From 287dced7c23c3861333f30cbcfeea306211725e8 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 1 Dec 2021 15:27:20 -0800
Subject: [PATCH 01/32] create new pr

---
 python/cudf/cudf/core/dataframe.py | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index c0cb6f1917f..c35ce1153df 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2633,6 +2633,13 @@ def insert(self, loc, name, value):
 
         self._data.insert(name, value, loc=loc)
 
+    def diff(self):
+        """
+        Calculates the difference of a Dataframe element compared with
+        another element in the Dataframe (default is element in previous row).
+        """
+        pass
+
     def drop(
         self,
         labels=None,

From 721cbaa5e8ef661f4125fe8b7d3c4175bda4d3c2 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 1 Dec 2021 17:40:50 -0800
Subject: [PATCH 02/32] docstrings

---
 python/cudf/cudf/core/dataframe.py | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index c35ce1153df..ea3e858bff5 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2636,7 +2636,21 @@ def insert(self, loc, name, value):
     def diff(self):
         """
         Calculates the difference of a Dataframe element compared with
-        another element in the Dataframe (default is element in previous row).
+        another element in the Dataframe, treating each column independently
+
+        Parameters
+        ----------
+        periods : int, default 1
+            Periods to shift for calculating difference,
+            accepts negative values.
+        axis : {0 or ‘index’, 1 or ‘columns’}, default 0
+            Take difference over rows (0) or columns (1).
+            Only row-wise (0) shift is supported.
+
+        Returns
+        -------
+        DataFrame
+            First differences of the DataFrame.
         """
         pass
 

From 4aa42c92497d3890cd50eae5a984d97b42652e38 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 1 Dec 2021 18:57:00 -0800
Subject: [PATCH 03/32] added df.diff() method. need to add tests

---
 python/cudf/cudf/core/dataframe.py | 5 +++--
 1 file changed, 3 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index ea3e858bff5..6dcace87674 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2633,7 +2633,7 @@ def insert(self, loc, name, value):
 
         self._data.insert(name, value, loc=loc)
 
-    def diff(self):
+    def diff(self, periods=1, axis=0):
         """
         Calculates the difference of a Dataframe element compared with
         another element in the Dataframe, treating each column independently
@@ -2652,7 +2652,8 @@ def diff(self):
         DataFrame
             First differences of the DataFrame.
         """
-        pass
+        result = self - self.shift(periods=periods)
+        return result
 
     def drop(
         self,

From e245a63826eec56c1f48277dabe0e06792391b51 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 2 Dec 2021 20:55:14 -0800
Subject: [PATCH 04/32] added checks for null values and non-numeric dtypes and
 axis

---
 python/cudf/cudf/core/dataframe.py | 23 +++++++++++++++++++++++
 1 file changed, 23 insertions(+)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 6dcace87674..3642d4ff493 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2651,7 +2651,30 @@ def diff(self, periods=1, axis=0):
         -------
         DataFrame
             First differences of the DataFrame.
+
+        Notes
+        -----
+        Diff currently only supports float and integer dtype columns with
+        no null values.
+
+        Examples
+        --------
+
         """
+
+        if not axis == 0:
+            raise NotImplementedError("Only axis=0 is supported.")
+
+        if self.isna().any().any():
+            raise AssertionError(
+                "Diff currently requires columns with no null values"
+            )
+
+        if not np.issubdtype(self.dtypes[0], np.number):
+            raise NotImplementedError(
+                "Diff currently only supports numeric dtypes"
+            )
+
         result = self - self.shift(periods=periods)
         return result
 

From 0f2bc48088fdf35e8ecae6571a07bc6812d96c68 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 2 Dec 2021 21:00:02 -0800
Subject: [PATCH 05/32] added tests, all passing. ready for initial review

---
 python/cudf/cudf/tests/test_dataframe.py | 47 ++++++++++++++++++++++++
 1 file changed, 47 insertions(+)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index d555b5c4033..f99232e7d87 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9039,3 +9039,50 @@ def test_pearson_corr_multiindex_dataframe():
     expected = gdf.to_pandas().groupby(level="a").corr("pearson")
 
     assert_eq(expected, actual)
+
+
+@pytest.mark.parametrize(
+    "data",
+    [
+        np.random.normal(-100, 100, 1000),
+        np.random.randint(-50, 50, 1000),
+        np.random.random_sample((4, 4)),
+        np.random.uniform(10.5, 75.5, (10,)),
+        np.array([1.123, 2.343, 5.890, 0.0]),
+    ],
+)
+@pytest.mark.parametrize("periods", [-1, -2, -3, -4, 1, 2, 3, 4])
+def test_diff_dataframe_valid(data, periods):
+    gdf = cudf.DataFrame(data)
+    pdf = gdf.to_pandas()
+
+    actual = gdf.diff(periods=periods, axis=0)
+    expected = pdf.diff(periods=periods, axis=0)
+
+    assert_eq(
+        expected, actual, check_dtype=False,
+    )
+
+
+def test_diff_dataframe_invalid_axis():
+    with pytest.raises(NotImplementedError, match="Only axis=0 is supported."):
+        gdf = cudf.DataFrame(np.random.random_sample((4, 4)))
+        gdf.diff(periods=1, axis=1)
+
+
+def test_diff_dataframe_null_columns():
+    with pytest.raises(
+        AssertionError,
+        match="Diff currently requires columns with no null values",
+    ):
+        gdf = cudf.DataFrame([1.123, 2.343, np.nan, None, 6.072, None])
+        gdf.diff(periods=4, axis=0)
+
+
+def test_diff_dataframe_non_numeric_dtypes():
+    with pytest.raises(
+        NotImplementedError,
+        match="Diff currently only supports numeric dtypes",
+    ):
+        gdf = cudf.DataFrame(["a", "b", "c", "d", "e"])
+        gdf.diff(periods=2, axis=0)

