compat with pandas 0.24 (#38)

* compat with pandas 0.24

* lint

* lint

* remove old envs

Author: TomAugspurger

.travis.yml

@@ -11,8 +11,6 @@ env:
     - secure: "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"
   matrix:
     - PYTHON=3.6 NUMPY=1.11
-    - PYTHON=3.5 NUMPY=1.9
-    - PYTHON=2.7 NUMPY=1.9
 before_install:
   - export PATH="$HOME/miniconda3/bin:$PATH"
 install:

cyberpandas/_utils.py

@@ -12,22 +12,19 @@ def to_bytes(n, length, byteorder='big'):
 
 
 def pack(ip):
-    # type: (int) -> bytes
     if six.PY2:
         return to_bytes(ip, length=16, byteorder='big')
     else:
         return ip.to_bytes(16, byteorder='big')
 
 
 def unpack(ip):
-    # type: (T.Tuple[int, int]) -> int, int
     # Recipe 3.5 from Python Cookbook 3rd ed. (p. 90)
     # int.from_bytes(data, 'big') for Py3+
     hi, lo = struct.unpack(">QQ", ip)
     return hi, lo
 
 
 def combine(hi, lo):
-    # type: (int, int) -> int
     """Combine the hi and lo bytes into the final ip address."""
     return (hi << 64) + lo

cyberpandas/base.py

@@ -12,8 +12,8 @@ def dtype(self):
         return self._dtype
 
     @classmethod
-    def _from_sequence(cls, scalars):
-        return cls(scalars)
+    def _from_sequence(cls, scalars, dtype=None, copy=False):
+        return cls(scalars, dtype=dtype)
 
     @classmethod
     def _from_factorized(cls, values, original):

cyberpandas/dtypes.py

@@ -2,7 +2,6 @@
 
 
 def is_ipv4(value):
-    # type: (Any) -> bool
     if isinstance(value, str):
         return value.count(".") == 3
     elif isinstance(value, bytes):
@@ -14,6 +13,5 @@ def is_ipv4(value):
 
 
 def is_ipv6(value):
-    # type: (Any) -> bool
     if isinstance(value, str):
         return value.count(":") == 7

cyberpandas/ip_array.py

@@ -29,6 +29,7 @@ class IPv4v6Base(object):
 IPv4v6Base.register(ipaddress.IPv6Address)
 
 
+@pd.api.extensions.register_extension_dtype
 class IPType(ExtensionDtype):
     name = 'ip'
     type = IPv4v6Base
@@ -44,6 +45,11 @@ def construct_from_string(cls, string):
             raise TypeError("Cannot construct a '{}' from "
                             "'{}'".format(cls, string))
 
+    @classmethod
+    def construct_array_type(cls):
+        return IPArray
+
+
 # -----------------------------------------------------------------------------
 # Extension Container
 # -----------------------------------------------------------------------------
@@ -69,15 +75,17 @@ class IPArray(NumPyBackedExtensionArrayMixin):
     ndim = 1
     can_hold_na = True
 
-    def __init__(self, values):
+    def __init__(self, values, dtype=None, copy=False):
         from .parser import _to_ip_array
 
         values = _to_ip_array(values)  # TODO: avoid potential copy
+        # TODO: dtype?
+        if copy:
+            values = values.copy()
         self.data = values
 
     @classmethod
     def from_pyints(cls, values):
-        # type: (T.Sequence[int]) -> 'IPArray'
         """Construct an IPArray from a sequence of Python integers.
 
         This can be useful for representing IPv6 addresses, which may
@@ -97,7 +105,7 @@ def from_pyints(cls, values):
 
     @classmethod
     def from_bytes(cls, bytestring):
-        """Create an IPArray from a bytestring.
+        r"""Create an IPArray from a bytestring.
 
         Parameters
         ----------
@@ -295,7 +303,7 @@ def to_pyints(self):
         return [combine(*map(int, x)) for x in self.data]
 
     def to_bytes(self):
-        """Serialize the IPArray as a Python bytestring.
+        r"""Serialize the IPArray as a Python bytestring.
 
         This and :meth:IPArray.from_bytes is the fastest way to roundtrip
         serialize and de-serialize an IPArray.
@@ -312,6 +320,13 @@ def to_bytes(self):
         """
         return self.data.tobytes()
 
+    def astype(self, dtype, copy=True):
+        if isinstance(dtype, IPType):
+            if copy:
+                self = self.copy()
+            return self
+        return super(IPArray, self).astype(dtype)
+
     # ------------------------------------------------------------------------
     # Ops
     # ------------------------------------------------------------------------
@@ -449,7 +464,6 @@ def isin(self, other):
         return mask
 
     def _isin_network(self, other):
-        # type: (Union[ipaddress.IPv4Network,ipaddress.IPv6Network]) -> ndarray
         """Check whether an array of addresses is contained in a network."""
         # A network is bounded below by 'network_address' and
         # above by 'broadcast_address'.
@@ -649,6 +663,7 @@ def mask(self, mask):
 # Accessor
 # -----------------------------------------------------------------------------
 
