static libmaus2::autoarray::AutoArray<char> loadFile(std::istream & in)
{
libmaus2::autoarray::AutoArray<char> C(1);
uint64_t p = 0;
while ( in )
{
in.read(C.begin() + p, C.size()-p);
if ( ! in.gcount() )
break;
p += in.gcount();
if ( p == C.size() )
{
libmaus2::autoarray::AutoArray<char> Cn(2*C.size(),false);
std::copy(C.begin(),C.end(),Cn.begin());
C = Cn;
}
}
libmaus2::autoarray::AutoArray<char> Cn(p,false);
std::copy(C.begin(),C.begin()+p,Cn.begin());
return Cn;
}
int_type underflow()
{
if ( gptr() < egptr() )
return static_cast<int_type>(*uptr());
assert ( gptr() == egptr() );
char * midptr = buffer.begin() + pushbackspace;
uint64_t const copyavail =
std::min(
// previously read
static_cast<uint64_t>(gptr()-eback()),
// space we have to copy into
static_cast<uint64_t>(midptr-buffer.begin())
);
::std::memmove(midptr-copyavail,gptr()-copyavail,copyavail);
stream.read(midptr, buffer.end()-midptr);
size_t const n = stream.gcount();
streamreadpos += n;
setg(midptr-copyavail, midptr, midptr+n);
if (!n)
return traits_type::eof();
return static_cast<int_type>(*uptr());
}
bool decodeBlock()
{
while ( FBO.file < index.Vfn.size() && FBO.block >= index.blocksPerFile[FBO.file] )
{
FBO.file++;
FBO.block = 0;
FBO.blockoffset = 0; // check this if we change the file format
FBO.offset = 0;
openFile();
}
if ( FBO.file == index.Vfn.size() )
{
PSGI.reset();
PISI.reset();
return false;
}
libmaus2::gamma::GammaDecoder< libmaus2::aio::SynchronousGenericInput<uint64_t> > GD(*PSGI);
uint64_t const bs = GD.decode() + 1;
B.ensureSize(bs);
for ( uint64_t i = 0; i < bs; ++i )
B[i] = GD.decode();
pa = B.begin();
pc = B.begin();
pe = B.begin() + bs;
FBO.block += 1;
return true;
}
void fillBuffer()
{
assert ( pc == pe );
if ( setpos )
{
// std::cerr << "Seeking to " << readpos << std::endl;
in.seekg(readpos);
in.clear();
}
if ( in.peek() >= 0 && readpos < endpos )
{
#if 0
std::cerr << "Filling block, readpos " << readpos
<< " stream at pos " << in.tellg()
<< " endpos " << endpos
<< std::endl;
#endif
uint64_t blocksize = sizeof(uint64_t) + sizeof(uint64_t);
// size of uncompressed buffer
uint64_t const n = ::libmaus2::util::NumberSerialisation::deserialiseNumber(in);
// size of compressed data
uint64_t const datasize = ::libmaus2::util::NumberSerialisation::deserialiseNumber(in);
// add to block size
blocksize += datasize;
if ( n > B.size() )
{
B = ::libmaus2::autoarray::AutoArray<char>(0,false);
B = ::libmaus2::autoarray::AutoArray<char>(n,false);
}
pa = B.begin();
pc = pa;
pe = pa + n;
::libmaus2::aio::IStreamWrapper wrapper(in);
::libmaus2::lz::IstreamSource< ::libmaus2::aio::IStreamWrapper> insource(wrapper,datasize);
try
{
SnappyCompress::uncompress(insource,B.begin(),n);
}
catch(std::exception const & ex)
{
libmaus2::exception::LibMausException lme;
lme.getStream() << "Failed to decompress snappy compressed data, comp=" << datasize << ", uncomp=" << n << ":\n" << ex.what() << "\n";
lme.finish();
throw lme;
}
readpos += blocksize;
}
}
int_type underflow()
{
if ( gptr() < egptr() )
return static_cast<int_type>(*uptr());
assert ( gptr() == egptr() );
char * midptr = buffer.begin() + pushbackspace;
uint64_t const copyavail =
