nodemgr.cpp

计算机英汉机器翻译系统中的英语词性标注方法实现

// Copyright 1993 Nildram Software. All rights reserved.
// May not be copied, except for backup purposes, without express written
// permission from Nildram Software
#include "stdafx.h"
#include <stdlib.h>

#include "nodemgr.h"

NodeMgr::NodeMgr(int nBlockSize, int itsize)
{
	if (itsize == 0)
	{
		AfxMessageBox("Error: Index Item Size Is Zero!");
		ASSERT(FALSE);
		exit(0);
	}
	m_nIndexSize = itsize + 4;
	m_nKeyWordSize = itsize;
	m_nBlockRecNum = (nBlockSize - 16) / m_nIndexSize; // 16 is the block header size

	m_plPointersOfBlock = new long[m_nBlockRecNum];
	m_pszKeyWordsOfBlock = new char *[m_nBlockRecNum];
	for (int x = 0; x < m_nBlockRecNum; x++)
		m_pszKeyWordsOfBlock[x] = new char[m_nIndexSize];
	m_lLeftNode = -1;
	m_lRightNode = -1;
	m_cItemNum = 0;
	m_nCurrItemPtr = -1;
	m_cBuffLabel = stLeaf;
}

NodeMgr::~NodeMgr(void)
{
	for (int x = 0; x < m_nBlockRecNum; x++)
		delete m_pszKeyWordsOfBlock[x];
	delete m_pszKeyWordsOfBlock;
	delete m_plPointersOfBlock;
}

void NodeMgr::read_data(char *buffer)
{
	char *strptr = buffer;
	m_lLeftNode = *((long *)strptr);
	strptr += 4;
	m_lRightNode = *((long *)strptr);
	strptr += 4;
	m_cItemNum = *((short *)strptr);
	strptr += 2;
	m_cBuffLabel = *((short *)strptr);
	strptr += 2;
	m_lSeemNoUse = *((long *)strptr);
	strptr += 4;
	for (int x = 0; x < m_cItemNum; x++)
	{
		m_plPointersOfBlock[x] = *((long *)strptr);
		strptr += 4;
		memcpy(m_pszKeyWordsOfBlock[x], strptr, m_nKeyWordSize);
		strptr += m_nKeyWordSize;
	}
	m_nCurrItemPtr = -1;
}

void NodeMgr::write_data(char *buffer)
{
	char *strptr = buffer;
	*((long *)strptr) = m_lLeftNode;
	strptr += 4;
	*((long *)strptr) = m_lRightNode;
	strptr += 4;
	*((short *)strptr) = m_cItemNum;
	strptr += 2;
	*((short *)strptr) = m_cBuffLabel;
	strptr += 2;
	*((long *)strptr) = ++m_lSeemNoUse;
	strptr += 4;

	static int nSmall8 = 0;
	static int nBig8 = 0;
	if ( m_cItemNum > 8 ) {
		nSmall8 ++;
	} else {
		nBig8 ++;
	}

	for ( int x = 0; x < m_cItemNum; x++ )
	{
		*((long *)strptr) = m_plPointersOfBlock[x];
		strptr += 4;
		memcpy(strptr, m_pszKeyWordsOfBlock[x], m_nKeyWordSize);
		strptr += m_nKeyWordSize;
	}
}


int NodeMgr::erase(int itemno)
{
	if ( m_cItemNum <= 1 )
	{
		m_cItemNum = 0;
		return noEmpty;
	}
	if ( itemno == -1 )
		itemno = m_nCurrItemPtr;
	if ( itemno < 0 || itemno >= m_cItemNum )
		return noNoSelect;
	char *tmpptr = m_pszKeyWordsOfBlock[itemno];
	for ( int x = itemno + 1; x < m_cItemNum; x++ )
	{
		m_pszKeyWordsOfBlock[x - 1] = m_pszKeyWordsOfBlock[x];
		m_plPointersOfBlock[x - 1] = m_plPointersOfBlock[x];
	}
	m_pszKeyWordsOfBlock[m_cItemNum - 1] = tmpptr;
	m_cItemNum--;
	if ( m_nCurrItemPtr >= itemno )
		m_nCurrItemPtr--;

