GtUword gt_bioseq_md5_to_index(GtBioseq *bs, const char *md5)
{
gt_assert(bs && md5 && gt_encseq_has_md5_support(bs->encseq));
if (!bs->md5_tab) {
bs->md5_tab = gt_encseq_get_md5_tab(bs->encseq, NULL);
}
return gt_md5_tab_map(bs->md5_tab, md5);
}
const char* gt_bioseq_get_md5_fingerprint(GtBioseq *bs, GtUword idx)
{
gt_assert(bs && idx < gt_bioseq_number_of_sequences(bs));
if (!bs->md5_tab) {
bs->md5_tab = gt_encseq_get_md5_tab(bs->encseq, NULL);
}
gt_assert(gt_md5_tab_get(bs->md5_tab, idx));
return gt_md5_tab_get(bs->md5_tab, idx);
}
GtSeqCol* gt_encseq_col_new(GtEncseq *encseq, GtError *err)
{
GtSeqCol *sc;
GtEncseqCol *esc;
gt_error_check(err);
gt_assert(encseq);
if (!gt_encseq_has_md5_support(encseq)) {
gt_error_set(err, "encoded sequence has no MD5 support");
return NULL;
}
sc = gt_seq_col_create(gt_encseq_col_class());
esc = gt_encseq_col_cast(sc);
esc->md5_tab = gt_encseq_get_md5_tab(encseq, err);
gt_assert(esc->md5_tab);
esc->encseq = gt_encseq_ref(encseq);
return sc;
}
const char* gt_region_mapping_get_md5_fingerprint(GtRegionMapping *rm,
GtStr *seqid,
const GtRange *range,
unsigned long *offset,
GtError *err)
{
const char *md5 = NULL;
int had_err;
unsigned long filenum, seqnum;
gt_error_check(err);
gt_assert(rm && seqid);
gt_assert(!rm->userawseq); /* not implemented */
gt_assert(!gt_md5_seqid_has_prefix(gt_str_get(seqid))); /* not implemented */
had_err = update_seq_col_if_necessary(rm, seqid, err);
if (!had_err) {
if (rm->usedesc) {
gt_assert(rm->seqid2seqnum_mapping);
had_err = gt_seqid2seqnum_mapping_map(rm->seqid2seqnum_mapping,
gt_str_get(seqid), range, &seqnum,
&filenum, offset, err);
if (!had_err)
md5 = gt_seq_col_get_md5_fingerprint(rm->seq_col, filenum, seqnum);
}
else if (rm->matchdesc) {
if (!rm->seq_col) {
if (rm->encseq) {
if (!(rm->seq_col = gt_encseq_col_new(rm->encseq, err)))
had_err = -1;
} else {
if (!(rm->seq_col = gt_bioseq_col_new(rm->sequence_filenames, err)))
had_err = -1;
}
}
if (!had_err)
(void) gt_seq_col_grep_desc_md5(rm->seq_col, &md5, seqid, err);
*offset = 1;
}
else if (rm->useseqno) {
GtMD5Tab *tab = NULL;
unsigned long seqno = GT_UNDEF_ULONG;
gt_assert(rm->encseq);
if (1 != sscanf(gt_str_get(seqid), "seq%lu", &seqno)) {
gt_error_set(err, "seqid '%s' does not have the form 'seqX' "
"where X is a sequence number in the encoded "
"sequence", gt_str_get(seqid));
had_err = -1;
}
gt_assert(had_err || seqno != GT_UNDEF_ULONG);
if (!had_err && seqno >= gt_encseq_num_of_sequences(rm->encseq)) {
gt_error_set(err, "trying to access sequence %lu, but encoded"
"sequence contains only %lu sequences",
seqno, gt_encseq_num_of_sequences(rm->encseq));
had_err = -1;
}
if (!had_err) {
tab = gt_encseq_get_md5_tab(rm->encseq, err);
if (!tab)
had_err = -1;
}
*offset = 1;
if (!had_err)
return gt_md5_tab_get(tab, seqno);
else
return NULL;
}
else {
if (!had_err)
md5 = gt_seq_col_get_md5_fingerprint(rm->seq_col, 0, 0);
*offset = 1;
}
}
return md5;
}