Fix Series.clip not to create a new DataFrame. (#1525)

Currently `Series.clip` creates a new DataFrame, so assigning back to the anchor DataFrame can't be executed without the option "compute.ops_on_diff_frames":

```py
>>> kdf = ks.DataFrame({"a": [1, 2, 6, 4, 4, 6, 4, 3, 7]})
>>> kdf['clip'] = kdf.a.clip(3, 6)
Traceback (most recent call last):
...
ValueError: Cannot combine the series or dataframe because it comes from a different dataframe. In order to allow this operation, enable 'compute.ops_on_diff_frames' option.
```

Author: ueshin

databricks/koalas/frame.py

@@ -46,14 +46,9 @@
 from pyspark.sql.readwriter import OptionUtils
 from pyspark.sql.types import (
     BooleanType,
-    ByteType,
-    DecimalType,
     DoubleType,
     FloatType,
-    IntegerType,
-    LongType,
     NumericType,
-    ShortType,
     StructType,
     StructField,
 )
@@ -5728,28 +5723,7 @@ def clip(self, lower: Union[float, int] = None, upper: Union[float, int] = None)
         if lower is None and upper is None:
             return self
 
-        numeric_types = (
-            DecimalType,
-            DoubleType,
-            FloatType,
-            ByteType,
-            IntegerType,
-            LongType,
-            ShortType,
-        )
-
-        def op(kser):
-            if isinstance(kser.spark_type, numeric_types):
-                scol = kser.spark_column
-                if lower is not None:
-                    scol = F.when(scol < lower, lower).otherwise(scol)
-                if upper is not None:
-                    scol = F.when(scol > upper, upper).otherwise(scol)
-                return scol.alias(kser._internal.data_spark_column_names[0])
-            else:
-                return kser
-
-        return self._apply_series_op(op)
+        return self._apply_series_op(lambda kser: kser.clip(lower=lower, upper=upper))
 
     def head(self, n: int = 5) -> "DataFrame":
         """

databricks/koalas/series.py

@@ -37,10 +37,11 @@
     BooleanType,
     DoubleType,
     FloatType,
+    IntegerType,
+    LongType,
+    NumericType,
     StringType,
     StructType,
-    LongType,
-    IntegerType,
 )
 from pyspark.sql.window import Window
 
@@ -1726,7 +1727,23 @@ def clip(self, lower: Union[float, int] = None, upper: Union[float, int] = None)
         instances of 'str' and 'int'" while `ks.Series(['a', 'b']).clip(0, 1)` will output the
         original Series, simply ignoring the incompatible types.
         """
-        return first_series(self.to_dataframe().clip(lower, upper))
+        if is_list_like(lower) or is_list_like(upper):
+            raise ValueError(
+                "List-like value are not supported for 'lower' and 'upper' at the " + "moment"
+            )
+
+        if lower is None and upper is None:
+            return self
+
+        if isinstance(self.spark_type, NumericType):
+            scol = self.spark_column
+            if lower is not None:
+                scol = F.when(scol < lower, lower).otherwise(scol)
+            if upper is not None:
+                scol = F.when(scol > upper, upper).otherwise(scol)
+            return self._with_new_scol(scol.alias(self._internal.data_spark_column_names[0]))
+        else:
+            return self
 
     def drop(
         self,

databricks/koalas/tests/test_dataframe.py

@@ -1387,6 +1387,10 @@ def test_clip(self):
         # Assert lower and upper
         self.assert_eq(kdf.clip(1, 3), pdf.clip(1, 3))
 
+        pdf["clip"] = pdf.A.clip(lower=1, upper=3)
+        kdf["clip"] = kdf.A.clip(lower=1, upper=3)
+        self.assert_eq(kdf, pdf)
+
         # Assert behavior on string values
         str_kdf = ks.DataFrame({"A": ["a", "b", "c"]}, index=np.random.rand(3))
         self.assert_eq(str_kdf.clip(1, 3), str_kdf)