fix(conversions): reject a Decimal whose scale is negative

pyiceberg/conversions.py

@@ -310,9 +310,11 @@ def _(primitive_type: DecimalType, value: Decimal) -> bytes:
         bytes: The byte representation of `value`.
     """
     _, digits, exponent = value.as_tuple()
-    exponent = abs(int(exponent))
-    if exponent != primitive_type.scale:
-        raise ValueError(f"Cannot serialize value, scale of value does not match type {primitive_type}: {exponent}")
+    # A Decimal carries the negated scale as its exponent, so a value with a negative
+    # scale (1E+2) must not be read as if it had the matching positive one
+    value_scale = -int(exponent)
+    if value_scale != primitive_type.scale:
+        raise ValueError(f"Cannot serialize value, scale of value does not match type {primitive_type}: {value_scale}")
     elif len(digits) > primitive_type.precision:
         raise ValueError(
             f"Cannot serialize value, precision of value is greater than precision of type {primitive_type}: {len(digits)}"

pyiceberg/expressions/literals.py

@@ -522,7 +522,7 @@ def to(self, type_var: IcebergType) -> Literal:  # type: ignore
 
     @to.register(DecimalType)
     def _(self, type_var: DecimalType) -> Literal[Decimal]:
-        if type_var.scale == abs(int(self.value.as_tuple().exponent)):
+        if type_var.scale == -int(self.value.as_tuple().exponent):
             return self
         raise ValueError(f"Could not convert {self.value} into a {type_var}")
 
@@ -628,7 +628,7 @@ def _(self, _: UUIDType) -> Literal[bytes]:
     @to.register(DecimalType)
     def _(self, type_var: DecimalType) -> Literal[Decimal]:
         dec = Decimal(self.value)
-        scale = abs(int(dec.as_tuple().exponent))
+        scale = -int(dec.as_tuple().exponent)
         if type_var.scale == scale:
             return DecimalLiteral(dec)
         else:

tests/expressions/test_literals.py

@@ -291,6 +291,9 @@ def test_decimal_to_decimal_conversion() -> None:
     with pytest.raises(ValueError) as e:
         _ = lit.to(DecimalType(9, 3))
     assert "Could not convert 34.11 into a decimal(9, 3)" in str(e.value)
+    with pytest.raises(ValueError) as e:
+        _ = literal(Decimal("1E+2")).to(DecimalType(9, 2))
+    assert "Could not convert 1E+2 into a decimal(9, 2)" in str(e.value)
 
 
 def test_timestamp_to_date() -> None:
@@ -947,6 +950,9 @@ def test_string_to_decimal_type_invalid_value() -> None:
     with pytest.raises(ValueError) as e:
         _ = literal("18.15").to(DecimalType(10, 0))
     assert "Could not convert 18.15 into a decimal(10, 0), scales differ 0 <> 2" in str(e.value)
+    with pytest.raises(ValueError) as e:
+        _ = literal("1E+2").to(DecimalType(10, 2))
+    assert "Could not convert 1E+2 into a decimal(10, 2), scales differ 2 <> -2" in str(e.value)
 
 
 def test_decimal_literal_increment() -> None:

tests/test_conversions.py

@@ -511,6 +511,13 @@ def __repr__(self) -> str:
     "primitive_type, value, expected_error_message",
     [
         (DecimalType(7, 3), Decimal("123.4567"), "Cannot serialize value, scale of value does not match type decimal(7, 3): 4"),
+        # A Decimal with a negative scale must not pass as its positive counterpart
+        (DecimalType(10, 2), Decimal("1E+2"), "Cannot serialize value, scale of value does not match type decimal(10, 2): -2"),
+        (
+            DecimalType(10, 2),
+            Decimal("100").normalize(),
+            "Cannot serialize value, scale of value does not match type decimal(10, 2): -2",
+        ),
         (
             DecimalType(18, 8),
             Decimal("123456789.123456789"),