Merge branch 'main' into bug_46673

Author: kapiliyer

doc/source/user_guide/sparse.rst

@@ -266,8 +266,8 @@ have no replacement.
 
 .. _sparse.scipysparse:
 
-Interaction with scipy.sparse
------------------------------
+Interaction with *scipy.sparse*
+-------------------------------
 
 Use :meth:`DataFrame.sparse.from_spmatrix` to create a :class:`DataFrame` with sparse values from a sparse matrix.
 

doc/source/user_guide/timeseries.rst

@@ -388,7 +388,7 @@ We subtract the epoch (midnight at January 1, 1970 UTC) and then floor divide by
 
 .. _timeseries.origin:
 
-Using the ``origin`` Parameter
+Using the ``origin`` parameter
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 Using the ``origin`` parameter, one can specify an alternative starting point for creation
@@ -1523,7 +1523,7 @@ or calendars with additional rules.
 
 .. _timeseries.advanced_datetime:
 
-Time series-related instance methods
+Time Series-related instance methods
 ------------------------------------
 
 Shifting / lagging
@@ -2601,7 +2601,7 @@ Transform nonexistent times to ``NaT`` or shift the times.
 
 .. _timeseries.timezone_series:
 
-Time zone series operations
+Time zone Series operations
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~
 
 A :class:`Series` with time zone **naive** values is

doc/source/user_guide/visualization.rst

@@ -3,7 +3,7 @@
 {{ header }}
 
 *******************
-Chart Visualization
+Chart visualization
 *******************
 
 This section demonstrates visualization through charting. For information on
@@ -1746,7 +1746,7 @@ Andrews curves charts:
 
    plt.close("all")
 
-Plotting directly with matplotlib
+Plotting directly with Matplotlib
 ---------------------------------
 
 In some situations it may still be preferable or necessary to prepare plots

doc/source/user_guide/window.rst

@@ -3,7 +3,7 @@
 {{ header }}
 
 ********************
-Windowing Operations
+Windowing operations
 ********************
 
 pandas contains a compact set of APIs for performing windowing operations - an operation that performs
@@ -490,7 +490,7 @@ For all supported aggregation functions, see :ref:`api.functions_expanding`.
 
 .. _window.exponentially_weighted:
 
-Exponentially Weighted window
+Exponentially weighted window
 -----------------------------
 
 An exponentially weighted window is similar to an expanding window but with each prior point

doc/source/whatsnew/v1.5.0.rst

@@ -1028,6 +1028,7 @@ Reshaping
 Sparse
 ^^^^^^
 - Bug in :meth:`Series.where` and :meth:`DataFrame.where` with ``SparseDtype`` failing to retain the array's ``fill_value`` (:issue:`45691`)
+- Bug in :meth:`SparseArray.unique` fails to keep original elements order (:issue:`47809`)
 -
 
 ExtensionArray

pandas/core/arrays/sparse/array.py

@@ -821,7 +821,7 @@ def shift(self: SparseArrayT, periods: int = 1, fill_value=None) -> SparseArrayT
 
     def _first_fill_value_loc(self):
         """
-        Get the location of the first missing value.
+        Get the location of the first fill value.
 
         Returns
         -------
@@ -834,14 +834,24 @@ def _first_fill_value_loc(self):
         if not len(indices) or indices[0] > 0:
             return 0
 
-        diff = indices[1:] - indices[:-1]
-        return np.searchsorted(diff, 2) + 1
+        # a number larger than 1 should be appended to
+        # the last in case of fill value only appears
+        # in the tail of array
+        diff = np.r_[np.diff(indices), 2]
+        return indices[(diff > 1).argmax()] + 1
 
     def unique(self: SparseArrayT) -> SparseArrayT:
         uniques = algos.unique(self.sp_values)
-        fill_loc = self._first_fill_value_loc()
-        if fill_loc >= 0:
-            uniques = np.insert(uniques, fill_loc, self.fill_value)
+        if len(self.sp_values) != len(self):
+            fill_loc = self._first_fill_value_loc()
+            # Inorder to align the behavior of pd.unique or
+            # pd.Series.unique, we should keep the original
+            # order, here we use unique again to find the
+            # insertion place. Since the length of sp_values
+            # is not large, maybe minor performance hurt
+            # is worthwhile to the correctness.
+            insert_loc = len(algos.unique(self.sp_values[:fill_loc]))
+            uniques = np.insert(uniques, insert_loc, self.fill_value)
         return type(self)._from_sequence(uniques, dtype=self.dtype)
 
     def _values_for_factorize(self):

pandas/tests/arrays/sparse/test_array.py

@@ -391,23 +391,36 @@ def test_setting_fill_value_updates():
 
 
 @pytest.mark.parametrize(
-    "arr, loc",
+    "arr,fill_value,loc",
     [
-        ([None, 1, 2], 0),
-        ([0, None, 2], 1),
-        ([0, 1, None], 2),
-        ([0, 1, 1, None, None], 3),
-        ([1, 1, 1, 2], -1),
-        ([], -1),
+        ([None, 1, 2], None, 0),
+        ([0, None, 2], None, 1),
+        ([0, 1, None], None, 2),
+        ([0, 1, 1, None, None], None, 3),
+        ([1, 1, 1, 2], None, -1),
+        ([], None, -1),
+        ([None, 1, 0, 0, None, 2], None, 0),
+        ([None, 1, 0, 0, None, 2], 1, 1),
+        ([None, 1, 0, 0, None, 2], 2, 5),
+        ([None, 1, 0, 0, None, 2], 3, -1),
+        ([None, 0, 0, 1, 2, 1], 0, 1),
+        ([None, 0, 0, 1, 2, 1], 1, 3),
     ],
 )
-def test_first_fill_value_loc(arr, loc):
-    result = SparseArray(arr)._first_fill_value_loc()
+def test_first_fill_value_loc(arr, fill_value, loc):
+    result = SparseArray(arr, fill_value=fill_value)._first_fill_value_loc()
     assert result == loc
 
