gistfile1.diff

diff --git a/src/extended/gff3_out_stream.c b/src/extended/gff3_out_stream.c
index 6dbbc23..f813e8f 100644
--- a/src/extended/gff3_out_stream.c
+++ b/src/extended/gff3_out_stream.c
@@ -18,6 +18,7 @@
 #include "extended/gff3_out_stream.h"
 #include "extended/gff3_visitor.h"
 #include "extended/node_stream_rep.h"
+#include "core/cstr_table.h"
 
 struct GtGFF3OutStream {
   const GtNodeStream parent_instance;
@@ -76,3 +77,11 @@ void gt_gff3_out_stream_set_fasta_width(GtNodeStream *gs,
   gt_assert(gff3_out_stream);
   gt_gff3_visitor_set_fasta_width(gff3_out_stream->gff3_visitor, fasta_width);
 }
+
+void gt_gff3_out_stream_retain_id_attributes(GtNodeStream *gs)
+{
+  GtGFF3OutStream *gff3_out_stream = gff3_out_stream_cast(gs);
+  gt_assert(gff3_out_stream);
+  gt_gff3_visitor_retain_id_attributes(gff3_out_stream->gff3_visitor);
+
+}
diff --git a/src/extended/gff3_out_stream.h b/src/extended/gff3_out_stream.h
index adc3eba..e004c5d 100644
--- a/src/extended/gff3_out_stream.h
+++ b/src/extended/gff3_out_stream.h
@@ -28,5 +28,7 @@ const GtNodeStreamClass* gt_gff3_out_stream_class(void);
 GtNodeStream*            gt_gff3_out_stream_new(GtNodeStream*, GtGenFile*);
 void                     gt_gff3_out_stream_set_fasta_width(GtNodeStream*,
                                                             unsigned long);
+void                     gt_gff3_out_stream_retain_id_attributes(
+                                                      GtNodeStream *);
 
 #endif
diff --git a/src/extended/gff3_visitor.c b/src/extended/gff3_visitor.c
index 8246680..02432f7 100644
--- a/src/extended/gff3_visitor.c
+++ b/src/extended/gff3_visitor.c
@@ -23,21 +23,26 @@
 #include "core/ma.h"
 #include "core/unused_api.h"
 #include "core/string_distri.h"
+#include "core/cstr_table.h"
+#include "core/str_api.h"
 #include "extended/genome_node.h"
 #include "extended/gff3_output.h"
 #include "extended/gff3_parser.h"
 #include "extended/gff3_visitor.h"
 #include "extended/node_visitor_rep.h"
+#include <stdbool.h>
 
 struct GtGFF3Visitor {
   const GtNodeVisitor parent_instance;
   bool version_string_shown,
+       retain_ids,
        fasta_directive_shown;
   GtStringDistri *id_counter;
   GtHashmap *gt_feature_node_to_id_array,
             *gt_feature_node_to_unique_id_str;
   unsigned long fasta_width;
   GtGenFile *outfp;
+  GtCstrTable *gt_used_ids;
 };
 
 typedef struct {
@@ -190,6 +195,30 @@ static GtStr* create_unique_id(GtGFF3Visitor *gff3_visitor, GtFeatureNode *gf)
   return id;
 }
 
+static void make_unique_id_string(GtStr *current_id, unsigned long counter)
+{
+    /* name => name.1 */
+    gt_str_append_char(current_id, '.');
+    gt_str_append_ulong(current_id, counter);
+}
+
+static void make_id_unique(GtGFF3Visitor *gff3_visitor, GtStr *id)
+{
+
+  GtCstrTable *used_ids = gff3_visitor->gt_used_ids;
+  unsigned long i = 0;
+  const char *id_string = gt_str_get(id);
+
+  while ( gt_cstr_table_get(used_ids, gt_str_get(id) )) {
+    /* TODO: add warning */
+    make_unique_id_string(id, ++i);
+    gt_str_set(id, id_string);
+  }
+  /* update table with the new id */
+  gt_cstr_table_add(used_ids, gt_str_get(id));
+
+}
+
 static int store_ids(GtGenomeNode *gn, void *data, GtError *err)
 {
   GtGFF3Visitor *gff3_visitor = (GtGFF3Visitor*) data;
@@ -197,32 +226,48 @@ static int store_ids(GtGenomeNode *gn, void *data, GtError *err)
   AddIDInfo add_id_info;
   int had_err = 0;
   GtStr *id;
+  const char *id_string;
+  bool has_id = false;
 
   gt_error_check(err);
   gt_assert(gn && gf && gff3_visitor);
+  id_string = gt_feature_node_get_attribute(gf, "ID");
 
-  if (gt_genome_node_has_children(gn) || gt_feature_node_is_multi(gf)) {
-    if (gt_feature_node_is_multi(gf)) {
-      id = gt_hashmap_get(gff3_visitor->gt_feature_node_to_unique_id_str,
-                          gt_feature_node_get_multi_representative(gf));
-      if (!id) { /* the representative does not have its own id */
-        id = create_unique_id(gff3_visitor,
-                              gt_feature_node_get_multi_representative(gf));
-      }
+  if (id_string) {
+    id = gt_str_new_cstr(id_string);
+    has_id = true;
+  }
+  else {
+    /* no id, but it's a multi feature. can this even happen? */
+    if (gt_feature_node_is_multi(gf) || gt_genome_node_has_children(gn) ) {
       if (gt_feature_node_get_multi_representative(gf) != gf) {
-        gt_hashmap_add(gff3_visitor->gt_feature_node_to_unique_id_str, gf,
-                       gt_str_ref(id));
+        id = gt_hashmap_get(gff3_visitor->gt_feature_node_to_unique_id_str,
+                          gt_feature_node_get_multi_representative(gf));
+        has_id = true;
       }
     }
-    else
-      id = create_unique_id(gff3_visitor, gf);
+    /* it doesnt have an id and it is not a child-feature */
+    else if ( ! gt_feature_node_get_attribute(gf, "Parent")) {
+        id = create_unique_id(gff3_visitor, gf);
+        has_id = true;
+    }
+
+  }
+  if (has_id) {
+    if (gff3_visitor->retain_ids) {
+      make_id_unique(gff3_visitor, id);
+    }
+    gt_hashmap_add(gff3_visitor->gt_feature_node_to_unique_id_str, gf,
+                   gt_str_ref(id));
 