From 2b81389376ad7c1be2aaed9ff063bd74cbcdac57 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 2 Dec 2021 21:05:13 -0800
Subject: [PATCH 06/32] added example to doctrings

---
 python/cudf/cudf/core/dataframe.py | 20 ++++++++++++++++++++
 1 file changed, 20 insertions(+)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 3642d4ff493..57a20e89dd8 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2659,6 +2659,26 @@ def diff(self, periods=1, axis=0):
 
         Examples
         --------
+        >>> import cudf
+        >>> gdf = cudf.DataFrame({'a': [1, 2, 3, 4, 5, 6],
+                                  'b': [1, 1, 2, 3, 5, 8],
+                                  'c': [1, 4, 9, 16, 25, 36]})
+        >>> gdf
+        a  b   c
+        0  1  1   1
+        1  2  1   4
+        2  3  2   9
+        3  4  3  16
+        4  5  5  25
+        5  6  8  36
+        >>> gdf.diff(periods=2)
+            a     b     c
+        0  <NA>  <NA>  <NA>
+        1  <NA>  <NA>  <NA>
+        2     2     1     8
+        3     2     2    12
+        4     2     3    16
+        5     2     5    20
 
         """
 

From 7cdb55a4a98dd793cbb062865e8cc90c7cafd360 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 2 Dec 2021 21:45:01 -0800
Subject: [PATCH 07/32] minor edits to tests

---
 python/cudf/cudf/tests/test_dataframe.py | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index f99232e7d87..252ce696662 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9044,10 +9044,10 @@ def test_pearson_corr_multiindex_dataframe():
 @pytest.mark.parametrize(
     "data",
     [
-        np.random.normal(-100, 100, 1000),
-        np.random.randint(-50, 50, 1000),
+        np.random.normal(-100, 100, (50, 50)),
+        np.random.randint(-50, 50, (25, 30)),
         np.random.random_sample((4, 4)),
-        np.random.uniform(10.5, 75.5, (10,)),
+        np.random.uniform(10.5, 75.5, (10, 6)),
         np.array([1.123, 2.343, 5.890, 0.0]),
     ],
 )

From 1e22293487f0f428e0736aa7c899a72380f37a49 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 2 Dec 2021 22:05:54 -0800
Subject: [PATCH 08/32] .

---
 python/cudf/cudf/tests/test_dataframe.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 252ce696662..29a1c4fba25 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9051,7 +9051,7 @@ def test_pearson_corr_multiindex_dataframe():
         np.array([1.123, 2.343, 5.890, 0.0]),
     ],
 )
-@pytest.mark.parametrize("periods", [-1, -2, -3, -4, 1, 2, 3, 4])
+@pytest.mark.parametrize("periods", range(-4, 5))
 def test_diff_dataframe_valid(data, periods):
     gdf = cudf.DataFrame(data)
     pdf = gdf.to_pandas()

From 3bd80173171e3c979ed22ce683f6e9f2a99b420f Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 8 Dec 2021 11:22:39 -0800
Subject: [PATCH 09/32] checks dtypes for all columns in a dataframe

---
 python/cudf/cudf/core/dataframe.py | 16 ++++++++++++++--
 1 file changed, 14 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 6f8825265d3..aa0a62b429b 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2690,12 +2690,24 @@ def diff(self, periods=1, axis=0):
                 "Diff currently requires columns with no null values"
             )
 
-        if not np.issubdtype(self.dtypes[0], np.number):
+        if not (np.issubdtype(i, np.number) for i in self.dtypes):
             raise NotImplementedError(
                 "Diff currently only supports numeric dtypes"
             )
 
-        result = self - self.shift(periods=periods)
+        try:
+            result = self - self.shift(periods=periods)
+        except TypeError as e:
+            if (
+                "sub operator not supported between"
+                "<class 'cudf.core.column.string.StringColumn'>"
+                "and <class 'cudf.core.column.string.StringColumn'>" in str(e)
+            ):
+                raise NotImplementedError(
+                    "Diff currently only supports numeric dtypes"
+                )
+            raise
+
         return result
 
     def drop(

From 496e837a5bc1e8ae1c9bdac013102116d4c54847 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 8 Dec 2021 12:12:13 -0800
Subject: [PATCH 10/32] split long string to multiple lines

---
 python/cudf/cudf/core/dataframe.py | 7 ++++---
 1 file changed, 4 insertions(+), 3 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 90ccd8ea30a..19e4f7bf9aa 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2699,9 +2699,10 @@ def diff(self, periods=1, axis=0):
             result = self - self.shift(periods=periods)
         except TypeError as e:
             if (
-                "sub operator not supported between"
-                "<class 'cudf.core.column.string.StringColumn'>"
-                "and <class 'cudf.core.column.string.StringColumn'>" in str(e)
+                "sub operator not supported between\
+                    <class 'cudf.core.column.string.StringColumn'> \
+                    and <class 'cudf.core.column.string.StringColumn'>"
+                in str(e)
             ):
                 raise NotImplementedError(
                     "Diff currently only supports numeric dtypes"

From 2767b753870d44fec4a4f90a0761e7c0d0b4e6bc Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 8 Dec 2021 14:55:09 -0800
Subject: [PATCH 11/32] addressed review: use has_nulls to check for nans

---
 python/cudf/cudf/core/dataframe.py       |  9 ++-------
 python/cudf/cudf/tests/test_dataframe.py | 13 +++----------
 2 files changed, 5 insertions(+), 17 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 19e4f7bf9aa..ad05bc5d32a 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2685,7 +2685,7 @@ def diff(self, periods=1, axis=0):
         if not axis == 0:
             raise NotImplementedError("Only axis=0 is supported.")
 
