deelx.h

一个方便实用的正则表达式类库

	int m_byes;
};

//
// Implementation
//
template <class CHART> CBoundaryElxT <CHART> :: CBoundaryElxT(int ntype, int byes)
{
	m_ntype = ntype;
	m_byes  = byes;
}

template <class CHART> int CBoundaryElxT <CHART> :: Match(CContext * pContext) const
{
	const CHART * pcsz  = (const CHART *)pContext->m_pMatchString;
	int npos = pContext->m_nCurrentPos;
	int tlen = pContext->m_pMatchStringLength;

	CHART chL = npos > 0    ? pcsz[npos - 1] : 0;
	CHART chR = npos < tlen ? pcsz[npos    ] : 0;

	int bsucc = 0;

	switch(m_ntype)
	{
	case BOUNDARY_FILE_BEGIN:
		bsucc = (npos <= 0);
		break;

	case BOUNDARY_FILE_END:
		bsucc = (npos >= tlen);
		break;

	case BOUNDARY_FILE_END_N:
		bsucc = (npos >= tlen) || (pcsz[tlen-1] == RCHART('\n') && (npos == tlen-1 || (pcsz[tlen-2] == RCHART('\r') && npos == tlen-2)));
		break;

	case BOUNDARY_LINE_BEGIN:
		bsucc = (npos <= 0   ) || (chL == RCHART('\n')) || ((chL == RCHART('\r')) && (chR != RCHART('\n')));
		break;

	case BOUNDARY_LINE_END:
		bsucc = (npos >= tlen) || (chR == RCHART('\r')) || ((chR == RCHART('\n')) && (chL != RCHART('\r')));
		break;

	case BOUNDARY_WORD_BEGIN:
		bsucc = ! IsWordChar(chL) &&   IsWordChar(chR);
		break;

	case BOUNDARY_WORD_END:
		bsucc =   IsWordChar(chL) && ! IsWordChar(chR);
		break;

	case BOUNDARY_WORD_EDGE:
		bsucc =   IsWordChar(chL) ?  ! IsWordChar(chR) : IsWordChar(chR);
		break;
	}

	return m_byes ? bsucc : ! bsucc;
}

template <class CHART> int CBoundaryElxT <CHART> :: MatchNext(CContext *) const
{
	return 0;
}

template <class CHART> inline int CBoundaryElxT <CHART> :: IsWordChar(CHART ch)
{
	return (ch >= RCHART('A') && ch <= RCHART('Z')) || (ch >= RCHART('a') && ch <= RCHART('z')) || (ch >= RCHART('0') && ch <= RCHART('9')) || (ch == RCHART('_'));
}

//
// Bracket
//
template <class CHART> class CBracketElxT : public ElxInterface  
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CBracketElxT(int nnumber, int bright);
	int CheckCaptureIndex(int & index, CContext * pContext) const;

public:
	int m_nnumber;
	int m_bright;

	CBufferT <CHART> m_szNamed;
};

template <class CHART> CBracketElxT <CHART> :: CBracketElxT(int nnumber, int bright)
{
	m_nnumber = nnumber;
	m_bright  = bright;
}

template <class CHART> inline int CBracketElxT <CHART> :: CheckCaptureIndex(int & index, CContext * pContext) const
{
	if( index >= pContext->m_capturestack.GetSize() )
		index  = pContext->m_capturestack.GetSize() - 4;

	while(index >= 0)
	{
		if(pContext->m_capturestack[index] == m_nnumber)
		{
			return 1;
		}

		index -= 4;
	}


	return 0;
}

//
// capturestack[index+0] => Group number
// capturestack[index+1] => Capture start pos
// capturestack[index+2] => Capture end pos
// capturestack[index+3] => Capture enclose z-index, zindex<0 means inner group with same name
//
template <class CHART> int CBracketElxT <CHART> :: Match(CContext * pContext) const
{
	// check, for named
	if(m_nnumber < 0) return 0;

	if( ! m_bright )
	{
		pContext->m_captureindex.Prepare(m_nnumber, -1);
		int index = pContext->m_captureindex[m_nnumber];

		// check
		if(CheckCaptureIndex(index, pContext) && pContext->m_capturestack[index+2] < 0)
		{
			pContext->m_capturestack[index+3] --;
			return 1;
		}

		// save
		pContext->m_captureindex[m_nnumber] = pContext->m_capturestack.GetSize();

		pContext->m_capturestack.Push(m_nnumber);
		pContext->m_capturestack.Push(pContext->m_nCurrentPos);
		pContext->m_capturestack.Push(-1);
		pContext->m_capturestack.Push( 0); // z-index
	}
	else
	{
		// check
		int index = pContext->m_captureindex[m_nnumber];

		if(CheckCaptureIndex(index, pContext))
		{
			if(pContext->m_capturestack[index + 3] < 0) // check inner group with same name
			{
				pContext->m_capturestack[index + 3] ++;
				return 1;
			}

