fix: raise when binding starts-with with an out-of-range literal

LiteralPredicate.bind folded AboveMax/BelowMin literals to AlwaysTrue or
AlwaysFalse for the comparison operations, but starts-with and
not-starts-with fell through and built a bound predicate wrapping the
sentinel itself, which then crashed evaluators. Raise a ValueError at
bind time instead, mirroring the Java implementation, which throws for
operations that have no defined fold.

Signed-off-by: Shahzeb Ali <126076773+Shahzeb-DataEngineer@users.noreply.github.com>

Author: Shahzeb-DataEngineer

pyiceberg/expressions/__init__.py

@@ -915,11 +915,15 @@ def bind(self, schema: Schema, case_sensitive: bool = True) -> BoundLiteralPredi
                 return AlwaysTrue()
             elif isinstance(self, (GreaterThan, GreaterThanOrEqual, EqualTo)):
                 return AlwaysFalse()
+            else:
+                raise ValueError(f"Cannot bind {type(self).__name__} with AboveMax literal")
         elif isinstance(lit, BelowMin):
             if isinstance(self, (GreaterThan, GreaterThanOrEqual, NotEqualTo)):
                 return AlwaysTrue()
             elif isinstance(self, (LessThan, LessThanOrEqual, EqualTo)):
                 return AlwaysFalse()
+            else:
+                raise ValueError(f"Cannot bind {type(self).__name__} with BelowMin literal")
 
         return self.as_bound(bound_term, lit)  # type: ignore
 

tests/expressions/test_expressions.py

@@ -295,6 +295,20 @@ def test_empty_bind_not_in(table_schema_simple: Schema) -> None:
     assert bound == AlwaysTrue()
 
 
+@pytest.mark.parametrize(
+    "expression, match",
+    [
+        (StartsWith("bar", "2147483648"), "AboveMax"),
+        (NotStartsWith("bar", "2147483648"), "AboveMax"),
+        (StartsWith("bar", "-2147483649"), "BelowMin"),
+        (NotStartsWith("bar", "-2147483649"), "BelowMin"),
+    ],
+)
+def test_bind_starts_with_out_of_range_literal(table_schema_simple: Schema, expression: BooleanExpression, match: str) -> None:
+    with pytest.raises(ValueError, match=match):
+        expression.bind(table_schema_simple)
+
+
 def test_bind_not_in_equal_term(table_schema_simple: Schema) -> None:
     bound = BoundNotIn(
         BoundReference(table_schema_simple.find_field(1), table_schema_simple.accessor_for_field(1)), {literal("hello")}