Skip to content

REF: remove compute_reduction #32087

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

Closed
wants to merge 2 commits into from
Closed

REF: remove compute_reduction #32087

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
2 commits
Select commit Hold shift + click to select a range
a5aa938
REF: get rid of apply fastpath
jbrockmendel Feb 18, 2020
1245473
Merge branch 'master' of https://github.com/pandas-dev/pandas into no…
jbrockmendel Feb 19, 2020
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
147 changes: 0 additions & 147 deletions pandas/_libs/reduction.pyx
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -27,126 +27,6 @@ cdef _check_result_array(object obj, Py_ssize_t cnt):
raise ValueError('Function does not reduce')


cdef class Reducer:
"""
Performs generic reduction operation on a C or Fortran-contiguous ndarray
while avoiding ndarray construction overhead
"""
cdef:
Py_ssize_t increment, chunksize, nresults
object dummy, f, labels, typ, ityp, index
ndarray arr

def __init__(self, ndarray arr, object f, axis=1, dummy=None, labels=None):
n, k = (<object>arr).shape

if axis == 0:
if not arr.flags.f_contiguous:
arr = arr.copy('F')

self.nresults = k
self.chunksize = n
self.increment = n * arr.dtype.itemsize
else:
if not arr.flags.c_contiguous:
arr = arr.copy('C')

self.nresults = n
self.chunksize = k
self.increment = k * arr.dtype.itemsize

self.f = f
self.arr = arr
self.labels = labels
self.dummy, self.typ, self.index, self.ityp = self._check_dummy(
dummy=dummy)

cdef _check_dummy(self, dummy=None):
cdef:
object index = None, typ = None, ityp = None

if dummy is None:
dummy = np.empty(self.chunksize, dtype=self.arr.dtype)

# our ref is stolen later since we are creating this array
# in cython, so increment first
Py_INCREF(dummy)

else:

# we passed a Series
typ = type(dummy)
index = dummy.index
dummy = dummy.values

if dummy.dtype != self.arr.dtype:
raise ValueError('Dummy array must be same dtype')
if len(dummy) != self.chunksize:
raise ValueError(f'Dummy array must be length {self.chunksize}')

return dummy, typ, index, ityp

def get_result(self):
cdef:
char* dummy_buf
ndarray arr, result, chunk
Py_ssize_t i
flatiter it
object res, name, labels
object cached_typ = None

arr = self.arr
chunk = self.dummy
dummy_buf = chunk.data
chunk.data = arr.data
labels = self.labels

result = np.empty(self.nresults, dtype='O')
it = <flatiter>PyArray_IterNew(result)

try:
for i in range(self.nresults):

# create the cached type
# each time just reassign the data
if i == 0:

if self.typ is not None:
# In this case, we also have self.index
name = labels[i]
cached_typ = self.typ(
chunk, index=self.index, name=name, dtype=arr.dtype)

# use the cached_typ if possible
if cached_typ is not None:
# In this case, we also have non-None labels
name = labels[i]

object.__setattr__(
cached_typ._data._block, 'values', chunk)
object.__setattr__(cached_typ, 'name', name)
res = self.f(cached_typ)
else:
res = self.f(chunk)

# TODO: reason for not squeezing here?
res = _extract_result(res, squeeze=False)
if i == 0:
# On the first pass, we check the output shape to see
# if this looks like a reduction.
_check_result_array(res, len(self.dummy))

PyArray_SETITEM(result, PyArray_ITER_DATA(it), res)
chunk.data = chunk.data + self.increment
PyArray_ITER_NEXT(it)
finally:
# so we don't free the wrong memory
chunk.data = dummy_buf

result = maybe_convert_objects(result)
return result


cdef class _BaseGrouper:
cdef _check_dummy(self, dummy):
# both values and index must be an ndarray!
Expand Down Expand Up @@ -588,30 +468,3 @@ cdef class BlockSlider:
# axis=1 is the frame's axis=0
arr.data = self.base_ptrs[i]
arr.shape[1] = 0


def compute_reduction(arr: np.ndarray, f, axis: int = 0, dummy=None, labels=None):
"""

Parameters
-----------
arr : np.ndarray
f : function
axis : integer axis
dummy : type of reduced output (series)
labels : Index or None
"""

# We either have both dummy and labels, or neither of them
if (labels is None) ^ (dummy is None):
raise ValueError("Must pass either dummy and labels, or neither")

if labels is not None:
# Caller is responsible for ensuring we don't have MultiIndex
assert labels.nlevels == 1

