Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

fix(python): ensure explicit "return_dtype" is respected by map_dicts #12436

Merged
merged 1 commit into from
Nov 14, 2023
Merged

fix(python): ensure explicit "return_dtype" is respected by map_dicts #12436

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
109 changes: 18 additions & 91 deletions py-polars/polars/expr/expr.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -9260,12 +9260,9 @@ def _remap_key_or_value_series(
raise ValueError(
f"remapping values for `map_dict` could not be converted to {dtype!r} without losing values in the conversion"
)

return s

# Use two functions to save unneeded work.
# This factors out allocations and branches.
def inner_with_default(s: Series) -> Series:
def inner_func(s: Series, default_value: Any = None) -> Series:
# Convert Series to:
# - multicolumn DataFrame, if Series is a Struct.
# - one column DataFrame in other cases.
Expand All @@ -9287,7 +9284,6 @@ def inner_with_default(s: Series) -> Series:
if return_dtype is None and isinstance(default, Expr)
else return_dtype
)

remap_key_s = _remap_key_or_value_series(
name=remap_key_column,
values=remapping.keys(),
Expand All @@ -9296,7 +9292,6 @@ def inner_with_default(s: Series) -> Series:
dtype_keys=input_dtype,
is_keys=True,
)

if return_dtype_:
# Create remap value Series with specified output dtype.
remap_value_s = pl.Series(
Expand All @@ -9318,97 +9313,29 @@ def inner_with_default(s: Series) -> Series:
is_keys=False,
)

default_parsed = self._from_pyexpr(
parse_as_expression(default, str_as_lit=True)
)
return (
(
df.lazy()
.join(
pl.DataFrame(
[
remap_key_s,
remap_value_s,
]
)
.lazy()
.with_columns(F.lit(True).alias(is_remapped_column)),
how="left",
left_on=column,
right_on=remap_key_column,
)
.select(
F.when(F.col(is_remapped_column).is_not_null())
.then(F.col(remap_value_column))
.otherwise(default_parsed)
.alias(column)
)
)
.collect(no_optimization=True)
.to_series()
remap_frame = pl.LazyFrame(data=[remap_key_s, remap_value_s]).with_columns(
F.lit(True).alias(is_remapped_column)
)

def inner(s: Series) -> Series:
column = s.name
input_dtype = s.dtype
remap_key_column = f"__POLARS_REMAP_KEY_{column}"
remap_value_column = f"__POLARS_REMAP_VALUE_{column}"
is_remapped_column = f"__POLARS_REMAP_IS_REMAPPED_{column}"

remap_key_s = _remap_key_or_value_series(
name=remap_key_column,
values=list(remapping.keys()),
dtype=input_dtype,
dtype_if_empty=input_dtype,
dtype_keys=input_dtype,
is_keys=True,
mapped = df.lazy().join(
other=remap_frame, how="left", left_on=column, right_on=remap_key_column
)

if return_dtype:
# Create remap value Series with specified output dtype.
remap_value_s = pl.Series(
remap_value_column,
remapping.values(),
dtype=return_dtype,
dtype_if_empty=input_dtype,
)
if default_value is None:
result_index = 1
else:
# Create remap value Series with same output dtype as remap key Series,
# if possible (if both are integers, both are floats or remap value
# Series is pl.Utf8 and remap key Series is pl.Categorical).
remap_value_s = _remap_key_or_value_series(
name=remap_value_column,
values=remapping.values(),
dtype=None,
dtype_if_empty=input_dtype,
dtype_keys=input_dtype,
is_keys=False,
expr_default = parse_as_expression(default_value, str_as_lit=True)
default_parsed = self._from_pyexpr(expr_default)
mapped = mapped.select(
F.when(F.col(is_remapped_column).is_not_null())
.then(F.col(remap_value_column))
.otherwise(default_parsed)
.alias(column)
)
result_index = 0

return (
(
s.to_frame()
.lazy()
.join(
pl.DataFrame(
[
remap_key_s,
remap_value_s,
]
)
.lazy()
.with_columns(F.lit(True).alias(is_remapped_column)),
how="left",
left_on=column,
right_on=remap_key_column,
)
)
.collect(no_optimization=True)
.to_series(1)
)
return mapped.collect(no_optimization=True).to_series(index=result_index)

func = inner_with_default if default is not None else inner
return self.map_batches(func)
remapping_func = partial(inner_func, default_value=default)
return self.map_batches(function=remapping_func, return_dtype=return_dtype)

@deprecate_renamed_function("map_batches", version="0.19.0")
def map(
Expand Down
18 changes: 18 additions & 0 deletions py-polars/tests/unit/test_exprs.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -768,6 +768,24 @@ def test_map_dict() -> None:
),
)

lf = pl.LazyFrame({"a": [1, 2, 3]})
assert_frame_equal(
lf.select(
pl.col("a").cast(pl.UInt8).map_dict({1: 11, 2: 22}, default=99)
).collect(),
pl.DataFrame({"a": [11, 22, 99]}, schema_overrides={"a": pl.UInt8}),
)

df = (
pl.LazyFrame({"a": ["one", "two"]})
.with_columns(pl.col("a").map_dict({"one": 1}, return_dtype=pl.UInt32))
.fill_null(999)
.collect()
)
assert_frame_equal(
df, pl.DataFrame({"a": [1, 999]}, schema_overrides={"a": pl.UInt32})
)


def test_lit_dtypes() -> None:
def lit_series(value: Any, dtype: pl.PolarsDataType | None) -> pl.Series:
Expand Down