std::min(
// previously read
static_cast<uint64_t>(gptr()-eback()),
// space we have to copy into
static_cast<uint64_t>(midptr-buffer.begin())
);
::std::memmove(midptr-copyavail,gptr()-copyavail,copyavail);
size_t n = 0;
bool done = false;
while ( ! done )
{
BgzfInflateInfo const info = stream.readAndInfo(midptr, buffer.end()-midptr);
n = info.uncompressed;
// non eof block
if ( n )
{
streamreadpos += n;
done = true;
}
else
{
// eof block at end of stream
if ( info.streameof )
{
done = true;
}
// intermediate empty block, skip it
else
{
}
}
}
setg(midptr-copyavail, midptr, midptr+n);
if (!n)
return traits_type::eof();
return static_cast<int_type>(*uptr());
}
MdStringComputationContext()
: T0(256,false), T1(256,false), nm(0)
{
std::fill(T0.begin(),T0.end(),4);
std::fill(T1.begin(),T1.end(),5);
T0['A'] = T0['a'] = T1['A'] = T1['a'] = 0;
T0['C'] = T0['c'] = T1['C'] = T1['c'] = 1;
T0['G'] = T0['g'] = T1['G'] = T1['g'] = 2;
T0['T'] = T0['t'] = T1['T'] = T1['t'] = 3;
auxvec.set("MD");
auxvec.set("NM");
}
ExternalMemoryIndexGenerator(std::string const & filename)
: Pstream(libmaus2::aio::InputOutputStreamFactoryContainer::constructUnique(filename,std::ios::in|std::ios::out|std::ios::trunc|std::ios::binary)), stream(*Pstream),
ic(0), flushed(false), writeCache(1024),
wa(writeCache.begin()), wc(wa), we(writeCache.end())
{
}
void get(uint64_t const i, libmaus2::bambam::GeneFlatFileEntry & entry) const
{
if ( i >= nl )
{
libmaus2::exception::LibMausException lme;
lme.getStream() << "GeneFlatFile::get(): line " << i << " is out of range." << std::endl;
lme.finish();
throw lme;
}
std::pair<uint64_t,uint64_t> P = LA->lineInterval(i);
while ( P.second != P.first && isspace(C[P.second-1]) )
--P.second;
entry.reset(C.begin() + P.first,C.begin() + P.second);
}
RLEncoderBaseTemplate(bit_writer_type & rwriter, uint64_t const rnumsyms, uint64_t const bufsize = 4ull*1024ull*1024ull)
: writer(rwriter), numsyms(rnumsyms), rlbuffer(bufsize), pa(rlbuffer.begin()), pc(pa), pe(rlbuffer.end()), cursym(0), curcnt(0),
indexwritten(false)
{
// std::cerr << "Writing RL file of length " << numsyms << std::endl;
writer.writeElias2(numsyms);
}
::libmaus2::util::Histogram::unique_ptr_type libmaus2::util::Utf8String::getHistogram(::libmaus2::autoarray::AutoArray<uint8_t> const & A)
{
#if defined(_OPENMP)
uint64_t const numthreads = omp_get_max_threads();
#else
uint64_t const numthreads = 1;
#endif
::libmaus2::autoarray::AutoArray<uint64_t> const partstarts = computePartStarts(A,numthreads);
uint64_t const numparts = partstarts.size()-1;
::libmaus2::util::Histogram::unique_ptr_type hist(new ::libmaus2::util::Histogram);
::libmaus2::parallel::OMPLock lock;
#if defined(_OPENMP)
#pragma omp parallel for
#endif
for ( int64_t t = 0; t < static_cast<int64_t>(numparts); ++t )
{
::libmaus2::util::Histogram::unique_ptr_type lhist(new ::libmaus2::util::Histogram);
uint64_t codelen = 0;
uint64_t const tcodelen = partstarts[t+1]-partstarts[t];
::libmaus2::util::GetObject<uint8_t const *> G(A.begin()+partstarts[t]);
while ( codelen != tcodelen )
(*lhist)(::libmaus2::util::UTF8::decodeUTF8(G,codelen));
lock.lock();
hist->merge(*lhist);
