feat: Count implicit tags

openedx_tagging/core/tagging/api.py

@@ -196,7 +196,7 @@ def get_object_tags(
     return tags
 
 
-def get_object_tag_counts(object_id_pattern: str) -> dict[str, int]:
+def get_object_tag_counts(object_id_pattern: str, count_implicit=False) -> dict[str, int]:
     """
     Given an object ID, a "starts with" glob pattern like
     "course-v1:foo+bar+baz@*", or a list of "comma,separated,IDs", return a
@@ -217,8 +217,17 @@ def get_object_tag_counts(object_id_pattern: str) -> dict[str, int]:
     qs = qs.exclude(taxonomy_id=None)  # The whole taxonomy was deleted
     qs = qs.exclude(taxonomy__enabled=False)  # The whole taxonomy is disabled
     qs = qs.exclude(tag_id=None, taxonomy__allow_free_text=False)  # The taxonomy exists but the tag is deleted
-    qs = qs.values("object_id").annotate(num_tags=models.Count("id")).order_by("object_id")
-    return {row["object_id"]: row["num_tags"] for row in qs}
+    if count_implicit:
+        tags = Tag.annotate_depth(Tag.objects.filter(pk=models.OuterRef("tag_id")))
+        qs = qs.annotate(tag_depth=models.Subquery(tags.values('depth')))
+        qs = qs.values("object_id").annotate(
+            num_tags=models.Count("id"),
+            num_implicit_tags=models.Sum("tag_depth"),
+        ).order_by("object_id")
+        return {row["object_id"]: row["num_tags"] + (row["num_implicit_tags"] or 0) for row in qs}
+    else:
+        qs = qs.values("object_id").annotate(num_tags=models.Count("id")).order_by("object_id")
+        return {row["object_id"]: row["num_tags"] for row in qs}
 
 
 def delete_object_tags(object_id: str):

openedx_tagging/core/tagging/rest_api/v1/views.py

@@ -496,9 +496,11 @@ class ObjectTagCountsView(
     **Retrieve Parameters**
         * object_id_pattern (required): - The Object ID to retrieve ObjectTags for. Can contain '*' at the end
           for wildcard matching, or use ',' to separate multiple object IDs.
+        * count_implicit (optional): If present, implicit parent/grandparent tags will be included in the counts
 
     **Retrieve Example Requests**
         GET api/tagging/v1/object_tag_counts/:object_id_pattern
+        GET api/tagging/v1/object_tag_counts/:object_id_pattern?count_implicit
 
     **Retrieve Query Returns**
         * 200 - Success
@@ -517,8 +519,9 @@ def retrieve(self, request, *args, **kwargs) -> Response:
         """
         # This API does NOT bother doing any permission checks as the # of tags is not considered sensitive information.
         object_id_pattern = self.kwargs["object_id_pattern"]
+        count_implicit = "count_implicit" in request.query_params
         try:
-            return Response(get_object_tag_counts(object_id_pattern))
+            return Response(get_object_tag_counts(object_id_pattern, count_implicit=count_implicit))
         except ValueError as err:
             raise ValidationError(err.args[0]) from err
 

tests/openedx_tagging/core/tagging/test_api.py

@@ -704,6 +704,29 @@ def test_get_object_tag_counts(self) -> None:
         assert tagging_api.get_object_tag_counts(f"{obj1},{obj2}") == {obj1: 1, obj2: 2}
         assert tagging_api.get_object_tag_counts("object_*") == {obj1: 1, obj2: 2}
 
