int gt_echodescriptionandsequence(const GtStrArray *filenametab,GtError *err)
{
GtSeqIterator *seqit;
char *desc = NULL;
const GtUchar *sequence;
unsigned long seqlen;
bool haserr = false;
int retval;
seqit = gt_seqiterator_sequence_buffer_new(filenametab, err);
if (!seqit)
return -1;
while (true)
{
retval = gt_seqiterator_next(seqit,
&sequence,
&seqlen,
&desc,
err);
if (retval < 0)
{
haserr = true;
break;
}
if (retval == 0)
{
break;
}
gt_symbolstring2fasta(stdout,desc,NULL,sequence,seqlen,70UL);
}
gt_seqiterator_delete(seqit);
return haserr ? -1 : 0;
}
int gt_testmaxpairs(const char *indexname,
GtUword samples,
unsigned int minlength,
GtUword substringlength,
GtLogger *logger,
GtError *err)
{
GtEncseq *encseq;
GtUword totallength = 0, dblen, querylen;
GtUchar *dbseq = NULL, *query = NULL;
bool haserr = false;
GtUword s;
GtArray *tabmaxquerymatches;
Maxmatchselfinfo maxmatchselfinfo;
GtEncseqLoader *el;
gt_logger_log(logger,"draw "GT_WU" samples",samples);
el = gt_encseq_loader_new();
gt_encseq_loader_do_not_require_des_tab(el);
gt_encseq_loader_do_not_require_ssp_tab(el);
gt_encseq_loader_do_not_require_sds_tab(el);
gt_encseq_loader_set_logger(el, logger);
encseq = gt_encseq_loader_load(el, indexname, err);
gt_encseq_loader_delete(el);
if (encseq == NULL)
{
haserr = true;
} else
{
totallength = gt_encseq_total_length(encseq);
}
if (!haserr)
{
if (substringlength > totallength/2)
{
substringlength = totallength/2;
}
dbseq = gt_malloc(sizeof *dbseq * substringlength);
query = gt_malloc(sizeof *query * substringlength);
}
for (s=0; s<samples && !haserr; s++)
{
dblen = samplesubstring(dbseq,encseq,substringlength);
querylen = samplesubstring(query,encseq,substringlength);
gt_logger_log(logger,"run query match for dblen="GT_WU""
",querylen= "GT_WU", minlength=%u",
dblen,
querylen,
minlength);
tabmaxquerymatches = gt_array_new(sizeof (Substringmatch));
if (gt_sarrquerysubstringmatch(dbseq,
dblen,
query,
(GtUword) querylen,
minlength,
gt_encseq_alphabet(encseq),
storemaxmatchquery,
tabmaxquerymatches,
logger,
err) != 0)
{
haserr = true;
break;
}
gt_logger_log(logger,"run self match for dblen="GT_WU""
",querylen= "GT_WU", minlength=%u",
dblen,
querylen,
minlength);
maxmatchselfinfo.results = gt_array_new(sizeof (Substringmatch));
maxmatchselfinfo.dblen = dblen;
maxmatchselfinfo.querylen = querylen;
maxmatchselfinfo.querymarkpos
= sequence2markpositions(&maxmatchselfinfo.numofquerysequences,
query,querylen);
if (sarrselfsubstringmatch(dbseq,
dblen,
query,
(GtUword) querylen,
minlength,
gt_encseq_alphabet(encseq),
storemaxmatchself,
&maxmatchselfinfo,
logger,
err) != 0)
{
haserr = true;
break;
}
gt_array_sort(tabmaxquerymatches,orderSubstringmatch);
gt_array_sort(maxmatchselfinfo.results,orderSubstringmatch);
if (!gt_array_equal(tabmaxquerymatches,maxmatchselfinfo.results,
orderSubstringmatch))
{
const GtUword width = 60UL;
printf("failure for query of length "GT_WU"\n",(GtUword) querylen);
printf("querymatches\n");
(void) gt_array_iterate(tabmaxquerymatches,showSubstringmatch,NULL,
err);
printf("dbmatches\n");
(void) gt_array_iterate(maxmatchselfinfo.results,showSubstringmatch,
NULL,err);
gt_symbolstring2fasta(stdout,"dbseq",
gt_encseq_alphabet(encseq),
dbseq,
(GtUword) dblen,
width);
gt_symbolstring2fasta(stdout,"queryseq",
gt_encseq_alphabet(encseq),
query,
(GtUword) querylen,
width);
exit(GT_EXIT_PROGRAMMING_ERROR);
}
gt_free(maxmatchselfinfo.querymarkpos);
printf("# numberofmatches="GT_WU"\n",gt_array_size(tabmaxquerymatches));
gt_array_delete(tabmaxquerymatches);
gt_array_delete(maxmatchselfinfo.results);
}
gt_free(dbseq);
gt_free(query);
gt_encseq_delete(encseq);
encseq = NULL;
return haserr ? -1 : 0;
}