Skip to content
This repository was archived by the owner on Feb 2, 2024. It is now read-only.

Df.at impl #738

Merged
AlexanderKalistratov merged 18 commits into from
Apr 23, 2020
Merged

Df.at impl #738

Show file tree
Hide file tree
Changes from 3 commits
Commits
Show all changes
18 commits
Select commit Hold shift + click to select a range
94a7a60
Df.at impl
Mar 19, 2020
cf1be41
fix check col
Mar 20, 2020
3f2db5c
pep
Mar 20, 2020
0fb45ea
fix limits+algo
Mar 26, 2020
20d2a35
pep
Mar 26, 2020
ecce368
Fix type checking
Apr 14, 2020
8947a62
fix attr
Apr 14, 2020
5f2d566
at return value (not array)
Apr 14, 2020
e98a1c2
add check type
Apr 14, 2020
928dae8
Merge branch 'master' into dfat
1e-to Apr 16, 2020
e39efb3
fix raise errors
Apr 16, 2020
a5d0662
add new example codegen
Apr 16, 2020
304ec98
add case of many idx
Apr 17, 2020
9ad570b
update example
Apr 17, 2020
c01a329
Add limitation
Apr 20, 2020
261b5b0
Merge branch 'master' into dfat
1e-to Apr 22, 2020
76ffabf
pep
Apr 22, 2020
71be518
Merge remote-tracking branch 'origin/dfat' into dfat
Apr 22, 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
39 changes: 39 additions & 0 deletions examples/dataframe/dataframe_at.py
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -0,0 +1,39 @@
# *****************************************************************************
# Copyright (c) 2020, Intel Corporation All rights reserved.
#
# Redistribution and use in source and binary forms, with or without
# modification, are permitted provided that the following conditions are met:
#
# Redistributions of source code must retain the above copyright notice,
# this list of conditions and the following disclaimer.
#
# Redistributions in binary form must reproduce the above copyright notice,
# this list of conditions and the following disclaimer in the documentation
# and/or other materials provided with the distribution.
#
# THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS "AS IS"
# AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT LIMITED TO,
# THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR
# PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT HOLDER OR
# CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL, SPECIAL,
# EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT LIMITED TO,
# PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS;
# OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY,
# WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT (INCLUDING NEGLIGENCE OR
# OTHERWISE) ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE,
# EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
# *****************************************************************************


import pandas as pd
from numba import njit


@njit
def dataframe_at():
df = pd.DataFrame({'A': [1.0, 2.0, 3.0, 1.0], 'B': [4, 5, 6, 7], 'C': ['a', 'b', 'c', 'd']})

return df.at[1, 'C'] # ['b']


print(dataframe_at())
112 changes: 111 additions & 1 deletion sdc/datatypes/hpat_pandas_dataframe_functions.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -38,7 +38,7 @@

from pandas.core.indexing import IndexingError

from numba import types
from numba import types, prange
from numba.special import literally
from numba.typed import List, Dict
from numba.errors import TypingError
Expand Down Expand Up @@ -1759,13 +1759,76 @@ def _df_getitem_unicode_idx_impl(self, idx):
ty_checker.raise_exc(idx, expected_types, 'idx')


def df_getitem_tuple_at_codegen(self, row, col):
"""
Example of generated implementation:
def _df_getitem_tuple_at_impl(self, idx):
row, _ = idx
check_row = False
for i in prange(len(self._dataframe.index)):

@AlexanderKalistratov AlexanderKalistratov Apr 16, 2020

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Do we actually need this? Wouldn't it be checked in Series.at?

if self._dataframe.index[i] == row:
check_row = True
if check_row:
data = self._dataframe._data[2]
res_data = pandas.Series(data, index=self._dataframe.index)
return res_data.at[row]
raise IndexingError('Index is out of bounds for axis')

@AlexanderKalistratov AlexanderKalistratov Apr 16, 2020

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

For which axis?

"""
func_lines = ['def _df_getitem_tuple_at_impl(self, idx):']
check_col = False
for i in range(len(self.columns)):
if self.columns[i] == col:
check_col = True
col_idx = i
if check_col == True: # noqa
func_lines += [' row, _ = idx',
' check_row = False',
' for i in prange(len(self._dataframe.index)):',
' if self._dataframe.index[i] == row:',
' check_row = True',
' if check_row:',
f' data = self._dataframe._data[{col_idx}]',
' res_data = pandas.Series(data, index=self._dataframe.index)',
' return res_data.at[row]',
" raise IndexingError('Index is out of bounds for axis')"]

else:
raise IndexingError('Index is out of bounds for axis')

@densmirn densmirn Mar 26, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I propose to change the code a bit

