Logo Search packages:      
Sourcecode: qt4-x11 version File versions

TermInfosWriter.h

/*------------------------------------------------------------------------------
* Copyright (C) 2003-2006 Ben van Klinken and the CLucene Team
* 
* Distributable under the terms of either the Apache License (Version 2.0) or 
* the GNU Lesser General Public License, as specified in the COPYING file.
------------------------------------------------------------------------------*/
#ifndef _lucene_index_TermInfosWriter_
#define _lucene_index_TermInfosWriter_

#if defined(_LUCENE_PRAGMA_ONCE)
# pragma once
#endif

#include "CLucene/store/Directory.h"
#include "FieldInfos.h"
#include "TermInfo.h"
#include "Term.h"

CL_NS_DEF(index)


      // This stores a monotonically increasing set of <Term, TermInfo> pairs in a
      // Directory.  A TermInfos can be written once, in order.  
      class TermInfosWriter :LUCENE_BASE{
      private:
            FieldInfos* fieldInfos;
            CL_NS(store)::IndexOutput* output;
            Term* lastTerm;
            TermInfo* lastTi;
            int64_t size;
            int64_t lastIndexPointer;
            bool isIndex;
            TermInfosWriter* other;
      
            //inititalize
            TermInfosWriter(CL_NS(store)::Directory* directory, const char* segment, FieldInfos* fis, int32_t interval, bool isIndex);
      public:
            /** The file format version, a negative number. */
            LUCENE_STATIC_CONSTANT(int32_t,FORMAT=-2);

            /**
            * Expert: The fraction of terms in the "dictionary" which should be stored
            * in RAM.  Smaller values use more memory, but make searching slightly
            * faster, while larger values use less memory and make searching slightly
            * slower.  Searching is typically not dominated by dictionary lookup, so
            * tweaking this is rarely useful.
            */
            int32_t indexInterval;// = 128

            /**
            * Expert: The fraction of {@link TermDocs} entries stored in skip tables,
            * used to accellerate {@link TermDocs#SkipTo(int32_t)}.  Larger values result in
            * smaller indexes, greater acceleration, but fewer accelerable cases, while
            * smaller values result in bigger indexes, less acceleration and more
            * accelerable cases. More detailed experiments would be useful here. 
            */
            int32_t skipInterval;// = 16

            TermInfosWriter(CL_NS(store)::Directory* directory, const char* segment, FieldInfos* fis, int32_t interval);

            ~TermInfosWriter();

            /**
            * Adds a new <Term, TermInfo> pair to the set.
            * Term must be lexicographically greater than all previous Terms added.
            * TermInfo pointers must be positive and greater than all previous.
            */
            void add(Term* term, const TermInfo* ti);

            /** Called to complete TermInfos creation. */
            void close();

      private:
        /** Helps constructors to initialize instances */
            void initialise(CL_NS(store)::Directory* directory, const char* segment, int32_t interval, bool IsIndex);
            void writeTerm(Term* term);
      };
CL_NS_END
#endif

Generated by  Doxygen 1.6.0   Back to index