80 lines
2.8 KiB
C++
80 lines
2.8 KiB
C++
#ifndef SCRAWLER_H
|
|
#define SCRAWLER_H
|
|
|
|
#include <QtWebKitWidgets>
|
|
#include "scrawlerdata.h"
|
|
class SCrawler : public QObject
|
|
{
|
|
Q_OBJECT
|
|
public:
|
|
enum E_SELECT
|
|
{
|
|
E_NAVER_CAFE_LIST = 0,
|
|
E_NAVER_CAFE_DATA,
|
|
E_NAVER_BLOG_LIST,
|
|
E_NAVER_BLOG_BODY,
|
|
E_NAVER_BLOG_REPLY,
|
|
E_DAUM_CAFE_LIST,
|
|
E_DAUM_CAFE_DATA,
|
|
E_DAUM_BLOG_LIST,
|
|
E_DAUM_BLOG_BODY,
|
|
E_DAUM_BLOG_REPLY
|
|
};
|
|
public:
|
|
SCrawler();
|
|
~SCrawler();
|
|
void load(QStringList _strlistArgv);
|
|
void saveFile();
|
|
static void Debug(QString _strFilename,QString _strData);
|
|
signals:
|
|
void finished();
|
|
private slots:
|
|
void saveResult(bool ok);
|
|
private:
|
|
int m_nSelect;
|
|
QString m_strReper;
|
|
QString m_strKeywordID;
|
|
|
|
SCrawlerData bodydata;
|
|
|
|
QWebPage *m_page;
|
|
QString m_strFile;
|
|
QString m_strUrl;
|
|
QString m_strTable;
|
|
QString m_strIp;
|
|
bool m_bUse;
|
|
bool m_bLast;
|
|
bool m_bError;
|
|
|
|
QString m_strProxyIP;
|
|
int m_nProxyPort;
|
|
|
|
QString SqlString(QString _str);
|
|
QString GetSafeUtf(QString _strData);
|
|
void saveFrameList(QWebFrame *frame);
|
|
void saveFrameCafeList(QWebFrame *frame);
|
|
void saveFrameUrl(QWebFrame *frame);
|
|
void saveFrameComment(QWebFrame *frame);
|
|
void saveFrameCafeUrl(QWebFrame *frame);
|
|
void saveFrameDaumBlogList(QWebFrame *frame);
|
|
void saveFrameDaumCafeList(QWebFrame *frame);
|
|
void saveFrameDaumBlogUrl(QWebFrame *frame);
|
|
void saveFrameDaumBlogComment(QWebFrame *frame);
|
|
void saveFrameDaumCafeUrl(QWebFrame *frame);
|
|
int GetNumber(QString _str);
|
|
bool getProxyList(QString &_str);
|
|
void setProxy();
|
|
void deleteProxy();
|
|
|
|
QWebElement Find(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind);
|
|
QWebElement FindMid(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind, const int _strStart, const int _strLength);
|
|
QWebElement FindMid(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind, const int _strStart);
|
|
QWebElement FindRight(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind);
|
|
QWebElement FindLeft(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind);
|
|
QList<QWebElement> FindAllMid(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind, const int _strStart, const int _strLength);
|
|
QWebElementCollection Finds(const QWebElement _FindElement,const QString _strElement,const QString _strAttrib,const QString _strFind);
|
|
void UpdateError(QString _strError);
|
|
};
|
|
|
|
#endif // SCRAWLER_H
|