lock.unlock();
}
return UNIQUE_PTR_MOVE(hist);
}
int_type underflow()
{
// if there is still data, then return it
if ( gptr() < egptr() )
return static_cast<int_type>(*(reinterpret_cast<uint8_t const *>(gptr())));
assert ( gptr() == egptr() );
// number of bytes for putback buffer
uint64_t const putbackcopy = std::min(
static_cast<uint64_t>(gptr() - eback()),
putbackspace
);
// copy bytes
#if 0
std::copy(
gptr()-putbackcopy,
gptr(),
buffer.begin() + putbackspace - putbackcopy
);
#endif
std::memmove(
buffer.begin() + putbackspace - putbackcopy,
gptr()-putbackcopy,
putbackcopy
);
// load data
uint64_t const uncompressedsize = stream.readPart(
buffer.begin()+putbackspace,
buffer.size()-putbackspace
);
// set buffer pointers
setg(
buffer.begin()+putbackspace-putbackcopy,
buffer.begin()+putbackspace,
buffer.begin()+putbackspace+uncompressedsize
);
if ( uncompressedsize )
return static_cast<int_type>(*(reinterpret_cast<uint8_t const *>(gptr())));
else
return traits_type::eof();
}
LinuxStreamingPosixFdOutputStreamBuffer(std::string const & fn, int64_t const rbuffersize)
: fd(doOpen(fn)), closefd(true),
optblocksize((rbuffersize < 0) ? getOptimalIOBlockSize(fd,std::string()) : rbuffersize),
buffersize(optblocksize),
buffer(buffersize,false), prevwrite(0,0)
{
setp(buffer.begin(),buffer.end()-1);
}
MemoryOutputStreamBuffer(std::string const & fn, int64_t const rbuffersize)
:
fd(doOpen(fn)),
buffersize((rbuffersize < 0) ? getDefaultBlockSize() : rbuffersize),
buffer(buffersize,false)
{
setp(buffer.begin(),buffer.end()-1);
}
BgzfParallelRecodeDeflateBase()
: B(getBgzfMaxBlockSize(),false),
pa(B.begin()),
pc(B.begin()),
pe(B.end())
{
}
/**
* buffer underflow callback
* @return next symbol
**/
typename base_type::int_type underflow()
{
if ( base_type::gptr() < base_type::egptr() )
return static_cast<typename base_type::int_type>(*uptr());
assert ( base_type::gptr() == base_type::egptr() );
char_type * midptr = buffer.begin() + pushbackspace;
uint64_t const copyavail =
std::min(
// previously read
static_cast<uint64_t>(base_type::gptr()-base_type::eback()),
// space we have to copy into
static_cast<uint64_t>(midptr-buffer.begin())
);
::std::memmove(midptr-copyavail,base_type::gptr()-copyavail,copyavail*sizeof(char_type));
if ( static_cast<int64_t>(stream.tellg()) == static_cast<int64_t>(0) )
{
stream.seekg(infilesize);
stream.clear();
}
uint64_t const rspace = stream.tellg();
uint64_t const toread = std::min(rspace,static_cast<uint64_t>(buffer.end()-midptr));
stream.seekg(-static_cast<int64_t>(toread),std::ios::cur);
stream.clear();
stream.read(midptr, toread);
size_t const n = stream.gcount();
assert ( n == toread );
std::reverse(midptr,midptr+n);
streamreadpos += n;
stream.seekg(-static_cast<int64_t>(toread),std::ios::cur);
stream.clear();
base_type::setg(midptr-copyavail, midptr, midptr+n);
if (!n)
return base_type::traits_type::eof();
return static_cast<typename base_type::int_type>(*uptr());
}
void exec()
{
pid = fork();
if ( pid < 0 )
{
::libmaus2::exception::LibMausException ex;
ex.getStream() << "failed to fork: " << strerror(errno);
ex.finish();
throw ex;
}
if ( ! pid )
{
signal(SIGCHLD,sigchildhandler);
