git-svn-id: svn://192.168.0.12/source@29 8346c931-da38-4b9b-9d4c-e48b93cbd075

This commit is contained in:
admin
2015-02-26 06:28:26 +00:00
parent aaebce6cf6
commit 21f33f69e2
3 changed files with 70 additions and 24 deletions

View File

@@ -13,7 +13,7 @@ TEMPLATE = app
SOURCES += main.cpp \ SOURCES += main.cpp \
scrawler.cpp \ scrawler.cpp \
scrawler_backup.cpp
HEADERS += \ HEADERS += \
scrawler.h scrawler.h

View File

@@ -76,6 +76,33 @@ void SCrawler::load(QStringList _strlistArgv)
m_nSelect = E_NAVER_BLOG_LIST; m_nSelect = E_NAVER_BLOG_LIST;
m_strKeywordID = _strlistArgv[4]; m_strKeywordID = _strlistArgv[4];
//cout << "ok"; //cout << "ok";
QFile file("proxy.txt");
if (file.open(QIODevice::ReadOnly | QIODevice::Text))
{
QVector <QStringList> vecProxy;
while (!file.atEnd())
{
QString str = QString(file.readLine());
if (str.isEmpty()) continue;
vecProxy.push_back(str.split(","));
}
if (vecProxy.size() > 0)
{
QStringList strList = vecProxy.at(rand()%vecProxy.size());
switch(strList.size())
{
case 1:
cout << "p : " << strList.at(0).toStdString() << endl;
QNetworkProxy::setApplicationProxy(*(new QNetworkProxy(QNetworkProxy::HttpProxy,strList.at(0))));
break;
case 2:
cout << "p : " << strList.at(0).toStdString() << endl;
QNetworkProxy::setApplicationProxy(*(new QNetworkProxy(QNetworkProxy::HttpProxy,strList.at(0),strList.at(1).toInt())));
break;
}
}
}
} }
if (_strlistArgv[1] == "blog_url") if (_strlistArgv[1] == "blog_url")
@@ -241,17 +268,16 @@ void SCrawler::saveFrameList(QWebFrame *frame)
m_bUse = true; m_bUse = true;
return; return;
} }
cout << "url : " << strUrl.toStdString();
QStringList strList = strUrl.split('/'); QStringList strList = strUrl.split('/');
if (strList.at(0).compare("blog.naver.com") != 0 ) { cout << " not" << endl; continue; }; if (strList.at(0).compare("blog.naver.com") != 0 ) { cout << "x http://" << strUrl.toStdString() <<endl; continue; };
QString strQuery = "select article_url from "; QString strQuery = "select article_url from ";
strQuery += m_strTable; strQuery += m_strTable;
strQuery += QString(" where article_url = '%1'").arg(strUrl); strQuery += QString(" where article_url = 'http://%1'").arg(strUrl);
sql.exec(strQuery); sql.exec(strQuery);
if (sql.size() == -1) if (sql.size() == 0 || sql.size() == -1)
{ {
QString str = Find(sub,"a","class","txt84").toPlainText(); QString str = Find(sub,"a","class","txt84").toPlainText();
str = GetSafeUtf(str); str = GetSafeUtf(str);
@@ -261,15 +287,16 @@ void SCrawler::saveFrameList(QWebFrame *frame)
QString strQuery = QString("insert into "); QString strQuery = QString("insert into ");
strQuery += m_strTable; strQuery += m_strTable;
