ai-content-maker/.venv/Lib/site-packages/pandas/tests/series/methods/test_compare.py

142 lines
4.5 KiB
Python
Raw Normal View History

2024-05-03 04:18:51 +03:00
import numpy as np
import pytest
import pandas as pd
import pandas._testing as tm
@pytest.mark.parametrize("align_axis", [0, 1, "index", "columns"])
def test_compare_axis(align_axis):
# GH#30429
s1 = pd.Series(["a", "b", "c"])
s2 = pd.Series(["x", "b", "z"])
result = s1.compare(s2, align_axis=align_axis)
if align_axis in (1, "columns"):
indices = pd.Index([0, 2])
columns = pd.Index(["self", "other"])
expected = pd.DataFrame(
[["a", "x"], ["c", "z"]], index=indices, columns=columns
)
tm.assert_frame_equal(result, expected)
else:
indices = pd.MultiIndex.from_product([[0, 2], ["self", "other"]])
expected = pd.Series(["a", "x", "c", "z"], index=indices)
tm.assert_series_equal(result, expected)
@pytest.mark.parametrize(
"keep_shape, keep_equal",
[
(True, False),
(False, True),
(True, True),
# False, False case is already covered in test_compare_axis
],
)
def test_compare_various_formats(keep_shape, keep_equal):
s1 = pd.Series(["a", "b", "c"])
s2 = pd.Series(["x", "b", "z"])
result = s1.compare(s2, keep_shape=keep_shape, keep_equal=keep_equal)
if keep_shape:
indices = pd.Index([0, 1, 2])
columns = pd.Index(["self", "other"])
if keep_equal:
expected = pd.DataFrame(
[["a", "x"], ["b", "b"], ["c", "z"]], index=indices, columns=columns
)
else:
expected = pd.DataFrame(
[["a", "x"], [np.nan, np.nan], ["c", "z"]],
index=indices,
columns=columns,
)
else:
indices = pd.Index([0, 2])
columns = pd.Index(["self", "other"])
expected = pd.DataFrame(
[["a", "x"], ["c", "z"]], index=indices, columns=columns
)
tm.assert_frame_equal(result, expected)
def test_compare_with_equal_nulls():
# We want to make sure two NaNs are considered the same
# and dropped where applicable
s1 = pd.Series(["a", "b", np.nan])
s2 = pd.Series(["x", "b", np.nan])
result = s1.compare(s2)
expected = pd.DataFrame([["a", "x"]], columns=["self", "other"])
tm.assert_frame_equal(result, expected)
def test_compare_with_non_equal_nulls():
# We want to make sure the relevant NaNs do not get dropped
s1 = pd.Series(["a", "b", "c"])
s2 = pd.Series(["x", "b", np.nan])
result = s1.compare(s2, align_axis=0)
indices = pd.MultiIndex.from_product([[0, 2], ["self", "other"]])
expected = pd.Series(["a", "x", "c", np.nan], index=indices)
tm.assert_series_equal(result, expected)
def test_compare_multi_index():
index = pd.MultiIndex.from_arrays([[0, 0, 1], [0, 1, 2]])
s1 = pd.Series(["a", "b", "c"], index=index)
s2 = pd.Series(["x", "b", "z"], index=index)
result = s1.compare(s2, align_axis=0)
indices = pd.MultiIndex.from_arrays(
[[0, 0, 1, 1], [0, 0, 2, 2], ["self", "other", "self", "other"]]
)
expected = pd.Series(["a", "x", "c", "z"], index=indices)
tm.assert_series_equal(result, expected)
def test_compare_unaligned_objects():
# test Series with different indices
msg = "Can only compare identically-labeled Series objects"
with pytest.raises(ValueError, match=msg):
ser1 = pd.Series([1, 2, 3], index=["a", "b", "c"])
ser2 = pd.Series([1, 2, 3], index=["a", "b", "d"])
ser1.compare(ser2)
# test Series with different lengths
msg = "Can only compare identically-labeled Series objects"
with pytest.raises(ValueError, match=msg):
ser1 = pd.Series([1, 2, 3])
ser2 = pd.Series([1, 2, 3, 4])
ser1.compare(ser2)
def test_compare_datetime64_and_string():
# Issue https://github.com/pandas-dev/pandas/issues/45506
# Catch OverflowError when comparing datetime64 and string
data = [
{"a": "2015-07-01", "b": "08335394550"},
{"a": "2015-07-02", "b": "+49 (0) 0345 300033"},
{"a": "2015-07-03", "b": "+49(0)2598 04457"},
{"a": "2015-07-04", "b": "0741470003"},
{"a": "2015-07-05", "b": "04181 83668"},
]
dtypes = {"a": "datetime64[ns]", "b": "string"}
df = pd.DataFrame(data=data).astype(dtypes)
result_eq1 = df["a"].eq(df["b"])
result_eq2 = df["a"] == df["b"]
result_neq = df["a"] != df["b"]
expected_eq = pd.Series([False] * 5) # For .eq and ==
expected_neq = pd.Series([True] * 5) # For !=
tm.assert_series_equal(result_eq1, expected_eq)
tm.assert_series_equal(result_eq2, expected_eq)
tm.assert_series_equal(result_neq, expected_neq)