static int fill_mapping(GtSeqid2SeqnumMapping *mapping, GtBioseq *bioseq,
GtSeqCol *seqcol, GT_UNUSED GtError *err)
{
GtUword i, j, nof_sequences, nof_files;
int had_err = 0;
gt_error_check(err);
gt_assert(mapping && (bioseq || seqcol) && !(bioseq && seqcol));
if (bioseq) {
nof_files = 1;
} else {
gt_assert(seqcol);
nof_files = gt_seq_col_num_of_files(seqcol);
}
for (j = 0; !had_err && j < nof_files; j++) {
if (bioseq)
nof_sequences = gt_bioseq_number_of_sequences(bioseq);
else {
gt_assert(seqcol);
nof_sequences = gt_seq_col_num_of_seqs(seqcol, j);
}
for (i = 0; !had_err && i < nof_sequences; i++) {
char *desc;
if (bioseq)
desc = gt_cstr_dup(gt_bioseq_get_description(bioseq, i));
else
desc = gt_seq_col_get_description(seqcol, j, i);
had_err = handle_description(mapping, desc, i, j, err);
gt_free(desc);
}
}
return had_err;
}
int gt_region_mapping_get_description(GtRegionMapping *rm, GtStr *desc,
GtStr *seqid, GtError *err)
{
int had_err = 0;
gt_error_check(err);
gt_assert(rm && desc && seqid);
if (rm->userawseq) {
gt_str_append_cstr(desc, "<rawseq>");
return 0;
}
had_err = update_seq_col_if_necessary(rm, seqid, err);
if (!had_err) {
if (gt_md5_seqid_has_prefix(gt_str_get(seqid))) {
had_err = gt_seq_col_md5_to_description(rm->seq_col, desc, seqid,
err);
}
return had_err;
}
if (!had_err) {
if (rm->usedesc) {
unsigned long filenum, seqnum;
gt_assert(rm->seqid2seqnum_mapping);
had_err = gt_seqid2seqnum_mapping_map(rm->seqid2seqnum_mapping,
gt_str_get(seqid), NULL, &seqnum,
&filenum, NULL, err);
if (!had_err) {
char *cdesc;
cdesc = gt_seq_col_get_description(rm->seq_col, filenum, seqnum);
gt_assert(cdesc);
gt_str_append_cstr(desc, cdesc);
gt_free(cdesc);
}
}
else if (rm->useseqno) {
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) {
unsigned long desclen;
const char *edesc;
edesc = gt_encseq_description(rm->encseq, &desclen, seqno);
gt_str_append_cstr_nt(desc, edesc, desclen);
}
} else if (rm->matchdesc) {
const char *md5;
/* XXX: not beautiful, but works -- this may be LOTS faster */
had_err = gt_seq_col_grep_desc_md5(rm->seq_col, &md5, seqid, err);
if (!had_err) {
GtStr *md5_seqid = gt_str_new_cstr(md5);
had_err = gt_seq_col_md5_to_description(rm->seq_col, desc, md5_seqid,
err);
gt_str_delete(md5_seqid);
}
} else {
if (!had_err) {
char *cdesc;
cdesc = gt_seq_col_get_description(rm->seq_col, 0, 0);
gt_assert(cdesc);
gt_str_append_cstr(desc, cdesc);
gt_free(cdesc);
}
}
}
return had_err;
}