Add NodeGroup.grouping_node #11613

arches/app/const.py

@@ -3,7 +3,8 @@
 
 IntegrityCheckDescriptions = {
     1005: "Nodes with ontologies found in graphs without ontologies",
-    1012: "Node Groups without matching nodes",
+    1012: "Node Groups without contained nodes",
+    1013: "Node Groups without a grouping node",
     1014: "Publication missing for language",
 }
 
@@ -12,6 +13,7 @@
 class IntegrityCheck(Enum):
     NODE_HAS_ONTOLOGY_GRAPH_DOES_NOT = 1005
     NODELESS_NODE_GROUP = 1012
+    NODEGROUP_WITHOUT_GROUPING_NODE = 1013
     PUBLICATION_MISSING_FOR_LANGUAGE = 1014
 
     def __str__(self):

arches/app/models/graph.py

@@ -1111,7 +1111,9 @@ def flatten_tree(tree, node_id_list=[]):
             if is_collector:
                 old_nodegroup_id = node.nodegroup_id
                 node.nodegroup = models.NodeGroup(
-                    pk=node.pk, cardinality=node.nodegroup.cardinality
+                    pk=node.pk,
+                    cardinality=node.nodegroup.cardinality,
+                    grouping_node=node,
                 )
                 if old_nodegroup_id not in nodegroup_map:
                     nodegroup_map[old_nodegroup_id] = node.nodegroup_id
@@ -1284,7 +1286,8 @@ def update_node(self, node):
             new_node.fieldname = node["fieldname"]
         self.populate_null_nodegroups()
 
-        # new_node will always have a nodegroup id even it if was set to None becuase populate_null_nodegroups
+        # new_node will always have a nodegroup id even if if was set to None
+        # because populate_null_nodegroups
         # will populate the nodegroup id with the parent nodegroup
         # add/remove a card if a nodegroup was added/removed
         if new_node.nodegroup_id != old_node.nodegroup_id:
@@ -1317,7 +1320,7 @@ def update_node(self, node):
 
     def delete_node(self, node=None):
         """
-        deletes a node and all if it's children from a graph
+        deletes a node and all of its children from a graph
 
         Arguments:
         node -- a node id or Node model to delete from the graph
@@ -1830,20 +1833,28 @@ def get_or_create_nodegroup(self, nodegroupid, nodegroups_list=[]):
         """
         get a nodegroup from an id by first looking through the nodes and cards associated with this graph.
         if not found then get the nodegroup instance from the database, otherwise return a new instance of a nodegroup
+        This is also the method responsible for setting the grouping_node_id
+        on the nodegroup when the first (collector) node is created.
 
         Keyword Arguments
 
         nodegroupid -- return a nodegroup with this id
         nodegroups_list -- list of nodegroups from which to filter
         """
 
+        found = None
         for nodegroup in nodegroups_list or self.get_nodegroups():
             if str(nodegroup.nodegroupid) == str(nodegroupid):
-                return nodegroup
-        try:
-            return models.NodeGroup.objects.get(pk=nodegroupid)
-        except models.NodeGroup.DoesNotExist:
-            return models.NodeGroup(pk=nodegroupid)
+                found = nodegroup
+                break
+        else:
+            try:
+                found = models.NodeGroup.objects.get(pk=nodegroupid)
+            except models.NodeGroup.DoesNotExist:
+                found = models.NodeGroup(pk=nodegroupid, grouping_node_id=nodegroupid)
+
+        found.grouping_node_id = found.nodegroupid
+        return found
 
     def get_root_nodegroup(self):
         """
@@ -2436,6 +2447,7 @@ def _update_source_nodegroup_hierarchy(nodegroup):
 
                 source_nodegroup.cardinality = nodegroup.cardinality
                 source_nodegroup.legacygroupid = nodegroup.legacygroupid
+                source_nodegroup.grouping_node_id = source_nodegroup.pk
 
