tanalyselink.cpp

本程序是VC为平台开发的股票资讯系统

// TAnalyseLink.cpp : implementation file
//

#include "stdafx.h"
#include "TAnalyseLink.h"

#ifdef _DEBUG
#define new DEBUG_NEW
#undef THIS_FILE
static char THIS_FILE[] = __FILE__;
#endif

/////////////////////////////////////////////////////////////////////////////
// TAnalyseLink
//-----------
  TAnalyseLink::TAnalyseLink()
 {
	  NewsDate=CTime::GetCurrentTime();
 }
//----------
CString  TAnalyseLink::ExtractLinkAndTitle(CString temp,bool Brief,bool Repeat)
{
   CString TempString,LinkString="",filter=FUrlCharacter.TitleFilter;
   CString TempBrief="简介：";
   TempBrief=TempBrief+'\n';
   bool Ahead=false;
   int pos=1,pos1=0;
   pos=FUrlCharacter.SearchKey.Find("&");     //pos=FUrlCharacter.SearchKey.AnsiPos("&");
   SearchHref=FUrlCharacter.SearchKey.Mid(0,pos);  // SearchHref=FUrlCharacter.SearchKey.SubString(1,pos-1);
   pos1=FUrlCharacter.SearchKey.Find("@");     //pos1=FUrlCharacter.SearchKey.AnsiPos("@");
   if (pos==-1||pos1==-1) return "SearchKeyError"; //if (pos==0||pos1==0) return "SearchKeyError";
   SearchA=FUrlCharacter.SearchKey.Mid(pos+1,pos1-pos-1); //SearchA=FUrlCharacter.SearchKey.SubString(pos+1,pos1-pos-1);
   pos=FUrlCharacter.SearchKey.Find("@@");  //pos=FUrlCharacter.SearchKey.AnsiPos("@@");
   if (pos==-1) TempKey="";          //if (pos==0) TempKey="";
   else {
       pos1=FUrlCharacter.SearchKey.GetLength();
       TempKey=FUrlCharacter.SearchKey.Mid(pos+1,pos1-pos-1);  
	   //TempKey=FUrlCharacter.SearchKey.SubString(pos+1,pos1-pos);
   }
   pos=FUrlCharacter.BaseLink.Find("@@");  // pos=FUrlCharacter.BaseLink.AnsiPos("@@");
   if (pos!=-1) {                  //if (pos!=0) {
       CString link,title;
        link=FUrlCharacter.BaseLink.Mid(0,pos); //link=FUrlCharacter.BaseLink.SubString(1,pos-1);
 //       title=FUrlCharacter.BaseLink.SubString(pos+2,
 //            FUrlCharacter.BaseLink.Length()-pos-1);
        title=FUrlCharacter.BaseLink.Mid(pos+2,
             FUrlCharacter.BaseLink.GetLength()-pos-2);

        TempBaseLink=link+NewsDate.Format(title);
   }
   else TempBaseLink=FUrlCharacter.BaseLink;
   pos=1;
   if (!FUrlCharacter.VerifyDate.IsEmpty()) {
   //    TempString=NewsDate.FormatString(FUrlCharacter.VerifyDate.c_str());
       TempString=
		   NewsDate.Format(FUrlCharacter.VerifyDate);
       if (temp.Find(TempString)==-1)      //if (temp.AnsiPos(TempString)==0)
           return "";
   }
   if (!filter.IsEmpty()&&filter[0]=='&') {
   // if (!filter.IsEmpty()&&filter[1]=='&')
   //  filter=filter.Delete(1,1);
       filter.Delete(0,1);
       filter=NewsDate.Format(filter);
   }
   else  filter="";
   if (!FUrlCharacter.StartPos.IsEmpty()) {
       pos=temp.Find(FUrlCharacter.StartPos); //pos=temp.AnsiPos(FUrlCharacter.StartPos);
       if (pos==-1) return "CharacterChanged";  //if (pos==0) return "CharacterChanged";
       temp.Delete(0,pos+1);               //temp=temp.Delete(1,pos);
   }
   if (!FUrlCharacter.FinishPos.IsEmpty()) {
      pos=temp.Find(FUrlCharacter.FinishPos);  //pos=temp.AnsiPos(FUrlCharacter.FinishPos);
      if (pos==-1) return "CharacterChanged";   //if (pos==0) return "CharacterChanged";
      temp=temp.Mid(0,pos+1);               //temp=temp.SubString(1,pos);
   }
   if (TempKey=="@") {
       TempKey="";
       if (Brief) { Ahead=true; Brief=false;}
   }
   else TempKey.Delete(0,1);       // TempKey=TempKey.Delete(1,1);
   if (TempKey.IsEmpty()) {
      pos=temp.Find(SearchHref);  //pos=temp.AnsiPos(SearchHref);
      if (pos==-1) return "CharacterChanged";  //if (pos==0)
      while (pos>-1) {
          CString sstr="";
          temp.Delete(0,pos); // temp=temp.Delete(1,pos-1);
          pos=temp.Find(SearchA); //  pos=temp.AnsiPos(SearchA);
          if (pos==-1)  break;      //if (pos==0)
   //     TempString=temp.SubString(1,pos+SearchA.GetLength()-1);
   //     temp=temp.Delete(1,pos);
          TempString=temp.Mid(0,pos+SearchA.GetLength());
          temp.Delete(0,pos+1);
          pos=temp.Find(SearchHref);  //pos=temp.AnsiPos(SearchHref);
          if (Brief) {
              TempBrief="";
              TempBrief=temp.Mid(3,pos-6);   //TempBrief=temp.SubString(4,pos-7);
              TempBrief="简介："+GetBrief(TempBrief)+'\n';
          }

