✅ test with other dtypes

Author: Marco Gorelli

doc/source/whatsnew/v1.1.0.rst

@@ -125,7 +125,8 @@ Numeric
 Conversion
 ^^^^^^^^^^
 - Bug in :class:`Series` construction from NumPy array with big-endian ``datetime64`` dtype (:issue:`29684`)
-- Bug in :meth:`DataFrame.replace` was changing other columns' dtypes when values in one column were being replaced with ``NaN`` (:issue:`30512`)
+- Bug in :meth:`DataFrame.replace` was casting all columns of the same dtype to the target value's dtype, even for columns which didn't contain
+ the value to be replaced (:issue:`30512`)
 -
 
 Strings

pandas/tests/frame/methods/test_replace.py

@@ -1357,9 +1357,82 @@ def test_replace_replacer_dtype(self, replacer):
  expected = pd.DataFrame([replacer])
  tm.assert_frame_equal(result, expected)
 
- def test_replace_with_nan(self):
+ def test_replace_real_dont_change_other_columns(self, any_real_dtype):
  # GH30512
- df = pd.DataFrame({"A": [np.nan, 3], "B": [1.096, 1.086], "C": [1, 2]})
- result = df.replace(to_replace={np.nan: None}).dtypes.drop("A")
+ dtype = any_real_dtype
+ value_a = 2
+ value_b = 3
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_string_dont_change_other_columns(self, string_dtype):
+ # GH30512
+ dtype = string_dtype
+ value_a = "a"
+ value_b = "b"
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_datetime64_dont_change_other_columns(self, datetime64_dtype):
+ # GH30512
+ dtype = datetime64_dtype
+ value_a = pd.Timestamp("2020-01-24")
+ value_b = pd.Timestamp("2020-01-25")
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_timedelta64_dont_change_other_columns(self, timedelta64_dtype):
+ # GH30512
+ dtype = timedelta64_dtype
+ value_a = pd.Timedelta(3)
+ value_b = pd.Timedelta(4)
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_bool_dont_change_other_columns(self):
+ # GH30512
+ dtype = bool
+ value_a = 1
+ value_b = 0
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_complex_dont_change_other_columns(self, complex_dtype):
+ # GH30512
+ dtype = complex_dtype
+ value_a = 1 + 2j
+ value_b = 1 + 3j
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_bytes_dont_change_other_columns(self, bytes_dtype):
+ # GH30512
+ dtype = bytes_dtype
+ value_a = b"a"
+ value_b = b"b"
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
+ expected = df.dtypes.drop("A")
+ tm.assert_series_equal(result, expected)
+
+ def test_replace_tz_aware_dont_change_other_columns(self):
+ # GH30512
+ dtype = "datetime64[ns, US/Eastern]"
+ value_a = Timestamp("20130102", tz="US/Eastern")
+ value_b = Timestamp("20130103", tz="US/Eastern")
+ df = pd.DataFrame({"A": [value_a], "B": [value_b]}, dtype=dtype)
+ result = df.replace(to_replace={value_a: None}).dtypes.drop("A")
  expected = df.dtypes.drop("A")
  tm.assert_series_equal(result, expected)