validation CHANGE checking list/leaflist duplicities
Optimize implementation and instead of comparing nodes to each other,
use hashtable to detect duplicities. To minimize memory consumption,
the hash tables are not stored permanently, but build when validating
the data tree.
diff --git a/src/parser_json.c b/src/parser_json.c
index 3d5879b..994fcda 100644
--- a/src/parser_json.c
+++ b/src/parser_json.c
@@ -1173,7 +1173,12 @@
}
/* validation successful */
- result->validity = LYD_VAL_OK;
+ if (result->schema->nodetype & (LYS_LIST | LYS_LEAFLIST)) {
+ /* postpone checking when there will be all list/leaflist instances */
+ result->validity = LYD_VAL_UNIQUE;
+ } else {
+ result->validity = LYD_VAL_OK;
+ }
if (!(*parent)) {
*parent = result;
@@ -1211,7 +1216,9 @@
struct lyd_node *result = NULL, *next = NULL, *iter = NULL;
struct unres_data *unres = NULL;
unsigned int len = 0, r;
+ int i;
struct attr_cont *attrs = NULL;
+ struct ly_set *set;
ly_errno = LY_SUCCESS;
@@ -1278,6 +1285,28 @@
goto error;
}
+ /* check for uniquness of top-level lists/leaflists because
+ * only the inner instances were tested in lyv_data_content() */
+ set = ly_set_new();
+ LY_TREE_FOR(result, iter) {
+ if (!(iter->schema->nodetype & (LYS_LIST | LYS_LEAFLIST)) || !(iter->validity & LYD_VAL_UNIQUE)) {
+ continue;
+ }
+
+ /* check each list/leaflist only once */
+ i = set->number;
+ if (ly_set_add(set, iter->schema, 0) != i) {
+ /* already checked */
+ continue;
+ }
+
+ if (lyv_data_unique(iter, result)) {
+ ly_set_free(set);
+ goto error;
+ }
+ }
+ ly_set_free(set);
+
/* check for missing top level mandatory nodes */
if (!(options & LYD_OPT_TRUSTED) && lyd_check_topmandatory(result, ctx, options)) {
goto error;