git-svn-id: svn://192.168.0.12/source@1 8346c931-da38-4b9b-9d4c-e48b93cbd075
This commit is contained in:
22
NaverBlogCrawler/NaverBlogCrawler.pro
Normal file
22
NaverBlogCrawler/NaverBlogCrawler.pro
Normal file
@@ -0,0 +1,22 @@
|
||||
#-------------------------------------------------
|
||||
#
|
||||
# Project created by QtCreator 2014-10-08T10:16:10
|
||||
#
|
||||
#-------------------------------------------------
|
||||
|
||||
QT += core gui
|
||||
QT += sql
|
||||
|
||||
greaterThan(QT_MAJOR_VERSION, 4): QT += widgets
|
||||
|
||||
TARGET = NaverBlogCrawler
|
||||
TEMPLATE = app
|
||||
|
||||
SOURCES += main.cpp\
|
||||
widget.cpp \
|
||||
smanage.cpp \
|
||||
snaverblogmanage.cpp \
|
||||
|
||||
HEADERS += widget.h \
|
||||
smanage.h \
|
||||
snaverblogmanage.h \
|
||||
10
NaverBlogCrawler/main.cpp
Normal file
10
NaverBlogCrawler/main.cpp
Normal file
@@ -0,0 +1,10 @@
|
||||
#include <QApplication>
|
||||
#include "widget.h"
|
||||
|
||||
int main(int argc, char *argv[])
|
||||
{
|
||||
QApplication a(argc, argv);
|
||||
Widget w;
|
||||
w.show();
|
||||
return a.exec();
|
||||
}
|
||||
97
NaverBlogCrawler/smanage.cpp
Normal file
97
NaverBlogCrawler/smanage.cpp
Normal file
@@ -0,0 +1,97 @@
|
||||
#include "smanage.h"
|
||||
#include "widget.h"
|
||||
|
||||
SManage::SManage(QObject *parent) :
|
||||
QObject(parent),m_pMain(0)
|
||||
{
|
||||
for(int i = 0; i < C_PROCESS_MAX ;i++)
|
||||
connect(&m_pro[i],SIGNAL(finished(int,QProcess::ExitStatus)),SLOT(processFinished(int,QProcess::ExitStatus)));
|
||||
}
|
||||
|
||||
void SManage::Start(QDate _StartDate,QDate _EndDate,QString _strKeyword,QString _strKeywordID,QString _strGroupID,int _nStart,int _nTime)
|
||||
{
|
||||
m_date = _StartDate;
|
||||
m_dateEnd = _EndDate;
|
||||
m_strKeyword = _strKeyword;
|
||||
m_bFinalLast = false;
|
||||
m_nTime = _nTime;
|
||||
m_strKeywordID = _strKeywordID;
|
||||
m_ncList = _nStart;
|
||||
m_strGroupID = _strGroupID;
|
||||
Start();
|
||||
}
|
||||
|
||||
void SManage::SetParent(Widget *_pWidget)
|
||||
{
|
||||
m_pMain = _pWidget;
|
||||
}
|
||||
|
||||
QString SManage::EncodetoUtf8(QString _str,bool _bExt)
|
||||
{
|
||||
QString strOut;
|
||||
for (int i = 0; i < _str.length(); i++)
|
||||
{
|
||||
QString strTran = _str.at(i);
|
||||
QByteArray byte = strTran.toUtf8();
|
||||
if (byte.length() == 3)
|
||||
{
|
||||
QByteArray byteHan = byte.toHex().toUpper();
|
||||
for (int j = 0; j < 3 ; 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++)
|
||||
{
|
||||
if (m_pro[i].state() == QProcess::Running)
|
||||
return true;
|
||||
}
|
||||
return false;
|
||||
}
|
||||
|
||||
void SManage::processFinished(int exitCode,QProcess::ExitStatus exitStatus)
|
||||
{
|
||||
QProcess *pPro = (QProcess*)sender();
|
||||
QString str = pPro->readAllStandardOutput();
|
||||
QStringList list = str.split("\n");
|
||||
foreach(QString log,list)
|
||||
{
|
||||
if (m_pMain)
|
||||
m_pMain->InsertLog(m_nID,log);
|
||||
else
|
||||
exit(0);
|
||||
}
|
||||
processFinished(pPro,str);
|
||||
}
|
||||
|
||||
void SManage::CheckLast()
|
||||
{
|
||||
if (m_bLast)
|
||||
{
|
||||
m_date = m_date.addDays(1);
|
||||
if (m_date > m_dateEnd)
|
||||
{
|
||||
m_bFinalLast = true;
|
||||
}
|
||||
m_ncList=1;
|
||||
}
|
||||
}
|
||||
51
