java/galago/src/galago/index/DocumentNameReader.java

   1 //
   2 // DocumentNameReader.java
   3 //
   4 // 28 November 2006 -- Trevor Strohman
   5 //
   6 // BSD License (http://galagosearch.org/license)
   7 //
   8
   9 package galago.index;
  10
  11 import java.io.BufferedInputStream;
  12 import java.io.DataInputStream;
  13 import java.io.FileInputStream;
  14 import java.io.IOException;
  15 import java.util.ArrayList;
  16 /**
  17  * Reads a binary file of document names produced by DocumentNameWriter.
  18  * The names are loaded into RAM for quick access.
  19  *
  20  * @author trevor
  21  */
  22 public class DocumentNameReader {
  23     private static class NameSlot {
  24         public String prefix;
  25         public int offset;
  26         public int footerWidth;
  27         public int[] footers;
  28     }
  29
  30     ArrayList<NameSlot> slots;
  31     int documentCount;
  32
  33     /** Creates a new instance of DocumentNameReader */
  34     public DocumentNameReader( String filename ) throws IOException {
  35         FileInputStream f = new FileInputStream( filename );
  36         DataInputStream input = new DataInputStream( new BufferedInputStream( f ) );
  37         slots = new ArrayList();
  38         read( input );
  39         input.close();
  40     }
  41
  42     private String getInSlot( NameSlot slot, int footerIndex ) {
  43         int footer = slot.footers[footerIndex-slot.offset];
  44         String prefix = slot.prefix;
  45         String documentName;
  46
  47         if( slot.footerWidth == 0 ) {
  48             documentName = slot.prefix;
  49         } else {
  50             String format = "%s-%0" + slot.footerWidth + "d";
  51             documentName = String.format( format, prefix, footer );
  52         }
  53
  54         return documentName;
  55     }
  56
  57     public String get( int index ) {
  58         assert index >= 0;
  59         assert index < documentCount;
  60
  61         if( index >= documentCount )
  62             return "unknown";
  63
  64         if( index < 0 )
  65             return "unknown";
  66
  67         int big = slots.size()-1;
  68         int small = 0;
  69
  70         while( big-small > 1 ) {
  71             int middle = small + (big-small)/2;
  72
  73             if( slots.get(middle).offset >= index )
  74                 big = middle;
  75             else
  76                 small = middle;
  77         }
  78
  79         NameSlot one = slots.get(small);
  80         NameSlot two = slots.get(big);
  81         String result = "";
  82
  83         if (two.offset <= index)
  84             result = getInSlot(two, index);
  85         else
  86             result = getInSlot(one, index);
  87
  88         return result;
  89     }
  90
  91     public void read( DataInputStream input ) throws IOException {
  92         int offset = 0;
  93
  94         // open a file
  95         while( input.available() > 0 ) {
  96             // read the prefix
  97             int prefixLength = input.readInt();
  98             byte[] prefixData = new byte[prefixLength];
  99             input.read(prefixData);
 100
 101             // read the footers
 102             int footerWidth = input.readInt();
 103             int footerCount = input.readInt();
 104             NameSlot slot = new NameSlot();
 105
 106             slot.footerWidth = footerWidth;
 107             slot.offset = offset;
 108             slot.prefix = new String(prefixData);
 109             slot.footers = new int[footerCount];
 110
 111             for( int i=0; i<footerCount; i++ ) {
 112                 slot.footers[i] = input.readInt();
 113             }
 114
 115             slots.add(slot);
 116             offset += slot.footers.length;
 117         }
 118
 119         documentCount = offset;
 120     }
 121 }