     /* for each child -> store the parent feature in the hash map */
     add_id_info.gt_feature_node_to_id_array =
-      gff3_visitor->gt_feature_node_to_id_array,
+       gff3_visitor->gt_feature_node_to_id_array,
     add_id_info.id = gt_str_get(id);
     had_err = gt_genome_node_traverse_direct_children(gn, &add_id_info, add_id,
                                                       err);
+    /* Q: needed? */
+    gt_str_delete(id);
   }
   return had_err;
 }
@@ -262,8 +307,9 @@ static int gff3_visitor_feature_node(GtNodeVisitor *gv, GtFeatureNode *fn,
 
   /* 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_genome_node_has_children((GtGenomeNode*) fn))
-    gt_genfile_xprintf(gff3_visitor->outfp, "%s\n", GFF_TERMINATOR);
+  /* if (gt_genome_node_has_children((GtGenomeNode*) fn)) */
+  gt_genfile_xprintf(gff3_visitor->outfp, "%s\n", GFF_TERMINATOR);
+  /* printf("%s\n", gt_feature_node_get_attribute(fn, "ID")); */
 
   return had_err;
 }
@@ -329,9 +375,18 @@ GtNodeVisitor* gt_gff3_visitor_new(GtGenFile *outfp)
     HASH_DIRECT, NULL, (GtFree) gt_str_delete);
   gff3_visitor->fasta_width = 0;
   gff3_visitor->outfp = outfp;
+  /* if retain_ids is set to true, hen gt_used_ids is .. used. */
+  gff3_visitor->gt_used_ids = gt_cstr_table_new();
+  gff3_visitor->retain_ids = false;
   return gv;
 }
 
+void gt_gff3_visitor_retain_id_attributes(GtNodeVisitor *gv)
+{
+    GtGFF3Visitor *gff3_visitor = gff3_visitor_cast(gv);
+    gff3_visitor->retain_ids = true;
+}
+
 void gt_gff3_visitor_set_fasta_width(GtNodeVisitor *gv,
                                      unsigned long fasta_width)
 {
diff --git a/src/extended/gff3_visitor.h b/src/extended/gff3_visitor.h
index 1939565..530371c 100644
--- a/src/extended/gff3_visitor.h
+++ b/src/extended/gff3_visitor.h
@@ -27,5 +27,6 @@ const GtNodeVisitorClass* gt_gff3_visitor_class(void);
 GtNodeVisitor*            gt_gff3_visitor_new(GtGenFile*);
 void                      gt_gff3_visitor_set_fasta_width(GtNodeVisitor*,
                                                           unsigned long);
+void                      gt_gff3_visitor_retain_id_attributes(GtNodeVisitor *);
 
 #endif
diff --git a/src/tools/gt_gff3.c b/src/tools/gt_gff3.c
index 74a1457..f498678 100644
--- a/src/tools/gt_gff3.c
+++ b/src/tools/gt_gff3.c
@@ -35,6 +35,7 @@
 typedef struct {
   bool sort,
        checkids,
+       retainids,
        mergefeat,
        addintrons,
        verbose,
@@ -91,6 +92,14 @@ static GtOptionParser* gt_gff3_option_parser_new(void *tool_arguments)
                            "parsing", &arguments->tidy, false);
   gt_option_parser_add_option(op, option);
 
+  /* -retainids */
+  option = gt_option_new_bool("retainids",
+                           "when available, use the original IDs provided"
+                           "in the source file\n"
+                           "(memory consumption is O(file_size))",
+                           &arguments->retainids, true);
+  gt_option_parser_add_option(op, option);
+
   /* -checkids */
   option = gt_option_new_bool("checkids",
                            "make sure the ID attributes are unique "
@@ -188,6 +197,7 @@ static int gt_gff3_runner(int argc, const char **argv, int parsed_args,
     gt_gff3_in_stream_show_progress_bar((GtGFF3InStream*) gff3_in_stream);
   if (arguments->checkids)
     gt_gff3_in_stream_check_id_attributes((GtGFF3InStream*) gff3_in_stream);
+
   last_stream = gff3_in_stream;
 
   /* set different type checker if necessary */
@@ -244,6 +254,9 @@ static int gt_gff3_runner(int argc, const char **argv, int parsed_args,
     gt_gff3_out_stream_set_fasta_width(gff3_out_stream, arguments->width);
   }
 
+  if (!had_err && arguments->retainids)
+    gt_gff3_out_stream_retain_id_attributes(gff3_out_stream);
+
   /* pull the features through the stream and free them afterwards */
   if (!had_err) {
     while (!(had_err = gt_node_stream_next(gff3_out_stream, &gn, err)) &&