NaverBlogCrawler/smanage.h
Normal file
51
NaverBlogCrawler/smanage.h
Normal file
@@ -0,0 +1,51 @@
|
||||
#ifndef SMANAGE_H
|
||||
#define SMANAGE_H
|
||||
|
||||
#include <QObject>
|
||||
#include <QSqlDatabase>
|
||||
#include <QDate>
|
||||
#include <QStringList>
|
||||
#include <QVector>
|
||||
#include <QProcess>
|
||||
|
||||
class Widget;
|
||||
class SManage : public QObject
|
||||
{
|
||||
Q_OBJECT
|
||||
public:
|
||||
explicit SManage(QObject *parent = 0);
|
||||
public:
|
||||
void Start(QDate _StartDate,QDate _EndDate,QString _strKeyword,QString _strKeywordID,QString _strGroupID,int _nStart,int _nTime);
|
||||
void SetParent(Widget *pWidget);
|
||||
virtual bool Update() = 0;
|
||||
signals:
|
||||
public slots:
|
||||
protected:
|
||||
virtual void Start() = 0;
|
||||
QString EncodetoUtf8(QString _str,bool _bExt=false);
|
||||
virtual void processFinished(QProcess *pPro,QString _strOut) = 0;
|
||||
bool UseProcess();
|
||||
void CheckLast();
|
||||
protected:
|
||||
Widget *m_pMain;
|
||||
QDate m_date,m_dateEnd;
|
||||
int m_nMode;
|
||||
QString m_strKeywordID;
|
||||
QString m_strGroupID;
|
||||
QString m_strKeyword;
|
||||
bool m_bFinalLast;
|
||||
int m_nTime;
|
||||
int m_nID;
|
||||
bool m_bLast;
|
||||
int m_ncList;
|
||||
int m_ncUrl;
|
||||
int m_nWait;
|
||||
static const int C_PROCESS_MAX = 1;
|
||||
QProcess m_pro[C_PROCESS_MAX];
|
||||
private:
|
||||
QVector <QStringList> m_vecList;
|
||||
private slots:
|
||||
void processFinished(int exitCode, QProcess::ExitStatus exitStatus);
|
||||
};
|
||||
|
||||
#endif // SMANAGE_H
|
||||
288
NaverBlogCrawler/snaverblogmanage.cpp
Normal file
288
NaverBlogCrawler/snaverblogmanage.cpp
Normal file
@@ -0,0 +1,288 @@
|
||||
#include "snaverblogmanage.h"
|
||||
#include "widget.h"
|
||||
#include <QSqlQuery>
|
||||
#include <QSqlError>
|
||||
#include <QFile>
|
||||
#include <QTextStream>
|
||||
|
||||
SNaverBlogManage::SNaverBlogManage(QObject *pObject) : SManage(pObject) , C_TABLE_URL("NAVER_BLOG_BODY_") , C_TABLE_COM("NAVER_BLOG_REPLY_")
|
||||
{
|
||||
m_nID = 0;
|
||||
}
|
||||
|
||||
QString SNaverBlogManage::makeGetListQuery(QString _str,QDate _date)
|
||||
{
|
||||
QString str;
|
||||
QString strDate = _date.toString("yyyyMMdd");
|
||||
//http://cafeblog.search.naver.com/search.naver?where=post&sm=tab_pge&query=%EC%95%84%EC%9D%B4%ED%8F%B0&st=date&date_option=6&date_from=20131103&date_to=20131103&dup_remove=1&post_blogurl=&post_blogurl_without=&srchby=all&nso=so%3Add%2Cp%3Afrom20131103to20131103&ie=utf8&start=11
|
||||
str = "http://cafeblog.search.naver.com/search.naver?where=post&sm=tab_pge&query=";
|
||||
str += EncodetoUtf8(_str,true);
|
||||
str += "&st=date&date_option=6&date_from=" + strDate + "&date_to=" + strDate ;
|
||||
str += "&dup_remove=1&post_blogurl=&post_blogurl_without=&srchby=all&nso=so%3Add%2Cp%3Afrom";
|
||||
str += strDate + "to" + strDate +"&ie=utf8&start=" + QString::number(m_ncList);
|
||||
return str;
|
||||
}
|
||||
|
||||
QString SNaverBlogManage::makeGetCommentQuery(QString _strUrl)
|
||||
{
|
||||
//http://blog.naver.com/kohaku3533/220149821481/CommentList.nhn?blogId=kohaku3533&logNo=220149821481¤tPage=&isMemolog=false&focusingCommentNo=&showLastPage=true&shortestContentAreaWidth=false;
|
||||
QStringList strList = _strUrl.split("/");
