⭐ 欢迎来到虫虫下载站! | 📦 资源下载 📁 资源专辑 ℹ️ 关于我们
⭐ 虫虫下载站

📄 deelx.h

📁 正则表达式库
💻 H
📖 第 1 页 / 共 5 页
字号:

		// skip '*'
		MoveNext();
		break;

	case RCHART('+'):
		nMin = 1;
		nMax = INT_MAX;

		// skip '+'
		MoveNext();
		break;

	default:
		bIsQuantifier = 0;
		break;
	}

	// do quantify
	if(bIsQuantifier)
	{
		// 0 times
		if(nMax == 0)
			return GetStockElx(STOCKELX_EMPTY);

		// fixed times
		if(nMin == nMax)
		{
			if(curr == CHART_INFO(RCHART('?'), 1) || curr == CHART_INFO(RCHART('+'), 1))
				MoveNext();

			return Keep(new CRepeatElx(pSimple, nMin));
		}

		// range times
		if(curr == CHART_INFO(RCHART('?'), 1))
		{
			MoveNext();
			return Keep(new CReluctantElx(pSimple, nMin, nMax));
		}
		else if(curr == CHART_INFO(RCHART('+'), 1))
		{
			MoveNext();
			return Keep(new CPossessiveElx(pSimple, nMin, nMax));
		}
		else
		{
			return Keep(new CGreedyElx(pSimple, nMin, nMax));
		}
	}

	return pSimple;
}

template <class CHART> ElxInterface * CBuilderT <CHART> :: BuildSimple(int & flags)
{
	CBufferT <CHART> fixed;

	while(curr != CHART_INFO(0, 1))
	{
		if(curr.type == 0)
		{
			if(next == CHART_INFO(RCHART('{'), 1) || next == CHART_INFO(RCHART('?'), 1) || next == CHART_INFO(RCHART('*'), 1) || next == CHART_INFO(RCHART('+'), 1))
			{
				if(fixed.GetSize() == 0)
				{
					fixed.Append(curr.ch, 1);
					MoveNext();
				}

				break;
			}
			else
			{
				fixed.Append(curr.ch, 1);
				MoveNext();
			}
		}
		else if(curr.type == 1)
		{
			CHART vch = curr.ch;

			// end of simple
			if(vch == RCHART(')') || vch == RCHART('|'))
				break;

			// has fixed already
			if(fixed.GetSize() > 0)
				break;

			// left parentheses
			if(vch == RCHART('('))
			{
				return BuildRecursive(flags);
			}

			// char set
			if( vch == RCHART('[') || vch == RCHART('.') || vch == RCHART('w') || vch == RCHART('W') ||
				vch == RCHART('s') || vch == RCHART('S') || vch == RCHART('d') || vch == RCHART('D')
			)
			{
				return BuildCharset(flags);
			}

			// boundary
			if( vch == RCHART('^') || vch == RCHART('$') || vch == RCHART('A') || vch == RCHART('Z') || vch == RCHART('z') ||
				vch == RCHART('b') || vch == RCHART('B') || vch == RCHART('G') // vch == RCHART('<') || vch == RCHART('>')
			)
			{
				return BuildBoundary(flags);
			}

			// backref
			if(vch == RCHART('\\') || vch == RCHART('k') || vch == RCHART('g'))
			{
				return BuildBackref(flags);
			}

			// treat vchar as char
			fixed.Append(curr.ch, 1);
			MoveNext();
		}
	}

	if(fixed.GetSize() > 0)
		return Keep(new CStringElxT <CHART> (fixed.GetBuffer(), fixed.GetSize(), flags & RIGHTTOLEFT, flags & IGNORECASE));
	else
		return GetStockElx(STOCKELX_EMPTY);
}

template <class CHART> ElxInterface * CBuilderT <CHART> :: BuildCharset(int & flags)
{
	// char
	CHART ch = curr.ch;

