From 62008ec521267d4f8e93c639ce60aefddc7042d6 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 12:36:56 -0400
Subject: [PATCH 1/9] ENH: Add skipna parameter to infer_dtype

---
 doc/source/whatsnew/v0.21.0.txt       |   2 +
 pandas/_libs/src/inference.pyx        | 134 ++++++++++++++++++--------
 pandas/tests/dtypes/test_inference.py |  33 ++++++-
 3 files changed, 129 insertions(+), 40 deletions(-)

diff --git a/doc/source/whatsnew/v0.21.0.txt b/doc/source/whatsnew/v0.21.0.txt
index 5a5ea827e74ad..d277e8d4b9434 100644
--- a/doc/source/whatsnew/v0.21.0.txt
+++ b/doc/source/whatsnew/v0.21.0.txt
@@ -24,6 +24,8 @@ New features
   <https://www.python.org/dev/peps/pep-0519/>`_ on most readers and writers (:issue:`13823`)
 - Added ``__fspath__`` method to :class:`~pandas.HDFStore`, :class:`~pandas.ExcelFile`,
   and :class:`~pandas.ExcelWriter` to work properly with the file system path protocol (:issue:`13823`)
+- Added ``skipna`` parameter :func:`~pandas.api.types.infer_dtype` to support
+  type inference in the presence of missing values (:issue:`17059`).
 
 
 .. _whatsnew_0210.enhancements.infer_objects:
diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 38e95fe6ee652..1b9f268f9a843 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -222,7 +222,7 @@ cdef _try_infer_map(v):
     return None
 
 
-def infer_dtype(object value):
+def infer_dtype(object value, bint skipna=False):
     """
     Effeciently infer the type of a passed val, or list-like
     array of values. Return a string describing the type.
@@ -230,6 +230,8 @@ def infer_dtype(object value):
     Parameters
     ----------
     value : scalar, list, ndarray, or pandas type
+    skipna : bool, default False
+        Ignore NaN values when inferring the type.
 
     Returns
     -------
@@ -272,6 +274,9 @@ def infer_dtype(object value):
     >>> infer_dtype(['foo', 'bar'])
     'string'
 
+    >>> infer_dtype(['a', np.nan, 'b'], skipna=True)
+    'string'
+
     >>> infer_dtype([b'foo', b'bar'])
     'bytes'
 
@@ -310,7 +315,6 @@ def infer_dtype(object value):
 
     >>> infer_dtype(pd.Series(list('aabc')).astype('category'))
     'categorical'
-
     """
     cdef:
         Py_ssize_t i, n
@@ -356,7 +360,7 @@ def infer_dtype(object value):
     values = values.ravel()
 
     # try to use a valid value
-    for i from 0 <= i < n:
+    for i in range(n):
         val = util.get_value_1d(values, i)
 
         # do not use is_nul_datetimelike to keep
@@ -403,11 +407,11 @@ def infer_dtype(object value):
             return 'datetime'
 
     elif is_date(val):
-        if is_date_array(values):
+        if is_date_array(values, skipna=skipna):
             return 'date'
 
     elif is_time(val):
-        if is_time_array(values):
+        if is_time_array(values, skipna=skipna):
             return 'time'
 
     elif is_decimal(val):
@@ -420,19 +424,19 @@ def infer_dtype(object value):
             return 'mixed-integer-float'
 
     elif util.is_bool_object(val):
-        if is_bool_array(values):
+        if is_bool_array(values, skipna=skipna):
             return 'boolean'
 
     elif PyString_Check(val):
-        if is_string_array(values):
+        if is_string_array(values, skipna=skipna):
             return 'string'
 
     elif PyUnicode_Check(val):
-        if is_unicode_array(values):
+        if is_unicode_array(values, skipna=skipna):
             return 'unicode'
 
     elif PyBytes_Check(val):
-        if is_bytes_array(values):
+        if is_bytes_array(values, skipna=skipna):
             return 'bytes'
 
     elif is_period(val):
@@ -593,10 +597,11 @@ cdef inline bint is_timedelta(object o):
     return PyDelta_Check(o) or util.is_timedelta64_object(o)
 
 
-cpdef bint is_bool_array(ndarray values):
+cpdef bint is_bool_array(ndarray values, bint skipna=False):
     cdef:
         Py_ssize_t i, n = len(values)
         ndarray[object] objbuf
+        object val
 
     if issubclass(values.dtype.type, np.bool_):
         return True
@@ -606,9 +611,16 @@ cpdef bint is_bool_array(ndarray values):
         if n == 0:
             return False
 