|
||||
QString strOut = _strUrl;
|
||||
strOut += "/CommentList.nhn?blogId=";
|
||||
strOut += strList.at(3);
|
||||
strOut += "&logNo=";
|
||||
strOut += strList.at(4);
|
||||
strOut += "¤tPage=&isMemolog=false&focusingCommentNo=&showLastPage=true&shortestContentAreaWidth=false";
|
||||
return strOut;
|
||||
}
|
||||
|
||||
void SNaverBlogManage::Start()
|
||||
{
|
||||
m_nMode = E_PROCESS_LIST_RUN;
|
||||
m_ncList = 1;
|
||||
m_bFinalLast = false;
|
||||
}
|
||||
|
||||
void SNaverBlogManage::processFinished(QProcess *pPro,QString _strOut)
|
||||
{
|
||||
switch(m_nMode)
|
||||
{
|
||||
case E_PROCESS_LIST_FINISH_WAIT:
|
||||
{
|
||||
m_bLast = false;
|
||||
if (_strOut.right(4) == "last" || m_ncList >= 991)
|
||||
m_bLast = true;
|
||||
m_strListURL.clear();
|
||||
QSqlQuery query;
|
||||
if(query.exec("SELECT URL FROM " + C_TABLE_URL + QString::number(m_nUrlTable) + " where ERROR is null"))
|
||||
{
|
||||
m_pMain->InsertLog(m_nID,query.lastError().text());
|
||||
}
|
||||
|
||||
while (query.next())
|
||||
m_strListURL.append(query.value(0).toString());
|
||||
m_ncUrl = 0;
|
||||
if (m_strListURL.size() == 0)
|
||||
{
|
||||
m_nMode = E_PROCESS_LIST_RUN;
|
||||
CheckLast();
|
||||
}
|
||||
else
|
||||
m_nMode = E_PROCESS_URL_RUN;
|
||||
break;
|
||||
}
|
||||
case E_PROCESS_URL_FINISH_WAIT:
|
||||
if (UseProcess() == false)
|
||||
{
|
||||
m_nMode = E_PROCESS_COMMENT_RUN;
|
||||
}
|
||||
break;
|
||||
case E_PROCESS_COMMENT_FINISH_WAIT:
|
||||
//if (UseProcess() == false)
|
||||
if (m_ncUrl >= m_strListURL.size())
|
||||
{
|
||||
m_nMode = E_PROCESS_LIST_RUN;
|
||||
CheckLast();
|
||||
m_bLast = false;
|
||||
}
|
||||
else
|
||||
m_nMode = E_PROCESS_URL_RUN;
|
||||
break;
|
||||
}
|
||||
}
|
||||
|
||||
bool SNaverBlogManage::Update()
|
||||
{
|
||||
if (m_bFinalLast) return m_bFinalLast;
|
||||
switch(m_nMode)
|
||||
{
|
||||
case E_PROCESS_LIST_RUN:
|
||||
if (UseProcess() == false)
|
||||
{
|
||||
m_strQuery = makeGetListQuery(m_strKeyword,m_date);
|
||||
m_pMain->InsertLog(m_nID,"Start : " + QString::number(m_ncList) + " Date : " + m_date.toString("yyyy-MM-dd"));
|
||||
//for(int i = 0; i < C_PROCESS_MAX ; i++)
|
||||
{
|
||||
m_pro[0].start("CrawlerProcess",QStringList()<< "naver" << "blog_list" << QString::number(m_nUrlTable) << m_strQuery << 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(m_nID,"(" + QString::number(m_ncUrl+1) + "/" + QString::number(m_strListURL.size()) + ")");
|
||||
for(int i = 0; i < C_PROCESS_MAX ; i++)
|
||||
{
|
||||
m_pro[0].start("CrawlerProcess",QStringList() << m_strListURL.at(m_ncUrl) << "url");
|
||||
}
|
||||
*/
|
||||
/*
|
||||
int i = 0;
|
||||
foreach(QString strUrl,m_strListURL)
|
||||
//for(int i = 0; i < C_PROCESS_MAX ; i++)
|
||||
{
|
||||
m_pro[i++].start("CrawlerProcess",QStringList() << strUrl << "url" << C_TABLE_URL + QString::number(m_nUrlTable));
|
||||
if (i >= C_PROCESS_MAX) break;
|
||||
}
|
||||
*/
|
||||
//m_pro[0].start("CrawlerProcess",QStringList() << m_strListURL.at(m_ncUrl) << "blog_url" << C_TABLE_URL + QString::number(m_nUrlTable));
|
||||
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "blog_url" << QString::number(m_nUrlTable) << m_strListURL.at(m_ncUrl) << makeGetCommentQuery(m_strListURL.at(m_ncUrl)) << "" );
|
||||
