图片多多媒体路径搜索

#ifndef _PARSEIMAGE_H_
#define _PARSEIMAGE_H_

//<img src=[ ' "]
class CImageTagParser
{
 const char* m_pNextPos;
public:

 enum
 {
  IMAGE_TYPE=0,
  MEDIA_TYPE
 };

 CImageTagParser(const char* pData)
 {
  m_pNextPos = pData;
 }
 ~CImageTagParser()
 {
 }
 const char* ParseNext(int &nLen,int &type);//0 图片，1 多媒体
private:
 const char* FindSRC(int &nLen);
 const char* FindSRCLink(int &nLen);
};

#endif

 

 

#include <stdio.h>
#include "ParseImage.h"

//<img src=[ ' "]

const char* CImageTagParser::ParseNext(int &nLen,int &type)
{
 nLen = 0;
 if( m_pNextPos == NULL )
  return NULL;
 const char* pTagLeft = NULL;

 int nTagMatchedStep = 0;
 for(  ; *m_pNextPos != '/0'; ++m_pNextPos)
 {
  char ch = *m_pNextPos;
  //embed
  if( ch == '<' )
  {
   if( ( *(m_pNextPos+1) == 'i' || *(m_pNextPos+1) == 'I')
    &&  *(m_pNextPos+2) != '/0'&& *(m_pNextPos+3) != '/0'  )
   {
    //确保有4个字符
    pTagLeft = m_pNextPos + 1;
    m_pNextPos += 4;
    /*if( *(pTagLeft) != 'i'&& *(pTagLeft) != 'I' )
    continue;*/
    pTagLeft++;
    if( *(pTagLeft) != 'm' && *(pTagLeft) != 'M' )
     continue;
    pTagLeft++;
    if( *(pTagLeft) != 'g' && *(pTagLeft) != 'G' )
     continue;
    //到这里说明是一个完整的<IMG,查找SRC
    type = IMAGE_TYPE;//图片
    return FindSRC(nLen);
   }
   else if( *(m_pNextPos+1) == 'e' || *(m_pNextPos+1) == 'E'
    &&  *(m_pNextPos+2) != '/0'&& *(m_pNextPos+3) != '/0' 
    && *(m_pNextPos+4) != '/0' && *(m_pNextPos+5) != '/0')
   {
    //确保有5个字符
    pTagLeft = m_pNextPos + 1;
    m_pNextPos += 6;
    /*if( *(pTagLeft) != 'e'&& *(pTagLeft) != 'E' )
    continue;*/
    pTagLeft++;
    if( *(pTagLeft) != 'm' && *(pTagLeft) != 'M' )
     continue;
    pTagLeft++;
    if( *(pTagLeft) != 'b' && *(pTagLeft) != 'B' )
     continue;
    pTagLeft++;
    if( *(pTagLeft) != 'e' && *(pTagLeft) != 'E' )
     continue;
    pTagLeft++;
    if( *(pTagLeft) != 'd' && *(pTagLeft) != 'D' )
     continue;
    //到这里说明是一个完整的<EMBED,查找SRC
    type = MEDIA_TYPE;//多媒体
    return FindSRC(nLen);
   }

  }
 }
 return NULL;
}

const char* CImageTagParser::FindSRC(int &nLen)
{
 for(  ; *m_pNextPos != '/0'; ++m_pNextPos)
 {
  if( *m_pNextPos == ' ' )
  {
   int nStep = 0;
   for( ; *m_pNextPos != '/0'; ++m_pNextPos )
   {
    switch(*m_pNextPos)
    {
    case 's':
    case 'S':
     if( nStep != 0 )
     {
      ++m_pNextPos;
      goto OUT_FOR;
     }
     nStep = 1;
     break;
    case 'r':
    case 'R':
     if( nStep != 1 )
     {
      ++m_pNextPos;
      goto OUT_FOR;
     }
     nStep = 2;
     break;
    case 'c':
    case 'C':
     if( nStep != 2 )
     {
      ++m_pNextPos;
      goto OUT_FOR;
     }
     nStep = 3;
     break;
    case ' ':
     if( nStep == 1 || nStep == 2)
     {
      ++m_pNextPos;
      goto OUT_FOR;
     }

     break;
    case '=':
     if( nStep != 3 )
     {
      ++m_pNextPos;
      goto OUT_FOR;
     }
     ++m_pNextPos;
     //找SRC=连接
     return FindSRCLink(nLen);
     break;
    default:
     goto OUT_FOR;

    }
   }
  }
OUT_FOR:
  ; 
 }
 return NULL;
}

const char* CImageTagParser::FindSRCLink(int &nLen)
{
 int nStep = 0;
 const char* pSrc = NULL;
 for(  ; *m_pNextPos != '/0'; ++m_pNextPos)
 {
  switch( *m_pNextPos )
  {
  case ' ':
   if( nStep != 0 && nStep != 1 )
   {
    //直接输出这段
    ++m_pNextPos;
    return pSrc;
   }
   nStep = 1;
   break;
  case '/'':
  case '"':
   if(nStep == 2 || nStep == 3 )
   {
    ++m_pNextPos;
    return pSrc;
   }
   nStep = 2;
   break;
  case '>':
   ++m_pNextPos;
   return pSrc;
  default:
   if( nStep != 3 )
    pSrc = m_pNextPos;
   nStep = 3;
   nLen++;
   break;
  }
 }
 return NULL;
}

/*
int _tmain(int argc, _TCHAR* argv[])
{
char * buf = new char[1024*1024];
if( buf == NULL )
return 0;

WIN32_FIND_DATAA wd;
HANDLE hFind =FindFirstFileA("biturldata//*.*", &wd);
if( hFind == INVALID_HANDLE_VALUE )
{
delete[] buf;
printf("failed to open file/n");
return 0;
}
do
{
if( wd.dwFileAttributes &FILE_ATTRIBUTE_DIRECTORY )
continue;
char szFilePath[256] = {0 };
strcpy(szFilePath, "biturldata//");
strcat(szFilePath, wd.cFileName);
printf("----------------------------%s----------------------------/n", wd.cFileName);
FILE* fp = fopen(szFilePath, "r");
if( fp != NULL )
{
memset(buf, 0, 1024 * 1024);
fread(buf, 1, 1024*1024, fp);
fclose(fp);
CImageTagParser p(buf);

int n = 0;
const char* t = NULL;
while( (t = p.ParseNext(n) ) != NULL  )
{
char szTest[1024] = { 0 };
memcpy(szTest, t, n);
printf("%s/n", szTest);
}
}
system("pause");
}while( FindNextFileA(hFind,&wd) );
FindClose(hFind);
return 0;
}
*/
//if( nTagMatchedStep < 4 )
//   { //没找到<IMG之前
//    switch( *m_pNextPos )
//    {
//    case '<':
//     nTagMatchedStep = nTagMatchedStep == 0 ? nTagMatchedStep + 1 : 0;
//     break;
//    case 'i':
//    case 'I':
//     nTagMatchedStep = nTagMatchedStep == 1 ? nTagMatchedStep + 1 : 0;
//     break;
//    case 'm':
//    case 'M':
//     nTagMatchedStep = nTagMatchedStep == 2 ? nTagMatchedStep + 1 : 0;
//     break;
//    case 'g':
//    case 'G':
//     nTagMatchedStep = nTagMatchedStep == 3 ? nTagMatchedStep + 1 : 0;
//     break;
//
//    }
//   }
//   else
//   {
//    //找到了img
//
//   }