Fix NotStartsWith residual evaluation to return correct result (#3503)

# Rationale for this change

- Fixes #3496

## Are these changes tested?

Yes - the added test fails if we revert this PR's change.

## Are there any user-facing changes?

Yes - this PR fixes a correctness issue of `NotStartsWith`.

---------

Co-authored-by: Kevin Liu <kevinjqliu@users.noreply.github.com>
Co-authored-by: Kevin Liu <kevin.jq.liu@gmail.com>

Author: ebyhr

pyiceberg/expressions/visitors.py

@@ -1905,10 +1905,7 @@ def visit_starts_with(self, term: BoundTerm, literal: LiteralValue) -> BooleanEx
             return AlwaysFalse()
 
     def visit_not_starts_with(self, term: BoundTerm, literal: LiteralValue) -> BooleanExpression:
-        if not self.visit_starts_with(term, literal):
-            return AlwaysTrue()
-        else:
-            return AlwaysFalse()
+        return ~self.visit_starts_with(term, literal)
 
     def visit_bound_predicate(self, predicate: BoundPredicate) -> BooleanExpression:
         """

tests/expressions/test_residual_evaluator.py

@@ -41,7 +41,7 @@
 from pyiceberg.schema import Schema
 from pyiceberg.transforms import DayTransform, IdentityTransform
 from pyiceberg.typedef import Record
-from pyiceberg.types import DoubleType, FloatType, IntegerType, NestedField, TimestampType
+from pyiceberg.types import DoubleType, FloatType, IntegerType, NestedField, StringType, TimestampType
 
 
 def test_identity_transform_residual() -> None:
@@ -249,3 +249,29 @@ def test_not_in_timestamp() -> None:
     ts_day += 3  # type: ignore
     residual = res_eval.residual_for(Record(ts_day))
     assert residual == AlwaysTrue()
+
+
+def test_starts_with() -> None:
+    schema = Schema(NestedField(1, "x", StringType()))
+    spec = PartitionSpec(PartitionField(1, 1001, IdentityTransform(), "x_part"))
+
+    predicate = StartsWith("x", "a")
+    res_eval = residual_evaluator_of(spec=spec, expr=predicate, case_sensitive=True, schema=schema)
+
+    assert res_eval.residual_for(Record("bb")) == AlwaysFalse()
+    assert res_eval.residual_for(Record("abc")) == AlwaysTrue()
+    assert res_eval.residual_for(Record("a")) == AlwaysTrue()
+    assert res_eval.residual_for(Record("zoo")) == AlwaysFalse()
+
+
+def test_not_starts_with() -> None:
+    schema = Schema(NestedField(1, "x", StringType()))
+    spec = PartitionSpec(PartitionField(1, 1001, IdentityTransform(), "x_part"))
+
+    predicate = NotStartsWith("x", "a")
+    res_eval = residual_evaluator_of(spec=spec, expr=predicate, case_sensitive=True, schema=schema)
+
+    assert res_eval.residual_for(Record("bb")) == AlwaysTrue()
+    assert res_eval.residual_for(Record("abc")) == AlwaysFalse()
+    assert res_eval.residual_for(Record("a")) == AlwaysFalse()
+    assert res_eval.residual_for(Record("zoo")) == AlwaysTrue()

tests/test_transforms.py

@@ -1030,7 +1030,6 @@ def test_projection_truncate_string_not_starts_with_longer_literal(bound_referen
     assert TruncateTransform(2).project("name", BoundNotStartsWith(term=bound_reference_str, literal=literal("hello"))) is None
 
 
-def test_projection_truncate_string_not_starts_with_shorter_literal(bound_reference_str: BoundReference) -> None:
 def test_projection_truncate_string_not_starts_with_equal_width_literal(bound_reference_str: BoundReference) -> None:
     # Valid projection: improve NOT STARTS WITH "he" to partition != "he".
     assert TruncateTransform(2).project(