imagedoc.cpp

网络图片收集软件

// imageDoc.cpp : implementation of the CImageDoc class
//

#include "stdafx.h"
#include "image.h"

#include "imageDoc.h"
#include "imageView.h"
#include "Progress.h"


#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif
#define SECTION	"LOCALSET"
#define ENTRY	"DIRECTORY"

/////////////////////////////////////////////////////////////////////////////
// CImageDoc

IMPLEMENT_DYNCREATE(CImageDoc, CDocument)

BEGIN_MESSAGE_MAP(CImageDoc, CDocument)
	//{{AFX_MSG_MAP(CImageDoc)
		// NOTE - the ClassWizard will add and remove mapping macros here.
		//    DO NOT EDIT what you see in these blocks of generated code!
	//}}AFX_MSG_MAP
END_MESSAGE_MAP()

/////////////////////////////////////////////////////////////////////////////
// CImageDoc construction/destruction

CImageDoc::CImageDoc()
{
	// TODO: add one-time construction code here
	//m_Option.bUseProxy=false;
	m_Option.date=0;
	m_Option.image_type=0;
	m_Option.max_length=0;
	m_Option.max_level=-1;//没有限制要下载的层数
	//m_Option.max_thread_num=-1;//没有限制要开的最大线程数
	m_Option.max_width=0;
	m_Option.min_length=0;
	m_Option.min_width=0;
	//m_Option.szProxy=_T("");
	localset.bProxy=FALSE;

	m_pProgress=NULL;
	m_pThread=NULL;
	localset.strDirectory=theApp.GetProfileString(SECTION,ENTRY,"d:\\zhang\\");
	localset.image_min_size=theApp.GetProfileInt(SECTION,_T("IMAGE_SIZE"),10);
	localset.max_thread=theApp.GetProfileInt(SECTION,_T("NUMOFTHREAD"),3);
	localset.unlike_level=theApp.GetProfileInt(SECTION,_T("LEVEL"),3);
	//localset.image_min_size=10;
	//localset.unlike_level=2;


	m_doneurl=new CDoneurl(&theApp.m_db);
	m_history=new CHistory(&theApp.m_db);
	m_todownloadurl=new CTodownloadurl(&theApp.m_db);
	m_log=new CLog(&theApp.m_db);
	m_save=new CSave(&theApp.m_db);
	m_exclusive=new CExclusive(&theApp.m_db);
	m_include=new CInclude(&theApp.m_db);

	m_log->Open();
	if(!(m_log->IsEOF()&&m_log->IsBOF()))
	{
		SITEINFO site;
		m_log->MoveFirst();
		while(!m_log->IsEOF())
		{
			site.bdownload=m_log->m_bdownload;
			site.imagetype=m_log->m_imagetype;
			site.lastnum=m_log->m_lastnum;
			site.lasttime=m_log->m_lasttime;
			site.max_height=m_log->m_max_height;
			site.max_width=m_log->m_max_width;
			site.min_width=m_log->m_min_width;
			site.min_height=m_log->m_min_height;
			site.priority=m_log->m_priority;
			site.sitename=m_log->m_sitename;
			site.status=m_log->m_status;
			site.m_exclusive_level=m_log->m_exclusive_level;
			

			site.url=m_log->m_url;

			site.max_level=m_log->m_max_level;
			m_siteinfo.Add(site);
			TRY{
				m_log->MoveNext();
			}
			CATCH(CDBException,e)
			{
				AfxMessageBox(e->m_strError);
			}
			END_CATCH

		}
		
	}
	m_log->Close();

}

CImageDoc::~CImageDoc()
{
	theApp.WriteProfileString(SECTION,ENTRY,localset.strDirectory);
	theApp.WriteProfileInt(SECTION,_T("IMAGE_SIZE"),localset.image_min_size);
	theApp.WriteProfileInt(SECTION,_T("LEVEL"),localset.unlike_level);
	theApp.WriteProfileInt(SECTION,_T("NUMOFTHREAD"),localset.max_thread);
	delete m_doneurl;
	delete m_history;
	delete m_todownloadurl;
	delete m_log;
	delete m_save;
	delete m_exclusive;
	delete m_include;
}

BOOL CImageDoc::OnNewDocument()
{
	//static bFirstTime=true;

	//是否是空的项目文件
	/*if(bFirstTime)
	{
		bFirstTime=false;

		CString strDefName;

	}
	*/
	if (!CDocument::OnNewDocument())
		return FALSE;

	// TODO: add reinitialization code here
	// (SDI documents will reuse this document)
	//以后要加上相关的判断 
	

	return TRUE;
}



/////////////////////////////////////////////////////////////////////////////
// CImageDoc serialization

void CImageDoc::Serialize(CArchive& ar)
{
	if (ar.IsStoring())
	{
		// TODO: add storing code here
	}
	else
	{
		// TODO: add loading code here
	}
}

/////////////////////////////////////////////////////////////////////////////
// CImageDoc diagnostics

#ifdef _DEBUG
void CImageDoc::AssertValid() const
{
	CDocument::AssertValid();
}

void CImageDoc::Dump(CDumpContext& dc) const
{
	CDocument::Dump(dc);
}
#endif //_DEBUG

/////////////////////////////////////////////////////////////////////////////
// CImageDoc commands


bool CImageDoc::ShouldGetLink(CString &strPage)
{
	//从数据库中查询strPage如果存在则返回false 否则返回true
	//return true;
	//下面的部分的东西是否需要跟函数ShouldQueuePage有关
	CString temp;
	temp.Format(_T("%s%s%s"),"select * from doneurl where url='",strPage,"'");

	if(m_doneurl->IsOpen())
		m_doneurl->Close();
	TRY
	{
		if(!m_doneurl->Open(CRecordset::snapshot,temp,CRecordset::readOnly))
		return false;
	}
	CATCH(CDBException,e)
	{
		AfxMessageBox(e->m_strError);
	}
	END_CATCH
	//if (!m_doneurl->Open()) {
	//	return false;
	//}
	//m_doneurl->MoveFirst();
	if(m_doneurl->IsBOF()&&m_doneurl->IsEOF())
	//if (m_doneurl-) 	
	{
		m_doneurl->Close();
		return true;
	}
	else
	{
		m_doneurl->Close();
		return false;
	}
}