m_nMode = E_PROCESS_URL_FINISH_WAIT;
|
||||
m_nWait = 0;
|
||||
}
|
||||
break;
|
||||
case E_PROCESS_COMMENT_RUN:
|
||||
if (UseProcess() == false)
|
||||
{
|
||||
//int i=0;
|
||||
//foreach(QString strUrl,m_strListURL)
|
||||
{
|
||||
//m_pro[i++].start("CrawlerProcess",QStringList() << makeGetCommentQuery(strUrl)<< "comment" << C_TABLE_COM + QString::number(m_nUrlTable));
|
||||
//if (i >= C_PROCESS_MAX) break;
|
||||
//m_ncUrl++;
|
||||
}
|
||||
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "blog_comm" << QString::number(m_nUrlTable) << makeGetCommentQuery(m_strListURL.at(m_ncUrl++)) << "" );
|
||||
//m_pro[0].start("CrawlerProcess",QStringList() << makeGetCommentQuery(m_strListURL.at(m_ncUrl++))<< "blog_comm" << C_TABLE_COM + QString::number(m_nUrlTable));
|
||||
m_nMode = E_PROCESS_COMMENT_FINISH_WAIT;
|
||||
m_nWait = 0;
|
||||
}
|
||||
break;
|
||||
case E_PROCESS_LIST_FINISH_WAIT:
|
||||
case E_PROCESS_URL_FINISH_WAIT:
|
||||
case E_PROCESS_COMMENT_FINISH_WAIT:
|
||||
m_nWait++;
|
||||
if (m_nWait > (1000000/m_nTime))
|
||||
{
|
||||
for(int i = 0; i < C_PROCESS_MAX ; i++)
|
||||
{
|
||||
m_pro[i].kill();
|
||||
m_pMain->InsertLog(m_nID,"Kill Process.");
|
||||
}
|
||||
if (m_nMode == E_PROCESS_LIST_FINISH_WAIT) return m_bFinalLast;
|
||||
QString strQuery = "update blog set ";
|
||||
strQuery += "Error ";
|
||||
strQuery += "='";
|
||||
strQuery += "Kill Process";
|
||||
strQuery += "'";
|
||||
strQuery += " where URL='";
|
||||
if (m_nMode == E_PROCESS_COMMENT_FINISH_WAIT)
|
||||
strQuery += m_strListURL.at(m_ncUrl-1);
|
||||
else
|
||||
strQuery += m_strListURL.at(m_ncUrl);
|
||||
strQuery += "'";
|
||||
QSqlQuery sql;
|
||||
sql.exec(strQuery);
|
||||
}
|
||||
break;
|
||||
}
|
||||
return m_bFinalLast;
|
||||
}
|
||||
|
||||
void SNaverBlogManage::MakeTables()
|
||||
{
|
||||
QString strQuery = "show tables";
|
||||
QSqlQuery query;
|
||||
query.exec(strQuery);
|
||||
int nUrlMax = -1;
|
||||
while (query.next())
|
||||
{
|
||||
QString str = query.value(0).toString();
|
||||
if (str.left(C_TABLE_URL.size()) == C_TABLE_URL.toUpper())
|
||||
{
|
||||
if (nUrlMax < str.mid(C_TABLE_URL.size()).toInt())
|
||||
nUrlMax = str.mid(C_TABLE_URL.size()).toInt();
|
||||
}
|
||||
}
|
||||
m_nUrlTable = nUrlMax + 1;
|
||||
strQuery = "Create table " + C_TABLE_URL + QString::number(m_nUrlTable)+ "(Url CHAR(128) not null primary key,keyword_id INT,PlatformTitle CHAR(128),PlatformID CHAR(64),ArticleTitle VARCHAR(128),ArticleID CHAR(32),Date DATETIME,Nickname CHAR(32),Data VARCHAR(18432),Error CHAR(32)) CHARSET=utf8";
|
||||
query.exec(strQuery);
|
||||
strQuery = "Create table " + C_TABLE_COM + QString::number(m_nUrlTable)+ "(Url CHAR(128) not null,Nickname CHAR(32),Data VARCHAR(1024),Parent CHAR(64),Date DATETIME,UrlReply VARCHAR(512),RowNum INT) CHARSET=utf8";
|
||||
query.exec(strQuery);
|
||||
/*
|
||||
strQuery = "Create table " + C_TABLE_URL + QString::number(m_nUrlTable)+ "(URL varchar(128) not null primary key , ID varchar(64) ,NICK varchar(64),TITLE varchar(256), DATA varchar(20480) , DATE char(32) , BLOG_ID varchar(64) , OTHER varchar(128), ERROR char(128)) CHARSET=utf8";
|
||||
query.exec(strQuery);
|
||||
