fix: raise FilterError on unknown metadata filter operators

Filtering with an unsupported operator (e.g. a typo like "gt" instead of
">", or "XOR" as a logical operator) raised a bare KeyError with no
context. Since filters.py is the shared filter engine used by
InMemoryDocumentStore and every external document store, this affected
all stores.

Both _comparison_condition and _logic_condition now validate the
operator and raise a FilterError that lists the valid operators,
consistent with how other malformed-filter cases are already handled.

Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>

Author: immuhammadfurqan

haystack/utils/filters.py

@@ -173,6 +173,9 @@ def _logic_condition(condition: dict[str, Any], document: Document | ByteStream)
         msg = f"'conditions' key missing in {condition}"
         raise FilterError(msg)
     operator: str = condition["operator"]
+    if operator not in LOGICAL_OPERATORS:
+        msg = f"Unknown logical operator '{operator}'. Valid operators are: {sorted(LOGICAL_OPERATORS)}"
+        raise FilterError(msg)
     conditions: list[dict[str, Any]] = condition["conditions"]
     return LOGICAL_OPERATORS[operator](document=document, conditions=conditions)
 
@@ -214,5 +217,8 @@ def _comparison_condition(condition: dict[str, Any], document: Document | ByteSt
     else:
         document_value = getattr(document, field)
     operator: str = condition["operator"]
+    if operator not in COMPARISON_OPERATORS:
+        msg = f"Unknown comparison operator '{operator}'. Valid operators are: {sorted(COMPARISON_OPERATORS)}"
+        raise FilterError(msg)
     filter_value: Any = condition["value"]
     return COMPARISON_OPERATORS[operator](filter_value=filter_value, value=document_value)

releasenotes/notes/filter-unknown-operator-error-8df4039c0db976ff.yaml

@@ -0,0 +1,8 @@
+---
+fixes:
+  - |
+    Metadata filters that use an unsupported operator now raise a clear ``FilterError``
+    that lists the valid operators, instead of a cryptic ``KeyError``. For example, a
+    typo such as ``{"field": "meta.year", "operator": "gt", "value": 2000}`` (instead of
+    ``">"``) previously surfaced as ``KeyError: 'gt'``. This applies to both comparison
+    operators and logical operators (``AND``, ``OR``, ``NOT``).

test/utils/test_filters.py

@@ -566,6 +566,12 @@ def test_document_matches_filter(filters, document, expected_result):
     pytest.param({"operator": "==", "value": "test"}, id="Missing condition field key"),
     pytest.param({"field": "meta.name", "value": "test"}, id="Missing condition operator key"),
     pytest.param({"field": "meta.name", "operator": "=="}, id="Missing condition value key"),
+    # Unknown operators
+    pytest.param({"field": "meta.page", "operator": "gt", "value": 10}, id="Unknown comparison operator"),
+    pytest.param(
+        {"operator": "XOR", "conditions": [{"field": "meta.page", "operator": "==", "value": 10}]},
+        id="Unknown logical operator",
+    ),
 ]