cpp/galago/include/BinnedFeatureIndex.hpp

   1
   2 //
   3 // BinnedFeatureIndex
   4 //
   5 // 5 January 2007 -- tds
   6 //
   7 // BSD License (http://galagosearch.org/license)
   8 //
   9
  10 #ifndef GALAGO_BINNEDFEATUREINDEX_HPP
  11 #define GALAGO_BINNEDFEATUREINDEX_HPP
  12
  13 #include "indri/Parameters.hpp"
  14 #include "BinOrderedBinnedIterator.hpp"
  15 #include "Stemmer.hpp"
  16 #include "Normalizer.hpp"
  17 #include "Query.hpp"
  18 #include "Stopper.hpp"
  19 #include "IndexReader.hpp"
  20
  21 #define BINNEDFEATUREINDEX_MAX_TERM_LENGTH (512)
  22 #define BINNEDFEATUREINDEX_VOCAB_GROUP (16)
  23 #define BINNEDFEATUREINDEX_BLOCK_SIZE (1<<16)
  24
  25 class BinnedFeatureIndex {
  26 private:
  27     IndexReader _reader;
  28
  29     Stemmer* stemmer;
  30     Stopper* stopper;
  31     Normalizer normalizer;
  32
  33     UINT64 _maximumDocumentCount;
  34     bool _requiresWeighting;
  35
  36     void _build_stemmer();
  37     void _build_stopper();
  38
  39 public:
  40     static const int blockSize = BINNEDFEATUREINDEX_BLOCK_SIZE;
  41     static const int vocabGroup = BINNEDFEATUREINDEX_VOCAB_GROUP;
  42     static const char maxTermLength = BINNEDFEATUREINDEX_MAX_TERM_LENGTH;
  43
  44 public:
  45     BinnedFeatureIndex();
  46     ~BinnedFeatureIndex();
  47
  48     void openRead( const std::string& indexPath );
  49     void close();
  50     BinOrderedBinnedIterator* getTerm( const std::string& term ) const;
  51     QueryTerm processTerm( const QueryTerm& term ) const;
  52     UINT64 maximumDocumentCount() const;
  53     bool requiresWeighting() const;
  54 };
  55
  56 #endif // GALAGO_BINNEDFEATUREINDEX_HPP