL_RUN,
E_PROCESS_URL_FINISH_WAIT,
};
private:
QString makeGetListQuery(QString _str,QDate _date,int _nPage);
private:
QString m_strListQuery;
QVector <QString> m_strListURL;
protected:
bool Update();
void Start();
void processFinished(QProcess *pPro,QString _strOut);
void ReLoadList();
};
#endif // SNAVERNEWSMANAGE_H

View File

@@ -11,6 +11,7 @@
#include "snavercafemanage.h"
#include "snaverblogmanage.h"
#include "sdaumcafemanage.h"
#include "snavernewsmanage.h"
#include <QApplication>
Widget::Widget(QWidget *parent) : QWidget(parent) , m_nMode(E_MODE_WAIT)
@@ -27,15 +28,19 @@ Widget::Widget(QWidget *parent) : QWidget(parent) , m_nMode(E_MODE_WAIT)
m_pNaverCafe = new SNaverCafeManage(this);
m_pNaverBlog = new SNaverBlogManage(this);
m_pDaumCafe = new SDaumCafeManage(this);
m_pNaverNews = new SNaverNewsManage(this);
m_pManage[0] = m_pNaverCafe;
m_pManage[1] = m_pNaverBlog;
m_pManage[2] = m_pDaumCafe;
m_pManage[2] = m_pDaumCafe;
m_pManage[3] = m_pNaverNews;
m_db = QSqlDatabase::addDatabase("QMYSQL");
m_db.setHostName("bigbird.iptime.org");
m_db.setUserName("admin");
m_db.setPassword("admin123");
m_db.setDatabaseName("concepters");
m_db.setConnectOptions("CLIENT_INTERACTIVE=999999999;");
m_db.setConnectOptions("MYSQL_OPT_RECONNECT=true;");
if (!m_db.open())
{
InsertLog("MySql Error...");
@@ -271,6 +276,7 @@ void Widget::RefreshButton()
case 0:str += ", Naver Cafe";break;
case 1:str += ", Naver Blog";break;
case 2:str += ", Daum Cafe"; break;
case 3:str += ", Naver News"; break;
}
m_pcb->addItem(str,query.value(7));
}

View File

@@ -14,6 +14,7 @@
class SNaverCafeManage;
class SNaverBlogManage;
class SDaumCafeManage;
class SNaverNewsManage;
class SManage;
#define SAFE_DELETE(p) {if(p) delete (p); (p) = NULL; }
@@ -41,7 +42,7 @@ private:
QLineEdit *m_pedUntilPage;
QTimer m_timer;
QSqlDatabase m_db;
static const int C_PLATFORM_MAX = 3;
static const int C_PLATFORM_MAX = 4;
SManage *m_pManage[C_PLATFORM_MAX];
QListWidget *m_pResultList;
QString m_strFileName;
@@ -50,6 +51,7 @@ private:
SNaverCafeManage *m_pNaverCafe;
SNaverBlogManage *m_pNaverBlog;
SDaumCafeManage *m_pDaumCafe;
SNaverNewsManage *m_pNaverNews;
int m_nStartTime,m_nRangeTime,m_nPlatform;
//QGroupBox *m_pgbManual;
QCheckBox *m_pcheckboxReal;