API: Added ExtensionArray constructor from scalars (pandas-dev#19913)

Author: TomAugspurger

pandas/core/arrays/base.py

@@ -18,6 +18,7 @@ class ExtensionArray(object):
     The interface includes the following abstract methods that must be
     implemented by subclasses:
 
+    * _constructor_from_sequence
     * __getitem__
     * __len__
     * dtype
@@ -56,6 +57,25 @@ class ExtensionArray(object):
     # '_typ' is for pandas.core.dtypes.generic.ABCExtensionArray.
     # Don't override this.
     _typ = 'extension'
+
+    # ------------------------------------------------------------------------
+    # Constructors
+    # ------------------------------------------------------------------------
+    @classmethod
+    def _constructor_from_sequence(cls, scalars):
+        """Construct a new ExtensionArray from a sequence of scalars.
+
+        Parameters
+        ----------
+        scalars : Sequence
+            Each element will be an instance of the scalar type for this
+            array, ``cls.dtype.type``.
+        Returns
+        -------
+        ExtensionArray
+        """
+        raise AbstractMethodError(cls)
+
     # ------------------------------------------------------------------------
     # Must be a Sequence
     # ------------------------------------------------------------------------

pandas/core/arrays/categorical.py

@@ -364,6 +364,10 @@ def __init__(self, values, categories=None, ordered=None, dtype=None,
         self._dtype = self._dtype.update_dtype(dtype)
         self._codes = coerce_indexer_dtype(codes, dtype.categories)
 
+    @classmethod
+    def _constructor_from_sequence(cls, scalars):
+        return cls(scalars)
+
     @property
     def categories(self):
         """The categories of this categorical.

pandas/tests/extension/base/constructors.py

@@ -9,6 +9,11 @@
 
 class BaseConstructorsTests(BaseExtensionTests):
 
+    def test_array_from_scalars(self, data):
+        scalars = [data[0], data[1], data[2]]
+        result = data._constructor_from_sequence(scalars)
+        assert isinstance(result, type(data))
+
     def test_series_constructor(self, data):
         result = pd.Series(data)
         assert result.dtype == data.dtype

pandas/tests/extension/decimal/array.py

@@ -32,6 +32,10 @@ def __init__(self, values):
 
         self.values = values
 
+    @classmethod
+    def _constructor_from_sequence(cls, scalars):
+        return cls(scalars)
+
     def __getitem__(self, item):
         if isinstance(item, numbers.Integral):
             return self.values[item]

pandas/tests/extension/json/array.py

@@ -33,11 +33,17 @@ def __init__(self, values):
                 raise TypeError
         self.data = values
 
+    @classmethod
+    def _constructor_from_sequence(cls, scalars):
+        return cls(scalars)
+
     def __getitem__(self, item):
         if isinstance(item, numbers.Integral):
             return self.data[item]
         elif isinstance(item, np.ndarray) and item.dtype == 'bool':
-            return type(self)([x for x, m in zip(self, item) if m])
+            return self._constructor_from_sequence([
+                x for x, m in zip(self, item) if m
+            ])
         else:
             return type(self)(self.data[item])
 
@@ -77,7 +83,7 @@ def isna(self):
     def take(self, indexer, allow_fill=True, fill_value=None):
         output = [self.data[loc] if loc != -1 else self._na_value
                   for loc in indexer]
-        return type(self)(output)
+        return self._constructor_from_sequence(output)
 
     def copy(self, deep=False):
         return type(self)(self.data[:])