삭제된 게시물에 접근시 alert 뜨는 문제 해결 및 오류처리

git-svn-id: svn://192.168.0.12/source@296 8346c931-da38-4b9b-9d4c-e48b93cbd075

CrawlerProcess/scrawler.cpp

@@ -9,6 +9,18 @@
 #include <QRegExp>
 #include <ctime>
 
+class SWebPage:public QWebPage
+{
+public:
+    SWebPage(QObject * parent = 0): QWebPage(parent){}
+protected:
+    void javaScriptAlert(QWebFrame * frame, const QString & msg){
+        std::cout << "deletedurl";
+        exit(1);
+    }
+    //bool javaScriptConfirm(QWebFrame * frame, const QString & msg){}
+};
+
 using namespace std;
 
 const int RETRY_MAX = 4;
@@ -22,7 +34,7 @@ struct SProxyList
 
 SCrawler::SCrawler():QObject()
 {
-    m_page = new QWebPage;
+    m_page = new SWebPage;
     m_nRetryCount = 0;
     m_bProcessed = false;
     connect(m_page, SIGNAL(loadFinished(bool)), this, SLOT(saveResult(bool)));
@@ -587,7 +599,7 @@ enum E_DATA
 bool SCrawler::saveFrameUrl(QWebFrame *frame)
 {    
     //static int cz = 0;
-  //  Debug(frame->frameName() + QString::number(cz++) + ".html",frame->toHtml());
+    //Debug(frame->frameName() + QString::number(cz++) + ".html",frame->toHtml());
 
     //QSqlQuery sql;
     if (frame->frameName().compare(QString("BuddyConnectIframe")) == 0)
@@ -601,6 +613,7 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
     {        
         QString str[E_DATA_MAX];
         QString sympathy;
+        QString numofReply;
         QString strProfile;
         QWebElement proTitle = Find(frame->documentElement(),"meta","property","og:article:author");
         str[E_DATA_PLATFORM_TITLE] = proTitle.attribute("content").split("|").at(1).trimmed();
@@ -661,6 +674,15 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
         }
         {
             QWebElement post = Find(frame->documentElement(),"div","id","postListBody");
+            {
+                QWebElement weCmt = post.findFirst("a[class^='pcol2 _cmtList']");
+                if (!weCmt.isNull())
+                {
+                    numofReply = weCmt.toPlainText().replace(",", "").trimmed();
+                    numofReply = numofReply.replace(QRegExp("[\\D]"), "");
+                }
+            }
+
             QWebElement post_top = Find(post,"table","class","post-top");
             {
                 QWebElement title = Find(post_top,"span","class","pcol1 itemSubjectBoldfont");
@@ -679,6 +701,7 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
                 }
             }
 
+
             {
                 QWebElement date = Find(post_top,"p","class","date fil5 pcol2 _postAddDate");
                 if(date.isNull())
@@ -762,8 +785,13 @@ bool SCrawler::saveFrameUrl(QWebFrame *frame)
         bodydata.setData(m_strUrl, bodydata.ARTICLE_URL);
         bodydata.setTable(m_strTable);
 
-        bodydata.setData(sympathy, bodydata.ARTICLE_HIT);
 
+
+        bodydata.setData(sympathy, bodydata.ARTICLE_HIT); //original data
+        /*
+        bodydata.setData(numofReply, bodydata.ARTICLE_ORDER);
+        bodydata.setData(sympathy, bodydata.REPLY_URL);
+        */
         bodydata.setData("naver", bodydata.PLATFORM_NAME);
         bodydata.setData("blog", bodydata.PLATFORM_FORM);
         bodydata.setData("body", bodydata.ARTICLE_FORM);
@@ -1047,7 +1075,7 @@ void SCrawler::saveFrameCafeUrl(QWebFrame *frame)
     if (frame->frameName() == "cafe_main")
     {
         {
-            QString strData,strDate,strNick,strID,strHits,strTitle;
+            QString strData,strDate,strNick,strID,strHits,strTitle,strReply,strLike;
             {
                 QWebElement group = Find(frame->documentElement(),"div","class","tbody m-tcol-c");
                 strData = SqlString(group.toPlainText().trimmed());
@@ -1096,6 +1124,15 @@ void SCrawler::saveFrameCafeUrl(QWebFrame *frame)
 
             }            
             strHits = Find(frame->documentElement(),"span","class","b m-tcol-c reply _rosReadcount").toPlainText();
+            {
+                strReply = frame->documentElement().findFirst("td.reply").toPlainText().replace(",", "").trimmed();
+                strReply = strReply.replace(QRegExp("[\\D]"), "");
+            }
+            {
+                strLike = frame->documentElement().findFirst("a#upArticleLink").toPlainText().replace(",", "").trimmed();
+            }
+
+
             if (strHits.isEmpty())
             {
                 strHits = Find(frame->documentElement(),"span","class","kin_count m-tcol-c _rosReadcount").toPlainText();
@@ -1114,7 +1151,9 @@ void SCrawler::saveFrameCafeUrl(QWebFrame *frame)
                 bodydata.setData("body", bodydata.ARTICLE_FORM);
                 bodydata.setData(m_strUrl.split("/").at(3), bodydata.PLATFORM_ID);
                 bodydata.setData(m_strKeywordID, bodydata.KEYWORD_ID);
+                bodydata.setData(strReply, bodydata.ARTICLE_ORDER);
                 bodydata.setData(strTitle, bodydata.ARTICLE_TITLE);
+                bodydata.setData(strLike, bodydata.REPLY_URL);
                 /*
                 QSqlQuery sql;
                 QString strQuery = "update ";
@@ -2112,7 +2151,7 @@ bool SCrawler::saveFrameNewsComment(QWebFrame *frame)
                 QString strID, strNick, strData, strDate;
                 strData = Find(sub_li, "span", "class", "u_cbox_contents").toPlainText();
                 strNick = strID = Find(sub_li, "span", "class", "u_cbox_name").toPlainText();
-                strDate = Find(comment_box, "span", "class", "u_cbox_date").toPlainText();
+                strDate = Find(sub_li, "span", "class", "u_cbox_date").toPlainText();
                 if(strDate.contains(":"))
                     strDate += ":00";
                 else
@@ -2401,6 +2440,9 @@ void SCrawler::deleteProxy()
     QSqlQuery sqlquery;
     QString strquery = "delete from Proxy where proxy = '" + m_strProxyIP + "' and port = " + QString::number(m_nProxyPort);
     if(sqlquery.exec(strquery.toUtf8()) == false)
+    {
         cout << "Error : " << strquery.toStdString() << endl;
+        cout << sqlquery.lastError().text().toStdString() << endl;
+    }
 }