-        for i in range(n):
-            if not util.is_bool_object(objbuf[i]):
-                return False
+        if skipna:
+            for i in range(n):
+                val = objbuf[i]
+                if not util._checknull(val) and not util.is_bool_object(val):
+                    return False
+        else:
+            for i in range(n):
+                val = objbuf[i]
+                if not util.is_bool_object(val):
+                    return False
         return True
     else:
         return False
@@ -639,6 +651,7 @@ cpdef bint is_integer_float_array(ndarray values):
     cdef:
         Py_ssize_t i, n = len(values)
         ndarray[object] objbuf
+        object value
 
     if issubclass(values.dtype.type, np.integer):
         return True
@@ -649,9 +662,8 @@ cpdef bint is_integer_float_array(ndarray values):
             return False
 
         for i in range(n):
-            if not (util.is_integer_object(objbuf[i]) or
-                    util.is_float_object(objbuf[i])):
-
+            val = objbuf[i]
+            if not (util.is_integer_object(val) or util.is_float_object(val)):
                 return False
         return True
     else:
@@ -679,10 +691,11 @@ cpdef bint is_float_array(ndarray values):
         return False
 
 
-cpdef bint is_string_array(ndarray values):
+cpdef bint is_string_array(ndarray values, bint skipna=False):
     cdef:
         Py_ssize_t i, n = len(values)
         ndarray[object] objbuf
+        object val
 
     if ((PY2 and issubclass(values.dtype.type, np.string_)) or
         not PY2 and issubclass(values.dtype.type, np.unicode_)):
@@ -693,18 +706,26 @@ cpdef bint is_string_array(ndarray values):
         if n == 0:
             return False
 
-        for i in range(n):
-            if not PyString_Check(objbuf[i]):
-                return False
+        if skipna:
+            for i in range(n):
+                val = objbuf[i]
+                if not util._checknull(val) and not PyString_Check(val):
+                    return False
+        else:
+            for i in range(n):
+                val = objbuf[i]
+                if not PyString_Check(val):
+                    return False
         return True
     else:
         return False
 
 
-cpdef bint is_unicode_array(ndarray values):
+cpdef bint is_unicode_array(ndarray values, bint skipna=False):
     cdef:
         Py_ssize_t i, n = len(values)
         ndarray[object] objbuf
+        object val
 
     if issubclass(values.dtype.type, np.unicode_):
         return True
@@ -714,18 +735,26 @@ cpdef bint is_unicode_array(ndarray values):
         if n == 0:
             return False
 
-        for i in range(n):
-            if not PyUnicode_Check(objbuf[i]):
-                return False
+        if skipna:
+            for i in range(n):
+                val = objbuf[i]
+                if not util._checknull(val) and not PyUnicode_Check(val):
+                    return False
+        else:
+            for i in range(n):
+                val = objbuf[i]
+                if not PyUnicode_Check(val):
+                    return False
         return True
     else:
         return False
 
 
-cpdef bint is_bytes_array(ndarray values):
+cpdef bint is_bytes_array(ndarray values, bint skipna=False):
     cdef:
         Py_ssize_t i, n = len(values)
         ndarray[object] objbuf
+        object val
 
     if issubclass(values.dtype.type, np.bytes_):
         return True
@@ -735,9 +764,16 @@ cpdef bint is_bytes_array(ndarray values):
         if n == 0:
             return False
 
-        for i in range(n):
-            if not PyBytes_Check(objbuf[i]):
-                return False
+        if skipna:
+            for i in range(n):
+                val = objbuf[i]
+                if not util._checknull(val) and not PyBytes_Check(val):
+                    return False
+        else:
+            for i in range(n):
+                val = objbuf[i]
+                if not PyBytes_Check(val):
+                    return False
         return True
     else:
         return False
@@ -856,23 +892,45 @@ cpdef bint is_timedelta_or_timedelta64_array(ndarray values):
     return null_count != n
 