-        if self.isna().any().any():
+        if (self[col].has_nulls for col in self._data) is True:
             raise AssertionError(
                 "Diff currently requires columns with no null values"
             )
@@ -2698,12 +2698,7 @@ def diff(self, periods=1, axis=0):
         try:
             result = self - self.shift(periods=periods)
         except TypeError as e:
-            if (
-                "sub operator not supported between\
-                    <class 'cudf.core.column.string.StringColumn'> \
-                    and <class 'cudf.core.column.string.StringColumn'>"
-                in str(e)
-            ):
+            if "sub operator not supported" in str(e):
                 raise NotImplementedError(
                     "Diff currently only supports numeric dtypes"
                 )
diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index a24c0e0cf55..32629185875 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9069,10 +9069,12 @@ def test_dataframe_add_suffix():
         np.random.random_sample((4, 4)),
         np.random.uniform(10.5, 75.5, (10, 6)),
         np.array([1.123, 2.343, 5.890, 0.0]),
+        [np.nan, None, np.nan, None],
+        {"a": [1.123, 2.343, np.nan, np.nan], "b": [None, 3, 9.08, None]},
     ],
 )
 @pytest.mark.parametrize("periods", range(-4, 5))
-def test_diff_dataframe_valid(data, periods):
+def test_diff_dataframe(data, periods):
     gdf = cudf.DataFrame(data)
     pdf = gdf.to_pandas()
 
@@ -9090,15 +9092,6 @@ def test_diff_dataframe_invalid_axis():
         gdf.diff(periods=1, axis=1)
 
 
-def test_diff_dataframe_null_columns():
-    with pytest.raises(
-        AssertionError,
-        match="Diff currently requires columns with no null values",
-    ):
-        gdf = cudf.DataFrame([1.123, 2.343, np.nan, None, 6.072, None])
-        gdf.diff(periods=4, axis=0)
-
-
 def test_diff_dataframe_non_numeric_dtypes():
     with pytest.raises(
         NotImplementedError,

From 72d14bd16ca5e4fd12d0b4c5cd0be7ef2f54f333 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 8 Dec 2021 15:01:59 -0800
Subject: [PATCH 12/32] removed nan-constraints check

---
 python/cudf/cudf/core/dataframe.py | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index ad05bc5d32a..05c314a50db 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2654,8 +2654,7 @@ def diff(self, periods=1, axis=0):
 
         Notes
         -----
-        Diff currently only supports float and integer dtype columns with
-        no null values.
+        Diff currently only supports float and integer dtype columns
 
         Examples
         --------
@@ -2685,11 +2684,6 @@ def diff(self, periods=1, axis=0):
         if not axis == 0:
             raise NotImplementedError("Only axis=0 is supported.")
 
-        if (self[col].has_nulls for col in self._data) is True:
-            raise AssertionError(
-                "Diff currently requires columns with no null values"
-            )
-
         if not (np.issubdtype(i, np.number) for i in self.dtypes):
             raise NotImplementedError(
                 "Diff currently only supports numeric dtypes"

From 798c642c8d4b64baa226bdb2605c2534627a36cb Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Thu, 9 Dec 2021 14:51:21 -0800
Subject: [PATCH 13/32] added tests for mix of numeric and non-numeric dtypes

---
 python/cudf/cudf/tests/test_dataframe.py | 15 +++++++++++++--
 1 file changed, 13 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 32629185875..7c883f7c34f 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9092,10 +9092,21 @@ def test_diff_dataframe_invalid_axis():
         gdf.diff(periods=1, axis=1)
 
 
-def test_diff_dataframe_non_numeric_dtypes():
+@pytest.mark.parametrize(
+    "data",
+    [
+        {
+            "int_col": [1, 2, 3, 4, 5],
+            "float_col": [1.0, 2.0, 3.0, 4.0, 5.0],
+            "string_col": ["a", "b", "c", "d", "e"],
+        },
+        ["a", "b", "c", "d", "e"],
+    ],
+)
+def test_diff_dataframe_numeric_and_non_numeric_dypes(data):
     with pytest.raises(
         NotImplementedError,
         match="Diff currently only supports numeric dtypes",
     ):
-        gdf = cudf.DataFrame(["a", "b", "c", "d", "e"])
+        gdf = cudf.DataFrame(data)
         gdf.diff(periods=2, axis=0)

From df2140e6a150259d26e6226f36bd42574d0a04d3 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Fri, 10 Dec 2021 13:46:24 -0800
Subject: [PATCH 14/32] addressed reviews by brandon

---
 python/cudf/cudf/core/dataframe.py | 11 ++---------
 1 file changed, 2 insertions(+), 9 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 05c314a50db..d6de31089f1 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2684,19 +2684,12 @@ def diff(self, periods=1, axis=0):
         if not axis == 0:
             raise NotImplementedError("Only axis=0 is supported.")
 
