WebResourceProvider

// WebResourceProvider.h : header file
//
// Written by Ravi Bhavnani <ravib@ravib.com>
// Copyright (c) 2002.  All Rights Reserved.
//
// This code may be used in compiled form in any way you desire. This
// file may be redistributed unmodified by any means PROVIDING it is 
// not sold for profit without the author's written consent, and 
// providing that this notice and the author's name and all copyright 
// notices remains intact. 
//
// An email letting me know how you are using it would be nice as well. 
//
// This code is provided "as is" with no expressed or implied warranty.
// The author accepts no liability for any damage/loss of business that
// this product may cause.
//
// A description of this object and instructions on how to use it are
// provided at http://www.codeproject.com/internet/WebResourceProvider.asp.
//
// Revision history:
//    See WebResourceProvider.cpp
//
//////////////////////////////////////////////////////////////////////

#ifndef _WebResourceProvider_h_
#define _WebResourceProvider_h_

//////////////////////////////////////////////////////////////////////
// CWebResourceProvider object
//

class CWebResourceProvider
{
/////////////////////////////////////////////////////////////////////////////
// Construction
public:
	CWebResourceProvider();
	virtual ~CWebResourceProvider();

/////////////////////////////////////////////////////////////////////////////
// Public methods (API)
public:
  void setAgent                       // Set agent name
    (CString strAgent)
    { m_strAgent = strAgent; }

  void setStatusWnd                   // Set status window
    (CWnd* pWndStatus)
    { m_pWndStatus = pWndStatus; }

  void fetchResource();               // Fetch resource

  DWORD getFetchStatus()              // Get fetch status
    { return m_dwFetchStatus; }

  CString getFetchError()             // Get fetch error message
    { return m_strFetchError; }

  CTime getFetchTime()                // Get time of last successful fetch
    { return m_tmFetchTime; }

  static bool urlExists               // checks if a URL exists
    (CString  strUrl);

/////////////////////////////////////////////////////////////////////////////
// Methods overridden by derived classes
public:
  virtual CString getName()           // Get provider name
    { return ""; }
  virtual CString getVersion()        // Get provider version
    { return ""; }
  virtual CString getCopyright()      // Get provider copyright
    { return ""; }
  virtual CString getOtherInfo()      // Get optional provider information
    { return ""; }

protected:
  virtual void constructUrl           // Construct the URL to be fetched
    (CString& strUrl) = 0;

  virtual bool isPost();              // Test if this is a POST request

  virtual void getPostData            // Get post data
    (CString& strPostData);

  virtual bool init();                // Initialize the provider

  virtual bool moreAvailable();       // Test if more data is available

  virtual void parseContent();        // Parse the fetched content

/////////////////////////////////////////////////////////////////////////////
// Helper methods used by derived classes
protected:
  bool at                             // check whether index is positioned
    (CString strText);                // at a string (case insensitive)

  bool atExact                        // check whether index is positioned
    (CString strText);                // at a string (case insensitive)

  bool skipTo                         // advance index to the next occurence
    (CString strText);                // of a string (case insensitive)

  bool skipToExact                    // advance index to the next occurence
    (CString strText);                // of a string (case sensitive)

  bool skipBackTo                     // retreat index to the previous occurence
    (CString strText);                // of a string (case insensitive)

  bool skipBackToExact                // retreat index to the previous occurence
    (CString strText);                // of a string (case sensitive)

  bool extractTo                      // extract text from current location to
    (CString  strTerminator,          // the start of a string and position index
     CString& strResult);             // just after string (case insensitive)

  bool extractToExact                 // extract text from current location to
    (CString  strTerminator,          // the start of a string and position index
     CString& strResult);             // just after string (case sensitive)

  void extractToEnd                   // extract text from current location to
    (CString& strResult);             // end of content

  long getIndex()                     // get current position
    { return m_nIndex; }

  void getLinks                       // get list of documents and images linked in content
    (CStringArray& documents,
     CStringArray& images);

  void resetIndex();                  // reset index to start of content

  CString replaceEvery                // case-sensitive text replacement
    (LPCTSTR  strTarget,
     LPCTSTR  strOccurence,
     LPCTSTR  strWith);

  void removeComments();              // remove comments from content

  void removeScripts();               // remove scripts from content

  void removeEnclosingAnchorTag       // remove enclosing <A> tag from string
    (CString& strText);

  void removeEnclosingQuotes          // remove enclosing quotes from string
    (CString& strText);

  void removeHtml                     // remove all HTML from string
    (CString& strText);

  void trim                           // remove leading and trailing whitespace
    (CString& strText);               // from string

/////////////////////////////////////////////////////////////////////////////
// Implementation
protected:
  static long findNoCase              // searches (case-insensitively) for a
    (CString  strString,              // substring within a string
     CString  strSubstring,
     long     nStart);

  long findStringInArray              // searches a string array for an exact match
    (CStringArray&  stringArray,
     CString        string);

  void getContent                     // retrieves the URL from the web
    (CString strUrl);

/////////////////////////////////////////////////////////////////////////////
// Members
protected:
  DWORD     m_dwFetchStatus;    // fetch status
  long      m_nIndex;           // current location
  CString   m_strAgent;         // agent name
  CString   m_strContent;       // fetched content
  CString   m_strFetchError;    // fetch error
  CString   m_strUrl;           // target url
  CWnd*     m_pWndStatus;       // status window
  CTime     m_tmFetchTime;      // fetch timestamp
};

#endif

// End WebResourceProvider.h