Fix namespace-is to search in namespace and its children (#1089)

Author: eikek

components/renku_data_services/namespace/models.py

@@ -20,6 +20,11 @@ class UnsavedGroup:
     name: str
     description: str | None = None
 
+    @property
+    def path(self) -> NamespacePath:
+        """Return the path of this group."""
+        return NamespacePath.from_strings(self.slug)
+
 
 @dataclass(kw_only=True)
 class Group(UnsavedGroup):

components/renku_data_services/search/solr_token.py

@@ -48,6 +48,11 @@ def from_str(input: str) -> SolrToken:
     return SolrToken(__escape_query(input))
 
 
+def prefix_search(input: str) -> SolrToken:
+    """Create a solr query part matching input prefixes."""
+    return SolrToken(f"{__escape_query(input)}*")
+
+
 def from_visibility(v: Visibility) -> SolrToken:
     """Create a solr query value for a visibility."""
     return SolrToken(v.value.lower())
@@ -103,6 +108,13 @@ def field_is_any(field: FieldName, value: Nel[SolrToken]) -> SolrToken:
         return field_is(field, SolrToken(f"({vs})"))
 
 
+def namespace_path_is_any(value: Nel[str]) -> SolrToken:
+    """Search for entities in the given namespaces or below."""
+    path_or_below = [x for k in [[from_str(e), prefix_search(f"{e}/")] for e in value] for x in k]
+    vs = fold_or(path_or_below)
+    return field_is(Fields.namespace_path, SolrToken(f"({vs})"))
+
+
 def type_is(et: EntityType) -> SolrToken:
     """Search for the type field."""
     return field_is(Fields.entity_type, from_entity_type(et))

components/renku_data_services/search/solr_user_query.py

@@ -354,7 +354,7 @@ async def visit_keyword_is(self, ft: KeywordIs) -> None:
 
     async def visit_namespace_is(self, ft: NamespaceIs) -> None:
         """Process the namespace-is segment."""
-        self.__append(st.field_is_any(Fields.namespace_path, ft.values.map(st.from_str)))
+        self.__append(st.namespace_path_is_any(ft.values))
 
     async def visit_created_by_is(self, ft: CreatedByIs) -> None:
         """Process the created-by segment."""

test/components/renku_data_services/search/test_db.py

@@ -1,15 +1,24 @@
 """Tests for the repository."""
 
+import uuid
 from datetime import datetime
 
 import pytest
 from ulid import ULID
 
 from renku_data_services.authz.models import Visibility
-from renku_data_services.base_models.core import NamespacePath, NamespaceSlug, ProjectPath, ProjectSlug
-from renku_data_services.data_connectors.models import CloudStorageCore, DataConnector
+from renku_data_services.base_models.core import (
+    AuthenticatedAPIUser,
+    NamespacePath,
+    NamespaceSlug,
+    ProjectPath,
+    ProjectSlug,
+)
+from renku_data_services.data_api.dependencies import DependencyManager
+from renku_data_services.data_connectors.models import CloudStorageCore, DataConnector, UnsavedDataConnector
 from renku_data_services.migrations.core import run_migrations_for_app
-from renku_data_services.namespace.models import ProjectNamespace, UserNamespace
+from renku_data_services.namespace.models import ProjectNamespace, UnsavedGroup, UserNamespace
+from renku_data_services.project.models import UnsavedProject
 from renku_data_services.search.db import SearchUpdatesRepo
 from renku_data_services.search.models import DeleteDoc
 from renku_data_services.solr.entity_documents import DataConnector as DataConnectorDoc
@@ -30,6 +39,71 @@
 )
 
 
+@pytest.mark.asyncio
+async def test_dc_in_group_project(app_manager_instance: DependencyManager) -> None:
+    run_migrations_for_app("common")
+
+    user_repo = app_manager_instance.kc_user_repo
+    group_repo = app_manager_instance.group_repo
+    proj_repo = app_manager_instance.project_repo
+    dc_repo = app_manager_instance.data_connector_repo
+    repo = SearchUpdatesRepo(app_manager_instance.config.db.async_session_maker)
+
+    user = AuthenticatedAPIUser(id=str(uuid.uuid4()), access_token="abc", first_name="Huhu")
+    u = await user_repo.get_or_create_user(user, user.id)
+    assert u
+
+    group = await group_repo.insert_group(user, UnsavedGroup(slug="grr1", name="Group Grr"))
+
+    proj1 = await proj_repo.insert_project(
+        user,
+        UnsavedProject(
+            namespace=group.slug,
+            name=f"proj of group {group.name}",
+            slug="proj-group-1",
+            visibility=Visibility.PUBLIC,
+            created_by=user.id,
+        ),
+    )
+    dc_in_proj = await dc_repo.insert_namespaced_data_connector(
+        user,
+        UnsavedDataConnector(
+            name="dc in group project",
+            slug="dc2",
+            visibility=Visibility.PUBLIC,
+            created_by=user.id,
+            namespace=proj1.path,
+            storage=CloudStorageCore(
+                storage_type="csc", configuration={}, source_path="", target_path="", readonly=True
+            ),
+        ),
+    )
+    dc_in_group = await dc_repo.insert_namespaced_data_connector(
+        user,
+        UnsavedDataConnector(
+            name="dc in group",
+            slug="dc1",
+            visibility=Visibility.PUBLIC,
+            created_by=user.id,
+            namespace=group.path,
+            storage=CloudStorageCore(
+                storage_type="csc", configuration={}, source_path="", target_path="", readonly=True
+            ),
+        ),
+    )
+
+    updates = await repo.select_next(10)
+    assert len(updates) == 5
+    e1 = next(e for e in updates if e.entity_type == "User")
+    assert e1.entity_id == user.id
+    e2 = next(e for e in updates if e.entity_type == "Project")
+    assert e2.entity_id == proj1.id
+    e3 = next(e for e in updates if e.entity_type == "Group")
+    assert e3.entity_id == group.id
+    e45 = {e.entity_id for e in updates if e.entity_type == "DataConnector"}
+    assert e45 == {str(dc_in_proj.id), str(dc_in_group.id)}
+
+
 @pytest.mark.asyncio
 async def test_data_connector_within_project(app_manager_instance):
     run_migrations_for_app("common")

test/components/renku_data_services/search/test_solr_user_query.py

@@ -88,7 +88,7 @@ async def test_from_term() -> None:
         Fields.keywords, Nel.of(st.from_str("k1"), st.from_str("w2"))
     )
     assert await to_solr(ctx, S.namespace_is("ns12")) == st.field_is_any(
-        Fields.namespace_path, Nel.of(st.from_str("ns12"))
+        Fields.namespace_path, Nel.of(st.from_str("ns12"), st.prefix_search("ns12/"))
     )
     assert await to_solr(ctx, S.created_by_is("12-34")) == st.field_is_any(
         Fields.created_by, Nel.of(st.from_str("12-34"))