-        if not (np.issubdtype(i, np.number) for i in self.dtypes):
+        if not all(is_numeric_dtype(i) for i in self.dtypes):
             raise NotImplementedError(
                 "Diff currently only supports numeric dtypes"
             )
 
-        try:
-            result = self - self.shift(periods=periods)
-        except TypeError as e:
-            if "sub operator not supported" in str(e):
-                raise NotImplementedError(
-                    "Diff currently only supports numeric dtypes"
-                )
-            raise
+        result = self - self.shift(periods=periods)
 
         return result
 

From 91412257b4cb4991ff17b4232acf98b1b1365031 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Mon, 13 Dec 2021 20:20:23 -0800
Subject: [PATCH 15/32] numeric types docs- fix

---
 python/cudf/cudf/core/dataframe.py | 6 ++----
 1 file changed, 2 insertions(+), 4 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index d6de31089f1..ce10f0aefdb 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2654,7 +2654,7 @@ def diff(self, periods=1, axis=0):
 
         Notes
         -----
-        Diff currently only supports float and integer dtype columns
+        Diff currently only supports numeric dtype columns
 
         Examples
         --------
@@ -2689,9 +2689,7 @@ def diff(self, periods=1, axis=0):
                 "Diff currently only supports numeric dtypes"
             )
 
-        result = self - self.shift(periods=periods)
-
-        return result
+        return self - self.shift(periods=periods)
 
     def drop(
         self,

From 79211f16a9dfc52d7fb9ed7516798e8c915ad9ac Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Mon, 13 Dec 2021 20:23:53 -0800
Subject: [PATCH 16/32] added test cases for decimal64 dtypes

---
 python/cudf/cudf/tests/test_dataframe.py | 25 ++++++++++++++++++++----
 1 file changed, 21 insertions(+), 4 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 7c883f7c34f..279d60bdabc 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9064,17 +9064,16 @@ def test_dataframe_add_suffix():
 @pytest.mark.parametrize(
     "data",
     [
-        np.random.normal(-100, 100, (50, 50)),
         np.random.randint(-50, 50, (25, 30)),
         np.random.random_sample((4, 4)),
-        np.random.uniform(10.5, 75.5, (10, 6)),
         np.array([1.123, 2.343, 5.890, 0.0]),
         [np.nan, None, np.nan, None],
+        [True, False, True, False, False],
         {"a": [1.123, 2.343, np.nan, np.nan], "b": [None, 3, 9.08, None]},
     ],
 )
 @pytest.mark.parametrize("periods", range(-4, 5))
-def test_diff_dataframe(data, periods):
+def test_diff_dataframe_numeric_dtypes(data, periods):
     gdf = cudf.DataFrame(data)
     pdf = gdf.to_pandas()
 
@@ -9086,6 +9085,24 @@ def test_diff_dataframe(data, periods):
     )
 
 
+@pytest.mark.parametrize(
+    ("precision", "scale"), [(5, 2), (4, 3), (8, 5), (3, 1), (6, 4)],
+)
+def test_diff_decimal64_dtype(precision, scale):
+    gdf = cudf.DataFrame(
+        np.random.uniform(10.5, 75.5, (10, 6)),
+        dtype=cudf.Decimal64Dtype(precision=precision, scale=scale),
+    )
+    pdf = gdf.to_pandas()
+
+    actual = gdf.diff()
+    expected = pdf.diff()
+
+    assert_eq(
+        expected, actual, check_dtype=False,
+    )
+
+
 def test_diff_dataframe_invalid_axis():
     with pytest.raises(NotImplementedError, match="Only axis=0 is supported."):
         gdf = cudf.DataFrame(np.random.random_sample((4, 4)))
@@ -9103,7 +9120,7 @@ def test_diff_dataframe_invalid_axis():
         ["a", "b", "c", "d", "e"],
     ],
 )
-def test_diff_dataframe_numeric_and_non_numeric_dypes(data):
+def test_diff_dataframe_non_numeric_dypes(data):
     with pytest.raises(
         NotImplementedError,
         match="Diff currently only supports numeric dtypes",

From 8df6881056235645e568ee3ef5b14670089f7ad3 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Fri, 7 Jan 2022 12:51:46 -0800
Subject: [PATCH 17/32] wip: moved binary_operator to DecimalBaseColumn

---
 python/cudf/cudf/core/column/decimal.py | 83 +++++++++++++------------
 1 file changed, 42 insertions(+), 41 deletions(-)

diff --git a/python/cudf/cudf/core/column/decimal.py b/python/cudf/cudf/core/column/decimal.py
index 7037b8e6f36..254882383c4 100644
--- a/python/cudf/cudf/core/column/decimal.py
+++ b/python/cudf/cudf/core/column/decimal.py
@@ -45,6 +45,48 @@ def as_decimal_column(
             return self
         return libcudf.unary.cast(self, dtype)
 
+    def binary_operator(self, op, other, reflect=False):
+        if reflect:
+            self, other = other, self
+
+        # Binary Arithmetics between decimal columns. `Scale` and `precision`
+        # are computed outside of libcudf
+        if op in ("add", "sub", "mul", "div"):
+            scale = _binop_scale(self.dtype, other.dtype, op)
+            output_type = Decimal64Dtype(
+                scale=scale, precision=Decimal64Dtype.MAX_PRECISION
+            )
+            result = libcudf.binaryop.binaryop(self, other, op, output_type)
+            result.dtype.precision = _binop_precision(
+                self.dtype, other.dtype, op
+            )
+        elif op in ("eq", "ne", "lt", "gt", "le", "ge"):
+            if not isinstance(
+                other,
+                (
+                    Decimal32Column,
+                    Decimal64Column,
+                    cudf.core.column.NumericalColumn,
+                    cudf.Scalar,
+                ),
+            ):
+                raise TypeError(
+                    f"Operator {op} not supported between"
+                    f"{str(type(self))} and {str(type(other))}"
+                )
+            if isinstance(
+                other, cudf.core.column.NumericalColumn
+            ) and not is_integer_dtype(other.dtype):
+                raise TypeError(
+                    f"Only decimal and integer column is supported for {op}."
+                )
+            if isinstance(other, cudf.core.column.NumericalColumn):
+                other = other.as_decimal_column(
+                    Decimal64Dtype(Decimal64Dtype.MAX_PRECISION, 0)
+                )
+            result = libcudf.binaryop.binaryop(self, other, op, bool)
+        return result
+
 
 class Decimal32Column(DecimalBaseColumn):
     dtype: Decimal32Dtype
@@ -156,47 +198,6 @@ def to_arrow(self):
             buffers=[mask_buf, data_buf],
         )
 