strQuery = "Create table " + C_TABLE_COM + QString::number(m_nUrlTable)+ "(URL varchar(128) not null, NICK varchar(64),DATA varchar(20480),PARENT varchar(64),DATE char(32),URL_COMMENT varchar(512) not null,URL_NICK varchar(128)) CHARSET=utf8";
|
||||
query.exec(strQuery);
|
||||
*/
|
||||
|
||||
m_pMain->setWindowTitle("NaverBlogCrawler " + QString::number(m_nUrlTable));
|
||||
}
|
||||
|
||||
void SNaverBlogManage::DropTables()
|
||||
{
|
||||
QString strQuery = "drop table ";
|
||||
QSqlQuery query;
|
||||
query.exec(strQuery + C_TABLE_URL + QString::number(m_nUrlTable));
|
||||
query.exec(strQuery + C_TABLE_COM + QString::number(m_nUrlTable));
|
||||
}
|
||||
|
||||
void SNaverBlogManage::Join()
|
||||
{
|
||||
m_pMain->InsertLog(m_nID,"Insert Article Data...");
|
||||
QString strQuery = "insert into "
|
||||
"data_" + m_strGroupID +
|
||||
"(platformname , platformform , articleform ,"
|
||||
"url , keyword_id , body_platformtitle , body_platformid , body_articletitle , body_articleid , body_date , body_nickname , body_data)"
|
||||
"select "
|
||||
"CONVERT('naver' USING utf8),"
|
||||
"CONVERT('blog' USING utf8),"
|
||||
"CONVERT('article' USING utf8),"
|
||||
"CONVERT(url USING utf8),"
|
||||
"CONVERT(keyword_id USING utf8),"
|
||||
"CONVERT(PlatformTitle USING utf8),"
|
||||
"CONVERT(PlatformID USING utf8),"
|
||||
"CONVERT(ArticleTitle USING utf8),"
|
||||
"CONVERT(ArticleID USING utf8),"
|
||||
"CONVERT(Date USING utf8),"
|
||||
"CONVERT(Nickname USING utf8),"
|
||||
"CONVERT(Data USING utf8)"
|
||||
"from NAVER_BLOG_BODY_" + QString::number(m_nUrlTable);
|
||||
QSqlQuery query;
|
||||
if (query.exec(strQuery) == false)
|
||||
{
|
||||
m_pMain->InsertLog(m_nID,query.lastError().text());
|
||||
return;
|
||||
}
|
||||
|
||||
m_pMain->InsertLog(m_nID,"Insert Reply Data...");
|
||||
|
||||
strQuery = "insert into "
|
||||
"data_" + m_strGroupID +
|
||||
"(platformname , platformform , articleform ,"
|
||||
"url , keyword_id , body_platformtitle , body_platformid , body_articletitle , body_articleid , body_date , body_nickname , body_data ,"
|
||||
"reply_nickname ,reply_data, reply_parent , reply_date ,reply_urlreply ,reply_rownum )"
|
||||
"select "
|
||||
"CONVERT('naver' USING utf8),"
|
||||
"CONVERT('blog' USING utf8),"
|
||||
"CONVERT('reply' USING utf8),"
|
||||
"CONVERT(_body.url USING utf8),"
|
||||
"CONVERT(_body.keyword_id USING utf8),"
|
||||
"CONVERT(_body.PlatformTitle USING utf8),"
|
||||
"CONVERT(_body.PlatformID USING utf8),"
|
||||
"CONVERT(_body.ArticleTitle USING utf8),"
|
||||
"CONVERT(_body.ArticleID USING utf8),"
|
||||
"CONVERT(_body.Date USING utf8),"
|
||||
"CONVERT(_body.Nickname USING utf8),"
|
||||
"CONVERT(_body.Data USING utf8),"
|
||||
"CONVERT(_reply.Nickname USING utf8),"
|
||||
"CONVERT(_reply.Data USING utf8),"
|
||||
"CONVERT(_reply.Parent USING utf8),"
|
||||
"CONVERT(_reply.Date USING utf8),"
|
||||
"CONVERT(_reply.UrlReply USING utf8),"
|
||||
"CONVERT(_reply.RowNum USING utf8) "
|
||||
"from " + C_TABLE_URL + QString::number(m_nUrlTable) + " _body INNER JOIN " + C_TABLE_COM + QString::number(m_nUrlTable) + " _reply ON _body.Url = _reply.Url";
|
||||
|
||||
if (query.exec(strQuery) == false)
|
||||
{
|
||||
m_pMain->InsertLog(m_nID,query.lastError().text());
|
||||
return;
|
||||
}
|
||||
m_pMain->InsertLog(m_nID,"Delete data ...");
|
||||