strQuery += QString(" set article_url='%1',platform_id='%2',platform_title='%3',keyword_id='%4'").arg("http://"+strUrl).arg(strUrl.split("/").at(1)).arg(str).arg(m_strKeywordID); strQuery += QString(" set platform_name='naver',platform_form='blog',article_form='body',article_url='%1',platform_id='%2',platform_title='%3',keyword_id='%4'").arg("http://"+strUrl).arg(strUrl.split("/").at(1)).arg(str).arg(m_strKeywordID);
QString strUtf8(strQuery.toUtf8()); QString strUtf8(strQuery.toUtf8());
if (sql.exec(strUtf8) == false) if (sql.exec(strUtf8) == false)
cout << "error : " << sql.lastError().text().toStdString(); cout << "error : " << sql.lastError().text().toStdString();
else else
cout << " ok" << endl; cout << "o ";
} }
else else
cout << " overlap" << endl; cout << "v ";
cout << "http://" << strUrl.toStdString() << endl;
m_bUse = true; m_bUse = true;
} }
@@ -395,7 +422,6 @@ void SCrawler::saveFrameUrl(QWebFrame *frame)
if ( str[E_DATA_DATE].isEmpty() == false) if ( str[E_DATA_DATE].isEmpty() == false)
{ {
str[E_DATA_DATE] += ":00"; str[E_DATA_DATE] += ":00";
cout << "str[E_DATA_DATE] = " << str[E_DATA_DATE].toStdString() << endl;
} }
else else
{ {
@@ -461,7 +487,7 @@ void SCrawler::saveFrameComment(QWebFrame *frame)
strNick = strParent = Find(element,"a","class","nick pcol2").toPlainText(); strNick = strParent = Find(element,"a","class","nick pcol2").toPlainText();
strDate = Find(element,"span","class","date fil5 pcol2").toPlainText(); strDate = Find(element,"span","class","date fil5 pcol2").toPlainText();
strComm = Find(element,"dd","class","comm pcol2").toPlainText(); strComm = Find(element,"dd","class","comm pcol2").toPlainText();
//strComm = GetSafeUtf(strComm); strComm = GetSafeUtf(strComm);
if (strComm.isEmpty()== false) if (strComm.isEmpty()== false)
{ {
strComm.replace("'","\\'"); strComm.replace("'","\\'");
@@ -469,13 +495,15 @@ void SCrawler::saveFrameComment(QWebFrame *frame)
strComm = strComm.trimmed(); strComm = strComm.trimmed();
//cout << strComm.toStdString() << endl; //cout << strComm.toStdString() << endl;
QSqlQuery query; QSqlQuery query;
query.prepare(QString("insert into " + m_strTable + " (article_url,article_nickname,article_data,article_date,article_parent,reply_url,article_order) VALUES (:URL,:NICK,:DATA,:DATE,:PARENT,:URLREPLY,:ROWNUM)").toUtf8()); query.prepare(QString("insert into " + m_strTable + " (article_url,article_nickname,article_data,article_date,article_parent,reply_url,article_order,platform_name,platform_form,article_form) VALUES (:URL,:NICK,:DATA,:DATE,:PARENT,:URLREPLY,:ROWNUM,'naver','blog','reply')").toUtf8());
/*
cout << "m_strTable = " << m_strTable.toStdString() << endl; cout << "m_strTable = " << m_strTable.toStdString() << endl;
cout << "url = " << strUrl.toStdString() << endl << "nickname = " << strNick.toStdString() << endl; cout << "url = " << strUrl.toStdString() << endl << "nickname = " << strNick.toStdString() << endl;