Suggested change
check_col = False
for i in range(len(self.columns)):
if self.columns[i] == col:
check_col = True
col_idx = i
if check_col == True: # noqa
func_lines += [' row, _ = idx',
' check_row = False',
' for i in prange(len(self._dataframe.index)):',
' if self._dataframe.index[i] == row:',
' check_row = True',
' if check_row:',
f' data = self._dataframe._data[{col_idx}]',
' res_data = pandas.Series(data, index=self._dataframe.index)',
' return res_data.at[row]',
" raise IndexingError('Index is out of bounds for axis')"]
else:
raise IndexingError('Index is out of bounds for axis')
for i in range(len(self.columns)):
if self.columns[i] == col:
func_lines += [
' row, _ = idx',
' check_row = False',
' for i in prange(len(self._dataframe.index)):',
' if self._dataframe.index[i] == row:',
' check_row = True',
' if check_row:',
f' data = self._dataframe._data[{i}]',
' res_data = pandas.Series(data, index=self._dataframe.index)',
' return res_data.at[row]',
' raise IndexingError("Index is out of bounds for axis")'
]
break
else:
raise IndexingError('Index is out of bounds for axis')


func_text = '\n'.join(func_lines)

global_vars = {'pandas': pandas,
'get_dataframe_data': get_dataframe_data,
'prange': prange,
'IndexingError': IndexingError}

return func_text, global_vars


def gen_df_getitem_tuple_at_impl(self, row, col):
func_text, global_vars = df_getitem_tuple_at_codegen(self, row, col)
loc_vars = {}
exec(func_text, global_vars, loc_vars)
_reduce_impl = loc_vars['_df_getitem_tuple_at_impl']

return _reduce_impl


@sdc_overload(operator.getitem)
def sdc_pandas_dataframe_accessor_getitem(self, idx):
if not isinstance(self, DataFrameGetitemAccessorType):
return None

accessor = self.accessor.literal_value

if accessor == 'at':
if isinstance(idx, types.Tuple) and isinstance(idx[1], types.Literal):
row = idx[0]

@AlexanderKalistratov AlexanderKalistratov Apr 10, 2020

Copy link
Copy Markdown
Collaborator

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Need to check that first item in idx is of the same type as index and second item is string

col = idx[1].literal_value

@densmirn densmirn Mar 26, 2020

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Shouldn't we add limitation about column supported only to be literal?

return gen_df_getitem_tuple_at_impl(self.dataframe, row, col)

raise TypingError('Operator getitem(). The index must be a row and literal column. Given: {}'.format(idx))

if accessor == 'iat':
if isinstance(idx, types.Tuple) and isinstance(idx[1], types.Literal):
col = idx[1].literal_value
Expand Down Expand Up @@ -1836,6 +1899,53 @@ def sdc_pandas_dataframe_iat_impl(self):
return sdc_pandas_dataframe_iat_impl


@sdc_overload_attribute(DataFrameType, 'at')
def sdc_pandas_dataframe_at(self):
"""
Intel Scalable Dataframe Compiler User Guide
********************************************

Pandas API: pandas.DataFrame.at

Examples
--------
.. literalinclude:: ../../../examples/dataframe/dataframe_at.py
:language: python
:lines: 28-
:caption: Access a single value for a row/column label pair.
:name: ex_dataframe_at

.. command-output:: python ./dataframe/dataframe_at.py
:cwd: ../../../examples

.. seealso::

:ref:`DataFrame.iat <pandas.DataFrame.iat>`
Access a single value for a row/column pair by integer position.

:ref:`DataFrame.loc <pandas.DataFrame.loc>`
Access a group of rows and columns by label(s).

:ref:`Series.at <pandas.Series.at>`
Access a single value using a label.

Intel Scalable Dataframe Compiler Developer Guide
*************************************************
Pandas DataFrame method :meth:`pandas.DataFrame.at` implementation.

.. only:: developer
Test: python -m sdc.runtests -k sdc.tests.test_dataframe.TestDataFrame.test_df_at*
"""

ty_checker = TypeChecker('Attribute at().')
ty_checker.check(self, DataFrameType)

def sdc_pandas_dataframe_at_impl(self):
return dataframe_getitem_accessor_init(self, 'at')

return sdc_pandas_dataframe_at_impl


@sdc_overload_method(DataFrameType, 'pct_change')
def pct_change_overload(df, periods=1, fill_method='pad', limit=None, freq=None):
"""
Expand Down
27 changes: 27 additions & 0 deletions sdc/tests/test_dataframe.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -1100,6 +1100,33 @@ def test_impl(df):
msg = 'Index is out of bounds for axis'
self.assertIn(msg, str(raises.exception))

def test_df_at(self):
def test_impl(df, n):
return df.at[n, 'C']

sdc_func = sdc.jit(test_impl)
idx = [3, 4, 1, 2, 0]
n_cases = [0, 2]
df = pd.DataFrame({"A": [3.2, 4.4, 7.0, 3.3, 1.0],
"B": [3, 4, 1, 0, 222],
"C": ['a', 'dd', 'c', '12', 'ddf']}, index=idx)
for n in n_cases:
self.assertEqual(sdc_func(df, n), test_impl(df, n))

def test_df_at_value_error(self):
def test_impl(df):
return df.at[1, 'D']
sdc_func = sdc.jit(test_impl)
idx = [3, 4, 1, 2, 0]
df = pd.DataFrame({"A": [3.2, 4.4, 7.0, 3.3, 1.0],
"B": [3, 4, 1, 0, 222],
"C": [3, 4, 2, 6, 1]}, index=idx)

with self.assertRaises(TypingError) as raises:
sdc_func(df)
msg = 'Index is out of bounds for axis'
self.assertIn(msg, str(raises.exception))

def test_df_head(self):
def get_func(n):
def impl(a):
Expand Down