-    def binary_operator(self, op, other, reflect=False):
-        if reflect:
-            self, other = other, self
-
-        # Binary Arithmetics between decimal columns. `Scale` and `precision`
-        # are computed outside of libcudf
-        if op in ("add", "sub", "mul", "div"):
-            scale = _binop_scale(self.dtype, other.dtype, op)
-            output_type = Decimal64Dtype(
-                scale=scale, precision=Decimal64Dtype.MAX_PRECISION
-            )  # precision will be ignored, libcudf has no notion of precision
-            result = libcudf.binaryop.binaryop(self, other, op, output_type)
-            result.dtype.precision = _binop_precision(
-                self.dtype, other.dtype, op
-            )
-        elif op in ("eq", "ne", "lt", "gt", "le", "ge"):
-            if not isinstance(
-                other,
-                (
-                    Decimal64Column,
-                    cudf.core.column.NumericalColumn,
-                    cudf.Scalar,
-                ),
-            ):
-                raise TypeError(
-                    f"Operator {op} not supported between"
-                    f"{str(type(self))} and {str(type(other))}"
-                )
-            if isinstance(
-                other, cudf.core.column.NumericalColumn
-            ) and not is_integer_dtype(other.dtype):
-                raise TypeError(
-                    f"Only decimal and integer column is supported for {op}."
-                )
-            if isinstance(other, cudf.core.column.NumericalColumn):
-                other = other.as_decimal_column(
-                    Decimal64Dtype(Decimal64Dtype.MAX_PRECISION, 0)
-                )
-            result = libcudf.binaryop.binaryop(self, other, op, bool)
-        return result
-
     def normalize_binop_value(self, other):
         if is_scalar(other) and isinstance(other, (int, np.int, Decimal)):
             return cudf.Scalar(Decimal(other))

From db742bbd875b5e78e803cc11d21bc801979688bc Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Fri, 7 Jan 2022 14:00:39 -0800
Subject: [PATCH 18/32] added checks for either decimal32 or decimal64

---
 python/cudf/cudf/core/column/decimal.py | 16 ++++++++++++----
 1 file changed, 12 insertions(+), 4 deletions(-)

diff --git a/python/cudf/cudf/core/column/decimal.py b/python/cudf/cudf/core/column/decimal.py
index 254882383c4..41b6f121cfb 100644
--- a/python/cudf/cudf/core/column/decimal.py
+++ b/python/cudf/cudf/core/column/decimal.py
@@ -53,9 +53,14 @@ def binary_operator(self, op, other, reflect=False):
         # are computed outside of libcudf
         if op in ("add", "sub", "mul", "div"):
             scale = _binop_scale(self.dtype, other.dtype, op)
-            output_type = Decimal64Dtype(
-                scale=scale, precision=Decimal64Dtype.MAX_PRECISION
-            )
+            if self.dtype == other.dtype == cudf.Decimal32Dtype:
+                output_type = Decimal32Dtype(
+                    scale=scale, precision=Decimal32Dtype.MAX_PRECISION
+                )
+            if self.dtype == other.dtype == cudf.Decimal64Dtype:
+                output_type = Decimal64Dtype(
+                    scale=scale, precision=Decimal64Dtype.MAX_PRECISION
+                )
             result = libcudf.binaryop.binaryop(self, other, op, output_type)
             result.dtype.precision = _binop_precision(
                 self.dtype, other.dtype, op
@@ -322,4 +327,7 @@ def _binop_precision(l_dtype, r_dtype, op):
     else:
         raise NotImplementedError()
     # TODO
-    return min(result, cudf.Decimal64Dtype.MAX_PRECISION)
+    if l_dtype == r_dtype == Decimal32Dtype:
+        return min(result, cudf.Decimal32Dtype.MAX_PRECISION)
+    if l_dtype == r_dtype == Decimal64Dtype:
+        return min(result, cudf.Decimal64Dtype.MAX_PRECISION)

From a29d1a324b020db9d6a0d93d220f9a9a1305ab74 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Tue, 18 Jan 2022 13:02:28 -0800
Subject: [PATCH 19/32] addressed michael's review comments

---
 python/cudf/cudf/core/dataframe.py       | 12 +++++++-----
 python/cudf/cudf/tests/test_dataframe.py |  4 ++--
 2 files changed, 9 insertions(+), 7 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 4c6c7a1d202..a4385237665 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2607,8 +2607,10 @@ def insert(self, loc, name, value, nan_as_null=None):
 
     def diff(self, periods=1, axis=0):
         """
-        Calculates the difference of a Dataframe element compared with
-        another element in the Dataframe, treating each column independently
+        First discrete difference of element.
+
+        Calculates the difference of a Dataframe element compared with another
+        element in the Dataframe (default is element in previous row).
 
         Parameters
         ----------
@@ -2626,7 +2628,7 @@ def diff(self, periods=1, axis=0):
 
         Notes
         -----
-        Diff currently only supports numeric dtype columns
+        Diff currently only supports numeric dtype columns.
 