cout << "data = " << strComm.toStdString() << endl; cout << "data = " << strComm.toStdString() << endl;
cout << "date = " << strDate.toStdString() << endl; cout << "date = " << strDate.toStdString() << endl;
cout << "urlReply = " << m_strUrl.toStdString() << endl; cout << "urlReply = " << m_strUrl.toStdString() << endl;
cout << "ronum = " << nCount << endl; cout << "ronum = " << nCount << endl;
*/
query.bindValue(":URL", strUrl.toUtf8()); query.bindValue(":URL", strUrl.toUtf8());
query.bindValue(":NICK",strNick.toUtf8()); query.bindValue(":NICK",strNick.toUtf8());
query.bindValue(":DATA",strComm.toUtf8()); query.bindValue(":DATA",strComm.toUtf8());
@@ -503,19 +531,22 @@ void SCrawler::saveFrameComment(QWebFrame *frame)
if (strComm.isEmpty() == false) if (strComm.isEmpty() == false)
{ {
// strComm = GetSafeUtf(strComm); strComm = GetSafeUtf(strComm);
strComm.replace("'","\\'"); strComm.replace("'","\\'");
strComm.replace("\"","\\\""); strComm.replace("\"","\\\"");
strComm = strComm.trimmed(); strComm = strComm.trimmed();
QSqlQuery query; QSqlQuery query;
cout << "m_strTable = " << m_strTable.toStdString() << endl; /*
cout << "m_strTable = " << m_strTable.toStdString() << endl;
cout << "url = " << strUrl.toStdString() << endl << "nickname = " << strNick.toStdString() << endl; cout << "url = " << strUrl.toStdString() << endl << "nickname = " << strNick.toStdString() << endl;
cout << "data = " << strComm.toStdString() << endl; cout << "data = " << strComm.toStdString() << endl;
cout << "date = " << strDate.toStdString() << endl; cout << "date = " << strDate.toStdString() << endl;
cout << "urlReply = " << m_strUrl.toStdString() << endl; cout << "urlReply = " << m_strUrl.toStdString() << endl;
cout << "ronum = " << nCount << endl; cout << "ronum = " << nCount << endl;
cout << "parent = " << strParent.toStdString() << endl; cout << "parent = " << strParent.toStdString() << endl;
query.prepare(QString("insert into " + m_strTable + " (article_url,article_nickname,article_data,article_date,article_parent,reply_url,article_order) VALUES (:URL,:NICK,:DATA,:DATE,:PARENT,:URLREPLY,:ROWNUM)").toUtf8()); */
query.prepare(QString("insert into " + m_strTable + " (article_url,article_nickname,article_data,article_date,article_parent,reply_url,article_order,platform_name,platform_form,article_form) VALUES (:URL,:NICK,:DATA,:DATE,:PARENT,:URLREPLY,:ROWNUM,'naver','blog','reply')").toUtf8());
query.bindValue(":URL",strUrl.toUtf8()); query.bindValue(":URL",strUrl.toUtf8());
query.bindValue(":NICK",strNick.toUtf8()); query.bindValue(":NICK",strNick.toUtf8());
query.bindValue(":DATA",strComm.toUtf8()); query.bindValue(":DATA",strComm.toUtf8());

