This is an automated email from the ASF dual-hosted git repository.
HyukjinKwon pushed a commit to branch master
in repository https://gitbox.apache.org/repos/asf/spark-connect-rust.git
The following commit(s) were added to refs/heads/master by this push:
new e541f64 [SPARK-60994][CONNECT][PYTHON] Type a decimal literal as
PySpark does
e541f64 is described below
commit e541f64c94e448d4b109686a3397469fbaf9c5cd
Author: Hyukjin Kwon <[email protected]>
AuthorDate: Fri Nov 6 16:18:01 2026 +0900
[SPARK-60994][CONNECT][PYTHON] Type a decimal literal as PySpark does
### What changes were proposed in this pull request?
[SPARK-60994](https://issues.apache.org/jira/browse/SPARK-60994): a literal
of a Python `Decimal` now has the precision 10 and scale 0 of `DecimalType()`,
as PySpark's `LiteralExpression` of a `Decimal` does. That covers `lit()` and
any other Column built from one, such as `F.col("a") == Decimal("1.5")`. The
server widens them to the precision and scale of the value. Before, the literal
had the precision and scale of the value itself.
### Why are the changes needed?
The literals, and the expressions over them, had other types than in
PySpark. Against a Spark 4.3.0 server:
| Column | PySpark 4.3.0 and now | Before |
| --- | --- | --- |
| `lit(Decimal("1.5"))` | `decimal(10,1)` | `decimal(2,1)` |
| `lit(Decimal(0))` | `decimal(10,0)` | `decimal(1,0)` |
| `lit(Decimal("-0.001"))` | `decimal(10,3)` | `decimal(3,3)` |
| `lit(Decimal("1.5")) * lit(Decimal("2.25"))` | `decimal(21,3)` |
`decimal(6,3)` |
So an expression could overflow where PySpark's does not. A literal of a
value with more than 10 digits, such as `Decimal("123456789012.5")`, keeps its
own precision in both.
### Does this PR introduce _any_ user-facing change?
Yes, decimal literals have the types of PySpark 4.3.0.
### How was this patch tested?
- Against a local Spark 4.3.0 server, the types and values of the literals
above give the same results as official PySpark 4.3.0.
- A new offline test checks, through a mock Connect server, the decimal
literal that `lit()` and a comparison send. A new live test checks the types
and values above. It also passes with official PySpark 4.3.0. Both fail without
this change.
- The lib tests and every integration test, e2e included, against the local
server (648 passed, and 123 of the transport crate), the build of all tests
with `wasm-udf`, the offline Python suites (721 passed, 9 skipped), and the
live drop-in and Arrow suites (77 passed) and live cases (10 passed). Clippy
reports no diagnostics in the changed lines.
### Was this patch authored or co-authored using generative AI tooling?
Generated-by: Claude Opus 5.5
This pull request and its description were written by Isaac.
Closes #196 from HyukjinKwon/SPARK-60994-decimal-literal.
Authored-by: Hyukjin Kwon <[email protected]>
Signed-off-by: Hyukjin Kwon <[email protected]>
---
crates/pyspark-rs/src/functions.rs | 15 ++++++---------
python/tests/test_dropin_e2e.py | 21 +++++++++++++++++++++
python/tests/test_dropin_offline.py | 27 +++++++++++++++++++++++++++
3 files changed, 54 insertions(+), 9 deletions(-)
diff --git a/crates/pyspark-rs/src/functions.rs
b/crates/pyspark-rs/src/functions.rs
index 6f24de8..392357b 100644
--- a/crates/pyspark-rs/src/functions.rs
+++ b/crates/pyspark-rs/src/functions.rs
@@ -261,19 +261,16 @@ fn scalar_literal(obj: &Bound<'_, PyAny>) ->
PyResult<Option<LiteralExpression>>
let py = obj.py();
// decimal.Decimal -> Decimal literal (before int/float: Decimal has
__float__/__int__).
+ // As PySpark's LiteralExpression of a Decimal, of DecimalType():
precision 10 and scale
+ // 0, which the server widens to those of the value, so
lit(Decimal("1.5")) is a
+ // decimal(10,1).
