static GtUword gt_mmsearch_extendright(const GtEncseq *dbencseq,
GtEncseqReader *esr,
GtReadmode readmode,
GtUword totallength,
GtUword dbend,
const GtQuerysubstring
*querysubstring,
GtUword matchlength)
{
GtUchar dbchar;
GtUword dbpos, querypos;
if (dbend < totallength)
{
gt_encseq_reader_reinit_with_readmode(esr,dbencseq,readmode,dbend);
}
for (dbpos = dbend, querypos = querysubstring->currentoffset + matchlength;
dbpos < totallength &&
querypos < querysubstring->queryrep->seqlen;
dbpos++, querypos++)
{
dbchar = gt_encseq_reader_next_encoded_char(esr);
if (ISSPECIAL(dbchar) ||
dbchar != gt_mmsearch_accessquery(querysubstring->queryrep,querypos))
{
break;
}
}
return dbpos - dbend;
}
static bool gt_mmsearch_isleftmaximal(const GtEncseq *dbencseq,
GtReadmode readmode,
GtUword dbstart,
const GtQuerysubstring *querysubstring)
{
GtUchar dbleftchar;
if (dbstart == 0 || querysubstring->currentoffset == 0)
{
return true;
}
dbleftchar = gt_encseq_get_encoded_char(dbencseq, /* Random access */
dbstart-1,
readmode);
if (ISSPECIAL(dbleftchar) ||
dbleftchar != gt_mmsearch_accessquery(querysubstring->queryrep,
querysubstring->currentoffset-1))
{
return true;
}
return false;
}
static void gt_querysubstringmatch(bool selfmatch,
const GtEncseq *dbencseq,
const ESASuffixptr *suftabpart,
GtReadmode readmode,
GtUword numberofsuffixes,
uint64_t queryunitnum,
GtQueryrepresentation *queryrep,
GtUword minmatchlength,
GtProcessquerymatch processquerymatch,
void *processquerymatchinfo,
GtQuerymatch *querymatchspaceptr)
{
GtMMsearchiterator *mmsi;
GtUword totallength, localqueryoffset = 0;
uint64_t localqueryunitnum = queryunitnum;
GtQuerysubstring querysubstring;
gt_assert(numberofsuffixes > 0);
totallength = gt_encseq_total_length(dbencseq);
querysubstring.queryrep = queryrep;
for (querysubstring.currentoffset = 0;
querysubstring.currentoffset <= queryrep->seqlen - minmatchlength;
querysubstring.currentoffset++)
{
GtUword dbstart;
mmsi = gt_mmsearchiterator_new(dbencseq,
suftabpart,
0, /* leftbound */
numberofsuffixes - 1, /* rightbound */
0, /* offset */
readmode,
&querysubstring,
minmatchlength);
while (gt_mmsearchiterator_next(&dbstart,mmsi))
{
if (gt_mmsearch_isleftmaximal(dbencseq,
readmode,
dbstart,
&querysubstring))
{
GtUword dbseqnum, dbseqstartpos, dbseqlen, extend;
extend = gt_mmsearch_extendright(dbencseq,
mmsi->esr,
readmode,
totallength,
dbstart + minmatchlength,
&querysubstring,
minmatchlength);
if (gt_encseq_has_multiseq_support(dbencseq))
{
dbseqnum = gt_encseq_seqnum(dbencseq,dbstart);
dbseqstartpos = gt_encseq_seqstartpos(dbencseq,dbseqnum);
dbseqlen = gt_encseq_seqlength(dbencseq,dbseqnum);
} else
{
dbseqnum = dbseqstartpos = dbseqlen = 0;
}
gt_querymatch_init(querymatchspaceptr,
minmatchlength + extend,
dbstart,
dbseqnum,
dbstart - dbseqstartpos,
dbseqlen,
0, /* score */
0, /* edist */
selfmatch,
localqueryunitnum,
minmatchlength + extend,
localqueryoffset,
queryrep->seqlen);
processquerymatch(processquerymatchinfo,querymatchspaceptr);
}
}
gt_mmsearchiterator_delete(mmsi);
mmsi = NULL;
if (gt_mmsearch_accessquery(queryrep,querysubstring.currentoffset)
== (GtUchar) SEPARATOR)
{
localqueryunitnum++;
localqueryoffset = 0;
} else
{
localqueryoffset++;
}
}
}
static int gt_querysubstringmatch_generic(
bool selfmatch,
const GtEncseq *dbencseq,
const ESASuffixptr *suftabpart,
GtReadmode readmode,
unsigned long numberofsuffixes,
uint64_t queryunitnum,
const GtQueryrep *queryrep,
unsigned long minmatchlength,
GtProcessquerymatch processquerymatch,
void *processquerymatchinfo,
GtQuerymatch *querymatchspaceptr,
GtError *err)
{
GtMMsearchiterator *mmsi;
unsigned long totallength, localqueryoffset = 0;
uint64_t localqueryunitnum = queryunitnum;
GtQuerysubstring querysubstring;
bool haserr = false;
gt_assert(numberofsuffixes > 0);
totallength = gt_encseq_total_length(dbencseq);
querysubstring.queryrep = queryrep;
for (querysubstring.offset = 0;
querysubstring.offset <= queryrep->length - minmatchlength;
querysubstring.offset++)
{
unsigned long dbstart;
mmsi = gt_mmsearchiterator_new_generic(dbencseq,
suftabpart,
0, /* leftbound */
numberofsuffixes-1, /* rightbound */
0, /* offset */
readmode,
&querysubstring,
minmatchlength);
while (!haserr && gt_mmsearchiterator_next(&dbstart,mmsi))
{
if (gt_mmsearch_isleftmaximal(dbencseq,
readmode,
dbstart,
&querysubstring))
{
unsigned long extend = gt_mmsearch_extendright(dbencseq,
mmsi->esr,
readmode,
totallength,
dbstart + minmatchlength,
&querysubstring,
minmatchlength);
gt_querymatch_fill(querymatchspaceptr,
minmatchlength + extend,
dbstart,
queryrep->readmode,
queryrep->reversecopy,
0, /* score */
0, /* edist */
selfmatch,
localqueryunitnum,
minmatchlength + extend,
localqueryoffset);
if (processquerymatch(processquerymatchinfo,
dbencseq,
querymatchspaceptr,
queryrep->sequence,
queryrep->length,
err) != 0)
{
haserr = true;
}
}
}
gt_mmsearchiterator_delete(mmsi);
mmsi = NULL;
if (!haserr)
{
if (gt_mmsearch_accessquery(queryrep,querysubstring.offset)
== (GtUchar) SEPARATOR)
{
localqueryunitnum++;
localqueryoffset = 0;
} else
{
localqueryoffset++;
}
}
}
return haserr ? -1 : 0;
}