void gt_kmercodeiterator_delete(GtKmercodeiterator *kmercodeiterator)
{
if (kmercodeiterator != NULL)
{
gt_encseq_reader_delete(kmercodeiterator->esr);
kmerstream_delete(kmercodeiterator->spwp);
gt_sequence_buffer_delete(kmercodeiterator->fb);
gt_free(kmercodeiterator);
}
}
void getencseqkmers(const GtEncseq *encseq,
GtReadmode readmode,
unsigned int kmersize,
void(*processkmercode)(void *,
unsigned long,
const GtKmercode *),
void *processkmercodeinfo)
{
unsigned long currentposition = 0, totallength;
Kmerstream *spwp;
GtUchar charcode;
GtEncseqReader *esr;
unsigned int numofchars, overshoot;
totallength = gt_encseq_total_length(encseq);
if (totallength < (unsigned long) kmersize)
{
return;
}
numofchars = gt_alphabet_num_of_chars(gt_encseq_alphabet(encseq));
spwp = kmerstream_new(numofchars,kmersize);
esr = gt_encseq_create_reader_with_readmode(encseq,readmode,0);
for (currentposition = 0; currentposition < (unsigned long) kmersize;
currentposition++)
{
charcode = gt_encseq_reader_next_encoded_char(esr);
GT_CHECKENCCHAR(charcode,encseq,currentposition,readmode);
spwp->windowwidth++;
updatespecialpositions(spwp,charcode,false,0);
spwp->cyclicwindow[spwp->windowwidth-1] = charcode;
}
kmerstream_newcode(&spwp->currentkmercode,spwp);
processkmercode(processkmercodeinfo,0,&spwp->currentkmercode);
for (currentposition = (unsigned long) kmersize; currentposition<totallength;
currentposition++)
{
charcode = gt_encseq_reader_next_encoded_char(esr);
GT_CHECKENCCHAR(charcode,encseq,currentposition,readmode);
shiftrightwithchar(spwp,charcode);
kmerstream_newcode(&spwp->currentkmercode,spwp);
processkmercode(processkmercodeinfo,currentposition + 1 - spwp->kmersize,
&spwp->currentkmercode);
}
gt_encseq_reader_delete(esr);
for (overshoot=0; overshoot<kmersize; overshoot++)
{
shiftrightwithchar(spwp,(GtUchar) WILDCARD);
kmerstream_newcode(&spwp->currentkmercode,spwp);
processkmercode(processkmercodeinfo,
overshoot + currentposition + 1 - spwp->kmersize,
&spwp->currentkmercode);
}
kmerstream_delete(spwp);
}
void gt_kmercodeiterator_reset(GtKmercodeiterator *kmercodeiterator,
GtReadmode readmode,
GtUword startpos)
{
GtUchar charcode;
const GtEncseq *encseq = kmercodeiterator->encseq;
GtUword kmersize = (GtUword) kmercodeiterator->spwp->kmersize;
gt_assert(!GT_ISDIRREVERSE(readmode) || startpos == 0);
kmercodeiterator->totallength = gt_encseq_total_length(encseq);
kmercodeiterator->startpos = startpos;
gt_assert(startpos < kmercodeiterator->totallength);
kmercodeiterator->fb = NULL;
if (kmercodeiterator->totallength - startpos < kmersize)
{
kmercodeiterator->inputexhausted = true;
gt_encseq_reader_delete(kmercodeiterator->esr);
kmercodeiterator->esr = NULL;
kmerstream_delete(kmercodeiterator->spwp);
kmercodeiterator->spwp = NULL;
} else
{
kmercodeiterator->inputexhausted = false;
kmercodeiterator->readmode = readmode;
gt_encseq_reader_reinit_with_readmode(kmercodeiterator->esr,
encseq,
readmode,
startpos);
kmerstream_reset(kmercodeiterator->spwp);
kmercodeiterator->hasprocessedfirst = false;
for (kmercodeiterator->currentposition = startpos;
kmercodeiterator->currentposition < startpos+(GtUword) kmersize;
kmercodeiterator->currentposition++)
{
charcode = gt_encseq_reader_next_encoded_char(kmercodeiterator->esr);
kmercodeiterator->spwp->windowwidth++;
kmerstream_updatespecialpositions(kmercodeiterator->spwp,charcode,
false,0);
kmercodeiterator->spwp->cyclicwindow[kmercodeiterator->
spwp->windowwidth-1] = charcode;
}
}
}
