WordPositionVector.h

00001 //
00002 // Copyright(C) 2005--2006 Institute of Computing Tech, Chinese Academy of Sciences. 
00003 // All rights reserved.
00004 // This file is part of FirteX (www.firtex.org)
00005 //
00006 // Use of the FirteX is subject to the terms of the software license set forth in 
00007 // the LICENSE file included with this software, and also available at
00008 // http://www.firtex.org/license.html
00009 //
00010 // Author       : 郭瑞杰(GuoRuijie)
00011 // Email        : ruijieguo@software.ict.ac.cn,ruijieguo@gmail.com
00012 // Created      : 2006/7/15
00013 //
00014 #ifndef _WORDPOSITIONVECTOR_H
00015 #define _WORDPOSITIONVECTOR_H
00016 
00017 #include "TermFreqVector.h"
00018 
00019 #if _MSC_VER > 1000
00020 #pragma once
00021 #endif // _MSC_VER > 1000
00022 
00023 #include "TermPositionVector.h"
00024 #include "../store/IndexInput.h"
00025 #include "../utility/DynamicArray.h"
00026 using namespace firtex::store;
00027 using namespace firtex::utility;
00028 
00029 namespace firtex
00030 {
00031         namespace index
00032         {
00033                 class CWordPositionVector : public CTermPositionVector
00034                 {
00035                         
00036                         class loc_t_nullvalue
00037                         {
00038                         public:
00039                                 inline operator loc_t()const {return -1;}
00040                         };
00041                         typedef struct _FREQ_POS 
00042                         {
00043                                 freq_t                                  freq;
00044                                 CDynamicArray<loc_t,loc_t_nullvalue>    positions;
00045                         }FREQ_POS;
00046                 public:
00047                         CWordPositionVector();
00048                         CWordPositionVector(const tchar* field,CIndexInput* pTVVInput);
00049                         virtual ~CWordPositionVector(void);
00050                 public:
00056                         void    open(const tchar* field,CIndexInput* pInput);
00060             count_t                     numTerms();
00061 
00065             count_t                     numDistinctTerms();
00066 
00072                         const CTerm*    getTerms();
00073 
00079                         const termid_t* getTermIDs();
00080 
00086                         const freq_t*   getTermFrequencies();
00087 
00094                         void                    getTermFreqVector(termid_t*& tids,freq_t*& freqs,count_t& size);
00095 
00101                         int                             indexOf(termid_t tid);
00102 
00104 
00110                         const loc_t* getTermPositions(const CTerm* term);
00111 
00118                         const loc_t* getTermPositions(termid_t tid);
00119 
00126                         const loc_t* getTermPositions(size_t index);
00127                 protected:
00133                         void    addField(CIndexOutput*  pOutput,CField* pField);
00134 
00135                 private:
00136                         CDynamicArray<FREQ_POS*,Const_NullValue<FREQ_POS*> >*   m_pTermDictionary;
00137                 };
00138         }
00139 }
00140 
00141 #endif