//说明strFilePath为strImage保存的完整路径
bool CImageDoc::GetImage(CString &strImage, CString &strFilePath)
{
	//下载strImage指定的url,把文件保存为strFileName
	BYTE *pbyBuffer=m_byBuffer;
	int nLen;
	//BOOL bRet=false;

	CInet::RESULTS ret;
	m_Inet.m_pProgress=this->m_pProgress;
	m_Inet.m_minsize=localset.image_min_size;
	ret=m_Inet.GetFile(strImage,&pbyBuffer,nLen,&m_Option.date);
	
	if(ret==CInet::SUCCESS)
	{
		
		m_Inet.SaveFile(strFilePath,pbyBuffer,nLen);
		//下面需要把下载的信息保存到数据库中
		//............
		return true;
	}//下载成功但是由于文件太大,换需要多次读


	
	return false ;

}

bool CImageDoc::GetPage(CString &strPage, int level)
{
	//下载strPage指定的url,分析页面内的链接，把叶面链接保存到数据库中
	//把图片的链接保存到m_arrImage中，在分析完页面后开始下载，m_arrImage
	//中的图片链接，当下载图片的线程数大于m_Option.max_thread_num时本函数
	//挂起。

	BYTE *pbyBuffer=m_byBuffer;
	int nLen;
	CInet::RESULTS ret;



	m_Inet.m_pProgress=this->m_pProgress;
		ret=m_Inet.GetPage(strPage,&pbyBuffer,nLen,true);

		if(ret==CInet::IMAGE_TYPE)
		{
			//本次下载的内容是图片不需要分析只需要保存
			CString strDirectory=m_Directory;
			strDirectory+=CInet::SplitFileName(strPage,CInet::FNAME);
			strDirectory+=".jpg";
			
			if((m_Option.max_level-m_Option.exclusive_level)>=(level+1))
				if(nLen>=localset.image_min_size*1000)
				m_Inet.SaveFile(strDirectory,pbyBuffer,nLen);

			return false;
		}

		//下载成功进行页面的分析
		if(ret==CInet::SUCCESS)
		{
			CHtmlParser Parser(m_Option,localset.unlike_level);
			Parser.SetLevel(level);
			Parser.SetPageURL(m_strServer,m_strObject);
			m_pProgress->SetActionTitle("正在分析网页："+strPage);
			
			
			//下面求下载的网页的相对目录并保存到m_parenturl中共Parser的MakeQualifiedUrl使用
			Parser.m_parenturl=strPage.SpanExcluding("?");
			CString strServer,strObject,strUser,strPassword;
			INTERNET_PORT nPort;
			DWORD dwServiceType;
			char szOut[1000];
			unsigned long ulLen=sizeof(szOut);

			if(AfxParseURLEx(Parser.m_parenturl,dwServiceType,strServer,strObject,nPort,
				strUser,strPassword,ICU_NO_ENCODE))
			{
				int kk=strObject.ReverseFind('/');
				if(kk!=-1)
					strObject=strObject.Left(kk+1);
				else
					strObject="/";

			strServer="http://"+strServer;
			::InternetCombineUrl(strServer,strObject,szOut,&ulLen,ICU_NO_ENCODE);
			Parser.m_parenturl=szOut;

			}

			if(nLen>MAX_INET_BUFFER)
				nLen=MAX_INET_BUFFER;

			pbyBuffer=m_byBuffer;

			//下面开始进行分析可能需要进行初始化处理
			//比如
			Parser.ParseText((char *)pbyBuffer,nLen);
			

			m_pProgress->SetActionTitle("对分析出的链接进行处理");

			//下面需要对分析的结果进行处理
			//把分析的图片结果放到m_arrImage中 
			m_arrImage.Append(Parser.GetImage());
	
			CString sql;

			for(int i=0;i<Parser.GetArrLink().GetSize();i++)
			{
				//URL_LEVEL er=Parser.m_url_level[i];
				MakeURLValid(Parser.GetArrLink()[i].m_url);
				
				if(ShouldQueuePage(Parser.GetArrLink()[i].m_url))
				{					
					sql.Format("insert into todownloadurl values('%s',%d,%s)",Parser.GetArrLink()[i].m_url,Parser.GetArrLink()[i].level,"1990-12-16");	
				
					TRY{
						theApp.m_db.ExecuteSQL(sql);						
					}
					CATCH(CDBException,e) {
					}
					END_CATCH				
				}
			}
		return true;
		}//	if(ret==CInet::SUCCESS)
	return false;

}

//本函数的实现需要使用数据库，即从数据库中查寻看是否有相同的文件有
bool CImageDoc::GenerateUniqueFileName(CString &strName)
{
	//int i;
	CString str;
	//BOOL b;
	//不确定是否周密
	str.Format("select * from history where filename like '%s.%%'",strName);

	if(m_history->IsOpen())
		m_history->Close();

	if(!m_history->Open(CRecordset::snapshot,str,CRecordset::readOnly))
		return false;
	//m_history->MoveFirst();
	//b=m_history->IsEOF();
	//b=m_history->IsBOF();
	if(!(m_history->IsEOF()&&m_history->IsBOF()))
	{
		//srand((unsigned)time(NULL));
		//i=rand();