cpp/galago/src/DocumentNameReader.cpp

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