			// save
			pContext->m_capturestack[index + 2] = pContext->m_nCurrentPos;
			pContext->m_capturestack[index + 3] = pContext->m_nParenZindex ++;
		}
	}

	return 1;
}

template <class CHART> int CBracketElxT <CHART> :: MatchNext(CContext * pContext) const
{
	int index = pContext->m_captureindex[m_nnumber];
	if( ! CheckCaptureIndex(index, pContext) )
	{
		return 0;
	}

	if( ! m_bright )
	{
		if(pContext->m_capturestack[index + 3] < 0)
		{
			pContext->m_capturestack[index + 3] ++;
			return 0;
		}

		pContext->m_capturestack.Restore(pContext->m_capturestack.GetSize() - 4);

		// to find
		CheckCaptureIndex(index, pContext);

		// new index
		pContext->m_captureindex[m_nnumber] = index;
	}
	else
	{
		if( pContext->m_capturestack[index + 2] >= 0 )
		{
			pContext->m_capturestack[index + 2] = -1;
			pContext->m_capturestack[index + 3] =  0;
		}
		else
		{
			pContext->m_capturestack[index + 3] --;
		}
	}

	return 0;
}

//
// Deletage
//
template <class CHART> class CDelegateElxT : public ElxInterface  
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CDelegateElxT(int ndata = 0);

public:
	ElxInterface * m_pelx;
	int m_ndata; // +0 : recursive to
	             // -3 : named recursive

	CBufferT <CHART> m_szNamed;
};

template <class CHART> CDelegateElxT <CHART> :: CDelegateElxT(int ndata)
{
	m_pelx  = 0;
	m_ndata = ndata;
}

template <class CHART> int CDelegateElxT <CHART> :: Match(CContext * pContext) const
{
	if(m_pelx != 0)
		return m_pelx->Match(pContext);
	else
		return 1;
}

template <class CHART> int CDelegateElxT <CHART> :: MatchNext(CContext * pContext) const
{
	if(m_pelx != 0)
		return m_pelx->MatchNext(pContext);
	else
		return 0;
}

//
// Empty
//
template <int x> class CEmptyElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CEmptyElxT();
};

typedef CEmptyElxT <0> CEmptyElx;

//
// Global
//
template <int x> class CGlobalElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CGlobalElxT();
};

typedef CGlobalElxT <0> CGlobalElx;

//
// Repeat
//
template <int x> class CRepeatElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CRepeatElxT(ElxInterface * pelx, int ntimes);

protected:
	int MatchFixed    (CContext * pContext) const;
	int MatchNextFixed(CContext * pContext) const;

public:
	ElxInterface * m_pelx;
	int m_nfixed;
};

typedef CRepeatElxT <0> CRepeatElx;

//
// Greedy
//
template <int x> class CGreedyElxT : public CRepeatElxT <x>
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CGreedyElxT(ElxInterface * pelx, int nmin = 0, int nmax = INT_MAX);

protected:
	int MatchVart    (CContext * pContext) const;
	int MatchNextVart(CContext * pContext) const;

public:
	int m_nvart;
};

typedef CGreedyElxT <0> CGreedyElx;

//
// Independent
//
template <int x> class CIndependentElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CIndependentElxT(ElxInterface * pelx);

public:
	ElxInterface * m_pelx;
};

typedef CIndependentElxT <0> CIndependentElx;

//
// List
//
template <int x> class CListElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CListElxT(int brightleft);

public:
	CBufferT <ElxInterface *> m_elxlist;
	int m_brightleft;
};

typedef CListElxT <0> CListElx;

//
// Posix Elx
//
template <class CHART> class CPosixElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CPosixElxT(const char * posix, int brightleft);

public:
	POSIX_FUNC m_posixfun;
	int m_brightleft;
	int m_byes;
};

//
// Implementation
//
template <class CHART> CPosixElxT <CHART> :: CPosixElxT(const char * posix, int brightleft)
{
	m_brightleft = brightleft;

	if(posix[1] == '^')
	{
		m_byes = 0;
		posix += 2;
	}
	else
	{
		m_byes = 1;
		posix += 1;
	}

