Filter container entities on dc_display_in_catalogue (#509)

* Filter container entities on dc_display_in_catalogue

* Handle edge case where display_name is None

via https://ministryofjustice.sentry.io/issues/5529696082/?project=4507181591101440&referrer=github-open-pr-bot

---------

Co-authored-by: Matt <38562764+LavMatt@users.noreply.github.com>

Author: MatMoore

lib/datahub-client/CHANGELOG.md

@@ -9,15 +9,21 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ## Unreleased
 
+### Changed
+
+- Search results and container entities are now filtered to those that include
+  a special tag, dc:display_in_catalogue. This allows the frontend to display
+  only the entities we think users are directly interested in, while still
+  ingesting intermediate tables (which are still relevant in the context of
+  lineage, governance, data quality etc.)
+- Return lists of objects for `SearchResult.tags` and
+  `SearchResult.tags_to_display` instead of strings.
+
 ### Added
 
 - Return domain metadata for Charts
 - Add `glossary_terms` list to `SearchResult`
 
-### Changed
-
-- Return lists of objects for `SearchResult.tags` and `SearchResult.tags_to_display` instead of strings.
-
 ## Removed
 
 - Removed all remaining references to Data Products

lib/datahub-client/data_platform_catalogue/client/datahub_client.py

@@ -339,7 +339,14 @@ def get_database_details(self, urn: str) -> Database:
                 )
             datasets = []
             if response["entities"]["total"] > 0:
-                datasets: list = response["entities"]["searchResults"]
+                datasets: list = [
+                    entity
+                    for entity in response["entities"]["searchResults"]
+                    if any(
+                        tag.urn == "urn:li:tag:dc_display_in_catalogue"
+                        for tag in parse_tags(entity["entity"])
+                    )
+                ]
 
             return Database(
                 urn=urn,

lib/datahub-client/data_platform_catalogue/client/graphql/getContainerDetails.graphql

@@ -30,7 +30,7 @@ query getContainer($urn: String!) {
     subTypes {
       typeNames
     }
-    entities(input: {start: 0, count: 500}) {
+    entities(input: { start: 0, count: 500 }) {
       total
       searchResults {
         entity {
@@ -171,6 +171,9 @@ fragment datasetDetails on Dataset {
   editableProperties {
     description
   }
+  tags {
+    ...globalTagsFields
+  }
 }
 
 fragment entityContainer on Container {

lib/datahub-client/data_platform_catalogue/client/graphql_helpers.py

@@ -36,7 +36,7 @@ def parse_owner(entity: dict[str, Any]) -> OwnerRef:
             else properties.get("fullName", "")
         )
         owner_details = OwnerRef(
-            display_name=display_name,
+            display_name=display_name or "",
             email=properties.get("email", ""),
             urn=owners[0].get("urn", ""),
         )

lib/datahub-client/tests/client/datahub/test_datahub_client.py

@@ -3,6 +3,7 @@
 from unittest.mock import MagicMock, patch
 
 import pytest
+
 from data_platform_catalogue.client.datahub_client import (
     DataHubCatalogueClient,
     InvalidDomain,
@@ -500,6 +501,95 @@ def test_get_chart_details(self, datahub_client, base_mock_graph):
             external_url="https://data.justice.gov.uk/prisons/public-protection/absconds",
         )
 
+    def test_get_database_details_filters_entities(
+        self, datahub_client, base_mock_graph
+    ):
+        urn = "urn:li:container:foo"
+        datahub_response = {
+            "container": {
+                "urn": "urn:li:container",
+                "type": "CONTAINER",
+                "platform": {"name": "platform"},
+                "parentContainers": {
+                    "count": 0,
+                },
+                "entities": {
+                    "total": 2,
+                    "searchResults": [
+                        {
+                            "entity": {
+                                "name": "DatasetToShow",
+                                "properties": {
+                                    "name": "DatasetToShow",
+                                    "description": "Dataset to show",
+                                },
+                                "tags": {
+                                    "tags": [
+                                        {
+                                            "tag": {
+                                                "urn": "urn:li:tag:dc_display_in_catalogue",
+                                                "properties": {
+                                                    "name": "dc:display_in_catalogue",
+                                                },
+                                            }
+                                        }
+                                    ]
+                                },
+                            }
+                        },
+                        {
+                            "entity": {
+                                "name": "DatasetToHide",
+                                "properties": {
+                                    "name": "DatasetToHide",
+                                    "description": "Dataset to hide",
+                                },
+                                "tags": {"tags": []},
+                            }
+                        },
+                    ],
+                },
+                "ownership": None,
+                "properties": {
+                    "name": "Some database",
+                    "description": "a test description",
+                    "customProperties": [],
+                    "lastModified": {"time": 0},
+                },
+            },
+            "extensions": {},
+        }
+        base_mock_graph.execute_graphql = MagicMock(return_value=datahub_response)
+
+        with patch(
+            "data_platform_catalogue.client.datahub_client.DataHubCatalogueClient.check_entity_exists_by_urn"
+        ) as mock_exists:
+            mock_exists.return_value = True
+            database = datahub_client.get_database_details(urn)
+            assert database.tables == [
+                {
+                    "entity": {
+                        "name": "DatasetToShow",
+                        "properties": {
+                            "description": "Dataset to show",
+                            "name": "DatasetToShow",
+                        },
+                        "tags": {
+                            "tags": [
+                                {
+                                    "tag": {
+                                        "properties": {
+                                            "name": "dc:display_in_catalogue",
+                                        },
+                                        "urn": "urn:li:tag:dc_display_in_catalogue",
+                                    },
+                                },
+                            ],
+                        },
+                    }
+                }
+            ]
+
     def test_upsert_table_and_database(
         self,
         datahub_client,