static void gt_seq_iterator_sequence_buffer_set_symbolmap(GtSeqIterator *si,
const GtUchar
*symbolmap)
{
GtSeqIteratorSequenceBuffer *seqit;
gt_assert(si);
seqit = gt_seq_iterator_sequence_buffer_cast(si);
gt_sequence_buffer_set_symbolmap(seqit->fb, symbolmap);
}
GtKmercodeiterator *gt_kmercodeiterator_filetab_new(
const GtStrArray *filenametab,
unsigned int numofchars,
unsigned int kmersize,
const GtUchar *symbolmap,
bool plainformat,
GtError *err)
{
GtKmercodeiterator *kmercodeiterator;
GtUchar charcode;
bool haserr = false;
int retval;
gt_error_check(err);
kmercodeiterator = gt_malloc(sizeof (*kmercodeiterator));
kmercodeiterator->esr = NULL;
kmercodeiterator->hasprocessedfirst = false;
kmercodeiterator->inputexhausted = false;
kmercodeiterator->spwp = kmerstream_new(numofchars,kmersize);
kmercodeiterator->totallength = 0;
if (plainformat)
{
kmercodeiterator->fb = gt_sequence_buffer_plain_new(filenametab);
} else
{
kmercodeiterator->fb = gt_sequence_buffer_new_guess_type(filenametab, err);
}
if (kmercodeiterator->fb == NULL)
{
haserr = true;
}
if (!haserr)
{
gt_sequence_buffer_set_symbolmap(kmercodeiterator->fb, symbolmap);
for (kmercodeiterator->currentposition = 0;
kmercodeiterator->currentposition < (unsigned long) kmersize;
kmercodeiterator->currentposition++)
{
retval = gt_sequence_buffer_next(kmercodeiterator->fb,&charcode,err);
if (retval < 0)
{
haserr = true;
break;
}
if (retval == 0)
{
kmercodeiterator->inputexhausted = true;
break;
}
kmercodeiterator->spwp->windowwidth++;
updatespecialpositions(kmercodeiterator->spwp,charcode,false,0);
kmercodeiterator->spwp->cyclicwindow[kmercodeiterator->
spwp->windowwidth-1] = charcode;
}
}
if (haserr)
{
gt_kmercodeiterator_delete(kmercodeiterator);
return NULL;
}
return kmercodeiterator;
}
static int testfullscan(const GtStrArray *filenametab,
const Encodedsequence *encseq,
Readmode readmode,
GtError *err)
{
Seqpos pos, totallength;
GtUchar ccscan = 0, ccra, ccsr;
GtSequenceBuffer *fb = NULL;
int retval;
bool haserr = false;
Encodedsequencescanstate *esr;
unsigned long long fullscanpbar = 0;
gt_error_check(err);
totallength = getencseqtotallength(encseq);
gt_progressbar_start(&fullscanpbar,(unsigned long long) totallength);
if (filenametab != NULL)
{
fb = gt_sequence_buffer_new_guess_type((GtStrArray*) filenametab, err);
if (!fb)
haserr = true;
if (!haserr)
gt_sequence_buffer_set_symbolmap(fb, getencseqAlphabetsymbolmap(encseq));
}
if (!haserr) {
esr = newEncodedsequencescanstate();
initEncodedsequencescanstate(esr,encseq,readmode,0);
for (pos=0; /* Nothing */; pos++)
{
if (filenametab != NULL && readmode == Forwardmode)
{
retval = gt_sequence_buffer_next(fb,&ccscan,err);
if (retval < 0)
{
haserr = true;
break;
}
if (retval == 0)
{
break;
}
} else
{
if (pos >= totallength)
{
break;
}
}
ccra = getencodedchar(encseq,pos,readmode); /* Random access */
if (filenametab != NULL && readmode == Forwardmode)
{
if (ccscan != ccra)
{
gt_error_set(err,"access=%s, position=" FormatSeqpos
": scan (readnextchar) = %u != "
"%u = random access",
encseqaccessname(encseq),
pos,
(unsigned int) ccscan,
(unsigned int) ccra);
haserr = true;
break;
}
}
ccsr = sequentialgetencodedchar(encseq,esr,pos,readmode);
if (ccra != ccsr)
{
gt_error_set(err,"access=%s, mode=%s: position=" FormatSeqpos
": random access = %u != %u = sequential read",
encseqaccessname(encseq),
showreadmode(readmode),
pos,
(unsigned int) ccra,
(unsigned int) ccsr);
haserr = true;
break;
}
fullscanpbar++;
}
gt_progressbar_stop();
}
if (!haserr)
{
if (pos != totallength)
{
gt_error_set(err,"sequence length must be " FormatSeqpos " but is "
FormatSeqpos,totallength,pos);
haserr = true;
}
}
freeEncodedsequencescanstate(&esr);
gt_sequence_buffer_delete(fb);
return haserr ? -1 : 0;
}