	// skip
	MoveNext();

	switch(ch)
	{
	case RCHART('.'):
		return GetStockElx(
			flags & RIGHTTOLEFT ?
			((flags & SINGLELINE) ? STOCKELX_DOT_ALL_RIGHTLEFT : STOCKELX_DOT_NOT_ALL_RIGHTLEFT) :
			((flags & SINGLELINE) ? STOCKELX_DOT_ALL : STOCKELX_DOT_NOT_ALL)
		);

	case RCHART('w'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_WORD_RIGHTLEFT : STOCKELX_WORD);

	case RCHART('W'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_WORD_RIGHTLEFT_NOT : STOCKELX_WORD_NOT);

	case RCHART('s'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_SPACE_RIGHTLEFT : STOCKELX_SPACE);

	case RCHART('S'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_SPACE_RIGHTLEFT_NOT : STOCKELX_SPACE_NOT);

	case RCHART('d'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_DIGITAL_RIGHTLEFT : STOCKELX_DIGITAL);

	case RCHART('D'):
		return GetStockElx(flags & RIGHTTOLEFT ? STOCKELX_DIGITAL_RIGHTLEFT_NOT : STOCKELX_DIGITAL_NOT);

	case RCHART('['):
		{
			CRangeElxT <CHART> * pRange;

			// create
			if(curr == CHART_INFO(RCHART(':'), 1))
			{
				CBufferT <char> posix;

				do {
					posix.Append(((curr.ch & (CHART)0xff) == curr.ch) ? (char)curr.ch : 0, 1);
					MoveNext();
				}
				while(curr.ch != RCHART(0) && curr != CHART_INFO(RCHART(']'), 1));

				MoveNext(); // skip ']'

				// posix
				return Keep(new CPosixElxT <CHART> (posix.GetBuffer(), flags & RIGHTTOLEFT));
			}
			else if(curr == CHART_INFO(RCHART('^'), 1))
			{
				MoveNext(); // skip '^'
				pRange = (CRangeElxT <CHART> *)Keep(new CRangeElxT <CHART> (flags & RIGHTTOLEFT, 0));
			}
			else
			{
				pRange = (CRangeElxT <CHART> *)Keep(new CRangeElxT <CHART> (flags & RIGHTTOLEFT, 1));
			}

			// parse
			while(curr != CHART_INFO(0, 1) && curr != CHART_INFO(RCHART(']'), 1))
			{
				ch = curr.ch;

				if(curr.type == 1 && (
					ch == RCHART('.') || ch == RCHART('w') || ch == RCHART('W') || ch == RCHART('s') || ch == RCHART('S') || ch == RCHART('d') || ch == RCHART('D') ||
					(ch == RCHART('[') && next == CHART_INFO(RCHART(':'), 1))
				))
				{
					pRange->m_embeds.Push(BuildCharset(flags));
				}
				else if(next == CHART_INFO(RCHART('-'), 1) && nex2.type == 0)
				{
					pRange->m_ranges.Push(ch); pRange->m_ranges.Push(nex2.ch);

					// next
					MoveNext();
					MoveNext();
					MoveNext();
				}
				else
				{
					pRange->m_chars.Push(ch);

					// next
					MoveNext();
				}
			}

			// skip ']'
			MoveNext();

			return pRange;
		}
	}

	return GetStockElx(STOCKELX_EMPTY);
}

template <class CHART> ElxInterface * CBuilderT <CHART> :: BuildRecursive(int & flags)
{
	// skip '('
	MoveNext();

	if(curr == CHART_INFO(RCHART('?'), 1))
	{
		ElxInterface * pElx = 0;

		// skip '?'
		MoveNext();

		int bNegative = 0;
		CHART named_end = RCHART('>');

		switch(curr.ch)
		{
		case RCHART('!'):
			bNegative = 1;

		case RCHART('='):
			{
				MoveNext(); // skip '!' or '='
				pElx = Keep(new CAssertElx(BuildAlternative(flags & ~RIGHTTOLEFT), !bNegative));
			}
			break;

		case RCHART('<'):
			switch(next.ch)
			{
			case RCHART('!'):
				bNegative = 1;

			case RCHART('='):
				MoveNext(); // skip '<'
				MoveNext(); // skip '!' or '='
				{
					pElx = Keep(new CAssertElx(BuildAlternative(flags | RIGHTTOLEFT), !bNegative));
				}
				break;

			default: // named group
				break;
			}
			// break if assertion // else named
			if(pElx != 0) break;

		case RCHART('P'):
			if(curr.ch == RCHART('P')) MoveNext(); // skip 'P'

		case RCHART('\''):
			if     (curr.ch == RCHART('<' )) named_end = RCHART('>' );
			else if(curr.ch == RCHART('\'')) named_end = RCHART('\'');
			MoveNext(); // skip '<' or '\''
			{
				// named number
				int nThisBackref = m_nNextNamed ++;