while ( true )
{
try
{
::libmaus2::network::SocketBase::unique_ptr_type recsock = seso->accept();
pid_t childpid = fork();
if ( childpid == 0 )
{
try
{
char const * ptr = data.begin();
char const * ptre = data.end();
uint64_t const bs = 4096;
while ( ptr != ptre )
{
uint64_t const rest = ptre-ptr;
uint64_t const towrite = std::min(bs,rest);
recsock->write(ptr,towrite);
ptr += towrite;
}
}
catch(std::exception const & ex)
{
std::cerr << ex.what() << std::endl;
}
_exit(0);
}
}
catch(std::exception const & ex)
{
std::cerr << "Error in SingleFileServer: " << ex.what() << std::endl;
}
}
_exit(0);
}
}
GammaRLEncoder(std::string const & filename, unsigned int const ralbits, uint64_t const n, uint64_t const rblocksize, uint64_t const rbufsize = 64*1024)
:
blocksize(rblocksize),
COS(filename), SGO(COS,rbufsize), GE(SGO),
A(blocksize), pa(A.begin()), pc(pa), pe(A.end()),
cursym(0), curcnt(0), indexwritten(false), albits(ralbits)
{
SGO.put(n);
SGO.put(albits);
}
SymBitEncoderBaseTemplate(bit_writer_type & rwriter, uint64_t const bufsize = 64*1024ull)
: writer(rwriter),
symcntruns(bufsize),
ra(symcntruns.begin()),
rc(symcntruns.begin()),
re(symcntruns.end()),
currun(std::numeric_limits<int64_t>::min(),false,0),
indexwritten(false)
{
}
GraphEdgeBlockBuffer(std::string const & filename, uint64_t const bufsize)
:
COS(filename),
B(bufsize,false),
pa(B.begin()),
pc(pa),
pe(B.end())
{
}
FastATwoBitTable()
: T(static_cast<size_t>(std::numeric_limits<unsigned char>::max())+1,false)
{
assert ( 3 < T.size() );
std::fill(T.begin(),T.end(),0);
T['a'] = T['A'] = 0;
T['c'] = T['C'] = 1;
T['g'] = T['G'] = 2;
T['t'] = T['T'] = 3;
}
void getPattern(pattern_type & pat, uint64_t i) const
{
assert ( i >= FI.low && i < FI.high );
uint64_t const j = i-FI.low;
uint64_t const offsetbase = longpointers [ designatorrank->rank1(j) ];
uint64_t const codepos = offsetbase + shortpointers[j];
uint8_t const * code = text.begin()+codepos;
::libmaus2::util::GetObject<uint8_t const *> G(code);
::libmaus2::parallel::SynchronousCounter<uint64_t> nextid(i);
CompactFastDecoderBase::decode(pat,G,nextid);
}
int_type underflow()
{
// if there is still data, then return it
if ( gptr() < egptr() )
return static_cast<int_type>(*uptr());
assert ( gptr() == egptr() );
// number of bytes for putback buffer
uint64_t const putbackcopy = std::min(
static_cast<uint64_t>(gptr() - eback()),
putbackspace
);
// copy bytes
std::copy(
gptr()-putbackcopy,
gptr(),
buffer.begin() + putbackspace - putbackcopy
);
// load data
uint64_t const uncompressedsize = fd->read(
buffer.begin()+putbackspace,
buffer.size()-putbackspace
);
// set buffer pointers
setgchecked(
buffer.begin()+putbackspace-putbackcopy,
buffer.begin()+putbackspace,
buffer.begin()+putbackspace+uncompressedsize);
symsread += uncompressedsize;
if ( uncompressedsize )
return static_cast<int_type>(*uptr());
else
return traits_type::eof();
}
MemoryInputOutputStreamBuffer(std::string const & fn, std::ios_base::openmode const cxxmode, int64_t const rbuffersize)
:
fd(doOpen(fn,cxxmode)),
buffersize(rbuffersize < 0 ? getDefaultBlockSize() : rbuffersize),
buffer(buffersize,false),
readpos(0),
writepos(0)
{
