parsenewslink.cpp

本程序是VC为平台开发的股票资讯系统

		filter=FUrlCharacter.SearchTime.Format(filter);
		if (temp.Find(filter)==-1)
			return FALSE;
		else return TRUE;
	}	
	else return TRUE;
}

CString  CParseNewsLink::ClearLink(CString temp)
{
	int pos,n;
	pos = temp.Find('<');
	n = temp.Find('>');
	if(n != -1) temp.Delete(n,1);
	if(pos != -1)
		temp = temp.Mid(0,pos-1);
	temp = ClearChar(temp,'"');
	temp = ClearChar(temp,_T("'"));
	temp = ClearChar(temp,_T("\\"));
	temp = ClearChar(temp,'\n');

	return temp;
}

CString  CParseNewsLink::ClearChar(CString temp,CString cc)
{
   int pos;
   while (1) {
       pos=temp.Find(cc);
       if (pos!=-1)
           temp.Delete(pos,1);
       else break;
   }
   return temp;

}

void  CParseNewsLink::Copy(LinkCharacter* temp)
{
	FUrlCharacter.SearchTime=temp->SearchTime;
	FUrlCharacter.BaseUrl=temp->BaseUrl;
	FUrlCharacter.LinkFilter=temp->LinkFilter;
	FUrlCharacter.TitleFilter=temp->TitleFilter;
	FUrlCharacter.ClearString=temp->ClearString;
	FUrlCharacter.VerifyDate=temp->VerifyDate;
	FUrlCharacter.StartPos=temp->StartPos;
	FUrlCharacter.FinishPos=temp->FinishPos;  
}

CString CParseNewsLink::GetNewsBody(CString content,CString title,CString key,CString cut)
{
	CString body,tmpstr,tmptit;
	BOOL isnull = FALSE;
	if(key.IsEmpty()) isnull = TRUE;
	else key.MakeLower();

	tmptit = title;
	tmptit.MakeLower();

	body = GetAllContent(content,tmptit);
	if(body.IsEmpty()) return _T("");
	tmpstr = body;
	if(tmpstr.Find(key) != -1 || tmptit.Find(key) != -1 || isnull){
		if(cut.IsEmpty())
			return body;
		else{
			body = CutNoUse(body,cut);
			return body;
		}
	}
	else
		return _T("");
}

CString CParseNewsLink::GetAllContent(CString html,CString title)
{
	int ki1,ki2,ki3,ki4,i;
	BOOL IsRtn;
	CString str,Ht,tmptit;

	Ht = html;
	str = Ht;
	Ht.MakeLower();

/*	ki1 = Ht.Find(_T("<a href="));
	while(ki1 >= 0)					//舍去 <a href=...> 链接内容
	{
		ki2 = Ht.Find(_T("</a>"),ki1);
		if(ki2<ki1)
			break;
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 4);
		str = str.Left(ki1) + str.Mid(ki2 + 4);
		ki1 = Ht.Find(_T("<a href="));
	}
*/
	ki1 = Ht.Find(_T("<a"));
	while(ki1 != -1)
	{
		ki2 = Ht.Find(_T("href"),ki1);
		ki3 = Ht.Find(_T("<"),ki1 + 1);
		ki4 = Ht.Find(_T(">"),ki1);
		if(ki2 > ki3 || ki2 > ki4)
			ki1 = Ht.Find(_T("<a"),ki1 + 1);
		else
		{
			ki3 = Ht.Find(_T("="),ki2);
			if(ki3 < ki1) break;
			ki4 = Ht.Find(_T("</a>"),ki3);
			if(ki4 < ki1) break;			
			Ht = Ht.Left(ki1) + Ht.Mid(ki4 + 4);
			str = str.Left(ki1) + str.Mid(ki4 + 4);
			ki1 = Ht.Find(_T("<a"));
		}
	}

	ki1 = Ht.Find(_T("<script"));
	while(ki1 >= 0)					//舍去 <script...> 内容
	{
		ki2 = Ht.Find(_T("</script>"),ki1);
		if(ki2<ki1)
			break;
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 9);
		str = str.Left(ki1) + str.Mid(ki2 + 9);
		ki1 = Ht.Find(_T("<script"));
	}

