java/galago/src/galago/retrieval/SparseFloatListReader.java

   1 /*
   2  * ScoreIndex
   3  *
   4  * October 2, 2007 -- Trevor Strohman
   5  *
   6  * BSD License (http://www.galagosearch.org/license)
   7  */
   8
   9 package galago.retrieval;
  10
  11 import galago.retrieval.extents.ScoreIterator;
  12 import galago.tupleflow.BufferedDataStream;
  13 import galago.tupleflow.VByteInput;
  14 import java.io.FileNotFoundException;
  15 import java.io.IOException;
  16
  17 /**
  18  *
  19  * @author trevor
  20  */
  21 public class SparseFloatListReader {
  22     public class ScoreListIterator implements ScoreIterator {
  23         IndexReader.Iterator iterator;
  24
  25         VByteInput stream;
  26         int documentCount;
  27         int index;
  28
  29         int currentDocument;
  30         double currentScore;
  31
  32         public ScoreListIterator( IndexReader.Iterator iterator ) throws IOException {
  33             this.iterator = iterator;
  34             load();
  35         }
  36
  37         void load() throws IOException {
  38             BufferedDataStream buffered = reader.blockStream(iterator);
  39             stream = new VByteInput( buffered );
  40             documentCount = stream.readInt();
  41             index = -1;
  42             currentDocument = 0;
  43
  44             if( documentCount > 0 )
  45                 read();
  46         }
  47
  48         void read() throws IOException {
  49             index += 1;
  50
  51             if( index < documentCount ) {
  52                 currentDocument += stream.readInt();
  53                 currentScore = stream.readFloat();
  54             }
  55         }
  56
  57         public void reset() throws IOException {
  58             currentDocument = 0;
  59             currentScore = 0;
  60             load();
  61         }
  62
  63         public int nextCandidate() {
  64             return currentDocument;
  65         }
  66
  67         public String currentTerm() {
  68             return iterator.currentTerm();
  69         }
  70
  71         public void nextTerm() throws IOException {
  72             iterator.nextTerm();
  73             load();
  74         }
  75
  76         public boolean hasMatch(int document) {
  77             return document == currentDocument;
  78         }
  79
  80         public void moveTo(int document) throws IOException {
  81             while( !isDone() && document > currentDocument )
  82                 read();
  83         }
  84
  85         public void movePast(int document) throws IOException {
  86             while( !isDone() && document >= currentDocument )
  87                 read();
  88         }
  89
  90         public double score(int document, int length) {
  91             if( document == currentDocument )
  92                 return currentScore;
  93             return Double.NEGATIVE_INFINITY;
  94         }
  95
  96         public boolean isDone() {
  97             return index >= documentCount;
  98         }
  99     }
 100
 101     IndexReader reader;
 102
 103     public SparseFloatListReader( String pathname ) throws FileNotFoundException, IOException {
 104         reader = new IndexReader( pathname );
 105     }
 106
 107     public ScoreListIterator getIterator() throws IOException {
 108         return new ScoreListIterator( reader.getIterator() );
 109     }
 110
 111     public ScoreListIterator getScores( String term ) throws IOException {
 112         IndexReader.Iterator iterator = reader.getTerm( term );
 113
 114         if( iterator != null )
 115             return new ScoreListIterator( iterator );
 116
 117         return null;
 118     }
 119
 120     void close() throws IOException {
 121         reader.close();
 122     }
 123 }