perf(access): provide bounded reference search

src/govoplan_access/backend/manifest.py

@@ -42,6 +42,7 @@ from govoplan_core.core.modules import (
     RoleTemplate,
 )
 from govoplan_core.core.people import CAPABILITY_ACCESS_PEOPLE_SEARCH
+from govoplan_core.core.references import CAPABILITY_ACCESS_REFERENCE_OPTIONS
 from govoplan_core.core.views import ViewSurface
 
 
@@ -538,6 +539,15 @@ def _access_semantic_directory(context: ModuleContext) -> object:
     )
 
 
+def _access_reference_options(context: ModuleContext) -> object:
+    del context
+    from govoplan_access.backend.reference_options import (
+        SqlAccessReferenceOptionProvider,
+    )
+
+    return SqlAccessReferenceOptionProvider()
+
+
 def _optional_identity_directory(context: ModuleContext) -> IdentityDirectory | None:
     if not context.registry.has_capability(CAPABILITY_IDENTITY_DIRECTORY):
         return None
@@ -648,6 +658,10 @@ manifest = ModuleManifest(
     optional_dependencies=("identity", "organizations", "tenancy", "idm"),
     provides_interfaces=(
         ModuleInterfaceProvider(name=CAPABILITY_ACCESS_PEOPLE_SEARCH, version="0.1.0"),
+        ModuleInterfaceProvider(
+            name=CAPABILITY_ACCESS_REFERENCE_OPTIONS,
+            version="0.1.0",
+        ),
         ModuleInterfaceProvider(
             name="auth.automation_principal",
             version="0.2.0",
@@ -725,6 +739,7 @@ manifest = ModuleManifest(
         CAPABILITY_ACCESS_ADMINISTRATION: _access_administration,
         CAPABILITY_ACCESS_GOVERNANCE_MATERIALIZER: _governance_materializer,
         CAPABILITY_ACCESS_PEOPLE_SEARCH: _people_search,
+        CAPABILITY_ACCESS_REFERENCE_OPTIONS: _access_reference_options,
         ACCESS_CONFIGURATION_CAPABILITY: _configuration_provider,
     },
     documentation=ACCESS_DOCUMENTATION,

