sadpandajoe commented on code in PR #44406:
URL: https://github.com/apache/superset/pull/44406#discussion_r4264573946
##########
tests/unit_tests/common/test_query_context_processor.py:
##########
@@ -126,25 +131,350 @@ def processor(mock_query_context):
return processor
-def test_query_cache_key_binds_annotation_data_to_requesting_user(processor):
- """The cache key for annotated queries must differ per requesting user."""
+def test_annotation_cache_key_binds_native_annotation_read_scope(processor) ->
None:
+ """The annotation cache key for NATIVE layers must differ when the
+ requester's ``can_read`` (Annotation) access differs -- not who they
are."""
query_obj = MagicMock()
query_obj.annotation_layers = [{"sourceType": "NATIVE", "name": "a",
"value": 1}]
- with (
- patch(
- "superset.common.query_context_processor.get_user_id",
- side_effect=[1, 2],
- ),
- patch("superset.common.query_context_processor.security_manager"),
- ):
- processor.query_cache_key(query_obj)
- processor.query_cache_key(query_obj)
+ # ``security_manager`` autodetects as an async spec under a bare
+ # ``patch()`` (its real object trips ``unittest.mock``'s coroutine
+ # inference), which would silently turn every attribute access into an
+ # ``AsyncMock`` returning a fresh unawaited coroutine per call -- always
+ # unequal to itself and never equal to a configured return value. Forcing
+ # ``new_callable=MagicMock`` keeps these synchronous, as the real object
+ # is.
+ with patch(
+ "superset.common.query_context_processor.security_manager",
+ new_callable=MagicMock,
+ ) as security_manager:
+ security_manager.can_access.side_effect = [True, False]
+ processor.annotation_cache_key(query_obj)
+ processor.annotation_cache_key(query_obj)
contexts = [
call.kwargs["annotation_context"] for call in
query_obj.cache_key.call_args_list
]
assert contexts[0] != contexts[1]
+def test_annotation_cache_key_shares_across_same_access_scope() -> None:
+ """Two distinct requesters (separate processor/query-object instances,
+ standing in for two different requests) with identical access scope must
+ produce identical annotation-context material. Reusing a single
+ processor/query_obj across both calls (as this test previously did)
+ would pass trivially regardless of whether the key is scope-based or
+ identity-based, since nothing about "who's asking" would ever vary."""
+ layer = {"sourceType": "NATIVE", "name": "a", "value": 1}
+ processor_a = QueryContextProcessor(MagicMock())
+ processor_b = QueryContextProcessor(MagicMock())
+ query_obj_a = MagicMock(annotation_layers=[layer])
+ query_obj_b = MagicMock(annotation_layers=[layer])
+
+ with patch(
+ "superset.common.query_context_processor.security_manager",
+ new_callable=MagicMock,
+ ) as security_manager:
+ security_manager.can_access.return_value = True
+ context_a = processor_a._annotation_cache_context(query_obj_a)
+ context_b = processor_b._annotation_cache_context(query_obj_b)
+
+ assert context_a == context_b
Review Comment:
This test never varies the requesting user: both processors run under the
same ambient identity, so putting `get_user_id()` back into the annotation
context would still give equal contexts and pass. That is the exact regression
this PR is meant to prevent (same-scope viewers sharing one entry).
Could it patch `get_user_id` to return different values for the two calls
and assert the contexts stay equal?
##########
superset/common/query_context_processor.py:
##########
@@ -470,72 +499,223 @@ def query_cache_key(self, query_obj: QueryObject,
**kwargs: Any) -> str | None:
)
return cache_key
- def _annotation_cache_context(self, query_obj: QueryObject) -> dict[str,
Any]:
+ def annotation_cache_key(self, query_obj: QueryObject) -> str | None:
"""
- Cache-key material binding cached annotation data to its security
- context.
-
- Annotation payloads are fetched per requesting user and stored on the
- same cache entry as the dataframe, so the key also binds the requesting
- user and, for chart-backed layers, the RLS clauses of the referenced
- chart's datasource.
+ Cache key for this query's annotation-layer payload, or ``None`` when
+ the query has no annotation layers or when some layer's access scope
+ could not be resolved (the payload is then never cached, see
+ :meth:`_annotation_source_scope`).
+
+ Annotation payloads are fetched under the requesting user's access
+ scope, which is a stricter security requirement than the dataframe
+ itself has. Keying them separately from :meth:`query_cache_key` keeps
+ that scoping from forcing every distinct viewer of an annotated chart
+ onto their own full copy of the (potentially much larger) shared
+ dataframe: users with the same access scope share this key too.