	ki1 = Ht.Find(_T("<option"));
	while(ki1 >= 0)					//舍去 <option...> 内容
	{
		ki2 = Ht.Find(_T("</option>"),ki1);
		if(ki2<ki1)
			break;
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 9);
		str = str.Left(ki1) + str.Mid(ki2 + 9);
		ki1 = Ht.Find(_T("<option"));
	}

	ki1 = Ht.Find(_T("<title"));
	while(ki1 >= 0)					//舍去 <title...> 内容
	{
		ki2 = Ht.Find(_T("</title>"),ki1);
		if(ki2<ki1)
			break;
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 8);
		str = str.Left(ki1) + str.Mid(ki2 + 8);
		ki1 = Ht.Find(_T("<title"));
	}

	ki1 = Ht.Find(_T("<body"));
	if(ki1 != -1)
		Ht = Ht.Mid(ki1);
	str = str.Mid(ki1);
	title.Replace(_T("\x09"),NULL);
	title.TrimLeft(_T(" "));
	title.TrimRight(_T(" "));
	i = Ht.Find(title);
	if(i == -1){
		i = Ht.Find(title.Left(1));
		if(i == -1) return _T("");
		Ht = Ht.Mid(i);
		str = str.Mid(i);
		ki2 = Ht.Find(_T("\n"));
		ki1 = Ht.Find(_T("<"));
		if(ki1 == -1 && ki2 == -1) return _T("");
		if(ki2 < ki1){IsRtn = TRUE;ki1 = ki2;}
		else IsRtn = FALSE;
		while(ki1 != -1){
			if(IsRtn){
				Ht.Delete(ki1,1);
				str.Delete(ki1,1);
				i = Ht.Find(title);
				if(i != -1) break;
			}
			else{
				ki2 = Ht.Find('>',ki1);
				if(ki2 == -1) break;
				Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 1);
				str = str.Left(ki1) + str.Mid(ki2 +1);
				i = Ht.Find(title);
				if(i != -1) break;
			}
			ki1 = Ht.Find('<',ki1);
			ki2 = Ht.Find(_T("\n"),ki1);
			if(ki2 < ki1){IsRtn = TRUE;ki1 = ki2;}
			else IsRtn = FALSE;
		}
		if(i == -1) return _T("");
	}
	ki1 = title.GetLength();
	title = str.Mid(i,ki1);
	Ht = Ht.Mid(i+ki1);
	str = str.Mid(i+ki1);
	ki1 = title.Find(_T("  "));
	while(ki1 != -1){
		title = title.Left(ki1 + 1) + title.Mid(ki1 + 2);
		ki1 = title.Find(_T("  "));
	}
	str = title + str;
	title.MakeLower();
	Ht = title + Ht;

	Ht.Replace(_T(">"),_T(">\n"));
	str.Replace(_T(">"),_T(">\n"));

	ki1 = Ht.Find(_T("<"));
	while(ki1 >= 0)					//舍去<>内容
	{
		ki2 = Ht.Find('>',ki1);
		if(ki2 == -1) break;
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 1);
		str = str.Left(ki1) + str.Mid(ki2 +1);
		ki1 = Ht.Find('<',ki1);
	}

	str.Replace(_T("&nbsp;"),_T(" "));
	str.Replace(_T("&nbsp"),_T(" "));
	str.Replace(_T("&gt;"),_T(">"));
	str.Replace(_T("&gt"),_T(">"));
	str.Replace(_T("&lt;"),_T("<"));
	str.Replace(_T("&lt"),_T("<"));
	str.Replace(_T("&quot;"),_T("\""));
	str.Replace(_T("&quot"),_T("\""));
	str.Replace(_T("&amp;"),_T("&"));
	str.Replace(_T("&amp"),_T("&"));
	str.Replace(_T("&copy;"),_T("(C)"));
	str.Replace(_T("&copy"),_T("(C)"));
	str.Replace(_T("&reg;"),_T("(R)"));
	str.Replace(_T("&reg"),_T("(R)"));
	str.Replace(_T("&acute;"),_T("'"));
	str.Replace(_T("&acute"),_T("'"));