	if     (!strncmp(posix, "alnum:", 6)) m_posixfun = ::isalnum ;
	else if(!strncmp(posix, "alpha:", 6)) m_posixfun = ::isalpha ;
	else if(!strncmp(posix, "ascii:", 6)) m_posixfun = ::isascii ;
	else if(!strncmp(posix, "cntrl:", 6)) m_posixfun = ::iscntrl ;
	else if(!strncmp(posix, "digit:", 6)) m_posixfun = ::isdigit ;
	else if(!strncmp(posix, "graph:", 6)) m_posixfun = ::isgraph ;
	else if(!strncmp(posix, "lower:", 6)) m_posixfun = ::islower ;
	else if(!strncmp(posix, "print:", 6)) m_posixfun = ::isprint ;
	else if(!strncmp(posix, "punct:", 6)) m_posixfun = ::ispunct ;
	else if(!strncmp(posix, "space:", 6)) m_posixfun = ::isspace ;
	else if(!strncmp(posix, "upper:", 6)) m_posixfun = ::isupper ;
	else if(!strncmp(posix, "xdigit:",7)) m_posixfun = ::isxdigit;
	else if(!strncmp(posix, "blank:", 6)) m_posixfun =   isblank ;
	else                                  m_posixfun = 0         ;
}

inline int isblank(int c)
{
	return c == 0x20 || c == '\t';
}

template <class CHART> int CPosixElxT <CHART> :: Match(CContext * pContext) const
{
	if(m_posixfun == 0) return 0;

	int tlen = pContext->m_pMatchStringLength;
	int npos = pContext->m_nCurrentPos;

	// check
	int at   = m_brightleft ? npos - 1 : npos;
	if( at < 0 || at >= tlen )
		return 0;

	CHART ch = ((const CHART *)pContext->m_pMatchString)[at];

	int bsucc = (*m_posixfun)(ch);

	if( ! m_byes )
		bsucc = ! bsucc;

	if( bsucc )
		pContext->m_nCurrentPos += m_brightleft ? -1 : 1;

	return bsucc;
}

template <class CHART> int CPosixElxT <CHART> :: MatchNext(CContext * pContext) const
{
	pContext->m_nCurrentPos -= m_brightleft ? -1 : 1;
	return 0;
}

//
// Possessive
//
template <int x> class CPossessiveElxT : public CGreedyElxT <x>
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CPossessiveElxT(ElxInterface * pelx, int nmin = 0, int nmax = INT_MAX);
};

typedef CPossessiveElxT <0> CPossessiveElx;

//
// Range Elx
//
template <class CHART> class CRangeElxT : public ElxInterface
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CRangeElxT(int brightleft, int byes);

public:
	int IsContainChar(CHART ch) const;

public:
	CBufferT <CHART> m_ranges;
	CBufferT <CHART> m_chars;
	CBufferT <ElxInterface *> m_embeds;

public:
	int m_brightleft;
	int m_byes;
};

//
// Implementation
//
template <class CHART> CRangeElxT <CHART> :: CRangeElxT(int brightleft, int byes)
{
	m_brightleft = brightleft;
	m_byes       = byes;
}

template <class CHART> int CRangeElxT <CHART> :: Match(CContext * pContext) const
{
	int tlen = pContext->m_pMatchStringLength;
	int npos = pContext->m_nCurrentPos;

	// check
	int at   = m_brightleft ? npos - 1 : npos;
	if( at < 0 || at >= tlen )
		return 0;

	CHART ch = ((const CHART *)pContext->m_pMatchString)[at];
	int bsucc = 0, i;

	// compare
	for(i=0; !bsucc && i<m_ranges.GetSize(); i+=2)
	{
		if(m_ranges[i] <= ch && ch <= m_ranges[i+1]) bsucc = 1;
	}

	for(i=0; !bsucc && i<m_chars.GetSize(); i++)
	{
		if(m_chars[i] == ch) bsucc = 1;
	}

	for(i=0; !bsucc && i<m_embeds.GetSize(); i++)
	{
		if(m_embeds[i]->Match(pContext))
		{
			pContext->m_nCurrentPos = npos;
			bsucc = 1;
		}
	}

	if( ! m_byes )
		bsucc = ! bsucc;

	if( bsucc )
		pContext->m_nCurrentPos += m_brightleft ? -1 : 1;

	return bsucc;
}

template <class CHART> int CRangeElxT <CHART> :: IsContainChar(CHART ch) const
{
	int bsucc = 0, i;

	// compare
	for(i=0; !bsucc && i<m_ranges.GetSize(); i+=2)
	{
		if(m_ranges[i] <= ch && ch <= m_ranges[i+1]) bsucc = 1;
	}

	for(i=0; !bsucc && i<m_chars.GetSize(); i++)
	{
		if(m_chars[i] == ch) bsucc = 1;
	}

	return bsucc;
}

template <class CHART> int CRangeElxT <CHART> :: MatchNext(CContext * pContext) const
{
	pContext->m_nCurrentPos -= m_brightleft ? -1 : 1;
	return 0;
}

//
// Reluctant
//
template <int x> class CReluctantElxT : public CRepeatElxT <x>
{
public:
	int Match    (CContext * pContext) const;
	int MatchNext(CContext * pContext) const;

public:
	CReluctantElxT(ElxInterface * pelx, int nmin = 0, int nmax = INT_MAX);