qim.cpp

我们一个小组

#include "Qim.h"
#include <fstream>

using namespace std ;
const int min_dynamic_word_id = 20769;

bool Qim::InsertDynamicPhrase(const string& py, const string& strPrase, double frequency)
{
	int id ;
	if (m_BiHashTable.GetIDAndInsertByWord(strPrase, id))
	{
		return m_DynamicPraseTable.InsertPhrase(py, id, frequency) ;
	}
	return false;
}

bool Qim::InsertDynamicPhrase1(string py, string strSlimcutpy, string strPhrase,double frequency)
{
	int id;
	if (m_BiHashTable.GetIDAndInsertByWord(strPhrase, id))
	{

		if( id >= min_dynamic_word_id)
		{
			m_DynamicPraseTable.InsertPhrase(py, id, frequency) ;
			m_DynamicPraseTable.InsertPhrase(strSlimcutpy,id,frequency);
		}
		
		return true;
	}
	return false;
}

bool Qim::InsertStaticPhrase(const string& py, const string& strPrase, double frequency)
{
	int id ;
	if (m_BiHashTable.GetIDByWord(strPrase, id))
	{
		m_StaticPhraseTable.InsertSingleWord(py, frequency, id) ;
		return true ;
	}
	else if( m_BiHashTable.GetIDAndInsertByWord(strPrase, id) )
	{
		m_StaticPhraseTable.InsertSingleWord(py, frequency, id) ;
		return true ;
	}

	return false ;
}
//Get the word by py, using longest-matching strategy, return the longest matching count
int Qim::GetStaticWordByLongestMatchedPy(const string& py, vector<string>& vecPhraseTable)
{
	vector<int> idTable ;

	vecPhraseTable.clear() ;

	int maxMatchedLength = m_StaticPhraseTable.GetWordByLongestMatchedPy(py, idTable) ;

	if(idTable.size()>0)
	{
		for (size_t i = 0; i<idTable.size(); i++)
		{
			int id = idTable[i] ;

			string word ;
			if( m_BiHashTable.GetWordByID(id, word) )
			{
				vecPhraseTable.push_back(word) ;
			}
		}
	}

	return maxMatchedLength ;
}

int Qim::GetDynamicPraseByLongestMatchedPy(const string& py, vector<string>& vecPhraseTable)
{
	vecPhraseTable.clear() ;

	vector<ID> vecResultID ;

	int longestMatchLength = m_DynamicPraseTable.GetPhraseByLongestMatchPy(py, vecResultID) ;

	for (size_t i = 0; i<vecResultID.size(); i++)
	{
		int id = vecResultID[i].getid() ;
		string phrase ;
		if ( m_BiHashTable.GetWordByID(id, phrase))
		{
			vecPhraseTable.push_back(phrase) ;
		}
	}

	return longestMatchLength ;

}

bool Qim::IncreaseStaticFrequency(const string& py, const string& strPhrase) 
{
	int id;
	if ( m_BiHashTable.GetIDByWord(strPhrase, id) )
	{
		m_StaticPhraseTable.IncreaseFrequency(py, id) ;
		return true ;
	}
	return false ;
}

bool Qim::IncreaseDynamicFrequency(const string& py, const string& strPhrase)
{
	int id ;
	if (m_BiHashTable.GetIDByWord(strPhrase, id))
	{
		return m_DynamicPraseTable.IncreaseFrequency(py, id) ;
	}
	return false ;
}

bool Qim::Save(const string& strStaticTableFilename, const string& strDynamicTableFilename, const string& strHashtableFilename)
{
	bool bRet[3] ;
	bRet[0] = m_StaticPhraseTable.Save(strStaticTableFilename) ;
	bRet[1] = m_DynamicPraseTable.Save(strDynamicTableFilename) ;
	bRet[2] = m_BiHashTable.Save(strHashtableFilename) ;

	return ( bRet[0] && bRet[1] && bRet[2] ) ;
}

bool Qim::Load(const string& strStaticTableFilename, const string& strDynamicTableFilename, const string& strHashtableFilename)
{
	bool bRet[3] ;
	bRet[0] = m_StaticPhraseTable.Load(strStaticTableFilename) ;
	bRet[1] = m_DynamicPraseTable.Load(strDynamicTableFilename) ;
	bRet[2] = m_BiHashTable.Load(strHashtableFilename) ;

	return ( bRet[0] && bRet[1] && bRet[2] ) ;
}

bool Qim::LoadFromRawFile (const string& strFileName)
{
	ifstream input(strFileName.c_str()) ;
	if (!input.is_open())
	{
		return false ;
	}

	string py ;
	string word ;
	double frequency ;

	while (input >> py)
	{
		input >> word >> frequency ;
		InsertStaticPhrase(py, word, frequency) ;
	}

	return true ;
}

#include <iostream>
#include <sstream>
bool Qim::LoadDynamicFromRawFile(const string& strFilename)
{
	ifstream in(strFilename.c_str()) ;
	if ( !in.is_open() )
	{
		return false ;
	}

	string strPhrase ;
	string strLine  ;

	while ( in >> strPhrase )
	{
		string strPy ;
		string strShortPy ;
		string temp ;

		getline( in, strLine) ;

		stringstream s(strLine.c_str()) ;

		while (s>>temp)
		{
			strPy += temp ;
			strShortPy += temp[0] ;
		}
		InsertDynamicPhrase1(strPy, strShortPy, strPhrase, 1) ;
	}

	in.close() ;
}