Make UnaryPredicate JSON Serializable and Add Unit Tests

pyiceberg/expressions/__init__.py

@@ -447,7 +447,20 @@ def bind(self, schema: Schema, case_sensitive: bool = True) -> BooleanExpression
     def as_bound(self) -> Type[BoundPredicate[L]]: ...
 
 
-class UnaryPredicate(UnboundPredicate[Any], ABC):
+class UnaryPredicate(IcebergBaseModel, UnboundPredicate[Any], ABC):
+    type: str
+
+    model_config = {"arbitrary_types_allowed": True}
+
+    def __init__(self, term: Union[str, UnboundTerm[Any]]):
+        unbound = _to_unbound_term(term)
+        super().__init__(term=unbound)
+
+    def __str__(self) -> str:
+        """Return the string representation of the UnaryPredicate class."""
+        # Sort to make it deterministic
+        return f"{str(self.__class__.__name__)}(term={str(self.term)})"
+
     def bind(self, schema: Schema, case_sensitive: bool = True) -> BoundUnaryPredicate[Any]:
         bound_term = self.term.bind(schema, case_sensitive)
         return self.as_bound(bound_term)
@@ -506,6 +519,8 @@ def as_unbound(self) -> Type[NotNull]:
 
 
 class IsNull(UnaryPredicate):
+    type: str = "is-null"
+
     def __invert__(self) -> NotNull:
         """Transform the Expression into its negated version."""
         return NotNull(self.term)
@@ -516,6 +531,8 @@ def as_bound(self) -> Type[BoundIsNull[L]]:
 
 
 class NotNull(UnaryPredicate):
+    type: str = "not-null"
+
     def __invert__(self) -> IsNull:
         """Transform the Expression into its negated version."""
         return IsNull(self.term)
@@ -558,6 +575,8 @@ def as_unbound(self) -> Type[NotNaN]:
 
 
 class IsNaN(UnaryPredicate):
+    type: str = "is-nan"
+
     def __invert__(self) -> NotNaN:
         """Transform the Expression into its negated version."""
         return NotNaN(self.term)
@@ -568,6 +587,8 @@ def as_bound(self) -> Type[BoundIsNaN[L]]:
 
 
 class NotNaN(UnaryPredicate):
+    type: str = "not-nan"
+
     def __invert__(self) -> IsNaN:
         """Transform the Expression into its negated version."""
         return IsNaN(self.term)

tests/expressions/test_expressions.py

@@ -694,7 +694,7 @@ def test_and() -> None:
     assert and_ == pickle.loads(pickle.dumps(and_))
 
     with pytest.raises(ValueError, match="Expected BooleanExpression, got: abc"):
-        null & "abc"  # type: ignore
+        null & "abc"
 
 
 def test_or() -> None:
@@ -711,7 +711,7 @@ def test_or() -> None:
     assert or_ == pickle.loads(pickle.dumps(or_))
 
     with pytest.raises(ValueError, match="Expected BooleanExpression, got: abc"):
-        null | "abc"  # type: ignore
+        null | "abc"
 
 
 def test_not() -> None:
@@ -780,6 +780,16 @@ def test_not_null() -> None:
     assert non_null == pickle.loads(pickle.dumps(non_null))
 
 
+def test_serialize_is_null() -> None:
+    pred = IsNull(term="foo")
+    assert pred.model_dump_json() == '{"term":"foo","type":"is-null"}'
+
+
+def test_serialize_not_null() -> None:
+    pred = NotNull(term="foo")
+    assert pred.model_dump_json() == '{"term":"foo","type":"not-null"}'
+
+
 def test_bound_is_nan(accessor: Accessor) -> None:
     # We need a FloatType here
     term = BoundReference[float](