 
-cpdef bint is_date_array(ndarray[object] values):
-    cdef Py_ssize_t i, n = len(values)
+cpdef bint is_date_array(ndarray[object] values, bint skipna=False):
+    cdef:
+        Py_ssize_t i, n = len(values)
+        object val
+
     if n == 0:
         return False
-    for i in range(n):
-        if not is_date(values[i]):
-            return False
+
+    if skipna:
+        for i in range(n):
+            val = values[i]
+            if not util._checknull(val) and not is_date(val):
+                return False
+    else:
+        for i in range(n):
+            val = values[i]
+            if not is_date(val):
+                return False
     return True
 
 
-cpdef bint is_time_array(ndarray[object] values):
-    cdef Py_ssize_t i, n = len(values)
+cpdef bint is_time_array(ndarray[object] values, bint skipna=False):
+    cdef:
+        Py_ssize_t i, n = len(values)
+        object val
+
     if n == 0:
         return False
-    for i in range(n):
-        if not is_time(values[i]):
-            return False
+
+    if skipna:
+        for i in range(n):
+            val = values[i]
+            if not util._checknull(val) and not is_time(val):
+                return False
+    else:
+        for i in range(n):
+            val = values[i]
+            if not is_time(val):
+                return False
     return True
 
 
diff --git a/pandas/tests/dtypes/test_inference.py b/pandas/tests/dtypes/test_inference.py
index d26ea047bb41f..dbde7ae5081d4 100644
--- a/pandas/tests/dtypes/test_inference.py
+++ b/pandas/tests/dtypes/test_inference.py
@@ -239,6 +239,9 @@ def test_infer_dtype_bytes(self):
         arr = arr.astype(object)
         assert lib.infer_dtype(arr) == compare
 
+        # object array of bytes with missing values
+        assert lib.infer_dtype([b'a', np.nan, b'c'], skipna=True) == compare
+
     def test_isinf_scalar(self):
         # GH 11352
         assert lib.isposinf_scalar(float('inf'))
@@ -444,6 +447,10 @@ def test_bools(self):
         result = lib.infer_dtype(arr)
         assert result == 'boolean'
 
+        arr = np.array([True, np.nan, False], dtype='O')
+        result = lib.infer_dtype(arr, skipna=True)
+        assert result == 'boolean'
+
     def test_floats(self):
         arr = np.array([1., 2., 3., np.float64(4), np.float32(5)], dtype='O')
         result = lib.infer_dtype(arr)
@@ -472,11 +479,26 @@ def test_decimals(self):
         result = lib.infer_dtype(arr)
         assert result == 'mixed'
 
+        arr = np.array([Decimal(1), Decimal('NaN'), Decimal(3)])
+        result = lib.infer_dtype(arr)
+        assert result == 'decimal'
+
+        arr = np.array([Decimal(1), np.nan, Decimal(3)], dtype='O')
+        result = lib.infer_dtype(arr)
+        assert result == 'decimal'
+
     def test_string(self):
         pass
 
     def test_unicode(self):
-        pass
+        arr = [u'a', np.nan, u'c']
+        result = lib.infer_dtype(arr)
+        assert result == 'mixed'
+
+        arr = [u'a', np.nan, u'c']
+        result = lib.infer_dtype(arr, skipna=True)
+        expected = 'unicode' if PY2 else 'string'
+        assert result == expected
 
     def test_datetime(self):
 
@@ -714,10 +736,17 @@ def test_is_datetimelike_array_all_nan_nat_like(self):
 
     def test_date(self):
 
-        dates = [date(2012, 1, x) for x in range(1, 20)]
+        dates = [date(2012, 1, day) for day in range(1, 20)]
         index = Index(dates)
         assert index.inferred_type == 'date'
 
+        dates = [date(2012, 1, day) for day in range(1, 20)] + [np.nan]
+        result = lib.infer_dtype(dates)
+        assert result == 'mixed'
+
+        result = lib.infer_dtype(dates, skipna=True)
+        assert result == 'date'
+
     def test_to_object_array_tuples(self):
         r = (5, 6)
         values = [r]

From 942265129df96f46f8a1b536de17265d143f632f Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 16:16:37 -0400
Subject: [PATCH 2/9] ENH: Refactor type inference Cython code

---
 pandas/_libs/src/inference.pyx | 532 +++++++++++++++++----------------
 1 file changed, 273 insertions(+), 259 deletions(-)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 1b9f268f9a843..82c2b125badac 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -597,222 +597,262 @@ cdef inline bint is_timedelta(object o):
     return PyDelta_Check(o) or util.is_timedelta64_object(o)
 