query.exec("delete from NAVER_BLOG_BODY_" + QString::number(m_nUrlTable) );
|
||||
query.exec("delete from NAVER_BLOG_REPLY_" + QString::number(m_nUrlTable) );
|
||||
m_pMain->InsertLog(m_nID,"Finish ... ");
|
||||
}
|
||||
39
NaverBlogCrawler/snaverblogmanage.h
Normal file
39
NaverBlogCrawler/snaverblogmanage.h
Normal file
@@ -0,0 +1,39 @@
|
||||
#ifndef SNAVERBLOGMANAGE_H
|
||||
#define SNAVERBLOGMANAGE_H
|
||||
#include "SManage.h"
|
||||
|
||||
class SNaverBlogManage : public SManage
|
||||
{
|
||||
public:
|
||||
enum E_PROCESS_STATE
|
||||
{
|
||||
E_PROCESS_LIST_RUN = 0,
|
||||
E_PROCESS_LIST_FINISH_WAIT,
|
||||
E_PROCESS_URL_RUN,
|
||||
E_PROCESS_URL_FINISH_WAIT,
|
||||
E_PROCESS_COMMENT_RUN,
|
||||
E_PROCESS_COMMENT_FINISH_WAIT,
|
||||
};
|
||||
public:
|
||||
SNaverBlogManage(QObject *pObject);
|
||||
void MakeTables();
|
||||
void DropTables();
|
||||
void SaveCsv(QString _strName);
|
||||
void Join();
|
||||
int GetTableNumber() {return m_nUrlTable;}
|
||||
private:
|
||||
QString makeGetListQuery(QString _str,QDate _date);
|
||||
QString makeGetCommentQuery(QString _strUrl);
|
||||
private:
|
||||
QString m_strQuery;
|
||||
QVector <QString> m_strListURL;
|
||||
const QString C_TABLE_URL;
|
||||
const QString C_TABLE_COM;
|
||||
int m_nUrlTable;
|
||||
protected:
|
||||
bool Update();
|
||||
void Start();
|
||||
void processFinished(QProcess *pPro,QString _strOut);
|
||||
};
|
||||
|
||||
#endif // SNAVERBLOGMANAGE_H
|
||||
6
NaverBlogCrawler/sprocess.cpp
Normal file
6
NaverBlogCrawler/sprocess.cpp
Normal file
@@ -0,0 +1,6 @@
|
||||
#include "sprocess.h"
|
||||
|
||||
SProcess::SProcess(QObject *parent) :
|
||||
QProcess(parent) , m_nState(0)
|
||||
{
|
||||
}
|
||||
25
NaverBlogCrawler/sprocess.h
Normal file
25
NaverBlogCrawler/sprocess.h
Normal file
@@ -0,0 +1,25 @@
|
||||
#ifndef SPROCESS_H
|
||||
#define SPROCESS_H
|
||||
|
||||
#include <QProcess>
|
||||
|
||||
class SProcess : public QProcess
|
||||
{
|
||||
Q_OBJECT
|
||||
public:
|
||||
enum STATE
|
||||
{
|
||||
STATE_WAIT = 0,
|
||||
STATE_RUNNING,
|
||||
};
|
||||
explicit SProcess(QObject *parent = 0);
|
||||
void SetState(int _nState) { m_nState = _nState; }
|
||||
int State() {return m_nState;}
|
||||
signals:
|
||||
public slots:
|
||||
private:
|
||||
int m_nState;
|
||||
|
||||
};
|
||||
|
||||
#endif // SPROCESS_H
|
||||
175
NaverBlogCrawler/widget.cpp
Normal file
175
NaverBlogCrawler/widget.cpp
Normal file
@@ -0,0 +1,175 @@
|
||||
#include "widget.h"
|
||||
#include <QHBoxLayout>
|
||||
#include <QSqlDatabase>
|
||||
#include <QSqlQuery>
|
||||
#include <QSqlError>
|
||||
#include <QDebug>
|
||||
#include <QDateEdit>
|
||||
#include <QFileDialog>
|
||||
#include "snaverblogmanage.h"
|
||||
|
||||
Widget::Widget(QWidget *parent)
|
||||
: QWidget(parent)
|
||||
{
|
||||
connect(&m_timer, SIGNAL(timeout()), this, SLOT(Update()));
|
||||
m_pcb = new QComboBox;
|
||||
m_pcb->setInsertPolicy(QComboBox::InsertAtCurrent);
|
||||
|
||||
QPushButton *pbtRefresh = new QPushButton("Refresh",this);
|
||||
QPushButton *pbtStart = new QPushButton("Start",this);
|
||||
QPushButton *pbtStop = new QPushButton("Stop",this);
|
||||
{
|
||||
QObject::connect(pbtStart,SIGNAL(clicked()),this,SLOT(StartButton()));
|
||||
QObject::connect(pbtStop,SIGNAL(clicked()),this,SLOT(StopButton()));
