Skip to content

armgit / dify

Public Imported from GitHub
0

Build Agentic workflows, RAG pipelines, with rich AI model and tool support on one collaborative workspace. Deploy on cloud, VPC, or self-hosted, so teams move from prototype to production without rebuilding the stack.

fix: fix list dataset not include own (#44795)

wangxiaoleia394d0444391
diff --git a/api/services/knowledge/dataset_service.py b/api/services/knowledge/dataset_service.py
index e5f48fecb28..cae3550ca15 100644
--- a/api/services/knowledge/dataset_service.py
+++ b/api/services/knowledge/dataset_service.py
@@ -151,7 +151,6 @@ class DatasetService:
tag_ids=None,
include_all=False,
accessible_dataset_ids: list[str] | None = None,
- include_own_datasets: bool = False,
creator_ids: list[str] | None = None,
*,
tags: TagTargetQuery,
@@ -161,7 +160,7 @@ class DatasetService:
if dify_config.RBAC_ENABLED and accessible_dataset_ids is not None:
accessible_filter: ColumnElement[bool] = Dataset.id.in_(accessible_dataset_ids)
- if include_own_datasets and user:
+ if user:
accessible_filter = sa.or_(Dataset.maintainer == user.id, accessible_filter)
query = query.where(accessible_filter)
@@ -276,7 +275,6 @@ class DatasetService:
tenant_id: str,
user=None,
accessible_dataset_ids: list[str] | None = None,
- include_own_datasets: bool = False,
*,
session: Session,
):
@@ -291,7 +289,7 @@ class DatasetService:
dataset_id for dataset_id in accessible_dataset_ids if dataset_id in requested_dataset_ids
]
accessible_filter: ColumnElement[bool] = Dataset.id.in_(accessible_dataset_ids)
- if include_own_datasets and user:
+ if user:
accessible_filter = sa.or_(Dataset.maintainer == user.id, accessible_filter)
stmt = stmt.where(accessible_filter)
diff --git a/api/services/knowledge/datasets/adapters.py b/api/services/knowledge/datasets/adapters.py
index 7248eedecb1..cbe8b542882 100644
--- a/api/services/knowledge/datasets/adapters.py
+++ b/api/services/knowledge/datasets/adapters.py
@@ -119,7 +119,7 @@ class SQLAlchemyDatasetOperations:
return {f"{model.model}:{model.provider.provider}" for model in models}
def list_datasets(
- self, context: RequestContext, query: DatasetListFilter, accessible_ids: list[str] | None, include_own: bool
+ self, context: RequestContext, query: DatasetListFilter, accessible_ids: list[str] | None
) -> DatasetPage:
page, limit = clamp_pagination(query.page, query.limit, 100)
with self._sessions() as session:
@@ -130,7 +130,6 @@ class SQLAlchemyDatasetOperations:
context.active_workspace_id,
user=account,
accessible_dataset_ids=accessible_ids,
- include_own_datasets=include_own,
session=session,
)
else:
@@ -146,7 +145,6 @@ class SQLAlchemyDatasetOperations:
creator_ids=list(query.creator_ids),
tags=self._tags,
accessible_dataset_ids=accessible_ids,
- include_own_datasets=include_own,
)
data = load_dataset_details(datasets, session=session)
partial_ids = [item["id"] for item in data if item["permission"] == "partial_members"]
diff --git a/api/services/knowledge/datasets/application.py b/api/services/knowledge/datasets/application.py
index 68db02c2f90..8d909c23969 100644
--- a/api/services/knowledge/datasets/application.py
+++ b/api/services/knowledge/datasets/application.py
@@ -31,13 +31,13 @@ class DatasetVisibility:
unrestricted: bool = True
whitelist_ids: Sequence[str] = ()
- def list_scope(self, *, rbac_enabled: bool) -> tuple[list[str] | None, bool]:
+ def list_scope(self, *, rbac_enabled: bool) -> list[str] | None:
if not rbac_enabled:
- return None, False
- # A restricted whitelist always takes precedence over default grants,
- # resource overrides and the creator's own-dataset fallback.
+ return None
+ # Maintainers always retain visibility of their own datasets. The
+ # whitelist takes precedence over grants only for other datasets.
if not self.unrestricted:
- return sorted(set(self.whitelist_ids)), False
+ return sorted(set(self.whitelist_ids))
read_keys = {"dataset.preview", "dataset.acl.preview", "dataset.full_access"}
default_read = bool(
read_keys.intersection(self.default_permissions) or read_keys.intersection(self.workspace_permissions)
@@ -47,7 +47,7 @@ class DatasetVisibility:
if default_read
else sorted(key for key, grants in self.overrides.items() if read_keys.intersection(grants))
)
