static int gff3_visitor_region_node(GtNodeVisitor *nv, GtRegionNode *rn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && rn);
gff3_version_string(nv);
if (!gff3_visitor->outstr) {
gt_file_xprintf(gff3_visitor->outfp, "%s %s "GT_WU" "GT_WU"\n",
GT_GFF_SEQUENCE_REGION,
gt_str_get(gt_genome_node_get_seqid((GtGenomeNode*) rn)),
gt_genome_node_get_start((GtGenomeNode*) rn),
gt_genome_node_get_end((GtGenomeNode*) rn));
} else {
gt_str_append_cstr(gff3_visitor->outstr, GT_GFF_SEQUENCE_REGION);
gt_str_append_cstr(gff3_visitor->outstr, " ");
gt_str_append_cstr(gff3_visitor->outstr,
gt_str_get(gt_genome_node_get_seqid((GtGenomeNode*) rn)));
gt_str_append_char(gff3_visitor->outstr, ' ');
gt_str_append_ulong(gff3_visitor->outstr,
gt_genome_node_get_start((GtGenomeNode*) rn));
gt_str_append_char(gff3_visitor->outstr, ' ');
gt_str_append_ulong(gff3_visitor->outstr,
gt_genome_node_get_end((GtGenomeNode*) rn));
gt_str_append_char(gff3_visitor->outstr, '\n');
}
return 0;
}
static int gff3_visitor_sequence_node(GtNodeVisitor *nv, GtSequenceNode *sn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && sn);
gff3_version_string(nv);
if (!gff3_visitor->fasta_directive_shown) {
if (!gff3_visitor->outstr)
gt_file_xprintf(gff3_visitor->outfp, "%s\n", GT_GFF_FASTA_DIRECTIVE);
else {
gt_str_append_cstr(gff3_visitor->outstr, GT_GFF_FASTA_DIRECTIVE);
gt_str_append_char(gff3_visitor->outstr, '\n');
}
gff3_visitor->fasta_directive_shown = true;
}
if (!gff3_visitor->outstr) {
gt_fasta_show_entry(gt_sequence_node_get_description(sn),
gt_sequence_node_get_sequence(sn),
gt_sequence_node_get_sequence_length(sn),
gff3_visitor->fasta_width, gff3_visitor->outfp);
} else {
gt_fasta_show_entry_str(gt_sequence_node_get_description(sn),
gt_sequence_node_get_sequence(sn),
gt_sequence_node_get_sequence_length(sn),
gff3_visitor->fasta_width, gff3_visitor->outstr);
}
return 0;
}
static int gff3_visitor_meta_node(GtNodeVisitor *nv, GtMetaNode *mn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && mn);
if (!gff3_visitor->version_string_shown) {
if (strncmp(gt_meta_node_get_directive(mn), GT_GFF_VERSION_DIRECTIVE,
strlen(GT_GFF_VERSION_DIRECTIVE)) == 0
|| strncmp(gt_meta_node_get_directive(mn), GT_GVF_VERSION_DIRECTIVE,
strlen(GT_GVF_VERSION_DIRECTIVE)) == 0) {
gff3_visitor->version_string_shown = true;
} else {
gff3_version_string(nv);
}
}
if (!gff3_visitor->outstr) {
gt_file_xprintf(gff3_visitor->outfp, "##%s %s\n",
gt_meta_node_get_directive(mn),
gt_meta_node_get_data(mn));
} else {
gt_str_append_cstr(gff3_visitor->outstr, "##");
gt_str_append_cstr(gff3_visitor->outstr, gt_meta_node_get_directive(mn));
gt_str_append_char(gff3_visitor->outstr, ' ');
gt_str_append_cstr(gff3_visitor->outstr, gt_meta_node_get_data(mn));
gt_str_append_char(gff3_visitor->outstr, '\n');
}
return 0;
}
GtNodeVisitor* gt_gff3_visitor_new_to_str(GtStr *outstr)
{
GtNodeVisitor *nv = gt_node_visitor_create(gt_gff3_visitor_class());
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_gff3_visitor_init(gff3_visitor);
gff3_visitor->outfp = NULL;
gff3_visitor->outstr = gt_str_ref(outstr);
return nv;
}
GtNodeVisitor* gt_gff3_visitor_new(GtFile *outfp)
{
GtNodeVisitor *nv = gt_node_visitor_create(gt_gff3_visitor_class());
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_gff3_visitor_init(gff3_visitor);
gff3_visitor->outfp = outfp;
gff3_visitor->outstr = NULL;
return nv;
}
static void gff3_visitor_free(GtNodeVisitor *nv)
{
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_assert(gff3_visitor);
gt_string_distri_delete(gff3_visitor->id_counter);
gt_hashmap_delete(gff3_visitor->feature_node_to_id_array);
gt_hashmap_delete(gff3_visitor->feature_node_to_unique_id_str);
gt_cstr_table_delete(gff3_visitor->used_ids);
}
static void gff3_version_string(GtNodeVisitor *nv)
{
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_assert(gff3_visitor);
if (!gff3_visitor->version_string_shown) {