if let Ok(dec_mod) = py.import("decimal") {
let decimal_cls = dec_mod.getattr("Decimal")?;
if obj.is_instance(&decimal_cls)? {
- let value: String = obj.str()?.extract()?;
- let as_tuple = obj.call_method0("as_tuple")?;
- let num_digits = as_tuple.get_item(1)?.len()? as i32;
- let exp: i64 = as_tuple.get_item(2)?.extract().unwrap_or(0);
- let scale = if exp < 0 { (-exp) as i32 } else { 0 };
- let precision = num_digits.max(scale).max(1);
return Ok(Some(LiteralExpression::Decimal {
- value,
- precision,
- scale,
+ value: obj.str()?.extract()?,
+ precision: 10,
+ scale: 0,
}));
}
}
diff --git a/python/tests/test_dropin_e2e.py b/python/tests/test_dropin_e2e.py
index e2adc58..e1e380a 100644
--- a/python/tests/test_dropin_e2e.py
+++ b/python/tests/test_dropin_e2e.py
@@ -824,6 +824,27 @@ def
test_maps_keep_their_entries_in_order_with_keys_of_their_type(session):
assert list(df.first()[0].items()) == [(2, "b"), (1, "a")]
+def test_decimal_literals_are_of_the_types_of_pyspark(session):
+ # The types of official PySpark 4.3.0: at least the precision 10 of
DecimalType().
+ from decimal import Decimal
+
+ df = session.range(1).select(
+ F.lit(Decimal("1.5")).alias("a"),
+ F.lit(Decimal(0)).alias("b"),
+ F.lit(Decimal("123456789012.5")).alias("c"),
+ F.lit(Decimal("-0.001")).alias("d"),
+ (F.lit(Decimal("1.5")) * F.lit(Decimal("2.25"))).alias("product"),
+ )
+ assert df.dtypes == [
+ ("a", "decimal(10,1)"), ("b", "decimal(10,0)"), ("c", "decimal(13,1)"),
+ ("d", "decimal(10,3)"), ("product", "decimal(21,3)"),
+ ]
+ assert tuple(df.first()) == (
+ Decimal("1.5"), Decimal("0"), Decimal("123456789012.5"),
Decimal("-0.001"),
+ Decimal("3.375"),
+ )
+
+
def test_ml_vector_functions_as_pyspark(session):
from pyspark.ml.functions import array_to_vector, vector_to_array
diff --git a/python/tests/test_dropin_offline.py
b/python/tests/test_dropin_offline.py
index ab23102..b3733b3 100644
--- a/python/tests/test_dropin_offline.py
+++ b/python/tests/test_dropin_offline.py
@@ -634,6 +634,33 @@ def test_show_checks_its_arguments_as_pyspark():
df._show_string(1, n=2)
+def
test_decimal_literal_is_of_the_default_decimal_type_as_pyspark(mock_connect):
+ # As PySpark's LiteralExpression of a Decimal, of DecimalType(): precision
10 and scale
+ # 0, which the server widens to those of the value, wherever the literal
is.
+ from decimal import Decimal
+
+ from pyspark.sql import functions as F
+
+ spark, service = mock_connect
+ spark.range(1).select(
+ F.lit(Decimal("1.5")), F.col("id") == Decimal("-0.001"),
F.lit(Decimal("12345678901.25"))
+ ).collect()
+ # ExecutePlanRequest.plan=3, Plan.root=1, Relation.project=3,
Project.expressions=3;
+ # Expression.literal=1, Literal.decimal=12 (value=1, precision=2, scale=3);
+ # .unresolved_function=3, UnresolvedFunction.arguments=2.
+ plan =
parse_message(parse_message(service.execute_requests[-1])[3][0])[1][0]
+ expressions = parse_message(parse_message(plan)[3][0])[3]
+
+ def decimal(literal):
+ return parse_message(parse_message(literal)[12][0])
+
+ first, compared, wide = (parse_message(e) for e in expressions)
+ compared = parse_message(parse_message(compared[3][0])[2][1])
+ assert decimal(first[1][0]) == {1: [b"1.5"], 2: [10], 3: [0]}
+ assert decimal(compared[1][0]) == {1: [b"-0.001"], 2: [10], 3: [0]}
+ assert decimal(wide[1][0]) == {1: [b"12345678901.25"], 2: [10], 3: [0]}
+
+
def test_dropin_module_imports():
# The thin re-export shims import cleanly.
import pyspark.sql.dataframe # noqa: F401
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]