void sort_run( indri::file::File& out, indri::file::File& in, size_t memory ) {
// read the data in
UINT64 length = in.size();
char* data = new char[length];
in.read( data, 0, length );
in.close();
qsort( data, length / 12, 12, sort_comparator );
out.write( data, 0, length );
delete[] data;
}
void convert_intscore_to_long_binary( indri::file::File& outfile, const char* infile ) {
std::ifstream in;
in.open( infile );
indri::file::SequentialWriteBuffer* outb = new indri::file::SequentialWriteBuffer( outfile, 1024*1024 );
while( !in.eof() ) {
int document;
double score;
in >> document
>> score;
outb->write( (const void*) &document, sizeof(UINT32) );
outb->write( (const void*) &score, sizeof(double) );
}
outb->flush();
delete outb;
outfile.close();
in.close();
}
void sort_file( indri::file::File& out, indri::file::File& in, size_t memory, int totalDocuments ) {
UINT64 length = in.size();
size_t rounded = (memory / 12) * 12;
UINT64 total = 0;
std::vector<std::string> temporaries;
while( length > total ) {
UINT64 chunk = lemur_compat::min<UINT64>( rounded, length - total );
indri::file::File tempIn;
indri::file::File tempOut;
std::string nameIn;
std::string nameOut;
tempIn.openTemporary( nameIn );
tempOut.openTemporary( nameOut );
// make a sorted run
copy_region( tempIn, in, total, chunk );
sort_run( tempOut, tempIn, memory );
tempIn.close();
tempOut.close();
lemur_compat::remove( nameIn.c_str() );
temporaries.push_back( nameOut );
total += chunk;
}
in.close();
merge_sorted_runs( out, temporaries, totalDocuments );
for( size_t i=0; i<temporaries.size(); i++ ) {
lemur_compat::remove( temporaries[i].c_str() );
}
}