	return 0;
}

int NodeMgr::insert(long ptr, char *pszKeyWord)
{
	int x = 0;
	if ( m_cItemNum > 0 )
	{
		x = getpos(pszKeyWord, 0, m_cItemNum - 1);
		if ( x != -1 )
		{
			if (memcmp(pszKeyWord, m_pszKeyWordsOfBlock[x], m_nKeyWordSize) == 0)
				//if (memcmp(pszKeyWord, m_pszKeyWordsOfBlock[x], m_nKeyWordSize) == 0)
				return noDuplicate;
		}
		else
			x = m_cItemNum;
	}

	char *tmpptr = m_pszKeyWordsOfBlock[m_cItemNum];
	for ( int y = m_cItemNum; y > x; y-- )
	{
		m_pszKeyWordsOfBlock[y] = m_pszKeyWordsOfBlock[y - 1];
		m_plPointersOfBlock[y] = m_plPointersOfBlock[y - 1];
	}
	m_pszKeyWordsOfBlock[x] = tmpptr;
	m_plPointersOfBlock[x] = ptr;
	memcpy(m_pszKeyWordsOfBlock[x], pszKeyWord, m_nKeyWordSize);
	if ( ++m_cItemNum == m_nBlockRecNum )
		return noFull;
	return 0;
}

int NodeMgr::getpos(char *pszKeyWord, int start, int stop)
{
	for (;;)
	{
		int checkpos = (start + stop) / 2;
		int cmp = memcmp(pszKeyWord, m_pszKeyWordsOfBlock[checkpos], m_nKeyWordSize);
		if (cmp == 0)
			return checkpos;
		if (cmp < 0)
			if (checkpos == start)
				return checkpos;
			else
			{
				stop = checkpos;
				continue;
			}
		else
			if (start == stop)
				return -1;
			else
			{
				start = checkpos + 1;
				continue;
			}
	}
}

long NodeMgr::find(char *pszKeyWord)
{
	if (m_cItemNum > 0)
	{
		int x = getpos(pszKeyWord, 0, m_cItemNum - 1);
		if (x != -1)
		{
			m_nCurrItemPtr = x;
			return m_plPointersOfBlock[x];
		}
		if (m_cBuffLabel == stLeaf)
		{
			m_nCurrItemPtr = m_cItemNum; 
			return -1;
		}
		m_nCurrItemPtr = m_cItemNum - 1;
		return m_plPointersOfBlock[m_cItemNum - 1];
	}
	return -1;
}

long NodeMgr::ffwd_core(void)
{
	if (m_cItemNum > 0)
	{
		if (m_cBuffLabel == stLeaf)
		{
			m_nCurrItemPtr = m_cItemNum; 
			return -1;
		}
		m_nCurrItemPtr = m_cItemNum - 1;
		return m_plPointersOfBlock[m_cItemNum - 1];
	}
	return -1;
}

long NodeMgr::retrieve(void)
{
	if (m_nCurrItemPtr >= m_cItemNum || m_nCurrItemPtr < 0)
		return -1;
	return m_plPointersOfBlock[m_nCurrItemPtr];
}

long NodeMgr::next(void)
{
    if (m_nCurrItemPtr == m_cItemNum - 1)
        m_nCurrItemPtr++;
    if (m_nCurrItemPtr > m_cItemNum - 1)
        return -1;
    return m_plPointersOfBlock[++m_nCurrItemPtr];
}

long NodeMgr::prev(void)
{
    if (m_nCurrItemPtr == 0)
        m_nCurrItemPtr--;
    if (m_nCurrItemPtr < 0)
        return -1;
    return m_plPointersOfBlock[--m_nCurrItemPtr];
}

void NodeMgr::reset(int st)
{
	m_lLeftNode = -1;
	m_lRightNode = -1;
	m_nCurrItemPtr = -1;
	m_cItemNum = 0;
	m_cBuffLabel = st;
	m_lSeemNoUse = 0;
}

void NodeMgr::split(char *block1, char *block2)
{
	int tmp_items = m_cItemNum;
	m_cItemNum /= 2;
	write_data(block1);
	m_cItemNum = tmp_items;
	for (int x = 0; x < tmp_items / 2; x++)
		erase(0);
	write_data(block2);
}