          if (!filter.IsEmpty())  {
             pos1=temp.Find(filter); //pos1=temp.AnsiPos(filter);
            if (pos1==-1) break;     //if (pos1==0)
            if (pos1>pos) continue;
          }
          if (Repeat)
              sstr= LinkAndTitle(TempString,"ada",true);
          else  sstr= LinkAndTitle(TempString,LinkString,true);
          if (sstr.IsEmpty()) {
             if (Ahead) {
                TempBrief="";
                TempBrief="<"+temp.Mid(0,pos+1);  //TempBrief="<"+temp.SubString(1,pos);
                TempBrief="简介："+GetBrief(TempBrief)+'\n';
             }
              continue;
          }
          LinkString=LinkString+sstr+TempBrief;
          if (Ahead) {
             TempBrief="";
             TempBrief="<"+temp.Mid(0,pos+1);   //TempBrief="<"+temp.SubString(1,pos);
             TempBrief="简介："+GetBrief(TempBrief)+'\n';
          }
      }
    }
    else  {
          pos1=0;
          CString TempTitle="",TempLink;
          int ll=TempKey.GetLength();
          pos=temp.Find(TempKey);    // pos=temp.AnsiPos(TempKey);
          if (pos==-1) return "CharacterChanged"; //if (pos==0)
          while (pos>1) {                     //while (pos>0)
                temp.Delete(0,pos+ll);  //temp=temp.Delete(1,pos+ll-1);
                pos=temp.Find(SearchA); //pos=temp.AnsiPos(SearchA);
                if (pos==-1)  break;     //if (pos==0)
         //       TempString=temp.SubString(1,pos+SearchA.GetLength()-1);
                TempString=temp.Mid(0,pos+SearchA.GetLength());
                temp.Delete(0,pos+1);    // temp=temp.Delete(1,pos);
                pos=temp.Find(TempKey);   //pos=temp.AnsiPos(TempKey);
                if (Brief) {
                   TempBrief="";
                   TempBrief=temp.Mid(3,pos-3); //TempBrief=temp.SubString(4,pos-4);
                   TempBrief="简介："+GetBrief(TempBrief)+'\n';
                }

                if (!filter.IsEmpty())  {
                    pos1=TempString.Find(filter);   //pos1=TempString.AnsiPos(filter);
                    if ((pos1==-1)||(pos1>pos)) continue; //if ((pos1==0)||(pos1>pos))
                }
                pos1=TempString.Find(SearchHref);   //pos1=TempString.AnsiPos(SearchHref);
                TempTitle=TempString.Mid(0,pos1-7);  //TempTitle=TempString.SubString(1,pos1-8);
                TempTitle=ClearTitle(TempTitle,FUrlCharacter.ClearString);
                TempString.Delete(0,pos1);  //TempString=TempString.Delete(1,pos1-1);
                TempLink=LinkAndTitle(TempString,LinkString,true);
                if (TempLink.IsEmpty()) continue;
                pos1=TempLink.Find('\n');   //pos1=TempLink.AnsiPos('\n');
                TempLink=TempLink.Mid(0,pos1+1);   //TempLink=TempLink.SubString(1,pos1);
                LinkString=LinkString+TempLink+TempTitle+'\n'+TempBrief;
          }
    }
    return LinkString;