# pass as an ndarray/ExtensionArray
labels = labels._values

reducer = Reducer(arr, f, axis=axis, dummy=dummy, labels=labels)
return reducer.get_result()
84 changes: 19 additions & 65 deletions pandas/core/apply.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -4,16 +4,10 @@

import numpy as np

from pandas._libs import reduction as libreduction
from pandas._typing import Axis
from pandas.util._decorators import cache_readonly

from pandas.core.dtypes.common import (
is_dict_like,
is_extension_array_dtype,
is_list_like,
is_sequence,
)
from pandas.core.dtypes.common import is_dict_like, is_list_like, is_sequence
from pandas.core.dtypes.generic import ABCSeries

from pandas.core.construction import create_series_with_explicit_dtype
Expand Down Expand Up @@ -221,15 +215,7 @@ def apply_empty_result(self):

def apply_raw(self):
""" apply to the values as a numpy array """
try:
result = libreduction.compute_reduction(self.values, self.f, axis=self.axis)
except ValueError as err:
if "Function does not reduce" not in str(err):
# catch only ValueError raised intentionally in libreduction
raise
# We expect np.apply_along_axis to give a two-dimensional result, or
# also raise.
result = np.apply_along_axis(self.f, self.axis, self.values)
result = np.apply_along_axis(self.f, self.axis, self.values)

# TODO: mixed type case
if result.ndim == 2:
Expand Down Expand Up @@ -265,51 +251,6 @@ def apply_broadcast(self, target: "DataFrame") -> "DataFrame":
return result

def apply_standard(self):

# try to reduce first (by default)
# this only matters if the reduction in values is of different dtype
# e.g. if we want to apply to a SparseFrame, then can't directly reduce

# we cannot reduce using non-numpy dtypes,
# as demonstrated in gh-12244
if (
self.result_type in ["reduce", None]
and not self.dtypes.apply(is_extension_array_dtype).any()
# Disallow dtypes where setting _index_data will break
# ExtensionArray values, see GH#31182
and not self.dtypes.apply(lambda x: x.kind in ["m", "M"]).any()
# Disallow complex_internals since libreduction shortcut raises a TypeError
and not self.agg_axis._has_complex_internals
):

values = self.values
index = self.obj._get_axis(self.axis)
labels = self.agg_axis
empty_arr = np.empty(len(index), dtype=values.dtype)

# Preserve subclass for e.g. test_subclassed_apply
dummy = self.obj._constructor_sliced(
empty_arr, index=index, dtype=values.dtype
)

try:
result = libreduction.compute_reduction(
values, self.f, axis=self.axis, dummy=dummy, labels=labels
)
except ValueError as err:
if "Function does not reduce" not in str(err):
# catch only ValueError raised intentionally in libreduction
raise
except TypeError:
# e.g. test_apply_ignore_failures we just ignore
if not self.ignore_failures:
raise
except ZeroDivisionError:
# reached via numexpr; fall back to python implementation
pass
else:
return self.obj._constructor_sliced(result, index=labels)

# compute the result using the series generator
results, res_index = self.apply_series_generator()

Expand Down Expand Up @@ -350,9 +291,23 @@ def wrap_results(
from pandas import Series

# see if we can infer the results
if len(results) > 0 and 0 in results and is_sequence(results[0]):
if (
len(results) > 0
and 0 in results
and is_sequence(results[0])
and (not isinstance(results[0], dict) or self.result_type == "expand")
):

return self.wrap_results_for_axis(results, res_index)
try:
return self.wrap_results_for_axis(results, res_index)
except ValueError as err:
# See e.g. test_agg_listlike_result
# FIXME: kludge for ragged array
if "arrays must all be same length" in str(err):
# Fall back to constructing Series
pass
else:
raise

# dict of scalars

Expand Down Expand Up @@ -440,9 +395,8 @@ def wrap_results_for_axis(

# we have a non-series and don't want inference
elif not isinstance(results[0], ABCSeries):
from pandas import Series

result = Series(results)
result = self.obj._constructor_sliced(results)
result.index = res_index

# we may want to infer results
Expand Down
36 changes: 1 addition & 35 deletions pandas/tests/groupby/test_bin_groupby.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -5,7 +5,7 @@

from pandas.core.dtypes.common import ensure_int64

from pandas import Index, Series, isna
from pandas import Series, isna
import pandas._testing as tm


Expand Down Expand Up @@ -111,37 +111,3 @@ def _ohlc(group):

class TestMoments:
pass


class TestReducer:
def test_int_index(self):
arr = np.random.randn(100, 4)

msg = "Must pass either dummy and labels, or neither"
# we must pass either both labels and dummy, or neither
with pytest.raises(ValueError, match=msg):
libreduction.compute_reduction(arr, np.sum, labels=Index(np.arange(4)))

with pytest.raises(ValueError, match=msg):
libreduction.compute_reduction(
arr, np.sum, axis=1, labels=Index(np.arange(100))
)

dummy = Series(0.0, index=np.arange(100))
result = libreduction.compute_reduction(
arr, np.sum, dummy=dummy, labels=Index(np.arange(4))
)
expected = arr.sum(0)
tm.assert_almost_equal(result, expected)

dummy = Series(0.0, index=np.arange(4))
result = libreduction.compute_reduction(
arr, np.sum, axis=1, dummy=dummy, labels=Index(np.arange(100))
)
expected = arr.sum(1)
tm.assert_almost_equal(result, expected)

result = libreduction.compute_reduction(
arr, np.sum, axis=1, dummy=dummy, labels=Index(np.arange(100))
)
tm.assert_almost_equal(result, expected)