- return ids, "dataset.create_and_management" in self.workspace_permissions
+ return ids
class DatasetOperations(Protocol):
@@ -55,7 +55,7 @@ class DatasetOperations(Protocol):
def visibility(self, context: RequestContext) -> DatasetVisibility: ...
def list_datasets(
- self, context: RequestContext, query: DatasetListFilter, accessible_ids: list[str] | None, include_own: bool
+ self, context: RequestContext, query: DatasetListFilter, accessible_ids: list[str] | None
) -> DatasetPage: ...
def embedding_models(self, workspace_id: str) -> set[str]: ...
def get_dataset(self, context: RequestContext, ref: DatasetRef) -> DatasetDetailRecord: ...
@@ -107,8 +107,8 @@ class DatasetApplicationService:
def list_datasets(self, context: RequestContext, query: DatasetListFilter) -> DatasetPage:
visibility = self._operations.visibility(context)
- ids, include_own = visibility.list_scope(rbac_enabled=self._rbac_enabled)
- result = self._operations.list_datasets(context, query, ids, include_own)
+ ids = visibility.list_scope(rbac_enabled=self._rbac_enabled)
+ result = self._operations.list_datasets(context, query, ids)
models = self._operations.embedding_models(context.active_workspace_id)
for item in result["data"]:
item["permission_keys"] = list(visibility.overrides.get(item["id"], visibility.default_permissions))
diff --git a/api/services/knowledge/datasets/upgrades.py b/api/services/knowledge/datasets/upgrades.py
index 5a04354230d..e8cdafaabe4 100644
--- a/api/services/knowledge/datasets/upgrades.py
+++ b/api/services/knowledge/datasets/upgrades.py
@@ -57,14 +57,13 @@ class KnowledgeUpgradeApplicationService:
dataset_ids = list(dict.fromkeys(job.old_dataset_id for job in jobs))
if not dataset_ids:
return []
- accessible_ids, include_own = self._datasets.visibility(context).list_scope(rbac_enabled=self._rbac_enabled)
+ accessible_ids = self._datasets.visibility(context).list_scope(rbac_enabled=self._rbac_enabled)
with self._sessions() as session:
datasets, _ = DatasetService.get_datasets_by_ids(
dataset_ids,
workspace_id,
user=load_actor(session, context),
accessible_dataset_ids=accessible_ids,
- include_own_datasets=include_own,
session=session,
)
visible_ids = [str(dataset.id) for dataset in datasets]
diff --git a/api/tests/unit_tests/services/knowledge/datasets/test_adapters.py b/api/tests/unit_tests/services/knowledge/datasets/test_adapters.py
index 5af19785fad..1224323a466 100644
--- a/api/tests/unit_tests/services/knowledge/datasets/test_adapters.py
+++ b/api/tests/unit_tests/services/knowledge/datasets/test_adapters.py
@@ -125,9 +125,7 @@ def test_listing_materializes_page_and_owner_scoped_partial_members(
DatasetPermission(tenant_id="other", dataset_id="dataset", account_id="foreign"),
]
)
- result = operations.list_datasets(
- CONTEXT, DatasetListFilter(page=0, limit=1000, ids=["dataset", "foreign"]), None, False
- )
+ result = operations.list_datasets(CONTEXT, DatasetListFilter(page=0, limit=1000, ids=["dataset", "foreign"]), None)
response = DatasetListResponse.model_validate(result).model_dump(mode="json")
assert response["total"] == 1
assert response["page"] == 1
@@ -159,7 +157,7 @@ def test_listing_upgrade_discovery_uses_server_availability(
KNOWLEDGE_FS_COMMUNITY_DEV_ENABLED=community_dev_enabled,
)
- result = operations.list_datasets(CONTEXT, DatasetListFilter(ids=["dataset"]), None, False)
+ result = operations.list_datasets(CONTEXT, DatasetListFilter(ids=["dataset"]), None)
discovery = result["data"][0]["knowledge_fs_upgrade"]
assert discovery["can_upgrade"] is can_upgrade
@@ -182,20 +180,29 @@ def test_listing_filters_tags_within_workspace(
TagBindingInput(tag_ids=(tag.id,), target_id="foreign" if foreign_tag else "dataset", type="knowledge"),
)
- result = operations.list_datasets(CONTEXT, DatasetListFilter(tag_ids=[tag.id]), None, False)
+ result = operations.list_datasets(CONTEXT, DatasetListFilter(tag_ids=[tag.id]), None)
response = DatasetListResponse.model_validate(result)
assert [row.id for row in response.data] == ([] if foreign_tag else ["dataset"])
assert response.total == (0 if foreign_tag else 1)
[email protected](("ids", "own"), [([], False), (["dataset"], False)])
[email protected]("ids", [[], ["dataset"]])
[email protected]("maintainer", ["actor", "other"])
def test_list_visibility_restricts_even_requested_ids(