         Examples
         --------
@@ -2635,7 +2637,7 @@ def diff(self, periods=1, axis=0):
                                   'b': [1, 1, 2, 3, 5, 8],
                                   'c': [1, 4, 9, 16, 25, 36]})
         >>> gdf
-        a  b   c
+           a  b   c
         0  1  1   1
         1  2  1   4
         2  3  2   9
@@ -2643,7 +2645,7 @@ def diff(self, periods=1, axis=0):
         4  5  5  25
         5  6  8  36
         >>> gdf.diff(periods=2)
-            a     b     c
+              a     b     c
         0  <NA>  <NA>  <NA>
         1  <NA>  <NA>  <NA>
         2     2     1     8
diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 6d3b49e862f..b90a64e928e 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9062,8 +9062,8 @@ def test_dataframe_add_suffix():
 @pytest.mark.parametrize(
     "data",
     [
-        np.random.randint(-50, 50, (25, 30)),
-        np.random.random_sample((4, 4)),
+        # np.random.seed(-50, 50, (25, 30)),
+        # np.random.random_sample.seed((4, 4)),
         np.array([1.123, 2.343, 5.890, 0.0]),
         [np.nan, None, np.nan, None],
         [True, False, True, False, False],

From 3dc90b0224783e80bfdb782449804c9ae20d13a5 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 19 Jan 2022 12:21:41 -0800
Subject: [PATCH 20/32] use const seed for random generated number -- cases

---
 python/cudf/cudf/tests/test_dataframe.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index b90a64e928e..b63f73c065c 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9062,8 +9062,8 @@ def test_dataframe_add_suffix():
 @pytest.mark.parametrize(
     "data",
     [
-        # np.random.seed(-50, 50, (25, 30)),
-        # np.random.random_sample.seed((4, 4)),
+        np.random.RandomState(seed=10).randint(-50, 50, (25, 30)),
+        np.random.RandomState(seed=10).random_sample((4, 4)),
         np.array([1.123, 2.343, 5.890, 0.0]),
         [np.nan, None, np.nan, None],
         [True, False, True, False, False],

From c8a424da19b4963c0ef3fec6c0e4defdc62056ac Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Mon, 24 Jan 2022 16:17:28 -0800
Subject: [PATCH 21/32] added check for periods>len(dataframe)

---
 python/cudf/cudf/core/dataframe.py | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 68bfbaf5699..c71b97428b6 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2660,6 +2660,12 @@ def diff(self, periods=1, axis=0):
                 "Diff currently only supports numeric dtypes"
             )
 
+        if abs(periods) > len(self):
+            df = cudf.DataFrame(
+                {name: ([cudf.NA] * len(self)) for name in self.columns}
+            )
+            return df
+
         return self - self.shift(periods=periods)
 
     def drop(

From b91a795f701247655124147b3846b9a5608e3c0f Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Mon, 24 Jan 2022 16:19:20 -0800
Subject: [PATCH 22/32] added test for decimal32dtype, all tests passing. ready
 for review

---
 python/cudf/cudf/tests/test_dataframe.py | 11 +++++++----
 1 file changed, 7 insertions(+), 4 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 4640cb122ea..437f96269ca 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9074,12 +9074,11 @@ def test_dataframe_add_suffix():
         np.random.RandomState(seed=10).randint(-50, 50, (25, 30)),
         np.random.RandomState(seed=10).random_sample((4, 4)),
         np.array([1.123, 2.343, 5.890, 0.0]),
-        [np.nan, None, np.nan, None],
         [True, False, True, False, False],
         {"a": [1.123, 2.343, np.nan, np.nan], "b": [None, 3, 9.08, None]},
     ],
 )
-@pytest.mark.parametrize("periods", range(-4, 5))
+@pytest.mark.parametrize("periods", range(-10, 10))
 def test_diff_dataframe_numeric_dtypes(data, periods):
     gdf = cudf.DataFrame(data)
     pdf = gdf.to_pandas()
@@ -9095,10 +9094,13 @@ def test_diff_dataframe_numeric_dtypes(data, periods):
 @pytest.mark.parametrize(
     ("precision", "scale"), [(5, 2), (4, 3), (8, 5), (3, 1), (6, 4)],
 )
-def test_diff_decimal64_dtype(precision, scale):
+@pytest.mark.parametrize(
+    "dtype", [cudf.Decimal32Dtype, cudf.Decimal64Dtype],
+)
+def test_diff_decimal_dtypes(precision, scale, dtype):
     gdf = cudf.DataFrame(
         np.random.uniform(10.5, 75.5, (10, 6)),
-        dtype=cudf.Decimal64Dtype(precision=precision, scale=scale),
+        dtype=dtype(precision=precision, scale=scale),
     )
     pdf = gdf.to_pandas()
 
@@ -9125,6 +9127,7 @@ def test_diff_dataframe_invalid_axis():
             "string_col": ["a", "b", "c", "d", "e"],
         },
         ["a", "b", "c", "d", "e"],
+        [np.nan, None, np.nan, None],
     ],
 )
 def test_diff_dataframe_non_numeric_dypes(data):

From 331945f4592445a0e3e8076890f5c21bb137f820 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Tue, 25 Jan 2022 09:21:37 -0800
Subject: [PATCH 23/32] use column_empty instead of cudf.NA to create df with
 all-nulls

Co-authored-by: Michael Wang <isVoid@users.noreply.github.com>
---
 python/cudf/cudf/core/dataframe.py | 5 ++---
 1 file changed, 2 insertions(+), 3 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index c71b97428b6..62aa2cff0f7 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2661,9 +2661,8 @@ def diff(self, periods=1, axis=0):
             )
 
         if abs(periods) > len(self):
-            df = cudf.DataFrame(
-                {name: ([cudf.NA] * len(self)) for name in self.columns}
-            )
+            df = cudf.DataFrame._from_data(
+            {name: column_empty(len(self), dtype=dtype, masked=True) for name, dtype in zip(self.columns, self.dtypes)})
             return df
 
         return self - self.shift(periods=periods)

From 559edc3a2940380da6a5c7858b56cd7eb1a5dd18 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Tue, 25 Jan 2022 12:00:36 -0800
Subject: [PATCH 24/32] apply bradley's suggestions to docstrings

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
---
 python/cudf/cudf/core/dataframe.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 62aa2cff0f7..308c9e1621e 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2606,8 +2606,8 @@ def diff(self, periods=1, axis=0):
         """
         First discrete difference of element.
 