src/govoplan_access/backend/reference_options.py

@@ -0,0 +1,262 @@
+from __future__ import annotations
+
+from collections.abc import Sequence
+
+from sqlalchemy import func, or_
+from sqlalchemy.orm import Session
+
+from govoplan_access.backend.db.models import Account, Group, User
+from govoplan_core.core.references import (
+    ReferenceOption,
+    ReferenceSearchPage,
+    ReferenceSearchRequest,
+)
+
+
+class SqlAccessReferenceOptionProvider:
+    """Principal-aware, bounded Access directory search."""
+
+    def search_reference_options(
+        self,
+        session: object,
+        principal: object,
+        *,
+        request: ReferenceSearchRequest,
+    ) -> ReferenceSearchPage:
+        db = _session(session)
+        tenant_id = str(request.tenant_id or "").strip()
+        if not tenant_id:
+            return ReferenceSearchPage()
+        limit = max(1, min(int(request.limit), 200))
+        offset = _cursor_offset(request.cursor)
+        selected = tuple(
+            dict.fromkeys(
+                str(value).strip()
+                for value in request.selected_values
+                if str(value).strip()
+            )
+        )[:200]
+        administrative = request.context.get("administrative") is True
+        query = str(request.query or "").strip().casefold()
+        if request.kind in {"user", "membership"}:
+            return _search_users(
+                db,
+                principal,
+                tenant_id=tenant_id,
+                kind=request.kind,
+                query=query,
+                selected=selected,
+                limit=limit,
+                offset=offset,
+                administrative=administrative,
+            )
+        if request.kind == "group":
+            return _search_groups(
+                db,
+                principal,
+                tenant_id=tenant_id,
+                query=query,
+                selected=selected,
+                limit=limit,
+                offset=offset,
+                administrative=administrative,
+            )
+        raise ValueError(f"Unsupported Access reference kind: {request.kind}")
+
+
+def _search_users(
+    session: Session,
+    principal: object,
+    *,
+    tenant_id: str,
+    kind: str,
+    query: str,
+    selected: Sequence[str],
+    limit: int,
+    offset: int,
+    administrative: bool,
+) -> ReferenceSearchPage:
+    value_column = User.id if kind == "membership" else User.account_id
+    base = (
+        session.query(User, Account)
+        .join(Account, Account.id == User.account_id)
+        .filter(User.tenant_id == tenant_id)
+    )
+    if not administrative:
+        account_id = str(getattr(principal, "account_id", "") or "")
+        if not account_id:
+            return ReferenceSearchPage()
+        base = base.filter(User.account_id == account_id)
+
+    selected_rows = (
+        base.filter(value_column.in_(selected)).all()
+        if selected
+        else []
+    )
+    search_query = base
+    if selected:
+        search_query = search_query.filter(value_column.notin_(selected))
+    if query:
+        search_query = search_query.filter(
+            or_(
+                func.lower(func.coalesce(User.display_name, "")).contains(
+                    query,
+                    autoescape=True,
+                ),
+                func.lower(User.email).contains(query, autoescape=True),
+                func.lower(Account.email).contains(query, autoescape=True),
+                func.lower(value_column).contains(query, autoescape=True),
+            )
+        )
+    rows = (
+        search_query.order_by(
+            func.lower(func.coalesce(User.display_name, User.email)).asc(),
+            value_column.asc(),
+        )
+        .offset(offset)
+        .limit(limit + 1)
+        .all()
+    )
+    has_more = len(rows) > limit
+    options = [
+        _user_option(user, account, kind=kind)
+        for user, account in rows[:limit]
+    ]
+    selected_by_value = {
+        _user_value(user, kind=kind): _user_option(user, account, kind=kind)
+        for user, account in selected_rows
+    }
+    options.extend(
+        selected_by_value[value]
+        for value in selected
+        if value in selected_by_value
+    )
+    return ReferenceSearchPage(
+        options=tuple(options),
+        next_cursor=f"offset:{offset + limit}" if has_more else None,
+        has_more=has_more,
+    )
+
+
+def _search_groups(
+    session: Session,
+    principal: object,
+    *,
+    tenant_id: str,
+    query: str,
+    selected: Sequence[str],
+    limit: int,
+    offset: int,
+    administrative: bool,
+) -> ReferenceSearchPage:
+    base = session.query(Group).filter(Group.tenant_id == tenant_id)
+    if not administrative:
+        permitted = tuple(
+            dict.fromkeys(
+                str(group_id)
+                for group_id in getattr(principal, "group_ids", ())
+                if str(group_id)
+            )
+        )
+        if not permitted:
+            return ReferenceSearchPage()
+        base = base.filter(Group.id.in_(permitted))
+
+    selected_rows = base.filter(Group.id.in_(selected)).all() if selected else []
+    search_query = base
+    if selected:
+        search_query = search_query.filter(Group.id.notin_(selected))
+    if query:
+        search_query = search_query.filter(
+            or_(
+                func.lower(Group.name).contains(query, autoescape=True),
+                func.lower(Group.slug).contains(query, autoescape=True),
+                func.lower(Group.id).contains(query, autoescape=True),
+            )
+        )
+    rows = (
+        search_query.order_by(func.lower(Group.name).asc(), Group.id.asc())
+        .offset(offset)
+        .limit(limit + 1)
+        .all()
+    )
+    has_more = len(rows) > limit
+    options = [_group_option(group) for group in rows[:limit]]
+    selected_by_value = {group.id: _group_option(group) for group in selected_rows}
+    options.extend(
+        selected_by_value[value]
+        for value in selected
+        if value in selected_by_value
+    )
+    return ReferenceSearchPage(
+        options=tuple(options),
+        next_cursor=f"offset:{offset + limit}" if has_more else None,
+        has_more=has_more,
+    )
+
+
+def _user_value(user: User, *, kind: str) -> str:
+    return user.id if kind == "membership" else user.account_id
+
+
+def _user_option(user: User, account: Account, *, kind: str) -> ReferenceOption:
+    inactive = not user.is_active or not account.is_active
+    value = _user_value(user, kind=kind)
+    description_parts = [
+        user.email,
+        "Inactive" if inactive else None,
+    ]
+    return ReferenceOption(
+        value=value,
+        label=user.display_name or user.email or value,
+        description=" · ".join(
+            part for part in description_parts if part
+        ) or None,
+        kind=kind,
+        availability="inactive" if inactive else "available",
+        disabled=inactive,
+        source_module="access",
+        provenance={
+            "tenant_id": user.tenant_id,
+            "membership_id": user.id,
+            "account_id": user.account_id,
+        },
+    )
+
+
+def _group_option(group: Group) -> ReferenceOption:
+    inactive = not group.is_active
+    return ReferenceOption(
+        value=group.id,
+        label=group.name or group.id,
+        description="Inactive" if inactive else None,
+        kind="group",
+        availability="inactive" if inactive else "available",
+        disabled=inactive,
+        source_module="access",
+        provenance={"tenant_id": group.tenant_id, "group_id": group.id},
+    )
+
+
+def _cursor_offset(cursor: str | None) -> int:
+    if cursor is None:
+        return 0
+    prefix = "offset:"
+    if not cursor.startswith(prefix):
+        raise ValueError("Invalid reference search cursor.")
+    try:
+        offset = int(cursor[len(prefix):])
+    except ValueError as exc:
+        raise ValueError("Invalid reference search cursor.") from exc
+    if offset < 0:
+        raise ValueError("Invalid reference search cursor.")
+    return offset
+
+
+def _session(session: object) -> Session:
+    if not isinstance(session, Session):
+        raise TypeError("Access reference search requires a SQLAlchemy Session")
+    return session
+
+
+__all__ = ["SqlAccessReferenceOptionProvider"]

