XSearch - a class that implements a search engine-style advanced search

// XStringSet.H: interface for the CIVStringSet class.
//
// Written 12 June 2002 by Scot T Brennecke
// Thanks to Moishe Halibard and Moshe Rubin for their article,
// "A Multiple Substring Search Algorithm" in the June 2002
// edition of C/C++ Users Journal.  This class is based on
// the algorthim therein described, but extended to return
// all strings and use MFC classes.
//
// Modified by Hans Dietrich  20 July 2005
//

#ifndef  XSTRINGSET_H 
#define  XSTRINGSET_H 

class CIVStringSet : public CStringArray
{
// Constructors / destructors
public:
	CIVStringSet(WORD wInitialWidth = 256);	// Initial width of FSM
	virtual ~CIVStringSet();

// Attributes
public:
	void	SetMatchCase(BOOL bFlag)  { m_bMatchCase = bFlag; }
	void	SetWholeWords(BOOL bFlag) { m_bWholeWords = bFlag; }
	void	SetFirstMatch(BOOL bFlag) { m_bFirstMatch = bFlag; }

// Operations
public:
	BOOL	Add(LPCTSTR pszWord);							// a single word
	UINT	FindFirstIn(CString strText, int & rnFirst);	// begin iteration
	UINT	FindNext(int & rnNext);							// continue interation

// Implementation
protected:
	DWORD	(* m_apnFSM)[128];	// Finite State Machine. Array of 128 char arrays
	size_t  m_nCurDim;			// Dimension of allocated width of FSM
	size_t  m_nUsedCols;		// Used portion of allocated width
	WORD	m_wMaxUsedState;	// largest state value used
	CString m_strSearch;		// Current search string
	UINT	m_nCurTextChar;		// Current position in search string
	BOOL	m_bMatchCase;		// TRUE = match case
	BOOL	m_bWholeWords;		// TRUE = match whole words only
	BOOL	m_bFirstMatch;		// TRUE = stop matching after first match
	BOOL	m_bMatchFound;		// TRUE = at least one word has been found

	BOOL	InsertWord(LPCTSTR pszWord, WORD wIndex);	// put the new word into 
														// the FSM for given index
	BOOL	SetColDim(size_t nNewDim);					// set the current width 
														// to at least nNewDim columns
	BOOL	IsWordChar(TCHAR c);						// returns TRUE if char is a 
														// word separator (a-z, 0-9, _)

private:
	// =====  Hide several base class members that shouldn't be used  =====
	//
	// Using these members could cause the index numbers of strings to change, 
	// throwing off the embedded index entries in the FSM
	void SetAt(int nIndex, LPCTSTR newElement);
	void SetAt(int nIndex, const CString & newElement);
	void SetAtGrow(int nIndex, LPCTSTR newElement);
	void SetAtGrow(int nIndex, const CString & newElement);
	int  Append(const CStringArray & src);
	void Copy(const CStringArray & src);
	void InsertAt(int nIndex, LPCTSTR newElement, int nCount = 1);
	void InsertAt(int nIndex, const CString & newElement, int nCount = 1);
	void RemoveAt(int nIndex, int nCount = 1);
	void InsertAt(int nStartIndex, CStringArray * pNewArray);
};

#endif // XSTRINGSET_H