int bamsplit(libmaus2::util::ArgInfo const & arginfo)
{
if ( isatty(STDIN_FILENO) )
{
::libmaus2::exception::LibMausException se;
se.getStream() << "Refusing read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
se.finish();
throw se;
}
int const level = libmaus2::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
uint64_t const n = arginfo.getValue<int>("n",getDefaultN());
std::string const prefix = arginfo.getUnparsedValue("prefix",getDefaultFilePrefix(arginfo));
libmaus2::bambam::BamDecoder bamdec(std::cin);
libmaus2::bambam::BamAlignment const & algn = bamdec.getAlignment();
libmaus2::bambam::BamHeader const & header = bamdec.getHeader();
::libmaus2::bambam::BamHeader::unique_ptr_type uphead(updateHeader(arginfo,header));
libmaus2::aio::OutputStreamInstance::unique_ptr_type COS;
libmaus2::bambam::BamWriter::unique_ptr_type writer;
uint64_t c = 0;
uint64_t f = 0;
while ( bamdec.readAlignment() )
{
if ( c++ % n == 0 )
{
writer.reset();
if ( COS )
COS->flush();
COS.reset();
std::ostringstream fnostr;
fnostr << prefix << "_" << std::setw(6) << std::setfill('0') << f++ << std::setw(0) << ".bam";
std::string const fn = fnostr.str();
libmaus2::aio::OutputStreamInstance::unique_ptr_type tCOS(new libmaus2::aio::OutputStreamInstance(fn));
COS = UNIQUE_PTR_MOVE(tCOS);
libmaus2::bambam::BamWriter::unique_ptr_type twriter(new libmaus2::bambam::BamWriter(*COS,*uphead,level));
writer = UNIQUE_PTR_MOVE(twriter);
if ( verbose )
std::cerr << "[V] opened file " << fn << std::endl;
}
algn.serialise(writer->getStream());
}
writer.reset();
if ( COS )
COS->flush();
COS.reset();
return EXIT_SUCCESS;
}
int main(int argc, char * argv[])
{
try
{
::libmaus::util::ArgInfo const arginfo(argc,argv);
for ( uint64_t i = 0; i < arginfo.restargs.size(); ++i )
if (
arginfo.restargs[i] == "-v"
||
arginfo.restargs[i] == "--version"
)
{
std::cerr << ::biobambam::Licensing::license();
return EXIT_SUCCESS;
}
else if (
arginfo.restargs[i] == "-h"
||
arginfo.restargs[i] == "--help"
)
{
std::cerr << ::biobambam::Licensing::license();
std::cerr << std::endl;
std::cerr << "Key=Value pairs:" << std::endl;
std::cerr << std::endl;
std::vector< std::pair<std::string,std::string> > V;
V.push_back ( std::pair<std::string,std::string> ( "div=<["+::biobambam::Licensing::formatNumber(getDefaultDiv())+"]>", "divisor" ) );
V.push_back ( std::pair<std::string,std::string> ( "level=<["+::biobambam::Licensing::formatNumber(getDefaultLevel())+"]>", libmaus::bambam::BamBlockWriterBaseFactory::getBamOutputLevelHelpText() ) );
V.push_back ( std::pair<std::string,std::string> ( "verbose=<["+::biobambam::Licensing::formatNumber(getDefaultVerbose())+"]>", "print progress report" ) );
V.push_back ( std::pair<std::string,std::string> ( "prefix=<["+getDefaultFilePrefix(arginfo)+"]>", "default output file prefix" ) );
::biobambam::Licensing::printMap(std::cerr,V);
std::cerr << std::endl;
return EXIT_SUCCESS;
}
return bamsplitmod(arginfo);
}
catch(std::exception const & ex)
{
std::cerr << ex.what() << std::endl;
return EXIT_FAILURE;
}
}
int bamsplitmod(libmaus::util::ArgInfo const & arginfo)
{
if ( isatty(STDIN_FILENO) )
{
::libmaus::exception::LibMausException se;
se.getStream() << "Refusing read binary data from terminal, please redirect standard input to pipe or file." << std::endl;
se.finish();
throw se;
}
int const level = libmaus::bambam::BamBlockWriterBaseFactory::checkCompressionLevel(arginfo.getValue<int>("level",getDefaultLevel()));
int const verbose = arginfo.getValue<int>("verbose",getDefaultVerbose());
uint64_t const div = arginfo.getValue<int>("div",getDefaultDiv());
std::string const prefix = arginfo.getUnparsedValue("prefix",getDefaultFilePrefix(arginfo));
if ( ! div )
{
::libmaus::exception::LibMausException se;
se.getStream() << "div cannot be 0." << std::endl;
se.finish();
throw se;
}
libmaus::bambam::BamDecoder bamdec(std::cin);
libmaus::bambam::BamAlignment const & algn = bamdec.getAlignment();
libmaus::bambam::BamHeader const & header = bamdec.getHeader();
::libmaus::bambam::BamHeader::unique_ptr_type uphead(updateHeader(arginfo,header));
libmaus::autoarray::AutoArray<libmaus::aio::CheckedOutputStream::unique_ptr_type> COS(div);
libmaus::autoarray::AutoArray<libmaus::bambam::BamWriter::unique_ptr_type> writers(div);
std::vector < std::string > filenames;
for ( uint64_t i = 0; i < div; ++i )
{
std::ostringstream ostr;
ostr << prefix << "_" << std::setw(6) << std::setfill('0') << i << std::setw(0) << ".bam";
libmaus::aio::CheckedOutputStream::unique_ptr_type tCOS(new libmaus::aio::CheckedOutputStream(ostr.str()));
COS[i] = UNIQUE_PTR_MOVE(tCOS);
libmaus::bambam::BamWriter::unique_ptr_type twriter(new libmaus::bambam::BamWriter(*COS[i],*uphead,level));
writers[i] = UNIQUE_PTR_MOVE(twriter);
}
uint64_t c = 0;
if ( verbose )
{
while ( bamdec.readAlignment() )
{
algn.serialise ( writers [ (c++) % div ] -> getStream() );
if ( ((c) & ((1ull<<20)-1)) == 0 )
std::cerr << "[V] " << c << std::endl;
}
std::cerr << "[V] " << c << std::endl;
}
else
{
while ( bamdec.readAlignment() )
algn.serialise ( writers [ (c++) % div ] -> getStream() );
}
for ( uint64_t i = 0; i < div; ++i )
{
writers[i].reset();
COS[i]->flush();
COS[i].reset();
}
return EXIT_SUCCESS;
}