}

//-------------------------
CString  TAnalyseLink::LinkAndTitle(CString temp,CString TempLink,bool Flag)
{
    int pos=0,pos1=0,length=SearchHref.GetLength();
    CString link,title;
    pos=temp.Find("javascript");    //pos=temp.AnsiPos("javascript");
    if (pos==-1) pos=temp.Find("Javascript");   //if (pos==0)
    if (pos==-1) pos=temp.Find("JavaScript");
    if (pos==-1) pos=temp.Find("JAVASCRIPT");
    if (pos!=-1) {                  // if (pos!=0)
        pos=temp.Find(">") ;     //pos=temp.AnsiPos(">") ;
        link=temp.Find(length,pos-length);    //link=temp.SubString(length+1,pos-length-1);
        pos=link.Find("../") ;     //pos=link.AnsiPos("../") ;
        pos1=link.Find("./");      //pos1=link.AnsiPos("./");
        if (pos!=-1)                 //if (pos!=0)
           link.Insert(pos,FUrlCharacter.BaseLink); //link=link.Insert(FUrlCharacter.BaseLink,pos);
        else if (pos1!=-1)             //if (pos1!=0)
           link.Insert(pos1,FUrlCharacter.BaseLink);// link=link.Insert(FUrlCharacter.BaseLink,pos1)
    }
    else {
        if (temp[length]=='"')  {     // if (temp[length+1]=='"')
          temp.Delete(0,length+1);       //temp=temp.Delete(1,length+1);
          pos=temp.Find('"');        //pos=temp.AnsiPos('"');
          if (pos<3) {                 // if (pos<4)
              temp.Delete(0,pos+1);   //temp=temp.Delete(1,pos);
              pos=temp.Find('"');   //pos=temp.AnsiPos('"');
          }
          link=temp.Mid(0,pos);     //link=temp.SubString(1,pos-1);
      }
   //    else if (temp.SubString(length+1,1)=="'") {
      else if (temp.Mid(length,1)=="'") {
      //      temp=temp.Delete(1,length+1);
              temp.Delete(0,length);
              pos=temp.Find("'");     //pos=temp.AnsiPos("'");
              if (pos<3) {            // if (pos<4)
                 temp.Delete(0,pos+1);    //temp=temp.Delete(1,pos);
                 pos=temp.Find("'");    //pos=temp.AnsiPos("'");
              }
              link=temp.Mid(0,pos);   //link=temp.SubString(1,pos-1);
      }
   //   else if (temp.SubString(length+1,1)=="\\")  {
       else if (temp.Mid(length,1)=="\\")  {
      //        temp=temp.Delete(1,length+1);            //=9
              temp=temp.Delete(0,length+1);
              pos=temp.Find("\\");          //pos=temp.AnsiPos("\\");
              if (pos<3) {                   // if (pos<4)
                 temp.Delete(0,pos+1);    //temp=temp.Delete(1,pos);
                 pos=temp.Find('"');    //pos=temp.AnsiPos('"');
              }
              link=temp.Mid(0,pos);  //link=temp.SubString(1,pos-1);
      }
      else {
           temp.Delete(0,length);    // temp=temp.Delete(1,length);
           pos=temp.Find(' ');     //pos=temp.AnsiPos(' ');
           pos1=temp.Find('>');      //pos1=temp.AnsiPos('>');
           if (pos<3||pos==-1||pos>pos1) pos=pos1;   //if (pos<4||pos==0||pos>pos1)
           link=temp.Mid(0,pos);             //link=temp.SubString(1,pos-1);
      }
    }
    link=ClearLink(link);
    if (TempLink.Find(link)!=-1){      //if (TempLink.AnsiPos(link)!=0)
        if (Flag)  return "";
        else return "Already Exist";
    }
    if (IsVidLink(link,FUrlCharacter.LinkFilter))    {
       pos=temp.Find(">");        //pos=temp.AnsiPos(">");
       //title=temp.Delete(1,pos);
       title=temp;
       title.Delete(0,pos+1);
   //    pos=temp.AnsiPos(SearchA);
       pos=title.Find(SearchA);
       title = title.Mid(0,pos);    //title=temp.SubString(1,pos-1);
       title=ClearTitle(title,FUrlCharacter.ClearString);
       if ((link.Find("http")!=-1)||(link.Find("HTTP")!=-1)||(link.Find("Http")!=-1))
//       if ((link.AnsiPos("http")!=0)||(link.AnsiPos("HTTP")!=0)||(link.AnsiPos("Http")!=0))
           link=link+'\n';
       else
           link=TempBaseLink+link+'\n';
       if (!IsVidTitle(title,FUrlCharacter.TitleFilter))
           title="";
       if (title.IsEmpty())
          return "";
       else
          return link+title+'\n';