View File

@@ -58,6 +58,13 @@ void SNaverBlogManage::processFinished(QProcess *pPro,QString _strOut)
if (_strOut.right(4) == "last" || m_ncList >= 991) if (_strOut.right(4) == "last" || m_ncList >= 991)
m_bLast = true; m_bLast = true;
m_strListURL.clear(); 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());
}
/*
QSqlQuery query; QSqlQuery query;
if(query.exec("SELECT ARTICLE_URL FROM " + C_TABLE_URL + QString::number(m_nUrlTable) + " where ERROR is null")) if(query.exec("SELECT ARTICLE_URL FROM " + C_TABLE_URL + QString::number(m_nUrlTable) + " where ERROR is null"))
{ {
@@ -66,6 +73,7 @@ void SNaverBlogManage::processFinished(QProcess *pPro,QString _strOut)
while (query.next()) while (query.next())
m_strListURL.append(query.value(0).toString()); m_strListURL.append(query.value(0).toString());
*/
m_ncUrl = 0; m_ncUrl = 0;
if (m_strListURL.size() == 0) if (m_strListURL.size() == 0)
{ {
@@ -108,7 +116,8 @@ bool SNaverBlogManage::Update()
m_pMain->InsertLog(m_nID,"Start : " + QString::number(m_ncList) + " Date : " + m_date.toString("yyyy-MM-dd")); 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++) //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_pro[0].start("CrawlerProcess",QStringList()<< "naver" << "blog_list" << m_strQuery << m_strGroupID << m_strKeywordID);
// m_pro[0].SetState(SProcess::STATE_RUNNING);
m_ncList+=10; m_ncList+=10;
} }
m_nMode = E_PROCESS_LIST_FINISH_WAIT; m_nMode = E_PROCESS_LIST_FINISH_WAIT;
@@ -119,13 +128,11 @@ bool SNaverBlogManage::Update()
if (UseProcess() == false) 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++) for(int i = 0; i < C_PROCESS_MAX ; i++)
{ {
m_pro[0].start("CrawlerProcess",QStringList() << m_strListURL.at(m_ncUrl) << "url"); m_pro[0].start("CrawlerProcess",QStringList() << m_strListURL.at(m_ncUrl) << "url");
} }
*/
/*
int i = 0; int i = 0;
foreach(QString strUrl,m_strListURL) foreach(QString strUrl,m_strListURL)
//for(int i = 0; i < C_PROCESS_MAX ; i++) //for(int i = 0; i < C_PROCESS_MAX ; i++)
@@ -135,7 +142,9 @@ bool SNaverBlogManage::Update()
} }
*/ */
//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() << 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_pMain->InsertLog(m_nID,"(" + QString::number(m_ncUrl+1) + "/" + QString::number(m_strListURL.size()) + ")");
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "blog_url" << m_strListURL.at(m_ncUrl) << m_strGroupID << "" );
// m_pro[0].SetState(SProcess::STATE_RUNNING);
m_nMode = E_PROCESS_URL_FINISH_WAIT; m_nMode = E_PROCESS_URL_FINISH_WAIT;
m_nWait = 0; m_nWait = 0;
} }
@@ -150,8 +159,10 @@ bool SNaverBlogManage::Update()
//if (i >= C_PROCESS_MAX) break; //if (i >= C_PROCESS_MAX) break;
//m_ncUrl++; //m_ncUrl++;
} }
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "blog_comm" << QString::number(m_nUrlTable) << makeGetCommentQuery(m_strListURL.at(m_ncUrl++)) << "" ); //m_pMain->InsertLog(m_nID,"(" + QString::number(m_ncUrl+1) + "/" + QString::number(m_strListURL.size()) + ")");
m_pro[0].start("CrawlerProcess",QStringList() << "naver" << "blog_comm" << makeGetCommentQuery(m_strListURL.at(m_ncUrl++)) << m_strGroupID << "" );
//m_pro[0].start("CrawlerProcess",QStringList() << makeGetCommentQuery(m_strListURL.at(m_ncUrl++))<< "blog_comm" << C_TABLE_COM + QString::number(m_nUrlTable)); //m_pro[0].start("CrawlerProcess",QStringList() << makeGetCommentQuery(m_strListURL.at(m_ncUrl++))<< "blog_comm" << C_TABLE_COM + QString::number(m_nUrlTable));
// m_pro[0].SetState(SProcess::STATE_RUNNING);
m_nMode = E_PROCESS_COMMENT_FINISH_WAIT; m_nMode = E_PROCESS_COMMENT_FINISH_WAIT;
m_nWait = 0; m_nWait = 0;
} }
@@ -162,12 +173,15 @@ bool SNaverBlogManage::Update()
m_nWait++; m_nWait++;
if (m_nWait > (1000000/m_nTime)) if (m_nWait > (1000000/m_nTime))
{ {
for(int i = 0; i < C_PROCESS_MAX ; i++) //for(int i = 0; i < C_PROCESS_MAX ; i++)
{ {
m_pro[i].kill(); {
m_pMain->InsertLog(m_nID,"Kill Process."); m_pro[0].kill();
m_pMain->InsertLog(m_nID,"Kill Process.");
}
} }
if (m_nMode == E_PROCESS_LIST_FINISH_WAIT) return m_bFinalLast; if (m_nMode == E_PROCESS_LIST_FINISH_WAIT) return m_bFinalLast;
/*
QString strQuery = "update blog set "; QString strQuery = "update blog set ";
strQuery += "Error "; strQuery += "Error ";
strQuery += "='"; strQuery += "='";
@@ -181,6 +195,7 @@ bool SNaverBlogManage::Update()
strQuery += "'"; strQuery += "'";
QSqlQuery sql; QSqlQuery sql;
sql.exec(strQuery); sql.exec(strQuery);
*/
} }
break; break;
} }