microsoft/semantic-kernel · error · VectorStoreOperationException
Call target node type '{type(node.func).__name__}' is not al
Error message
Call target node type '{type(node.func).__name__}' is not allowed in filter expressions. What it means
Thrown by _eval_Call (in_memory.py:307-309) when a Call node's func is neither an ast.Name nor an ast.Attribute (e.g. calling the result of a subscript or another call). This is defense-in-depth: the parse-time Call validation (in_memory.py:796-806) rejects such constructs earlier with a different message, so this branch is only reachable if the allowlist has been subclassed/relaxed to admit chained or subscript calls without teaching the evaluator to handle them.
Source
Thrown at python/semantic_kernel/connectors/in_memory.py:307
if node.func.attr == "contains":
if len(args) != 1:
raise VectorStoreOperationException("Method 'contains' expects exactly one argument.")
return args[0] in target
try:
func = getattr(target, node.func.attr)
except AttributeError as e:
raise VectorStoreOperationException(
f"Method '{node.func.attr}' is not available in filter expressions."
) from e
if not callable(func):
raise VectorStoreOperationException(
f"Attribute '{node.func.attr}' is not callable in filter expressions."
)
return func(*args)
raise VectorStoreOperationException(
f"Call target node type '{type(node.func).__name__}' is not allowed in filter expressions."
)
def _compare(self, operator_node: ast.AST, left: Any, right: Any) -> bool:
"""Evaluate a comparison operator."""
if isinstance(operator_node, ast.Eq):
return left == right
if isinstance(operator_node, ast.NotEq):
return left != right
if isinstance(operator_node, ast.Lt):
return left < right
if isinstance(operator_node, ast.LtE):
return left <= right
if isinstance(operator_node, ast.Gt):
return left > right
if isinstance(operator_node, ast.GtE):
return left >= right
if isinstance(operator_node, ast.In):View on GitHub (pinned to c028a0c7dc)
Solutions
- Do not relax allowed_filter_ast_nodes to permit chained/subscript/lambda call targets.
- If you genuinely need such calls, override _eval_Call in your subclass to handle the extra target types safely.
- Keep the parse-time Call validation intact so unsupported call shapes are rejected before evaluation.
Defensive patterns
Strategy: validation
Validate before calling
import ast
SAFE_NODES = {
ast.Expression, ast.Lambda, ast.arguments, ast.arg, ast.Compare, ast.BoolOp,
ast.UnaryOp, ast.And, ast.Or, ast.Not, ast.Eq, ast.NotEq, ast.Lt, ast.LtE,
ast.Gt, ast.GtE, ast.In, ast.NotIn, ast.Is, ast.IsNot, ast.Name, ast.Load,
ast.Attribute, ast.Subscript, ast.Slice, ast.Constant, ast.List, ast.Tuple,
ast.Set, ast.Dict, ast.BinOp, ast.Add, ast.Sub, ast.Mult, ast.Div, ast.Mod,
ast.FloorDiv, ast.Call,
}
def preflight_filter(expr: str, *, max_len: int = 2048, max_nodes: int = 128) -> None:
if len(expr) > max_len:
raise ValueError("filter too long")
try:
tree = ast.parse(expr, mode="eval")
except SyntaxError as e:
raise ValueError(f"invalid python: {e}") from e
if not (isinstance(tree, ast.Expression) and isinstance(tree.body, ast.Lambda)):
raise ValueError("filter must be a lambda expression")
blocked = {"__class__", "__globals__", "__subclasses__", "__builtins__", "__code__"}
for n in ast.walk(tree):
if isinstance(n, ast.Attribute) and n.attr in blocked:
raise ValueError(f"blocked attribute: {n.attr}")
if type(n) not in SAFE_NODES:
raise ValueError(f"disallowed node: {type(n).__name__}")
if sum(1 for _ in ast.walk(tree)) > max_nodes:
raise ValueError("filter too complex")
Try / catch
try:
results = await collection.search(search_type=SearchType.VECTOR, options=opts)
except VectorStoreOperationException as e:
logger.warning("filter rejected: %s", e.__cause__ or e)
results = None Prevention
- Treat allowed_filter_ast_nodes as read-only unless you also extend the evaluator.
- Run the full filter test suite after any allowlist change.
When it happens
Trigger: Subclassing InMemoryCollection and adding ast.Subscript, ast.Call, or ast.Lambda to allowed_filter_ast_nodes so a filter like `lambda x: x[0]()` or `lambda x: (lambda y: y)(x)` passes parsing, then reaches an evaluator that cannot dispatch the call target.
Common situations: Extending the filter sandbox via subclassing without extending _eval_Call; importing filters from an untrusted source that the subclass is more permissive about.
Related errors
- Attribute '{node.func.attr}' is not callable in filter expre
- Comparison operator '{type(operator_node).__name__}' is not
- Addition in filter expressions is only allowed for numeric v
- Multiplication in filter expressions is only allowed for num
- Sequence repetition in filter expressions exceeds the maximu
AI-assisted analysis of microsoft/semantic-kernel@c028a0c7dc (2026-08-13).
Data as JSON: /api/errors/07c4318e69e5dcba.
Report an issue: GitHub.