|
||||
QObject::connect(pbtRefresh,SIGNAL(clicked()),this,SLOT(RefreshButton()));
|
||||
}
|
||||
|
||||
QVBoxLayout *vlayout = new QVBoxLayout;
|
||||
{
|
||||
QHBoxLayout *hlayout = new QHBoxLayout;
|
||||
hlayout->addWidget(m_pcb,Qt::AlignLeft);
|
||||
hlayout->addWidget(pbtRefresh);
|
||||
vlayout->addLayout(hlayout);
|
||||
}
|
||||
|
||||
m_pedTime = new QLineEdit(this);
|
||||
m_pedTime->setText(QString("500"));
|
||||
|
||||
{
|
||||
QHBoxLayout *hlayout = new QHBoxLayout;
|
||||
hlayout->addWidget(m_pedTime);
|
||||
hlayout->addWidget(pbtStart);
|
||||
hlayout->addWidget(pbtStop);
|
||||
vlayout->addLayout(hlayout);
|
||||
}
|
||||
|
||||
{
|
||||
QHBoxLayout *hlayout = new QHBoxLayout;
|
||||
for (int i = 0; i < C_CRAWLER_MAX; i++ )
|
||||
hlayout->addWidget(&m_aResultList[i]);
|
||||
vlayout->addLayout(hlayout);
|
||||
}
|
||||
|
||||
setLayout(vlayout);
|
||||
m_pNaverBlog = new SNaverBlogManage(this);
|
||||
m_pManage[0] = m_pNaverBlog;
|
||||
|
||||
m_db = QSqlDatabase::addDatabase("QMYSQL");
|
||||
m_db.setHostName("bigbird.iptime.org");
|
||||
m_db.setUserName("admin");
|
||||
m_db.setPassword("admin123");
|
||||
m_db.setDatabaseName("concepters");
|
||||
if (!m_db.open())
|
||||
{
|
||||
InsertLog(0,"MySql Error...");
|
||||
return;
|
||||
}
|
||||
|
||||
for (int i = 0 ; i < C_CRAWLER_MAX ; i++)
|
||||
m_pManage[i]->SetParent(this);
|
||||
|
||||
m_pNaverBlog->MakeTables();
|
||||
RefreshButton();
|
||||
}
|
||||
|
||||
Widget::~Widget()
|
||||
{
|
||||
m_pNaverBlog->DropTables();
|
||||
m_db.close();
|
||||
m_pNaverBlog->SetParent(0);
|
||||
}
|
||||
|
||||
void Widget::InsertLog(int _nSelect,QString str)
|
||||
{
|
||||
if (_nSelect >= C_CRAWLER_MAX) return;
|
||||
|
||||
QTime time = QTime::currentTime();
|
||||
QString strOut = time.toString("[hh:mm:ss] ") + str;
|
||||
m_aResultList[_nSelect].addItem(strOut);
|
||||
QDate date = QDate::currentDate();
|
||||
|
||||
QFile file(date.toString(Qt::ISODate)+"_"+QString::number(_nSelect)+".log");
|
||||
if (!file.open(QIODevice::WriteOnly | QIODevice::Text | QIODevice::Append))
|
||||
return;
|
||||
|
||||
QTextStream out(&file);
|
||||
out << strOut << "\n";
|
||||
file.close();
|
||||
|
||||
if (m_aResultList[_nSelect].count() > 1024)
|
||||
{
|
||||
m_aResultList[_nSelect].removeItemWidget(m_aResultList[_nSelect].item(0));
|
||||
QListWidgetItem* item = m_aResultList[_nSelect].takeItem(0);
|
||||
delete item;
|
||||
}
|
||||
m_aResultList[_nSelect].setCurrentRow( m_aResultList[_nSelect].count() - 1 );
|
||||
m_aResultList[_nSelect].repaint();
|
||||
}
|
||||
|
||||
void Widget::StartButton()
|
||||
{
|
||||
m_timer.stop();
|
||||
QSqlQuery query;
|
||||
query.exec("UPDATE crawling set state = '" + QString::number(m_pNaverBlog->GetTableNumber()) + "' where id = '" + m_pcb->currentData().toString() + "'");
|
||||
query.exec("SELECT _keyword.start,_keyword.end, _keyword.searches,_keyword.id,_datagroup.id "
|
||||
"FROM crawling _crawling INNER JOIN keyword _keyword ON _crawling.keyword_id = _keyword.id "
|
||||
"inner join datagroup _datagroup on _crawling.datagroup_id = _datagroup.id "
|
||||
"where _crawling.id = '" + m_pcb->currentData().toString() + "'");
|
||||
query.next();
|
||||
QString str = + " , ";
|
||||
