Skip to content

Commit 94cd107

Browse files
mdwintmdwint
committed
test: unroll parametrized tests for clarity
1 parent 57f95e1 commit 94cd107

File tree

1 file changed

+68
-62
lines changed

1 file changed

+68
-62
lines changed

‎tests/table/test_upsert.py‎

Lines changed: 68 additions & 62 deletions
Original file line numberDiff line numberDiff line change
@@ -15,7 +15,6 @@
1515
# specific language governing permissions and limitations
1616
# under the License.
1717
from pathlib import PosixPath
18-
from typing import Any
1918

2019
import pyarrow as pa
2120
import pytest
@@ -24,7 +23,7 @@
2423

2524
from pyiceberg.catalog import Catalog
2625
from pyiceberg.exceptions import NoSuchTableError
27-
from pyiceberg.expressions import AlwaysTrue, And, BooleanExpression, EqualTo, In, IsNaN, IsNull, Or, Reference
26+
from pyiceberg.expressions import AlwaysTrue, And, EqualTo, In, IsNaN, IsNull, Or, Reference
2827
from pyiceberg.expressions.literals import DoubleLiteral, LongLiteral
2928
from pyiceberg.io.pyarrow import schema_to_pyarrow
3029
from pyiceberg.schema import Schema
@@ -444,73 +443,80 @@ def test_create_match_filter_single_condition() -> None:
444443
)
445444

446445

447-
@pytest.mark.parametrize(
448-
"data, expected",
449-
[
450-
pytest.param(
451-
[{"x": 1.0}, {"x": 2.0}, {"x": 3.0}],
452-
In(Reference(name="x"), {DoubleLiteral(1.0), DoubleLiteral(2.0), DoubleLiteral(3.0)}),
453-
id="single-column-without-null",
446+
def test_create_match_filter_single_column_without_null() -> None:
447+
data = [{"x": 1.0}, {"x": 2.0}, {"x": 3.0}]
448+
449+
schema = pa.schema([pa.field("x", pa.float64())])
450+
table = pa.Table.from_pylist(data, schema=schema)
451+
452+
expr = create_match_filter(table, join_cols=["x"])
453+
454+
assert expr == In(Reference(name="x"), {DoubleLiteral(1.0), DoubleLiteral(2.0), DoubleLiteral(3.0)})
455+
456+
457+
def test_create_match_filter_single_column_with_null() -> None:
458+
data = [
459+
{"x": 1.0},
460+
{"x": 2.0},
461+
{"x": None},
462+
{"x": 4.0},
463+
{"x": float("nan")},
464+
]
465+
schema = pa.schema([pa.field("x", pa.float64())])
466+
table = pa.Table.from_pylist(data, schema=schema)
467+
468+
expr = create_match_filter(table, join_cols=["x"])
469+
470+
assert expr == Or(
471+
left=IsNull(term=Reference(name="x")),
472+
right=Or(
473+
left=IsNaN(term=Reference(name="x")),
474+
right=In(Reference(name="x"), {DoubleLiteral(1.0), DoubleLiteral(2.0), DoubleLiteral(4.0)}),
454475
),
455-
pytest.param(
456-
[{"x": 1.0}, {"x": 2.0}, {"x": None}, {"x": 4.0}, {"x": float("nan")}],
457-
Or(
458-
left=IsNull(term=Reference(name="x")),
459-
right=Or(
460-
left=IsNaN(term=Reference(name="x")),
461-
right=In(Reference(name="x"), {DoubleLiteral(1.0), DoubleLiteral(2.0), DoubleLiteral(4.0)}),
462-
),
476+
)
477+
478+
479+
def test_create_match_filter_multi_column_with_null() -> None:
480+
data = [
481+
{"x": 1.0, "y": 9.0},
482+
{"x": 2.0, "y": None},
483+
{"x": None, "y": 7.0},
484+
{"x": 4.0, "y": float("nan")},
485+
{"x": float("nan"), "y": 0.0},
486+
]
487+
schema = pa.schema([pa.field("x", pa.float64()), pa.field("y", pa.float64())])
488+
table = pa.Table.from_pylist(data, schema=schema)
489+
490+
expr = create_match_filter(table, join_cols=["x", "y"])
491+
492+
assert expr == Or(
493+
left=Or(
494+
left=And(
495+
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(1.0)),
496+
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(9.0)),
497+
),
498+
right=And(
499+
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(2.0)),
500+
right=IsNull(term=Reference(name="y")),
463501
),
464-
id="single-column-with-null",
465502
),
466-
pytest.param(
467-
[
468-
{"x": 1.0, "y": 9.0},
469-
{"x": 2.0, "y": None},
470-
{"x": None, "y": 7.0},
471-
{"x": 4.0, "y": float("nan")},
472-
{"x": float("nan"), "y": 0.0},
473-
],
474-
Or(
475-
left=Or(
476-
left=And(
477-
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(1.0)),
478-
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(9.0)),
479-
),
480-
right=And(
481-
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(2.0)),
482-
right=IsNull(term=Reference(name="y")),
483-
),
503+
right=Or(
504+
left=And(
505+
left=IsNull(term=Reference(name="x")),
506+
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(7.0)),
507+
),
508+
right=Or(
509+
left=And(
510+
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(4.0)),
511+
right=IsNaN(term=Reference(name="y")),
484512
),
485-
right=Or(
486-
left=And(
487-
left=IsNull(term=Reference(name="x")),
488-
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(7.0)),
489-
),
490-
right=Or(
491-
left=And(
492-
left=EqualTo(term=Reference(name="x"), literal=DoubleLiteral(4.0)),
493-
right=IsNaN(term=Reference(name="y")),
494-
),
495-
right=And(
496-
left=IsNaN(term=Reference(name="x")),
497-
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(0.0)),
498-
),
499-
),
513+
right=And(
514+
left=IsNaN(term=Reference(name="x")),
515+
right=EqualTo(term=Reference(name="y"), literal=DoubleLiteral(0.0)),
500516
),
501517
),
502-
id="multi-column-with-null",
503518
),
504-
],
505-
)
506-
def test_create_match_filter(data: list[dict[str, Any]], expected: BooleanExpression) -> None:
507-
schema = pa.schema([pa.field("x", pa.float64()), pa.field("y", pa.float64())])
508-
table = pa.Table.from_pylist(data, schema=schema)
509-
join_cols = sorted({col for record in data for col in record})
510-
511-
expr = create_match_filter(table, join_cols)
512-
513-
assert expr == expected
519+
)
514520

515521

516522
def test_upsert_with_duplicate_rows_in_table(catalog: Catalog) -> None:

0 commit comments

Comments
 (0)