	str.Replace(_T("\xA1\xA1"),_T("  "));

	for(i=0;i<92;i++)
		str.Replace(ResChar[i],NULL);

	return str;
}

CString CParseNewsLink::ClearStr(CString html)
{
	CString Ht;
	int ki1,ki2;
	Ht = html;
	ki1 = Ht.Find('<');
	while(ki1 >= 0)					//舍去<>内容
	{
		ki2 = Ht.Find('>',ki1);
		Ht = Ht.Left(ki1) + Ht.Mid(ki2 + 1);
		ki1 = Ht.Find('<',ki1);
	}

	return Ht;
}

CString CParseNewsLink::CutNoUse(CString body, CString snouse)
{
	int ki,ki1,ki2,ki3,i,len;
	CString line,rcut,cut,news,nouse,tmpline,begin,end;
	COleDateTime m_Date;
	BOOL bSuccess;

	if(snouse.IsEmpty())
		return body;

	rcut = snouse;
	rcut.Replace(_T("\\7c"),_T("|"));
	ki = body.Find('\n');
	while(ki != -1)
	{
		line = body.Left(ki);
		cut = rcut;
		ki1 = cut.Find('&');
		while(ki1 != -1 && !line.IsEmpty())
		{
			i = 0;
			nouse = cut.Left(ki1);
			ki2 = nouse.Find('^');
			while(ki2 != -1)
			{
				ki3 = nouse.Find('^',ki2 + 1);
				if(ki3 != -1){
					tmpline = nouse.Mid(ki2 + 1,ki3 - ki2 - 1);
					begin.Format(_T("%c"),atoi(tmpline));
					end = nouse.Mid(ki3 + 1);
					nouse = begin + end;
				}else
					break;
				ki2 = nouse.Find('^');
			}
			len = line.GetLength();
			while(i<len && line.GetAt(i) == ' ') i++;
			ki2 = nouse.Find('%');
			if(ki2 != -1)
			{
				begin = nouse.Left(ki2);
				end = nouse.Mid(ki2 + 1);
				if(begin.IsEmpty() || line.Find(begin) == i)
				{
					if(end.Find(_T("@Date@")) == 0)
					{
						if(end.GetLength() > 6)
						{
							end = end.Mid(6);
							ki3 = line.Find(end);
							if(ki3 != -1)
							{
								tmpline = line.Left(ki3);
								tmpline.Replace(begin,NULL);
							}
							else tmpline.Empty();
						}
						else
						{
							end.Empty();
							tmpline = line.Mid(begin.GetLength());
						}
						try
						{
							bSuccess = m_Date.ParseDateTime(tmpline);
						}
						catch(...)
						{
							bSuccess = FALSE;
						}
						if(bSuccess)
						{
							line = begin + end;
							break;
						}
					}
					else
					{
						ki3 = line.Find(end);
						if(ki3 != -1)
						{
							nouse = line.Mid(i,ki3 - i + end.GetLength());
							line.Replace(nouse,NULL);
							break;
						}
					}
				}
			}
			else
			{
				ki2 = nouse.Find('#');
				if(ki2 != -1)
				{
					nouse = nouse.Mid(1);
					tmpline = line;
					tmpline.MakeLower();
					if(tmpline.Find(nouse) == i)
					{
						line = _T("");break;
					}
				}
				else
				{
					if(line.Find(nouse) == i)
					{
						line.Replace(nouse,NULL);break;
					}
				}
			}
			cut = cut.Mid(ki1 + 1);
			ki1 = cut.Find('&');
		}
		if(!line.IsEmpty()){
			tmpline = line;
			tmpline.Replace(_T(" "),NULL);
			if(!tmpline.IsEmpty()){news += line;news += '\n';}
		}
		body = body.Mid(ki + 1);
		ki = body.Find('\n');
	}

	return news;
}