tests/test_reference_options.py

@@ -0,0 +1,153 @@
+from __future__ import annotations
+
+import unittest
+from types import SimpleNamespace
+
+from sqlalchemy import create_engine
+from sqlalchemy.orm import Session
+
+from govoplan_access.backend.db.models import Account, Group, User
+from govoplan_access.backend.reference_options import (
+    SqlAccessReferenceOptionProvider,
+)
+from govoplan_core.core.references import ReferenceSearchRequest
+from govoplan_core.db.base import Base
+
+
+class AccessReferenceOptionProviderTests(unittest.TestCase):
+    def setUp(self) -> None:
+        self.engine = create_engine("sqlite+pysqlite:///:memory:")
+        Base.metadata.create_all(
+            self.engine,
+            tables=[Account.__table__, User.__table__, Group.__table__],
+        )
+        self.session = Session(self.engine)
+        for index in range(120):
+            account = Account(
+                id=f"account-{index:03}",
+                email=f"person-{index:03}@example.test",
+                normalized_email=f"person-{index:03}@example.test",
+            )
+            self.session.add(account)
+            self.session.add(
+                User(
+                    id=f"membership-{index:03}",
+                    tenant_id="tenant-1",
+                    account_id=account.id,
+                    email=account.email,
+                    display_name=f"Person {index:03}",
+                )
+            )
+        for index in range(75):
+            self.session.add(
+                Group(
+                    id=f"group-{index:03}",
+                    tenant_id="tenant-1",
+                    slug=f"group-{index:03}",
+                    name=f"Group {index:03}",
+                )
+            )
+        self.session.commit()
+        self.provider = SqlAccessReferenceOptionProvider()
+        self.admin = SimpleNamespace(
+            tenant_id="tenant-1",
+            account_id="account-000",
+            group_ids=frozenset(),
+        )
+
+    def tearDown(self) -> None:
+        self.session.close()
+        self.engine.dispose()
+
+    def test_large_directory_search_is_bounded_and_paged(self) -> None:
+        first = self.provider.search_reference_options(
+            self.session,
+            self.admin,
+            request=ReferenceSearchRequest(
+                kind="membership",
+                tenant_id="tenant-1",
+                limit=25,
+                context={"administrative": True},
+            ),
+        )
+        second = self.provider.search_reference_options(
+            self.session,
+            self.admin,
+            request=ReferenceSearchRequest(
+                kind="membership",
+                tenant_id="tenant-1",
+                limit=25,
+                cursor=first.next_cursor,
+                context={"administrative": True},
+            ),
+        )
+
+        self.assertEqual(25, len(first.options))
+        self.assertTrue(first.has_more)
+        self.assertEqual("offset:25", first.next_cursor)
+        self.assertEqual("membership-000", first.options[0].value)
+        self.assertEqual("membership-025", second.options[0].value)
+
+    def test_search_and_selected_values_do_not_materialize_the_directory(self) -> None:
+        page = self.provider.search_reference_options(
+            self.session,
+            self.admin,
+            request=ReferenceSearchRequest(
+                kind="membership",
+                tenant_id="tenant-1",
+                query="PERSON 119",
+                selected_values=("membership-005", "removed-membership"),
+                limit=10,
+                context={"administrative": True},
+            ),
+        )
+
+        self.assertEqual(
+            ["membership-119", "membership-005"],
+            [option.value for option in page.options],
+        )
+        self.assertFalse(page.has_more)
+
+    def test_non_administrators_only_search_their_permitted_references(self) -> None:
+        principal = SimpleNamespace(
+            tenant_id="tenant-1",
+            account_id="account-004",
+            group_ids=frozenset({"group-007"}),
+        )
+
+        users = self.provider.search_reference_options(
+            self.session,
+            principal,
+            request=ReferenceSearchRequest(
+                kind="user",
+                tenant_id="tenant-1",
+            ),
+        )
+        groups = self.provider.search_reference_options(
+            self.session,
+            principal,
+            request=ReferenceSearchRequest(
+                kind="group",
+                tenant_id="tenant-1",
+            ),
+        )
+
+        self.assertEqual(["account-004"], [option.value for option in users.options])
+        self.assertEqual(["group-007"], [option.value for option in groups.options])
+
+    def test_invalid_cursor_is_rejected(self) -> None:
+        with self.assertRaisesRegex(ValueError, "Invalid reference search cursor"):
+            self.provider.search_reference_options(
+                self.session,
+                self.admin,
+                request=ReferenceSearchRequest(
+                    kind="group",
+                    tenant_id="tenant-1",
+                    cursor="page:2",
+                    context={"administrative": True},
+                ),
+            )
+
+
+if __name__ == "__main__":
+    unittest.main()