gt_file_xprintf(gff3_visitor->outfp, "%s %u\n", GT_GFF_VERSION_PREFIX,
GT_GFF_VERSION);
gff3_visitor->version_string_shown = true;
}
}
static int gff3_visitor_comment_node(GtNodeVisitor *nv, GtCommentNode *cn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && cn);
gff3_version_string(nv);
gt_file_xprintf(gff3_visitor->outfp, "#%s\n",
gt_comment_node_get_comment(cn));
return 0;
}
static int gff3_visitor_region_node(GtNodeVisitor *nv, GtRegionNode *rn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && rn);
gff3_version_string(nv);
gt_file_xprintf(gff3_visitor->outfp, "%s %s %lu %lu\n",
GT_GFF_SEQUENCE_REGION,
gt_str_get(gt_genome_node_get_seqid((GtGenomeNode*) rn)),
gt_genome_node_get_start((GtGenomeNode*) rn),
gt_genome_node_get_end((GtGenomeNode*) rn));
return 0;
}
static void gff3_version_string(GtNodeVisitor *nv)
{
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gt_assert(gff3_visitor);
if (!gff3_visitor->version_string_shown) {
if (!gff3_visitor->outstr) {
gt_file_xprintf(gff3_visitor->outfp, "%s %u\n", GT_GFF_VERSION_PREFIX,
GT_GFF_VERSION);
} else {
gt_str_append_cstr(gff3_visitor->outstr, GT_GFF_VERSION_PREFIX);
gt_str_append_cstr(gff3_visitor->outstr, " ");
gt_str_append_uint(gff3_visitor->outstr, GT_GFF_VERSION);
gt_str_append_char(gff3_visitor->outstr, '\n');
}
gff3_visitor->version_string_shown = true;
}
}
GtNodeVisitor* gt_gff3_visitor_new(GtFile *outfp)
{
GtNodeVisitor *nv = gt_node_visitor_create(gt_gff3_visitor_class());
GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(nv);
gff3_visitor->version_string_shown = false;
gff3_visitor->fasta_directive_shown = false;
gff3_visitor->id_counter = gt_string_distri_new();
gff3_visitor->feature_node_to_id_array =
gt_hashmap_new(GT_HASH_DIRECT, NULL, (GtFree) gt_array_delete);
gff3_visitor->feature_node_to_unique_id_str =
gt_hashmap_new(GT_HASH_DIRECT, NULL, (GtFree) gt_str_delete);
gff3_visitor->fasta_width = 0;
gff3_visitor->outfp = outfp;
gff3_visitor->used_ids = gt_cstr_table_new();
gff3_visitor->retain_ids = false;
return nv;
}
static int gff3_visitor_feature_node(GtNodeVisitor *nv, GtFeatureNode *fn,
GtError *err)
{
GtGFF3Visitor *gff3_visitor;
int had_err;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gff3_version_string(nv);
had_err = gt_feature_node_traverse_children(fn, gff3_visitor, store_ids, true,
err);
if (!had_err) {
if (gt_feature_node_is_tree(fn)) {
had_err = gt_feature_node_traverse_children(fn, gff3_visitor,
gff3_show_feature_node, true,
err);
}
else {
/* got a DAG -> traverse in topologically sorted depth first fashion to
make sure that the 'Parent' attributes are shown in correct order */
had_err =
gt_feature_node_traverse_children_top(fn, gff3_visitor,
gff3_show_feature_node, err);
}
}
/* reset hashmaps */
gt_hashmap_reset(gff3_visitor->feature_node_to_id_array);
gt_hashmap_reset(gff3_visitor->feature_node_to_unique_id_str);
/* show terminator, if the feature has children (otherwise it is clear that
the feature is complete, because no ID attribute has been shown) */
if (gt_feature_node_has_children(fn) ||
(gff3_visitor->retain_ids && gt_feature_node_get_attribute(fn, "ID"))) {
if (!gff3_visitor->outstr)
gt_file_xprintf(gff3_visitor->outfp, "%s\n", GT_GFF_TERMINATOR);
else {
gt_str_append_cstr(gff3_visitor->outstr, GT_GFF_TERMINATOR);
gt_str_append_char(gff3_visitor->outstr, '\n');
}
}
return had_err;
}
static int gff3_visitor_comment_node(GtNodeVisitor *nv, GtCommentNode *cn,
GT_UNUSED GtError *err)
{
GtGFF3Visitor *gff3_visitor;
gt_error_check(err);
gff3_visitor = gff3_visitor_cast(nv);
gt_assert(nv && cn);
gff3_version_string(nv);
if (!gff3_visitor->outstr) {
gt_file_xprintf(gff3_visitor->outfp, "#%s\n",
gt_comment_node_get_comment(cn));
} else {
gt_str_append_char(gff3_visitor->outstr, '#');
gt_str_append_cstr(gff3_visitor->outstr, gt_comment_node_get_comment(cn));
gt_str_append_char(gff3_visitor->outstr, '\n');
}
return 0;
}