 
-cpdef bint is_bool_array(ndarray values, bint skipna=False):
+cdef class Validator:
+
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
-        object val
+        Py_ssize_t n
+        object dtype
+        bint skipna
 
-    if issubclass(values.dtype.type, np.bool_):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
+    def __cinit__(self, Py_ssize_t n, object dtype=None, bint skipna=False):
+        self.n = n
+        self.dtype = dtype if dtype is not None else np.dtype(np.object_)
+        self.skipna = skipna
 
-        if n == 0:
+    cdef bint validate(self, object[:] values):
+        if not self.n:
             return False
 
-        if skipna:
-            for i in range(n):
-                val = objbuf[i]
-                if not util._checknull(val) and not util.is_bool_object(val):
-                    return False
+        if self.is_array_typed():
+            return True
+        elif self.dtype.type == np.object_:
+            if self.skipna:
+                return self._validate_skipna(values)
+            else:
+                return self._validate(values)
         else:
-            for i in range(n):
-                val = objbuf[i]
-                if not util.is_bool_object(val):
-                    return False
-        return True
-    else:
+            return False
+
+    cdef bint _validate(self, object[:] values):
+        cdef:
+            Py_ssize_t i
+            Py_ssize_t n = self.n
+
+        for i in range(n):
+            if not self.is_valid(values[i]):
+                return False
+        return self.finalize()
+
+    cdef bint _validate_skipna(self, object[:] values):
+        cdef:
+            Py_ssize_t i
+            Py_ssize_t n = self.n
+
+        for i in range(n):
+            if not self.is_valid_skipna(values[i]):
+                return False
+        return self.finalize()
+
+    cdef bint is_valid(self, object value) except -1:
+        return self.is_value_typed(value)
+
+    cdef bint is_valid_skipna(self, object value) except -1:
+        return self.is_valid(value) or self.is_valid_null(value)
+
+    cdef bint is_value_typed(self, object value) except -1:
+        raise NotImplementedError(
+            'is_value_typed must be implemented in subclasses'
+        )
+
+    cdef bint is_valid_null(self, object value) except -1:
+        return util._checknull(value)
+
+    cdef bint is_array_typed(self) except -1:
         return False
 
+    cdef bint finalize(self):
+        return True
+
+
+cdef class BoolValidator(Validator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_bool_object(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.bool_)
+
+
+cpdef bint is_bool_array(ndarray values, bint skipna=False):
+    cdef:
+        BoolValidator validator = BoolValidator(
+            len(values),
+            values.dtype,
+            skipna=skipna
+        )
+    return validator.validate(values)
+
+
+cdef class IntegerValidator(Validator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_integer_object(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.integer)
+
 
 cpdef bint is_integer_array(ndarray values):
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
+        IntegerValidator validator = IntegerValidator(
+            len(values),
+            values.dtype,
+        )
+    return validator.validate(values)
 
-    if issubclass(values.dtype.type, np.integer):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class IntegerFloatValidator(Validator):
 
-        for i in range(n):
-            if not util.is_integer_object(objbuf[i]):
-                return False
-        return True
-    else:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_integer_object(value) or util.is_float_object(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.integer)
 
 
 cpdef bint is_integer_float_array(ndarray values):
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
-        object value
+        IntegerFloatValidator validator = IntegerFloatValidator(
+            len(values),
+            values.dtype,
+        )
+    return validator.validate(values)
 
-    if issubclass(values.dtype.type, np.integer):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class FloatValidator(Validator):
 
-        for i in range(n):
-            val = objbuf[i]
-            if not (util.is_integer_object(val) or util.is_float_object(val)):
-                return False
-        return True
-    else:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_float_object(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.floating)
 
 
 cpdef bint is_float_array(ndarray values):
-    cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
+    cdef FloatValidator validator = FloatValidator(len(values), values.dtype)
+    return validator.validate(values)
 
-    if issubclass(values.dtype.type, np.floating):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class StringValidator(Validator):
 
-        for i in range(n):
-            if not util.is_float_object(objbuf[i]):
-                return False
-        return True
-    else:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return PyString_Check(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return (
+            PY2 and issubclass(self.dtype.type, np.string_)
+        ) or not PY2 and issubclass(self.dtype.type, np.unicode_)
 
 
 cpdef bint is_string_array(ndarray values, bint skipna=False):
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
-        object val
+        StringValidator validator = StringValidator(
+            len(values),
+            values.dtype,
+            skipna=skipna,
+        )
+    return validator.validate(values)
 
