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.

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

refactor(python): Update Series.to_numpy to handle Decimal/Time types in Rust #14296

Merged
merged 3 commits into from
Feb 7, 2024
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
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
12 changes: 3 additions & 9 deletions py-polars/polars/series/series.py
Original file line number Diff line number Diff line change
Expand Up @@ -4354,20 +4354,14 @@ def temporal_dtype_to_numpy(dtype: PolarsDataType) -> Any:
if (
use_pyarrow
and _PYARROW_AVAILABLE
and dtype != Object
and (dtype == Time or not dtype.is_temporal())
and dtype not in (Object, Datetime, Duration, Date)
):
return self.to_arrow().to_numpy(
zero_copy_only=zero_copy_only, writable=writable
)

if dtype in (Time, Decimal):
# There are no native NumPy "time" or "decimal" dtypes
raise_no_zero_copy()
return np.array(self.to_list(), dtype="object", copy=False)

if self.null_count() == 0:
if dtype.is_numeric():
if dtype.is_integer() or dtype.is_float():
np_array = self._view(ignore_nulls=True)
elif dtype == Boolean:
raise_no_zero_copy()
Expand All @@ -4386,7 +4380,7 @@ def temporal_dtype_to_numpy(dtype: PolarsDataType) -> Any:
else:
raise_no_zero_copy()
np_array = self._s.to_numpy()
if dtype.is_temporal():
if dtype in (Datetime, Duration, Date):
np_dtype = temporal_dtype_to_numpy(dtype)
np_array = np_array.view(np_dtype)

Expand Down
70 changes: 41 additions & 29 deletions py-polars/src/conversion/chunked_array.rs
Original file line number Diff line number Diff line change
Expand Up @@ -141,16 +141,21 @@ impl ToPyObject for Wrap<&DatetimeChunked> {

impl ToPyObject for Wrap<&TimeChunked> {
fn to_object(&self, py: Python) -> PyObject {
let utils = UTILS.as_ref(py);
let convert = utils.getattr(intern!(py, "_to_python_time")).unwrap();
let iter = self
.0
.into_iter()
.map(|opt_v| opt_v.map(|v| convert.call1((v,)).unwrap()));
let iter = time_to_pyobject_iter(py, self.0);
PyList::new(py, iter).into_py(py)
}
}

pub(crate) fn time_to_pyobject_iter<'a>(
py: Python<'a>,
ca: &'a TimeChunked,
) -> impl ExactSizeIterator<Item = Option<&'a PyAny>> {
let utils = UTILS.as_ref(py);
let convert = utils.getattr(intern!(py, "_to_python_time")).unwrap();
ca.0.into_iter()
.map(|opt_v| opt_v.map(|v| convert.call1((v,)).unwrap()))
}

impl ToPyObject for Wrap<&DateChunked> {
fn to_object(&self, py: Python) -> PyObject {
let utils = UTILS.as_ref(py);
Expand All @@ -165,29 +170,36 @@ impl ToPyObject for Wrap<&DateChunked> {

impl ToPyObject for Wrap<&DecimalChunked> {
fn to_object(&self, py: Python) -> PyObject {
let utils = UTILS.as_ref(py);
let convert = utils.getattr(intern!(py, "_to_python_decimal")).unwrap();
let py_scale = (-(self.0.scale() as i32)).to_object(py);
// if we don't know precision, the only safe bet is to set it to 39
let py_precision = self.0.precision().unwrap_or(39).to_object(py);
let iter = self.0.into_iter().map(|opt_v| {
opt_v.map(|v| {
// TODO! use AnyValue so that we have a single impl.
const N: usize = 3;
let mut buf = [0_u128; N];
let n_digits = decimal_to_digits(v.abs(), &mut buf);
let buf = unsafe {
std::slice::from_raw_parts(
buf.as_slice().as_ptr() as *const u8,
N * std::mem::size_of::<u128>(),
)
};
let digits = PyTuple::new(py, buf.iter().take(n_digits));
convert
.call1((v.is_negative() as u8, digits, &py_precision, &py_scale))
.unwrap()
})
});
let iter = decimal_to_pyobject_iter(py, self.0);
PyList::new(py, iter).into_py(py)
}
}

pub(crate) fn decimal_to_pyobject_iter<'a>(
py: Python<'a>,
ca: &'a DecimalChunked,
) -> impl ExactSizeIterator<Item = Option<&'a PyAny>> {
let utils = UTILS.as_ref(py);
let convert = utils.getattr(intern!(py, "_to_python_decimal")).unwrap();
let py_scale = (-(ca.scale() as i32)).to_object(py);
// if we don't know precision, the only safe bet is to set it to 39
let py_precision = ca.precision().unwrap_or(39).to_object(py);
ca.into_iter().map(move |opt_v| {
opt_v.map(|v| {
// TODO! use AnyValue so that we have a single impl.
const N: usize = 3;
let mut buf = [0_u128; N];
let n_digits = decimal_to_digits(v.abs(), &mut buf);
let buf = unsafe {
std::slice::from_raw_parts(
buf.as_slice().as_ptr() as *const u8,
N * std::mem::size_of::<u128>(),
)
};
let digits = PyTuple::new(py, buf.iter().take(n_digits));
convert
.call1((v.is_negative() as u8, digits, &py_precision, &py_scale))
.unwrap()
})
})
}
2 changes: 1 addition & 1 deletion py-polars/src/conversion/mod.rs
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
pub(crate) mod any_value;
mod chunked_array;
pub(crate) mod chunked_array;

use std::fmt::{Display, Formatter};
use std::hash::{Hash, Hasher};
Expand Down
13 changes: 13 additions & 0 deletions py-polars/src/series/export.rs
Original file line number Diff line number Diff line change
Expand Up @@ -4,6 +4,7 @@ use polars_core::prelude::*;
use pyo3::prelude::*;
use pyo3::types::PyList;

use crate::conversion::chunked_array::{decimal_to_pyobject_iter, time_to_pyobject_iter};
use crate::error::PyPolarsErr;
use crate::prelude::{ObjectValue, *};
use crate::{arrow_interop, raise_err, PySeries};
Expand Down Expand Up @@ -185,6 +186,12 @@ impl PySeries {
},
Date => date_series_to_numpy(py, s),
Datetime(_, _) | Duration(_) => temporal_series_to_numpy(py, s),
Time => {
let ca = s.time().unwrap();
let iter = time_to_pyobject_iter(py, ca);
let np_arr = PyArray1::from_iter(py, iter.map(|v| v.into_py(py)));
np_arr.into_py(py)
},
String => {
let ca = s.str().unwrap();
let np_arr = PyArray1::from_iter(py, ca.into_iter().map(|s| s.into_py(py)));
Expand All @@ -200,6 +207,12 @@ impl PySeries {
let np_arr = PyArray1::from_iter(py, ca.iter_str().map(|s| s.into_py(py)));
np_arr.into_py(py)
},
Decimal(_, _) => {
let ca = s.decimal().unwrap();
let iter = decimal_to_pyobject_iter(py, ca);
let np_arr = PyArray1::from_iter(py, iter.map(|v| v.into_py(py)));
np_arr.into_py(py)
},
#[cfg(feature = "object")]
Object(_, _) => {
let ca = s
Expand Down
Loading