- operations: SQLAlchemyDatasetOperations, ids: list[str], own: bool, monkeypatch: pytest.MonkeyPatch
+ operations: SQLAlchemyDatasetOperations,
+ ids: list[str],
+ maintainer: str,
+ monkeypatch: pytest.MonkeyPatch,
+ sqlite_session_factory: sessionmaker[Session],
) -> None:
apply_config_overrides(monkeypatch, RBAC_ENABLED=True)
- result = operations.list_datasets(CONTEXT, DatasetListFilter(ids=["dataset"]), ids, own)
- assert result["total"] == len(ids)
+ with sqlite_session_factory.begin() as session:
+ row = session.get(Dataset, "dataset")
+ assert row is not None
+ row.maintainer = maintainer
+ result = operations.list_datasets(CONTEXT, DatasetListFilter(ids=["dataset"]), ids)
+ assert result["total"] == (1 if maintainer == "actor" else len(ids))
@pytest.mark.parametrize(
diff --git a/api/tests/unit_tests/services/knowledge/datasets/test_application.py b/api/tests/unit_tests/services/knowledge/datasets/test_application.py
index 9041397de01..cf902e26f71 100644
--- a/api/tests/unit_tests/services/knowledge/datasets/test_application.py
+++ b/api/tests/unit_tests/services/knowledge/datasets/test_application.py
@@ -48,26 +48,26 @@ def service(operations: MagicMock, access: MagicMock) -> DatasetApplicationServi
([], [], {"hidden": ["dataset.preview"]}),
],
)
-def test_restricted_whitelist_wins_over_all_other_grants(
+def test_restricted_whitelist_takes_precedence_over_grants(
defaults: list[str], workspace: list[str], overrides: dict[str, list[str]]
) -> None:
visibility = DatasetVisibility(defaults, workspace, overrides, False, ["allowed"])
- assert visibility.list_scope(rbac_enabled=True) == (["allowed"], False)
+ assert visibility.list_scope(rbac_enabled=True) == ["allowed"]
@pytest.mark.parametrize(
("visibility", "expected"),
[
- (DatasetVisibility(default_permissions=["dataset.preview"]), (None, False)),
- (DatasetVisibility(workspace_permissions=["dataset.acl.preview"]), (None, False)),
- (DatasetVisibility(overrides={"a": ["dataset.preview"], "b": ["dataset.edit"]}), (["a"], False)),
- (DatasetVisibility(workspace_permissions=["dataset.create_and_management"]), ([], True)),
- (DatasetVisibility(unrestricted=False), ([], False)),
+ (DatasetVisibility(default_permissions=["dataset.preview"]), None),
+ (DatasetVisibility(workspace_permissions=["dataset.acl.preview"]), None),
+ (DatasetVisibility(overrides={"a": ["dataset.preview"], "b": ["dataset.edit"]}), ["a"]),
+ (DatasetVisibility(workspace_permissions=["dataset.create_and_management"]), []),
+ (DatasetVisibility(unrestricted=False), []),
],
)
-def test_visibility_scope(visibility: DatasetVisibility, expected: tuple[list[str] | None, bool]) -> None:
+def test_visibility_scope(visibility: DatasetVisibility, expected: list[str] | None) -> None:
assert visibility.list_scope(rbac_enabled=True) == expected
- assert visibility.list_scope(rbac_enabled=False) == (None, False)
+ assert visibility.list_scope(rbac_enabled=False) is None
def test_list_applies_visibility_and_reports_model_availability(
@@ -92,7 +92,7 @@ def test_list_applies_visibility_and_reports_model_availability(
}
operations.embedding_models.return_value = set()
result = service.list_datasets(CONTEXT, DatasetListFilter(ids=["a"]))
- operations.list_datasets.assert_called_once_with(CONTEXT, DatasetListFilter(ids=["a"]), ["a"], False)
+ operations.list_datasets.assert_called_once_with(CONTEXT, DatasetListFilter(ids=["a"]), ["a"])
assert result["data"][0]["embedding_available"] is False
assert result["data"][0]["permission_keys"] == ["dataset.edit"]
assert result["data"][1]["embedding_available"] is True
diff --git a/api/tests/unit_tests/services/knowledge/test_dataset_service.py b/api/tests/unit_tests/services/knowledge/test_dataset_service.py
index 1dd2b98d504..f4bd9699b25 100644
--- a/api/tests/unit_tests/services/knowledge/test_dataset_service.py
+++ b/api/tests/unit_tests/services/knowledge/test_dataset_service.py
@@ -675,7 +675,6 @@ class TestDatasetServiceDatasetCases:
tenant_id="tenant-1",
user=user,
accessible_dataset_ids=[accessible.id],
- include_own_datasets=True,
tags=application_tags,
)
@@ -720,7 +719,6 @@ class TestDatasetServiceDatasetCases:
"tenant-1",
user=user,
accessible_dataset_ids=[accessible.id, "not-requested"],
- include_own_datasets=True,
session=sqlite_session,
)