webdatafile.h

此源码功能是捕获网页中的链接并进行分析

// WebDataFile.h: interface for the CWebDataFile class.
//
//////////////////////////////////////////////////////////////////////

#ifndef _WEBDATAFILE_H_050404_
#define _WEBDATAFILE_H_050404_

#include <stdio.h>
#include <string>

using namespace std;

class CWebDataFile  
{
protected:
	bool m_bRawFormat;
	bool m_bPageError;
	bool m_bFileOpen;
	char * m_sFileName;
	FILE * m_fRaw;
	int m_nReadFileLen;

	char * m_sDataBuffer;
	int m_nBufferLen;
	char * m_sUnzipBuffer;

	int m_nDataLen;
	int m_nProcessedLen;
	bool m_bEOF;

	bool GetARawWebPage(string sFileName);

public:
	// The default buffer length to allocate: 32MB
	// You can change it first.
	// It assert that a Web page is less than the length.
	// It can be configured lager for good performance.
	static int m_nDefaultBufferLen;

protected:
	void ReadFromFile();

public:
	CWebDataFile();
	virtual ~CWebDataFile();

	// This function must be called before any other functions.
	// It's to allocate the data buffer, return FALSE in the 
	// case of insufficient memory.
	bool InitDataBuffer(unsigned int nLen = m_nDefaultBufferLen);

	// Close the Web data file.
	void CloseFile();
	// Open a Web data file.
	bool OpenFile(const char * sFileName);

	// Get a Web page for further processing.
	// return NULL if failed (end of file).
	bool GetAWebPage(string sFileName);

	// Get the processed file len.
	int GetProcessedFileLen();
	// Seek to processed length.
	void SeekToLength(int nLen);
	
	bool IsRawFormat(){
		return m_bRawFormat;
	}
	bool IsPageError(){
		return m_bPageError;
	}
};

#endif // _WEBDATAFILE_H_050404_