str += query.value(1).toString() + " ~ ";
|
||||
str += query.value(2).toString() + " -> ";
|
||||
str += query.value(3).toString();
|
||||
m_timer.start(m_pedTime->text().trimmed().toInt());
|
||||
m_pManage[0]->Start(QDate::fromString(query.value(0).toString(),"yyyy-MM-dd"),
|
||||
QDate::fromString(query.value(1).toString(),"yyyy-MM-dd"),
|
||||
query.value(2).toString().trimmed(),// keyword
|
||||
query.value(3).toString().trimmed(),// keyword_id
|
||||
query.value(4).toString().trimmed(),
|
||||
1,
|
||||
m_pedTime->text().trimmed().toInt());
|
||||
}
|
||||
|
||||
void Widget::StopButton()
|
||||
{
|
||||
m_timer.stop();
|
||||
}
|
||||
|
||||
void Widget::Update()
|
||||
{
|
||||
if (m_db.isOpen() == false)
|
||||
{
|
||||
if (m_db.open())
|
||||
{
|
||||
InsertLog(0,"MySql Open Error...");
|
||||
m_timer.stop();
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
int nCount = 0;
|
||||
for (int i = 0 ; i < C_CRAWLER_MAX ; i++)
|
||||
nCount += m_pManage[i]->Update();
|
||||
if (nCount == C_CRAWLER_MAX)
|
||||
{
|
||||
m_timer.stop();
|
||||
m_pNaverBlog->Join();
|
||||
}
|
||||
}
|
||||
|
||||
void Widget::RefreshButton()
|
||||
{
|
||||
m_pcb->clear();
|
||||
QSqlQuery query("SELECT _keyword.searches,_keyword.start,_keyword.end, _datagroup.name , _crawling.id "
|
||||
"FROM crawling _crawling INNER JOIN keyword _keyword ON _crawling.keyword_id = _keyword.id "
|
||||
"inner join datagroup _datagroup on _crawling.datagroup_id = _datagroup.id where _crawling.state is null");
|
||||
while (query.next())
|
||||
{
|
||||
QString str;
|
||||
str += query.value(0).toString() + " , ";
|
||||
str += query.value(1).toString() + " ~ ";
|
||||
str += query.value(2).toString() + " -> ";
|
||||
str += query.value(3).toString();
|
||||
m_pcb->addItem(str,query.value(4));
|
||||
}
|
||||
}
|
||||
49
NaverBlogCrawler/widget.h
Normal file
49
NaverBlogCrawler/widget.h
Normal file
@@ -0,0 +1,49 @@
|
||||
#ifndef WIDGET_H
|
||||
#define WIDGET_H
|
||||
|
||||
#include <QWidget>
|
||||
#include <QPushButton>
|
||||
#include <QLineEdit>
|
||||
#include <QTimer>
|
||||
#include <QDateEdit>
|
||||
#include <QListWidget>
|
||||
#include <QSqlDatabase>
|
||||
#include <QComboBox>
|
||||
|
||||
class SNaverBlogManage;
|
||||
class SManage;
|
||||
|
||||
#define SAFE_DELETE(p) {if(p) delete (p); (p) = NULL; }
|
||||
|
||||
class Widget : public QWidget
|
||||
{
|
||||
Q_OBJECT
|
||||
public:
|
||||
Widget(QWidget *parent = 0);
|
||||
~Widget();
|
||||
public:
|
||||
void InsertLog(int _nSelect,QString str);
|
||||
private:
|
||||
QLineEdit *m_pedTime;
|
||||
QTimer m_timer;
|
||||
QSqlDatabase m_db;
|
||||
static const int C_CRAWLER_MAX = 1;
|
||||
SManage *m_pManage[C_CRAWLER_MAX];
|
||||
QListWidget m_aResultList[C_CRAWLER_MAX];
|
||||
QString m_strFileName;
|
||||
QComboBox *m_pcb;
|
||||
QVector <QString> m_vecSelect;
|
||||
|
||||
SNaverBlogManage *m_pNaverBlog;
|
||||
private:
|
||||
QString makeCafeGetListQuery(QString _str,QDate _date,int _nPage);
|
||||
//void SaveBlog(QString _strName);
|
||||
//void SaveCafe(QString _strName);
|
||||
private slots:
|
||||
void RefreshButton();
|
||||
void StartButton();
|
||||
void StopButton();
|
||||
void Update();
|
||||
};
|
||||
|
||||
#endif // WIDGET_H
|
||||
Reference in New Issue
Block a user