"""
- source_rls: dict[str, list[str] | None] = {}
- source_versions: dict[str, tuple[str, str]] = {}
- metadata_versions: dict[str, str] = {}
+ if not query_obj or not query_obj.annotation_layers:
+ return None
+ if (annotation_context := self._annotation_cache_context(query_obj))
is None:
+ return None
+ return self.query_cache_key(query_obj,
annotation_context=annotation_context)
+
+ def _annotation_cache_context(
+ self, query_obj: QueryObject
+ ) -> dict[str, Any] | None:
+ """
+ Cache-key material binding annotation data to its security *scope* so
+ users with the same access share a cache entry and users with a
+ different scope — or no access — never read each other's data.
+
+ * NATIVE layers: the ``can_read`` permission on ``Annotation``, the
+ only user-dependent dimension of these global records.
+ * Chart-backed (``line``/``table``) layers: see
+ :meth:`_annotation_source_scope`, recorded per *layer* (not per
+ referenced chart) since two layers can reference the same chart
+ with different overrides, and so different scopes.
+
+ Returns ``None`` when any layer's scope cannot be resolved, meaning
+ the payload must not be cached at all.
+ """
+ context: dict[str, Any] = {}
+
+ if any(
+ layer.get("sourceType") == "NATIVE" for layer in
query_obj.annotation_layers
+ ):
+ context["annotation_read"] = security_manager.can_access(
+ "can_read", "Annotation"
+ )
+
+ source_scope: list[dict[str, Any]] = []
for layer in query_obj.annotation_layers:
if (
layer.get("sourceType")
not in ANNOTATION_SOURCE_TYPES_WITH_CHART_REFERENCE
):
continue
- layer_value = layer.get("value")
+ if (scope := self._annotation_source_scope(layer)) is None:
+ return None
+ source_scope.append({"value": str(layer.get("value")), **scope})
+ if source_scope:
+ context["source_scope"] = source_scope
+
+ return context
+
+ def _annotation_source_scope(self, layer: dict[str, Any]) -> dict[str,
Any] | None:
+ """
+ Access and data-identity cache-key material for one chart-backed
+ annotation layer, or ``None`` when it cannot be resolved.
+
+ ``None`` means "do not cache": a missing chart, a chart with neither
+ a saved query context nor a resolvable datasource, or any error while
+ deriving the scope would otherwise collapse an authorized and a
+ denied requester onto one shared identity, letting the latter read
+ the former's warmed payload. The caller fetches live instead.
+
+ ``access`` keeps a user denied the referenced chart's datasource from
+ reading an authorized user's cached payload. When the chart has a
+ saved query context, this runs the *same* authorization path
+ :meth:`get_viz_annotation_data` executes
+ (``QueryContext.raise_for_access``) rather than the coarser
+ :meth:`SecurityManager.can_access_datasource` — a requester whose
+ access comes from a dashboard/viewer-promiscuous-mode bypass (which
+ depends on the chart's own saved ``form_data``, e.g. its
+ ``slice_id``/``dashboardId``) would otherwise still fail that coarser,
+ context-free check and collapse onto the same denied scope as a
+ genuinely unauthorized requester, letting the latter read the
+ former's cached payload. ``data_key`` is the annotation chart's own
+ query cache key(s) — derived from the same, override-applied query
+ objects actually executed (see :meth:`_apply_annotation_overrides`),
+ so it captures the datasource version, RLS clauses, and any per-user
+ Jinja/virtual-dataset RLS material exactly as the live fetch would,
+ including material an override only introduces at a finer grain.
+ Reusing this logic avoids re-deriving it and automatically inherits
+ any future correctness fixes made there.
+ """
+ layer_value = layer.get("value")
+ try:
chart = (
ChartDAO.find_by_id(layer_value) if layer_value is not None
else None
)
- annotation_datasource: Datasource | None = (
- chart.resolved_datasource if chart else None
- )
- if chart and isinstance(chart.query_context, str):
+ if chart is None:
+ return None
+
+ # The saved query context is what the live fetch executes and
+ # authorizes against, so it is consulted first, independent of
+ # the chart's own datasource pointer: a chart can have its
+ # ``datasource_id`` cleared while its saved query context still
+ # targets a valid dataset, and the live fetch would still work.