// empty get buffer
setg(buffer.end(),buffer.end(),buffer.end());
// empty put buffer
setp(buffer.begin(),buffer.end()-1);
}
int_type underflow()
{
// if there is still data, then return it
if ( gptr() < egptr() )
return static_cast<int_type>(*uptr());
assert ( gptr() == egptr() );
// load data
size_t const g = doRead(buffer.begin(),buffersize);
// set buffer pointers
setg(buffer.begin(),buffer.begin(),buffer.begin()+g);
// update start of buffer position
readpos += g;
if ( g )
return static_cast<int_type>(*uptr());
else
return traits_type::eof();
}
GammaPDDecoder(std::string const & rfn, uint64_t const blocksize = 4096)
: fn(rfn), POSI(new libmaus2::aio::OutputStreamInstance(fn)),
PSGO(new libmaus2::aio::SynchronousGenericOutput<uint64_t>(*POSI,4096)),
metafn(fn + ".meta"),
PMETA(new libmaus2::aio::OutputStreamInstance(metafn)),
B(blocksize,false),
pa(B.begin()),
pc(B.begin()),
pe(B.end()),
headerlength(sizeof(uint64_t)),
valueswritten(0),
flushed(false)
{
}
/**
* read out the data in the decompressed block
*
* @param ldata buffer for storing the decompressed block
* @param n size of buffer ldata in bytes
* @return the number of uncompressed bytes in the buffer
**/
uint64_t read(char * const ldata, uint64_t const n)
{
state = bgzfinflateblockstate_idle;
if ( n < getBgzfMaxBlockSize() )
{
::libmaus2::exception::LibMausException se;
se.getStream() << "BgzfInflate::decompressBlock(): provided buffer is too small: " << n << " < " << getBgzfMaxBlockSize();
se.finish(false);
throw se;
}
if ( failed() )
throw getException();
uint64_t const ndata = blockinfo.uncompressed;
std::copy ( data.begin(), data.begin() + ndata, reinterpret_cast<uint8_t *>(ldata) );
blockinfo = ::libmaus2::lz::BgzfInflateInfo(0,0,true);
return ndata;
}
::libmaus2::autoarray::AutoArray< std::pair<int64_t,uint64_t> > libmaus2::util::Utf8String::getHistogramAsArray(::libmaus2::autoarray::AutoArray<uint8_t> const & A)
{
#if defined(_OPENMP)
uint64_t const numthreads = omp_get_max_threads();
#else
uint64_t const numthreads = 1;
#endif
::libmaus2::autoarray::AutoArray<uint64_t> const partstarts = computePartStarts(A,numthreads);
uint64_t const numparts = partstarts.size()-1;
::libmaus2::parallel::OMPLock lock;
::libmaus2::parallel::PosixMutex mutex;
::libmaus2::util::ExtendingSimpleCountingHash<uint64_t,uint64_t> ESCH(8u);
typedef HistogramThread< ::libmaus2::util::GetObject<uint8_t const *> > thread_type;
typedef thread_type::unique_ptr_type thread_ptr_type;
::libmaus2::autoarray::AutoArray< ::libmaus2::util::GetObject<uint8_t const *>::unique_ptr_type > getters(numparts);
::libmaus2::autoarray::AutoArray<thread_ptr_type> threads(numparts);
for ( uint64_t i = 0; i < numparts; ++i )
{
::libmaus2::util::GetObject<uint8_t const *>::unique_ptr_type tgettersi(
new ::libmaus2::util::GetObject<uint8_t const *>(A.begin()+partstarts[i])
);
getters[i] = UNIQUE_PTR_MOVE(tgettersi);
thread_ptr_type tthreadsi(new thread_type(*getters[i],
partstarts[i+1]-partstarts[i],mutex,ESCH,i));
threads[i] = UNIQUE_PTR_MOVE(tthreadsi);
}
for ( uint64_t i = 0; i < numparts; ++i )
{
threads[i]->join();
threads[i].reset();
}
::libmaus2::autoarray::AutoArray< std::pair<int64_t,uint64_t> > R(ESCH.size(),false);