+    def test_get_object_tag_counts_implicit(self) -> None:
+        """
+        Basic test of get_object_tag_counts, including implicit (parent) tags
+
+        Note that:
+            - "DPANN" is "Archaea > DPANN" (2 tags, 1 implicit), and
+            - "Chordata" is "Eukaryota > Animalia > Chordata" (3 tags, 2 implicit)
+        """
+        obj1 = "object_id1"
+        obj2 = "object_id2"
+        other = "other_object"
+        # Give each object 1-2 tags:
+        tagging_api.tag_object(object_id=obj1, taxonomy=self.taxonomy, tags=["DPANN"])
+        tagging_api.tag_object(object_id=obj2, taxonomy=self.taxonomy, tags=["Chordata"])
+        tagging_api.tag_object(object_id=obj2, taxonomy=self.free_text_taxonomy, tags=["has a notochord"])
+        tagging_api.tag_object(object_id=other, taxonomy=self.free_text_taxonomy, tags=["other"])
+
+        assert tagging_api.get_object_tag_counts(obj1, count_implicit=True) == {obj1: 2}
+        assert tagging_api.get_object_tag_counts(obj2, count_implicit=True) == {obj2: 4}
+        assert tagging_api.get_object_tag_counts(f"{obj1},{obj2}", count_implicit=True) == {obj1: 2, obj2: 4}
+        assert tagging_api.get_object_tag_counts("object_*", count_implicit=True) == {obj1: 2, obj2: 4}
+        assert tagging_api.get_object_tag_counts(other, count_implicit=True) == {other: 1}
+
     def test_get_object_tag_counts_deleted_disabled(self) -> None:
         """
         Test that get_object_tag_counts doesn't "count" disabled taxonomies or
@@ -726,6 +749,9 @@ def test_get_object_tag_counts_deleted_disabled(self) -> None:
         self.free_text_taxonomy.enabled = False
         self.free_text_taxonomy.save()
         assert tagging_api.get_object_tag_counts("object_*") == {obj1: 1, obj2: 1}
+        # Also check the result with count_implicit:
+        # "English" has no implicit tags but "Chordata" has two, so we expect these totals:
+        assert tagging_api.get_object_tag_counts("object_*", count_implicit=True) == {obj1: 1, obj2: 3}
 
         # But, by the way, if we re-enable the taxonomy and restore the tag, the counts return:
         self.free_text_taxonomy.enabled = True

tests/openedx_tagging/core/tagging/test_views.py

@@ -1028,10 +1028,14 @@ def test_get_counts(self):
         # Course 7 Unit 2
         api.tag_object(object_id="course07-unit02-problem01", taxonomy=self.free_text_taxonomy, tags=["b"])
         api.tag_object(object_id="course07-unit02-problem02", taxonomy=self.free_text_taxonomy, tags=["c", "d"])
-        api.tag_object(object_id="course07-unit02-problem03", taxonomy=self.free_text_taxonomy, tags=["N", "M", "x"])
-
-        def check(object_id_pattern: str):
-            result = self.client.get(OBJECT_TAG_COUNTS_URL.format(object_id_pattern=object_id_pattern))
+        api.tag_object(object_id="course07-unit02-problem03", taxonomy=self.free_text_taxonomy, tags=["N", "M"])
+        api.tag_object(object_id="course07-unit02-problem03", taxonomy=self.taxonomy, tags=["Mammalia"])
+
+        def check(object_id_pattern: str, count_implicit=False):
+            url = OBJECT_TAG_COUNTS_URL.format(object_id_pattern=object_id_pattern)
+            if count_implicit:
+                url += "?count_implicit"
+            result = self.client.get(url)
             assert result.status_code == status.HTTP_200_OK
             return result.data
 
@@ -1044,6 +1048,20 @@ def check(object_id_pattern: str):
                 "course07-unit01-problem01": 3,
                 "course07-unit01-problem02": 2,
             }
+        with self.assertNumQueries(1):
+            assert check(object_id_pattern="course07-unit02-*") == {
+                "course07-unit02-problem01": 1,
+                "course07-unit02-problem02": 2,
+                "course07-unit02-problem03": 3,
+            }
+        with self.assertNumQueries(1):
+            assert check(object_id_pattern="course07-unit02-*", count_implicit=True) == {
+                "course07-unit02-problem01": 1,
+                "course07-unit02-problem02": 2,
+                # "Mammalia" includes 1 explicit + 3 implicit tags: "Eukaryota > Animalia > Chordata > Mammalia"
+                # so problem03 has 2 free text tags and "4" life on earth tags:
+                "course07-unit02-problem03": 6,
+            }
         with self.assertNumQueries(1):
             assert check(object_id_pattern="course07-unit*") == {
                 "course07-unit01-problem01": 3,