-        Calculates the difference of a Dataframe element compared with another
-        element in the Dataframe (default is element in previous row).
+        Calculates the difference of a DataFrame element compared with another
+        element in the DataFrame (default is element in previous row).
 
         Parameters
         ----------

From f33b931c41526815a412a7ab53422c404d02ced6 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Tue, 25 Jan 2022 12:01:05 -0800
Subject: [PATCH 25/32] add dots to indicate continuation in docstring examples

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
---
 python/cudf/cudf/core/dataframe.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 308c9e1621e..147a58e4b80 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2631,8 +2631,8 @@ def diff(self, periods=1, axis=0):
         --------
         >>> import cudf
         >>> gdf = cudf.DataFrame({'a': [1, 2, 3, 4, 5, 6],
-                                  'b': [1, 1, 2, 3, 5, 8],
-                                  'c': [1, 4, 9, 16, 25, 36]})
+        ...                       'b': [1, 1, 2, 3, 5, 8],
+        ...                       'c': [1, 4, 9, 16, 25, 36]})
         >>> gdf
            a  b   c
         0  1  1   1

From 2dcabe99cbc6c5cb97019d3463c6abef6b67e026 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Tue, 25 Jan 2022 13:44:09 -0800
Subject: [PATCH 26/32] added checks for periods as integer, and axis

---
 python/cudf/cudf/core/dataframe.py | 20 +++++++++++++++-----
 1 file changed, 15 insertions(+), 5 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 147a58e4b80..2c4979bfc12 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -33,6 +33,7 @@
     is_datetime_dtype,
     is_dict_like,
     is_dtype_equal,
+    is_integer,
     is_list_dtype,
     is_list_like,
     is_numeric_dtype,
@@ -2614,7 +2615,7 @@ def diff(self, periods=1, axis=0):
         periods : int, default 1
             Periods to shift for calculating difference,
             accepts negative values.
-        axis : {0 or ‘index’, 1 or ‘columns’}, default 0
+        axis : {0 or 'index', 1 or 'columns'}, default 0
             Take difference over rows (0) or columns (1).
             Only row-wise (0) shift is supported.
 
@@ -2651,18 +2652,27 @@ def diff(self, periods=1, axis=0):
         5     2     5    20
 
         """
+        if not is_integer(periods):
+            if not (isinstance(periods, float) and isinstance(periods, int)):
+                raise ValueError("periods must be an integer")
+            periods = int(periods)
 
-        if not axis == 0:
-            raise NotImplementedError("Only axis=0 is supported.")
+        axis = self._get_axis_from_axis_arg(axis)
+        if axis != 0:
+            raise NotImplementedError("Only axis=0 is currently supported.")
 
         if not all(is_numeric_dtype(i) for i in self.dtypes):
             raise NotImplementedError(
-                "Diff currently only supports numeric dtypes"
+                "DataFrame.diff only supports numeric dtypes"
             )
 
         if abs(periods) > len(self):
             df = cudf.DataFrame._from_data(
-            {name: column_empty(len(self), dtype=dtype, masked=True) for name, dtype in zip(self.columns, self.dtypes)})
+                {
+                    name: column_empty(len(self), dtype=dtype, masked=True)
+                    for name, dtype in zip(self.columns, self.dtypes)
+                }
+            )
             return df
 
         return self - self.shift(periods=periods)

From e6b2400d933d0b91c4039c08b595968ed01c7a01 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Tue, 25 Jan 2022 13:45:05 -0800
Subject: [PATCH 27/32] minor test-fixes. ready for review

---
 python/cudf/cudf/tests/test_dataframe.py | 8 ++++----
 1 file changed, 4 insertions(+), 4 deletions(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index c1f163cbdc2..bc9f30ac587 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9084,7 +9084,7 @@ def test_dataframe_add_suffix():
         {"a": [1.123, 2.343, np.nan, np.nan], "b": [None, 3, 9.08, None]},
     ],
 )
-@pytest.mark.parametrize("periods", range(-10, 10))
+@pytest.mark.parametrize("periods", (-5, -1, 0, 1, 5))
 def test_diff_dataframe_numeric_dtypes(data, periods):
     gdf = cudf.DataFrame(data)
     pdf = gdf.to_pandas()
@@ -9105,7 +9105,7 @@ def test_diff_dataframe_numeric_dtypes(data, periods):
 )
 def test_diff_decimal_dtypes(precision, scale, dtype):
     gdf = cudf.DataFrame(
-        np.random.uniform(10.5, 75.5, (10, 6)),
+        np.random.default_rng(seed=42).uniform(10.5, 75.5, (10, 6)),
         dtype=dtype(precision=precision, scale=scale),
     )
     pdf = gdf.to_pandas()
@@ -9120,7 +9120,7 @@ def test_diff_decimal_dtypes(precision, scale, dtype):
 
 def test_diff_dataframe_invalid_axis():
     with pytest.raises(NotImplementedError, match="Only axis=0 is supported."):
-        gdf = cudf.DataFrame(np.random.random_sample((4, 4)))
+        gdf = cudf.DataFrame(np.array([1.123, 2.343, 5.890, 0.0]))
         gdf.diff(periods=1, axis=1)
 
 
@@ -9139,7 +9139,7 @@ def test_diff_dataframe_invalid_axis():
 def test_diff_dataframe_non_numeric_dypes(data):
     with pytest.raises(
         NotImplementedError,
-        match="Diff currently only supports numeric dtypes",
+        match="DataFrame.diff only supports numeric dtypes.",
     ):
         gdf = cudf.DataFrame(data)
         gdf.diff(periods=2, axis=0)