 
 @pytest.mark.parametrize(
-    "arr", [[1, 2, np.nan, np.nan], [1, np.nan, 2, np.nan], [1, 2, np.nan]]
+    "arr",
+    [
+        [1, 2, np.nan, np.nan],
+        [1, np.nan, 2, np.nan],
+        [1, 2, np.nan],
+        [np.nan, 1, 0, 0, np.nan, 2],
+        [np.nan, 0, 0, 1, 2, 1],
+    ],
 )
 @pytest.mark.parametrize("fill_value", [np.nan, 0, 1])
 def test_unique_na_fill(arr, fill_value):

pandas/tests/scalar/timedelta/test_timedelta.py

@@ -14,6 +14,7 @@
     iNaT,
 )
 from pandas._libs.tslibs.dtypes import NpyDatetimeUnit
+from pandas.compat import IS64
 from pandas.errors import OutOfBoundsTimedelta
 
 import pandas as pd
@@ -690,6 +691,7 @@ def test_round_implementation_bounds(self):
         with pytest.raises(OverflowError, match=msg):
             Timedelta.max.ceil("s")
 
+    @pytest.mark.xfail(not IS64, reason="Failing on 32 bit build", strict=False)
     @given(val=st.integers(min_value=iNaT + 1, max_value=lib.i8max))
     @pytest.mark.parametrize(
         "method", [Timedelta.round, Timedelta.floor, Timedelta.ceil]

pandas/tests/scalar/timestamp/test_unary_ops.py

@@ -21,6 +21,7 @@
 )
 from pandas._libs.tslibs.dtypes import NpyDatetimeUnit
 from pandas._libs.tslibs.period import INVALID_FREQ_ERR_MSG
+from pandas.compat import IS64
 import pandas.util._test_decorators as td
 
 import pandas._testing as tm
@@ -297,6 +298,7 @@ def test_round_implementation_bounds(self):
         with pytest.raises(OverflowError, match=msg):
             Timestamp.max.ceil("s")
 
+    @pytest.mark.xfail(not IS64, reason="Failing on 32 bit build", strict=False)
     @given(val=st.integers(iNaT + 1, lib.i8max))
     @pytest.mark.parametrize(
         "method", [Timestamp.round, Timestamp.floor, Timestamp.ceil]

web/pandas/config.yml

@@ -118,17 +118,27 @@ sponsors:
     url: https://www.twosigma.com/
     logo: /static/img/partners/two_sigma.svg
     kind: partner
-    description: "Phillip Cloud, Jeff Reback"
-  - name: "Ursa Labs"
-    url: https://ursalabs.org/
-    logo: /static/img/partners/ursa_labs.svg
+    description: "Jeff Reback"
+  - name: "Voltron Data"
+    url: https://voltrondata.com/
+    logo: /static/img/partners/voltron_data.svg
     kind: partner
-    description: "Wes McKinney, Joris Van den Bossche"
+    description: "Joris Van den Bossche"
   - name: "d-fine GmbH"
     url: https://www.d-fine.com/en/
     logo: /static/img/partners/dfine.svg
     kind: partner
     description: "Patrick Hoefler"
+  - name: "Quansight"
+    url: https://quansight.com/
+    logo: /static/img/partners/quansight_labs.svg
+    kind: partner
+    description: "Marco Gorelli"
+  - name: "Nvidia"
+    url: https://www.nvidia.com
+    logo: /static/img/partners/nvidia.svg
+    kind: partner
+    description: "Matthew Roeschke"
   - name: "Tidelift"
     url: https://tidelift.com
     logo: /static/img/partners/tidelift.svg
@@ -139,6 +149,11 @@ sponsors:
     logo: /static/img/partners/czi.svg
     kind: regular
     description: "<i>pandas</i> is funded by the Essential Open Source Software for Science program of the Chan Zuckerberg Initiative. The funding is used for general maintenance, improve extension types, and a efficient string type."
+  - name: "Bodo"
+    url: https://www.bodo.ai/
+    logo: /static/img/partners/bodo.svg
+    kind: regular
+    description: "Bodo's parallel computing platform uses pandas API, and Bodo financially supports pandas development to help improve pandas, in particular the pandas API"
   inkind:  # not included in active so they don't appear in the home page
   - name: "OVH"
     url: https://us.ovhcloud.com/
@@ -152,10 +167,13 @@ sponsors:
     kind: partner
   - name: "Anaconda"
     url: https://www.anaconda.com/
-    logo: /static/img/partners/anaconda.svg
     kind: partner
   - name: "RStudio"
     url: https://www.rstudio.com/
-    logo: /static/img/partners/r_studio.svg
     kind: partner
-    description: "Wes McKinney"
+  - name: "Ursa Labs"
+    url: https://ursalabs.org/
+    kind: partner
+  - name: "Gousto"
+    url: https://www.gousto.co.uk/
+    kind: partner