uint64_t p = 0;
for ( ::libmaus2::util::ExtendingSimpleCountingHash<uint64_t,uint64_t>::key_type const * ita =
ESCH.begin(); ita != ESCH.end(); ++ita )
if ( *ita != ::libmaus2::util::ExtendingSimpleCountingHash<uint64_t,uint64_t>::unused() )
R [ p++ ] = std::pair<int64_t,uint64_t>(*ita,ESCH.getCount(*ita));
std::sort(R.begin(),R.end());
return R;
}
/**
* seek to absolute position
**/
::std::streampos seekpos(::std::streampos sp, ::std::ios_base::openmode /* which */)
{
// flush write buffer before seeking anywhere
checkWriteBuffer();
// seek
off_t const off = doSeek(sp,SEEK_SET);
if ( off == static_cast<off_t>(-1) )
return -1;
// empty get buffer
setg(buffer.end(),buffer.end(),buffer.end());
// empty put buffer
setp(buffer.begin(),buffer.end()-1);
// set positions
readpos = off;
writepos = off;
return off;
}
/**
* decompress the currenctly buffered block
*
* @return number of uncompressed bytes in block, zero for EOF or failure
**/
uint64_t decompressBlock()
{
state = bgzfinflateblockstate_decompressed_block;
if ( failed() )
return 0;
if ( ! blockinfo.uncompressed )
return 0;
try
{
BgzfInflateBase::decompressBlock(
reinterpret_cast<char *>(data.begin()),
std::make_pair(blockinfo.compressed,blockinfo.uncompressed)
);
return blockinfo.uncompressed;
}
catch(libmaus2::exception::LibMausException const & lex)
{
libmaus2::exception::LibMausException::unique_ptr_type tex(lex.uclone());
ex = UNIQUE_PTR_MOVE(tex);
return 0;
}
catch(std::exception const & lex)
{
libmaus2::exception::LibMausException::unique_ptr_type tex(new libmaus2::exception::LibMausException);
ex = UNIQUE_PTR_MOVE(tex);
ex->getStream() << lex.what();
ex->finish(false);
return 0;
}
catch(...)
{
libmaus2::exception::LibMausException::unique_ptr_type tex(new libmaus2::exception::LibMausException);
ex = UNIQUE_PTR_MOVE(tex);
ex->getStream() << "BgzfInflateBlock::decompressBlock(): unknown exception caught";
ex->finish(false);
return 0;
}
}
virtual size_t compress(char const * input, size_t inputLength, libmaus2::autoarray::AutoArray<char> & output)
{
zintf->z_deflateReset();
if ( inputLength > inputBound )
{
inputBound = inputLength;
outputBound = zintf->z_deflateBound(inputBound);
}
if ( outputBound > output.size() )
output = libmaus2::autoarray::AutoArray<char>(outputBound,false);
// maximum number of output bytes
zintf->setAvailOut(output.size());
// next compressed output byte
zintf->setNextOut(reinterpret_cast<Bytef *>(output.begin()));
// number of bytes to be compressed
zintf->setAvailIn(inputLength);
// data to be compressed
zintf->setNextIn(const_cast<Bytef *>(reinterpret_cast<Bytef const *>(input)));
int const retcode = zintf->z_deflate(Z_FINISH);
// std::cerr << "avail_out=" << strm.avail_out << std::endl;
// std::cerr << "avail_in=" << strm.avail_in << std::endl;
// call deflate
if ( retcode != Z_STREAM_END )
{
libmaus2::exception::LibMausException se;
se.getStream() << "deflate() failed: " << retcode << ", " << zError(retcode) << std::endl;
se.finish(false /* do not translate stack trace */);
throw se;
}
uint64_t const compsize = output.size() - zintf->getAvailOut();
return compsize;
}