				CListElx    * pList  = (CListElx    *)Keep(new CListElx(flags & RIGHTTOLEFT));
				CBracketElx * pleft  = (CBracketElx *)Keep(new CBracketElx(-1, flags & RIGHTTOLEFT ? 1 : 0));
				CBracketElx * pright = (CBracketElx *)Keep(new CBracketElx(-1, flags & RIGHTTOLEFT ? 0 : 1));

				// save name
				CBufferT <CHART> & name = pleft->m_szNamed;
				CBufferT <char> num;

				while(curr.ch != RCHART(0) && curr.ch != named_end)
				{
					name.Append(curr.ch, 1);
					num .Append(((curr.ch & (CHART)0xff) == curr.ch) ? (char)curr.ch : 0, 1);
					MoveNext();
				}
				MoveNext(); // skip '>' or '\''

				// check <num>
				unsigned int number;
				char * str = num.GetBuffer();

				if( ReadDec(str, number) ? ( *str == '\0') : 0 )
				{
					pleft ->m_nnumber = number;
					pright->m_nnumber = number;

					name.Release();
				}

				// left, center, right
				pList->m_elxlist.Push(pleft);
				pList->m_elxlist.Push(BuildAlternative(flags));
				pList->m_elxlist.Push(pright);

				// for recursive
				m_namedlist.Prepare(nThisBackref);
				m_namedlist[nThisBackref] = pList;

				pElx = pList;
			}
			break;

		case RCHART('>'):
			{
				MoveNext(); // skip '>'
				pElx = Keep(new CIndependentElx(BuildAlternative(flags)));
			}
			break;

		case RCHART('R'):
			MoveNext(); // skip 'R'
			while(curr.ch != RCHART(0) && isspace(curr.ch)) MoveNext(); // skip space

			if(curr.ch == RCHART('<') || curr.ch == RCHART('\''))
			{
				named_end = curr.ch == RCHART('<') ? RCHART('>') : RCHART('\'');
				CDelegateElx * pDelegate = (CDelegateElx *)Keep(new CDelegateElx(-3));

				MoveNext(); // skip '<' or '\\'

				// save name
				CBufferT <CHART> & name = pDelegate->m_szNamed;
				CBufferT <char> num;

				while(curr.ch != RCHART(0) && curr.ch != named_end)
				{
					name.Append(curr.ch, 1);
					num .Append(((curr.ch & (CHART)0xff) == curr.ch) ? (char)curr.ch : 0, 1);
					MoveNext();
				}
				MoveNext(); // skip '>' or '\''

				// check <num>
				unsigned int number;
				char * str = num.GetBuffer();

				if( ReadDec(str, number) ? ( *str == '\0') : 0 )
				{
					pDelegate->m_ndata = number;
					name.Release();
				}

				m_recursivelist.Push(pDelegate);
				pElx = pDelegate;
			}
			else
			{
				CBufferT <char> rto;
				while(curr.ch != RCHART(0) && curr.ch != RCHART(')'))
				{
					rto.Append(((curr.ch & (CHART)0xff) == curr.ch) ? (char)curr.ch : 0, 1);
					MoveNext();
				}

				unsigned int rtono = 0;
				char * str = rto.GetBuffer();
				ReadDec(str, rtono);

				CDelegateElx * pDelegate = (CDelegateElx *)Keep(new CDelegateElx(rtono));

				m_recursivelist.Push(pDelegate);
				pElx = pDelegate;
			}
			break;

		case RCHART('('):
			{
				CConditionElx * pConditionElx = (CConditionElx *)Keep(new CConditionElx());

				// condition
				ElxInterface * & pCondition = pConditionElx->m_pelxask;

				if(next == CHART_INFO(RCHART('?'), 1))
				{
					pCondition = BuildRecursive(flags);
				}
				else // named, assert or number
				{
					MoveNext(); // skip '('
					int pos0 = curr.pos;

					// save elx condition
					pCondition = Keep(new CAssertElx(BuildAlternative(flags), 1));

					// save name
					pConditionElx->m_szNamed.Append(m_pattern.GetBuffer() + pos0, curr.pos - pos0, 1);