+ annotation_query_context = chart.get_query_context()
+ if annotation_query_context is not None:
+ self._apply_annotation_overrides(annotation_query_context,
layer)
try:
- source: dict[str, Any] = json.loads(chart.query_context)[
- "datasource"
- ]
- annotation_datasource = DatasourceDAO.get_datasource(
- source["type"], source["id"]
- )
- except (
- ValueError,
- TypeError,
- KeyError,
- DatasourceNotFound,
- DatasourceTypeNotSupportedError,
- DatasourceValueIsIncorrect,
- ):
- annotation_datasource = None
+ annotation_query_context.raise_for_access()
+ access: Any = True
+ except SupersetSecurityException:
+ access = False
+ data_key: Any = [
+ annotation_query_context.query_cache_key(query_object)
+ for query_object in annotation_query_context.queries
+ ]
+ annotation_datasource: Any =
annotation_query_context.datasource
+ else:
+ # Fall back to the RLS-clause identity when the chart has no
+ # saved query context to key on. resolved_datasource, not
+ # datasource: the latter is pinned to table-backed
+ # datasources and resolves to None for a semantic-view-backed
+ # chart.
+ datasource = chart.resolved_datasource
+ if datasource is None:
+ return None
+ access = security_manager.can_access_datasource(datasource)
+ data_key = security_manager.get_rls_cache_key(datasource)
+ annotation_datasource = datasource
+ scope: dict[str, Any] = {"access": access, "data_key": data_key}
+ # Semantic-view results are versioned by the layer's metadata
+ # generation and result discriminator, which the datasource-level
+ # key material above does not carry.
if isinstance(annotation_datasource, SemanticView):
- metadata_versions[str(layer_value)] = (
- annotation_datasource.metadata_generation
- )
- source_rls[str(layer.get("value"))] = (
- security_manager.get_rls_cache_key(annotation_datasource)
- if annotation_datasource
- else None
+ scope["metadata_version"] =
annotation_datasource.metadata_generation
+ if (
+ discriminator :=
annotation_datasource.result_cache_discriminator
+ ) is not None:
+ scope["result_version"] = discriminator
+ except Exception: # noqa: BLE001 pylint: disable=broad-except
+ # Derivation can fail well beyond SupersetException: the DAO
+ # lookup and lazy ``datasource`` load are themselves live DB
+ # queries, and the RLS lookup / a virtual dataset's
+ # get_extra_cache_keys() renders Jinja, either of which can raise
+ # a driver/template error. None of that should ever 500 the whole
+ # chart-data request; fail closed instead. A fallback scope
+ # shared by every requester whose derivation failed would still
+ # be a shared identity, so the answer is "no caching", not a
+ # substitute key.
+ logger.warning(
+ "Could not derive annotation cache key for chart %s; "
+ "annotation data for this request will not be cached",
+ layer_value,
+ exc_info=True,
)
- if isinstance(annotation_datasource, SemanticView):
- discriminator: tuple[str, str] | None = (
- annotation_datasource.result_cache_discriminator
+ return None
+ return scope
+
+ def _get_annotation_data_cached(
+ self,
+ query_obj: QueryObject,
+ cache_key: str | None,
+ force_query: bool,
+ force_cached: bool | None,
+ timeout: int | None,
+ datasource_uid: str | None,
+ ) -> dict[str, Any]:
+ """
+ Fetch this query's annotation-layer payload, cached under its own
+ (user/RLS-scoped) entry, separate from the shared dataframe cache.
+
+ A ``None`` ``cache_key`` means some layer's access scope could not be
+ resolved (see :meth:`_annotation_source_scope`), so there is no key
+ under which a cached payload could be safely shared: an unresolved
+ scope looks the same for an authorized and a denied requester. The
+ payload is then fetched live and never cached. That holds under
+ ``force_cached`` too: computing here beats failing the read-back of
+ an otherwise successful async task.
+ """
+ if cache_key is None:
+ return self.get_annotation_data(query_obj)
Review Comment:
When an annotation layer's scope can't be resolved, this live fetch runs
even under `force_cached`. In the async probe (`_run_async` only catches the
cache-load error), a failed fetch now surfaces as an uncaught
`ChartDataQueryFailedError` instead of falling through to the task path.
Concretely: user A warms the shared dataframe for a chart whose annotation
layer references a chart that user B can't see. B loads it in async mode, hits
the warm dataframe, the live fetch fails with "chart not found", and B gets a
server error where the old code (cache miss, then task) and the sync path
return a 400.
Should the `force_cached` probe return a cache-load miss for an unresolved
scope instead of computing, so the request still goes through the normal task
path?
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
