MultiTermDocs.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
00012 // Created      : 2006/1/4
00013 //
00014 #ifndef _MULTITERMDOCS_H
00015 #define _MULTITERMDOCS_H
00016 
00017 #if _MSC_VER > 1000
00018 #pragma once
00019 #endif // _MSC_VER > 1000
00020 
00021 
00022 #include "TermDocs.h"
00023 #include "BarrelsInfo.h"
00024 #include "../utility/PriorityQueue.h"
00025 #include <list>
00026 using namespace firtex::utility;
00027 using namespace std;
00028 
00029 namespace firtex
00030 {
00031         namespace index
00032         {
00033                 class CBarrelTermDocsEntry
00034                 {
00035                 public:
00036                         CBarrelTermDocsEntry(CBarrelInfo* barrelInfoA,CTermDocs* termDocsA)
00037                         {
00038                                 barrelInfo = new CBarrelInfo(*barrelInfoA);
00039                                 termDocs = termDocsA;
00040                         }
00041                         ~CBarrelTermDocsEntry()
00042                         {
00043                                 delete barrelInfo;
00044                                 delete termDocs;
00045                         }
00046                 protected:
00047                         CBarrelTermDocsEntry(){}
00048                 public:
00049                         CBarrelInfo*    barrelInfo;
00050                         CTermDocs*              termDocs;
00051 
00052                         friend class CMultiTermDocs;
00053                 };
00054 
00055                 class CMultiTermDocs :  public CTermDocs
00056                 {
00057                         class CTermDocsQueue:public CPriorityQueue<CBarrelTermDocsEntry*>
00058                         {
00059                         public:
00060                                 CTermDocsQueue(size_t size) 
00061                                 {
00062                                         initialize(size,false);
00063                                 }
00064                         protected:
00065                                 bool lessThan(CBarrelTermDocsEntry* o1, CBarrelTermDocsEntry* o2) 
00066                                 {
00067                                         return (o1->barrelInfo->minDocID() + o1->termDocs->doc()) < (o2->barrelInfo->minDocID() + o2->termDocs->doc());
00068                                 }
00069                         };
00070                 public:
00071                         CMultiTermDocs(void);
00072                         virtual ~CMultiTermDocs(void);
00073                 public:
00077                         bool  seek(CTerm* term);                        
00078                                                 
00082                         docid_t doc();
00083 
00087                         count_t freq();
00088 
00093                         bool next();
00094 
00102                         count_t next(docid_t*& docs, count_t*& freqs);
00103 
00111                         bool skipTo(docid_t target,docid_t& nearTarget);
00112 
00113                         freq_t  docFreq();
00114 
00118                         void  close();
00119 
00124                         void    add(CBarrelInfo* pBarrelInfo,CTermDocs* pTermDocs);
00125                 protected:
00126                         void initQueue();
00127                 protected:
00128                         list<CBarrelTermDocsEntry*>             m_barrelTermDocs;                       
00129                         CBarrelTermDocsEntry*                   m_current;
00130                         int                                                             m_cursor;
00131                         CTermDocsQueue*                                 m_pTermDocsQueue;
00132                 };
00133 
00134         }
00135 }
00136 
00137 #endif