cpp/galago/src/DocumentNameReader.cpp

   1
   2 //
   3 // DocumentNameReader
   4 //
   5 // 8 January 2007 -- tds
   6 //
   7 // BSD License (http://galagosearch.org/license)
   8 //
   9
  10 #include "DocumentNameReader.hpp"
  11
  12 void DocumentNameReader::getInSlot( std::string& result, const NameSlot& slot, int footerIndex ) const {
  13     char buffer[1024];
  14     int footer = slot.footers[footerIndex-slot.offset];
  15
  16     strncpy( buffer, slot.prefix.c_str(), sizeof buffer );
  17
  18     char* start = buffer + slot.prefix.size();
  19     char* end = start + slot.footerWidth + 1;
  20
  21     *start = '-';
  22     *end = 0;
  23         start++;
  24
  25     for( char* current = end-1; current >= start; current-- ) {
  26         char digit = (footer % 10) + '0';
  27         footer /= 10;
  28         *current = digit;
  29     }
  30
  31     result = buffer;
  32 }
  33
  34 std::string DocumentNameReader::operator[] ( int index ) const {
  35     return get(index);
  36 }
  37
  38 std::string DocumentNameReader::get( int index ) const {
  39     assert( index >= 0 );
  40     assert( index < _documentCount );
  41
  42     if( index >= _documentCount )
  43         return "unknown";
  44
  45     if( index < 0 )
  46         return "unknown";
  47
  48     int big = _slots.size()-1;
  49     int small = 0;
  50
  51     while( big-small > 1 ) {
  52         int middle = small + (big-small)/2;
  53
  54         if( _slots[middle].offset >= index )
  55             big = middle;
  56         else
  57             small = middle;
  58     }
  59
  60     const NameSlot& one = _slots[small];
  61     const NameSlot& two = _slots[big];
  62     std::string result;
  63
  64     if (two.offset <= index)
  65         getInSlot(result, two, index);
  66     else
  67         getInSlot(result, one, index);
  68
  69     return result;
  70 }
  71
  72 void DocumentNameReader::read( const std::string& filename ) {
  73     int offset = 0;
  74
  75     // mmap the file
  76     indri::file::File inputFile;
  77
  78         inputFile.openRead( filename );
  79         UINT64 length = inputFile.size();
  80     indri::file::FileMapping* mapping = inputFile.mapFileRead();
  81         const UINT8* region = (const UINT8*) mapping->region;
  82         const UINT8* end = region + length;
  83
  84     while( region < end ) {
  85         UINT32 prefixLength = UncompressedRead::read_u32( region );
  86
  87                 char* prefixData = new char[prefixLength+1];
  88         memcpy( prefixData, region, prefixLength );
  89                 prefixData[prefixLength] = 0;
  90                 region += prefixLength;
  91
  92         UINT32 footerWidth = UncompressedRead::read_u32( region );
  93         UINT32 footerCount = UncompressedRead::read_u32( region );
  94
  95         NameSlot slot;
  96         slot.footerWidth = footerWidth;
  97         slot.offset = offset;
  98         slot.prefix = prefixData;
  99
 100         for( int i=0; i<footerCount; i++ ) {
 101             UINT32 footer = UncompressedRead::read_u32( region );
 102             slot.footers.push_back( footer );
 103         }
 104
 105         _slots.push_back(slot);
 106         offset += slot.footers.size();
 107         delete[] prefixData;
 108     }
 109
 110     _documentCount = offset;
 111
 112     delete mapping;
 113     inputFile.close();
 114 }
 115