                 if nodegroup.parentnodegroup_id:
                     nodegroup_parent_node = models.Node.objects.get(
@@ -2480,7 +2492,7 @@ def _update_source_nodegroup_hierarchy(nodegroup):
             # graph ( the graph mapped to `self` ); we iterate over the item attributes and map
             # them to source item. If the item does not have a `source_identifier` attribute, it
             # has been newly created; we update the `graph_id` to match the source graph. We are
-            # not saving in this block so updates can accur in any order.
+            # not saving in this block so updates can occur in any order.
             for future_widget in list(editable_future_graph.widgets.values()):
                 source_widget = future_widget.source_identifier
 

arches/app/models/migrations/11613_nodegroup_grouping_node.py

@@ -0,0 +1,69 @@
+# Generated by Django 5.1.3 on 2024-11-19 09:29
+
+import django.db.models.deletion
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("models", "11408_loadstaging_sortorder"),
+    ]
+
+    def add_grouping_node(apps, schema_editor):
+        NodeGroup = apps.get_model("models", "NodeGroup")
+        nodegroups_with_nodes = NodeGroup.objects.filter(node__gt=0).distinct()
+        for nodegroup in nodegroups_with_nodes:
+            nodegroup.grouping_node_id = nodegroup.pk
+        NodeGroup.objects.bulk_update(nodegroups_with_nodes, ["grouping_node_id"])
+
+        PublishedGraph = apps.get_model("models", "PublishedGraph")
+        published_graphs = PublishedGraph.objects.all()
+        for published_graph in published_graphs:
+            for nodegroup_dict in published_graph.serialized_graph["nodegroups"]:
+                nodegroup_dict["grouping_node_id"] = nodegroup_dict["nodegroupid"]
+        PublishedGraph.objects.bulk_update(published_graphs, ["serialized_graph"])
+
+    def remove_grouping_node(apps, schema_editor):
+        PublishedGraph = apps.get_model("models", "PublishedGraph")
+        published_graphs = PublishedGraph.objects.all()
+        for published_graph in published_graphs:
+            for nodegroup_dict in published_graph.serialized_graph["nodegroups"]:
+                nodegroup_dict.pop("grouping_node_id", None)
+        PublishedGraph.objects.bulk_update(published_graphs, ["serialized_graph"])
+
+    operations = [
+        migrations.AddField(
+            model_name="nodegroup",
+            name="grouping_node",
+            field=models.OneToOneField(
+                blank=True,
+                db_column="groupingnodeid",
+                null=True,
+                on_delete=django.db.models.deletion.SET_NULL,
+                related_name="grouping_node_nodegroup",
+                to="models.node",
+            ),
+        ),
+        migrations.AlterField(
+            model_name="nodegroup",
+            name="cardinality",
+            field=models.CharField(
+                blank=True, choices=[("1", "1"), ("n", "n")], default="1", max_length=1
+            ),
+        ),
+        migrations.AlterField(
+            model_name="nodegroup",
+            name="parentnodegroup",
+            field=models.ForeignKey(
+                blank=True,
+                db_column="parentnodegroupid",
+                null=True,
+                on_delete=django.db.models.deletion.CASCADE,
+                related_name="children",
+                related_query_name="child",
+                to="models.nodegroup",
+            ),
+        ),
+        migrations.RunPython(add_grouping_node, remove_grouping_node),
+    ]

arches/app/models/migrations/11613_nodegroup_grouping_node_constraints.py

@@ -0,0 +1,33 @@
+# Generated by Django 5.1.3 on 2024-11-19 09:33
+
+from django.db import migrations, models
+
+
+class Migration(migrations.Migration):
+
+    dependencies = [
+        ("models", "11613_nodegroup_grouping_node"),
+    ]
+
+    operations = [
+        migrations.AddConstraint(
+            model_name="node",
+            constraint=models.CheckConstraint(
+                condition=models.Q(
+                    ("istopnode", True), ("nodegroup__isnull", False), _connector="OR"
+                ),
+                name="has_nodegroup_or_istopnode",
+            ),
+        ),
+        migrations.AddConstraint(
+            model_name="nodegroup",
+            constraint=models.CheckConstraint(
+                condition=models.Q(
+                    ("grouping_node", models.F("pk")),
+                    ("grouping_node__isnull", True),
+                    _connector="OR",
+                ),
+                name="grouping_node_matches_pk_or_null",
+            ),
+        ),
+    ]

arches/app/models/migrations/7787_relational_data_model.py

@@ -9,7 +9,7 @@ class Migration(migrations.Migration):
 
