开通博客赚积分
发布资源赚积分
分类

源码开发语言/平台
当前位置：首页 > 源码/资料 > Internet/网络编程 > WEB源码(ASP,PHP,...) > 查看源码
CheckLinks.cpp
资源名称：htmlpars.zip　[点击查看]
上传用户：zexelpump
上传日期：2007-01-04
资源大小：22k
文件大小：3k
源码类别：

WEB源码(ASP,PHP,...)

开发平台：

Visual C++

CheckLinks.cpp：源码内容
  1. #include <windows.h>
  2. #include <iostream>
  3. #include "inet.h"
  4. #include "html.h"
  6. using namespace std;
  9. int main( int argc, char *argv[] )
  10. {
  12.     if ( argc < 2 )
  13.     {
  14.         cout << "syntax: CheckLinks <HTML filename or wildcard>" << endl;
  15. return 0;
  16.     }
  18.     HRESULT hRes;
  19.     hRes = OleInitialize( NULL );
  20.     if( !SUCCEEDED( hRes ) ) 
  21.         return 1;     
  24.     WinInet inet;
  25.     HTMLParser *php = HTMLParser::Create();
  26.     WIN32_FIND_DATA wfd;
  28. // make sure we got an internet connection
  29. if ( !inet.IsConnected() )
  30. {
  31. cerr << "Could not connect to the internet..." << endl;
  32. php->Release();
  33. OleUninitialize();
  34. return 2;
  35. }
  37. // make sure we connected to IE4 properly
  38. if ( !php->IsConnected() )
  39. {
  40. cerr << "Could not connect to Internet Explorer 4..." << endl;
  41. php->Release();
  42. OleUninitialize();
  43. return 3;
  44. }
  46. // cache the directory specified on the command line
  47. string strDir;
  48. char szDrive[_MAX_DRIVE];
  49. char szDir[_MAX_PATH];
  51. _splitpath(argv[1], szDrive, szDir, NULL, NULL );
  52. if ( szDrive[0] || szDir[0] )
  53. {
  54. strDir = szDrive;
  55. strDir += szDir;
  56. }
  59.     HANDLE hFind = FindFirstFile(argv[1], &wfd);
  61.     if ( hFind != INVALID_HANDLE_VALUE )
  62.     {
  63. string strFile;
  65.         do
  66.         {
  67.             if ( !(wfd.dwFileAttributes & FILE_ATTRIBUTE_DIRECTORY) )
  68.             {
  69.                 strFile = strDir + wfd.cFileName;
  71.                 if ( php->LoadHTMLFile(strFile.c_str()) )
  72.                 {
  73.                     cout << php->GetLinkCount() << " links" << endl;
  74.                     for ( int i = 0; i < php->GetLinkCount(); i++ )
  75.                     {
  76.                         string strURL;
  77.                         if (php->GetLinkURL(i, strURL) )
  78.                         {
  79.                             if ( !inet.CheckLink(strURL.c_str()) )
  80.                                 cout << "DEAD LINK: " << strURL << endl;
  81.                             else
  82.                                 cout << "OK: " << strURL << endl;
  83.                         }
  84.                         else
  85.                             cout << "ERROR getting link" << endl;
  86.                     }
  88.                     cout << php->GetImageCount() << " images" << endl;
  89.                     for ( i = 0; i < php->GetImageCount(); i++ )
  90.                     {
  91.                         string strURL;
  92.                         if (php->GetImageURL(i, strURL) )
  93.                         {
  94.                             if ( !inet.CheckLink(strURL.c_str()) )
  95.                                 cout << "DEAD IMG: " << strURL << endl;
  96.                             else
  97.                                 cout << "OK: " << strURL << endl;
  98.                         }
  99.                         else
  100.                             cout << "ERROR getting image" << endl;
  101.                     }
  102.                 }
  103.                 else
  104.                     cout << "Error loading html file " << strFile << endl;
  105.             }
  107.         } while ( FindNextFile(hFind, &wfd));
  109.         FindClose(hFind);
  110.     }
  113. php->Release();
  115. OleUninitialize();
  118.     return( 0 ); 
  120. }