					// save number
					CBufferT <char> numstr;
					while(pos0 < curr.pos)
					{
						CHART ch = m_pattern[pos0];
						numstr.Append(((ch & (CHART)0xff) == ch) ? (char)ch : 0, 1);
						pos0 ++;
					}

					unsigned int number;
					char * str = numstr.GetBuffer();

					// valid group number
					if( ReadDec(str, number) ? ( *str == '\0') : 0 )
					{
						pConditionElx->m_nnumber = number;
						pCondition = 0;
					}
					else // maybe elx, maybe named
					{
						pConditionElx->m_nnumber = -1;
						m_namedconditionlist.Push(pConditionElx);
					}

					MoveNext(); // skip ')'
				}

				// alternative
				{
					int newflags = flags;

					pConditionElx->m_pelxyes = BuildList(newflags);
				}

				if(curr.ch == RCHART('|'))
				{
					MoveNext(); // skip '|'

					pConditionElx->m_pelxno = BuildAlternative(flags);
				}
				else
				{
					pConditionElx->m_pelxno = 0;
				}

				pElx = pConditionElx;
			}
			break;

		default:
			while(curr.ch != RCHART(0) && isspace(curr.ch)) MoveNext(); // skip space

			if(curr.ch >= RCHART('0') && curr.ch <= RCHART('9')) // recursive (?1) => (?R1)
			{
				CBufferT <char> rto;
				while(curr.ch != RCHART(0) && curr.ch != RCHART(')'))
				{
					rto.Append(((curr.ch & (CHART)0xff) == curr.ch) ? (char)curr.ch : 0, 1);
					MoveNext();
				}

				unsigned int rtono = 0;
				char * str = rto.GetBuffer();
				ReadDec(str, rtono);

				CDelegateElx * pDelegate = (CDelegateElx *)Keep(new CDelegateElx(rtono));

				m_recursivelist.Push(pDelegate);
				pElx = pDelegate;
			}
			else
			{
				// flag
				int newflags = flags;
				while(curr != CHART_INFO(0, 1) && curr.ch != RCHART(':') && curr.ch != RCHART(')') && curr != CHART_INFO(RCHART('('), 1))
				{
					int tochange = 0;

					switch(curr.ch)
					{
					case RCHART('i'):
					case RCHART('I'):
						tochange = IGNORECASE;
						break;

					case RCHART('s'):
					case RCHART('S'):
						tochange = SINGLELINE;
						break;

					case RCHART('m'):
					case RCHART('M'):
						tochange = MULTILINE;
						break;

					case RCHART('g'):
					case RCHART('G'):
						tochange = GLOBAL;
						break;

					case RCHART('-'):
						bNegative = 1;
						break;
					}

					if(bNegative)
						newflags &= ~tochange;
					else
						newflags |=  tochange;

					// move to next char
					MoveNext();
				}

				if(curr.ch == RCHART(':') || curr == CHART_INFO(RCHART('('), 1))
				{
					// skip ':'
					if(curr.ch == RCHART(':')) MoveNext();

					pElx = BuildAlternative(newflags);
				}
				else
				{
					// change parent flags
					flags = newflags;

					pElx = GetStockElx(STOCKELX_EMPTY);
				}
			}
			break;
		}

		MoveNext(); // skip ')'

		return pElx;
	}
	else
	{
		// group and number
		CListElx * pList = (CListElx *)Keep(new CListElx(flags & RIGHTTOLEFT));
		int nThisBackref = ++ m_nMaxNumber;

		// left, center, right
		pList->m_elxlist.Push(Keep(new CBracketElx(nThisBackref, flags & RIGHTTOLEFT ? 1 : 0)));
		pList->m_elxlist.Push(BuildAlternative(flags));
		pList->m_elxlist.Push(Keep(new CBracketElx(nThisBackref, flags & RIGHTTOLEFT ? 0 : 1)));

		// for recursive
		m_grouplist.Prepare(nThisBackref);
		m_grouplist[nThisBackref] = pList;

		// right
		MoveNext(); // skip ')' 

		return pList;
	}
}

template <class CHART> ElxInterface * CBuilderT <CHART> :: BuildBoundary(int 

⌨️ 快捷键说明

复制代码 Ctrl + C
搜索代码 Ctrl + F
全屏模式 F11
切换主题 Ctrl + Shift + D
显示快捷键 ?
增大字号 Ctrl + =
减小字号 Ctrl + -