+
 @pd.api.extensions.register_series_accessor("ip")
 class IPAccessor:
 

cyberpandas/ip_methods.py

@@ -8,7 +8,6 @@
 
 
 def _as_int(ip):
-    # type: (Union[int, str, IPv4Address, IPv6Address]) -> int
     if isinstance(ip, six.string_types):
         ip = ipaddress.ip_address(ip)
     return int(ip)

cyberpandas/mac_array.py

@@ -3,11 +3,13 @@
 import numpy as np
 import six
 
-from pandas.api.extensions import ExtensionDtype, take
+from pandas.api.extensions import (
+    ExtensionDtype, take, register_extension_dtype)
 
 from .base import NumPyBackedExtensionArrayMixin
 
 
+@register_extension_dtype
 class MACType(ExtensionDtype):
     """Dtype for MAC Address Data."""
     name = 'mac'
@@ -24,6 +26,10 @@ def construct_from_string(cls, string):
             raise TypeError("Cannot construct a '{}' from "
                             "'{}'".format(cls, string))
 
+    @classmethod
+    def construct_array_type(cls):
+        return MACArray
+
 
 class MACArray(NumPyBackedExtensionArrayMixin):
     """Array for MAC Address data.
@@ -38,9 +44,13 @@ class MACArray(NumPyBackedExtensionArrayMixin):
     ndim = 1
     can_hold_na = True
 
-    def __init__(self, values, copy=True):
+    def __init__(self, values, copy=True, dtype=None):
         # TODO: parse hex / strings
         self.data = np.array(values, dtype='uint64', copy=copy)
+        if isinstance(dtype, str):
+            MACType.construct_array_type(dtype)
+        elif dtype:
+            assert isinstance(dtype, MACType)
 
     @classmethod
     def _from_ndarray(cls, data, copy=False):
@@ -119,6 +129,13 @@ def take_nd(self, indexer, allow_fill=True, fill_value=None):
     def copy(self, deep=False):
         return type(self)(self.data.copy())
 
+    def astype(self, dtype, copy=True):
+        if isinstance(dtype, type(self.dtype)):
+            if copy:
+                self = self.copy()
+            return self
+        return super().astype(dtype, copy)
+
 
 def _format(mac):
     # https://stackoverflow.com/a/36883363/1889400

tests/ip/test_interface.py

@@ -48,6 +48,14 @@ def data_for_grouping():
     ])
 
 
+@pytest.fixture
+def data_repeated(data):
+    def gen(count):
+        for _ in range(count):
+            yield data
+    return gen
+
+
 @pytest.fixture
 def na_cmp():
     """Binary operator for comparing NA values.
@@ -78,7 +86,13 @@ class TestConstructors(base.BaseConstructorsTests):
 
 
 class TestReshaping(base.BaseReshapingTests):
-    pass
+    @pytest.mark.skip("We consider 0 to be NA.")
+    def test_stack(self):
+        pass
+
+    @pytest.mark.skip("We consider 0 to be NA.")
+    def test_unstack(self):
+        pass
 
 
 class TestGetitem(base.BaseGetitemTests):
@@ -94,3 +108,15 @@ class TestMethods(base.BaseMethodsTests):
     @pytest.mark.xfail(reason='upstream')
     def test_value_counts(data, dropna):
         pass
+
+    @pytest.mark.skip(reason='0 for NA')
+    def test_combine_le(self, data_repeated):
+        super().test_combine_le(data_repeated)
+
+    @pytest.mark.skip(reason='No __add__')
+    def test_combine_add(self, data_repeated):
+        super().test_combine_add(data_repeated)
+
+    @pytest.mark.xfail(reason="buggy comparison of v4 and v6")
+    def test_searchsorted(self, data_for_sorting, as_series):
+        return super().test_searchsorted(data_for_sorting, as_series)

tests/mac/test_interface.py

@@ -49,6 +49,14 @@ def data_for_grouping():
     ])
 
 
+@pytest.fixture
+def data_repeated(data):
+    def gen(count):
+        for _ in range(count):
+            yield data
+    return gen
+
+
 @pytest.fixture
 def na_cmp():
     """Binary operator for comparing NA values.
@@ -83,6 +91,14 @@ class TestReshaping(base.BaseReshapingTests):
     def test_concat_mixed_dtypes(self):
         pass
 
+    @pytest.mark.skip(reason="0 for null")
+    def test_stack(self):
+        pass
+
+    @pytest.mark.skip(reason="0 for null")
+    def test_unstack(self):
+        pass
+
 
 class TestGetitem(base.BaseGetitemTests):
     pass
@@ -96,3 +112,11 @@ class TestMethods(base.BaseMethodsTests):
     @pytest.mark.xfail(reason='upstream')
     def test_value_counts(data, dropna):
         pass
+
+    @pytest.mark.skip(reason="buggy comparison")
+    def test_combine_le(self, data_repeated):
+        super().test_combine_le(data_repeated)
+
+    @pytest.mark.skip(reason="TODO")
+    def test_hash_pandas_object_works(self):
+        pass