     operations = [
         migrations.RunSQL(
-            """
+            r"""
             create extension if not exists "unaccent";
 
             create or replace function __arches_slugify(

arches/app/models/models.py

@@ -709,14 +709,27 @@ class Meta:
 class NodeGroup(models.Model):
     nodegroupid = models.UUIDField(primary_key=True)
     legacygroupid = models.TextField(blank=True, null=True)
-    cardinality = models.TextField(blank=True, default="1")
+    cardinality = models.CharField(
+        max_length=1, blank=True, default="1", choices={"1": "1", "n": "n"}
+    )
     parentnodegroup = models.ForeignKey(
         "self",
         db_column="parentnodegroupid",
         blank=True,
         null=True,
         on_delete=models.CASCADE,
+        related_name="children",
+        related_query_name="child",
     )  # Allows nodegroups within nodegroups
+    grouping_node = models.OneToOneField(
+        "Node",
+        db_column="groupingnodeid",
+        blank=True,
+        null=True,
+        # models.RESTRICT might be better, but revisit after future graph refactor.
+        on_delete=models.SET_NULL,
+        related_name="grouping_node_nodegroup",
+    )
 
     def __init__(self, *args, **kwargs):
         super(NodeGroup, self).__init__(*args, **kwargs)
@@ -726,6 +739,13 @@ def __init__(self, *args, **kwargs):
     class Meta:
         managed = True
         db_table = "node_groups"
+        constraints = [
+            models.CheckConstraint(
+                condition=Q(grouping_node=models.F("pk"))
+                | Q(grouping_node__isnull=True),
+                name="grouping_node_matches_pk_or_null",
+            )
+        ]
 
         default_permissions = ()
         permissions = (
@@ -874,15 +894,18 @@ def __init__(self, *args, **kwargs):
     def clean(self):
         if not self.alias:
             Graph.objects.get(pk=self.graph_id).create_node_alias(self)
+        if self.pk == self.source_identifier_id:
+            self.source_identifier_id = None
 
     def save(self, **kwargs):
         if not self.alias:
-            self.clean()
             add_to_update_fields(kwargs, "alias")
             add_to_update_fields(kwargs, "hascustomalias")
         if self.pk == self.source_identifier_id:
-            self.source_identifier_id = None
             add_to_update_fields(kwargs, "source_identifier_id")
+
+        self.clean()
+
         super(Node, self).save()
 
     class Meta:
@@ -895,6 +918,10 @@ class Meta:
             models.UniqueConstraint(
                 fields=["alias", "graph"], name="unique_alias_graph"
             ),
+            models.CheckConstraint(
+                condition=Q(istopnode=True) | Q(nodegroup__isnull=False),
+                name="has_nodegroup_or_istopnode",
+            ),
         ]
 
 

arches/management/commands/validate.py

@@ -120,6 +120,11 @@ def handle(self, *args, **options):
             ),
             fix_action=DELETE_QUERYSET,
         )
+        self.check_integrity(
+            check=IntegrityCheck.NODEGROUP_WITHOUT_GROUPING_NODE,  # 1013
+            queryset=models.NodeGroup.objects.filter(node__gt=0, grouping_node=None),
+            fix_action=None,
+        )
         self.check_integrity(
             check=IntegrityCheck.PUBLICATION_MISSING_FOR_LANGUAGE,  # 1014
             queryset=(

releases/8.0.0.md

@@ -20,6 +20,7 @@ Arches 8.0.0 Release Notes
 ### Additional highlights
 - Add session-based REST APIs for login, logout [#11261](https://github.com/archesproject/arches/issues/11261)
 - Improve handling of longer model names [#11317](https://github.com/archesproject/arches/issues/11317)
+- New column `NodeGroup.grouping_node`: one-to-one field to the grouping node [#11613](https://github.com/archesproject/arches/issues/11613)
 - Support more expressive plugin URLs [#11320](https://github.com/archesproject/arches/issues/11320)
 - Make node aliases not nullable [#10437](https://github.com/archesproject/arches/issues/10437)
 - Concepts API no longer responds with empty body for error conditions [#11519](https://github.com/archesproject/arches/issues/11519)