From 3d59ba2d624ee44fe8605d6cf8d1b5e8eceff2f5 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Tue, 25 Jan 2022 13:53:14 -0800
Subject: [PATCH 28/32] fixed regex issues, all tests passing now

---
 python/cudf/cudf/core/dataframe.py       | 2 +-
 python/cudf/cudf/tests/test_dataframe.py | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 2c4979bfc12..439af3de222 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -2659,7 +2659,7 @@ def diff(self, periods=1, axis=0):
 
         axis = self._get_axis_from_axis_arg(axis)
         if axis != 0:
-            raise NotImplementedError("Only axis=0 is currently supported.")
+            raise NotImplementedError("Only axis=0 is supported.")
 
         if not all(is_numeric_dtype(i) for i in self.dtypes):
             raise NotImplementedError(
diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index bc9f30ac587..417c2f76b29 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9139,7 +9139,7 @@ def test_diff_dataframe_invalid_axis():
 def test_diff_dataframe_non_numeric_dypes(data):
     with pytest.raises(
         NotImplementedError,
-        match="DataFrame.diff only supports numeric dtypes.",
+        match="DataFrame.diff only supports numeric dtypes",
     ):
         gdf = cudf.DataFrame(data)
         gdf.diff(periods=2, axis=0)

From e0722aea84b09333fb5151d3014dde805ecd3f08 Mon Sep 17 00:00:00 2001
From: Sheilah <kirui.sheilah@gmail.com>
Date: Wed, 26 Jan 2022 09:05:12 -0800
Subject: [PATCH 29/32] using pandas is_integer() and float() instead

---
 python/cudf/cudf/core/dataframe.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/python/cudf/cudf/core/dataframe.py b/python/cudf/cudf/core/dataframe.py
index 439af3de222..e973468a322 100644
--- a/python/cudf/cudf/core/dataframe.py
+++ b/python/cudf/cudf/core/dataframe.py
@@ -20,6 +20,7 @@
 from numba import cuda
 from nvtx import annotate
 from pandas._config import get_option
+from pandas.core.dtypes.common import is_float, is_integer
 from pandas.io.formats import console
 from pandas.io.formats.printing import pprint_thing
 
@@ -33,7 +34,6 @@
     is_datetime_dtype,
     is_dict_like,
     is_dtype_equal,
-    is_integer,
     is_list_dtype,
     is_list_like,
     is_numeric_dtype,
@@ -2653,7 +2653,7 @@ def diff(self, periods=1, axis=0):
 
         """
         if not is_integer(periods):
-            if not (isinstance(periods, float) and isinstance(periods, int)):
+            if not (is_float(periods) and periods.is_integer()):
                 raise ValueError("periods must be an integer")
             periods = int(periods)
 

From cdf5187166fb46cc4bccffe1b8b7bf2fcce427a8 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Thu, 3 Feb 2022 21:56:20 -0800
Subject: [PATCH 30/32] omit unnecessary extra tests.

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
---
 python/cudf/cudf/tests/test_dataframe.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index a3b7bd2373f..32511d99a98 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9094,7 +9094,7 @@ def test_diff_dataframe_numeric_dtypes(data, periods):
 
 
 @pytest.mark.parametrize(
-    ("precision", "scale"), [(5, 2), (4, 3), (8, 5), (3, 1), (6, 4)],
+    ("precision", "scale"), [(5, 2), (8, 5)],
 )
 @pytest.mark.parametrize(
     "dtype", [cudf.Decimal32Dtype, cudf.Decimal64Dtype],

From 50b5085c6b3881abc18b941c66fb970155b60aa4 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Thu, 3 Feb 2022 21:58:23 -0800
Subject: [PATCH 31/32] only use context manager around function that raises

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
---
 python/cudf/cudf/tests/test_dataframe.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index 32511d99a98..f3c2d233617 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9115,8 +9115,8 @@ def test_diff_decimal_dtypes(precision, scale, dtype):
 
 
 def test_diff_dataframe_invalid_axis():
+    gdf = cudf.DataFrame(np.array([1.123, 2.343, 5.890, 0.0]))
     with pytest.raises(NotImplementedError, match="Only axis=0 is supported."):
-        gdf = cudf.DataFrame(np.array([1.123, 2.343, 5.890, 0.0]))
         gdf.diff(periods=1, axis=1)
 
 

From 575c11857c7e00396989ff81dc740db2913338e1 Mon Sep 17 00:00:00 2001
From: Sheilah Kirui <71867292+skirui-source@users.noreply.github.com>
Date: Thu, 3 Feb 2022 21:59:14 -0800
Subject: [PATCH 32/32] context manager around function only

Co-authored-by: Bradley Dice <bdice@bradleydice.com>
---
 python/cudf/cudf/tests/test_dataframe.py | 2 +-
 1 file changed, 1 insertion(+), 1 deletion(-)

diff --git a/python/cudf/cudf/tests/test_dataframe.py b/python/cudf/cudf/tests/test_dataframe.py
index f3c2d233617..f3996d1f4bd 100644
--- a/python/cudf/cudf/tests/test_dataframe.py
+++ b/python/cudf/cudf/tests/test_dataframe.py
@@ -9133,11 +9133,11 @@ def test_diff_dataframe_invalid_axis():
     ],
 )
 def test_diff_dataframe_non_numeric_dypes(data):
+    gdf = cudf.DataFrame(data)
     with pytest.raises(
         NotImplementedError,
         match="DataFrame.diff only supports numeric dtypes",
     ):
-        gdf = cudf.DataFrame(data)
         gdf.diff(periods=2, axis=0)