-    if ((PY2 and issubclass(values.dtype.type, np.string_)) or
-        not PY2 and issubclass(values.dtype.type, np.unicode_)):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class UnicodeValidator(Validator):
 
-        if skipna:
-            for i in range(n):
-                val = objbuf[i]
-                if not util._checknull(val) and not PyString_Check(val):
-                    return False
-        else:
-            for i in range(n):
-                val = objbuf[i]
-                if not PyString_Check(val):
-                    return False
-        return True
-    else:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return PyUnicode_Check(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.unicode_)
 
 
 cpdef bint is_unicode_array(ndarray values, bint skipna=False):
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
-        object val
+        UnicodeValidator validator = UnicodeValidator(
+            len(values),
+            values.dtype,
+            skipna=skipna,
+        )
+    return validator.validate(values)
 
-    if issubclass(values.dtype.type, np.unicode_):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class BytesValidator(Validator):
 
-        if skipna:
-            for i in range(n):
-                val = objbuf[i]
-                if not util._checknull(val) and not PyUnicode_Check(val):
-                    return False
-        else:
-            for i in range(n):
-                val = objbuf[i]
-                if not PyUnicode_Check(val):
-                    return False
-        return True
-    else:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return PyBytes_Check(value)
+
+    cdef bint is_array_typed(self) except -1:
+        return issubclass(self.dtype.type, np.bytes_)
 
 
 cpdef bint is_bytes_array(ndarray values, bint skipna=False):
     cdef:
-        Py_ssize_t i, n = len(values)
-        ndarray[object] objbuf
-        object val
+        BytesValidator validator = BytesValidator(
+            len(values),
+            values.dtype,
+            skipna=skipna
+        )
+    return validator.validate(values)
 
-    if issubclass(values.dtype.type, np.bytes_):
-        return True
-    elif values.dtype == np.object_:
-        objbuf = values
 
-        if n == 0:
-            return False
+cdef class TemporalValidator(Validator):
 
-        if skipna:
-            for i in range(n):
-                val = objbuf[i]
-                if not util._checknull(val) and not PyBytes_Check(val):
-                    return False
-        else:
-            for i in range(n):
-                val = objbuf[i]
-                if not PyBytes_Check(val):
-                    return False
-        return True
-    else:
-        return False
+    cdef Py_ssize_t generic_null_count
+
+    def __cinit__(self, Py_ssize_t n, object dtype=None, bint skipna=False):
+        self.n = n
+        self.dtype = dtype if dtype is not None else np.dtype(np.object_)
+        self.skipna = skipna
+        self.generic_null_count = 0
+
+    cdef bint is_valid(self, object value) except -1:
+        return self.is_value_typed(value) or self.is_valid_null(value)
+
+    cdef bint is_value_typed(self, object value) except -1:
+        raise NotImplementedError()
+
+    cdef bint is_valid_null(self, object value) except -1:
+        raise NotImplementedError()
+
+    cdef bint is_valid_skipna(self, object value) except -1:
+        cdef:
+            bint is_typed_null = self.is_valid_null(value)
+            bint is_generic_null = util._checknull(value)
+        self.generic_null_count += is_typed_null and is_generic_null
+        return self.is_value_typed(value) or is_typed_null or is_generic_null
+
+    cdef bint finalize(self):
+        return self.generic_null_count != self.n
+
+
+cdef class DatetimeValidator(TemporalValidator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_datetime(value)
+
+    cdef bint is_valid_null(self, object value) except -1:
+        return is_null_datetime64(value)
 
 
 cpdef bint is_datetime_array(ndarray[object] values):
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
+    cdef:
+        DatetimeValidator validator = DatetimeValidator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
 
-    # return False for all nulls
-    for i in range(n):
-        v = values[i]
-        if is_null_datetime64(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not is_datetime(v):
-            return False
-    return null_count != n
 
+cdef class Datetime64Validator(DatetimeValidator):
 
-cpdef bint is_datetime64_array(ndarray values):
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_datetime64_object(value)
 
-    # return False for all nulls
-    for i in range(n):
-        v = values[i]
-        if is_null_datetime64(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not util.is_datetime64_object(v):
-            return False
-    return null_count != n
+
+cpdef bint is_datetime64_array(ndarray values):
+    cdef:
+        Datetime64Validator validator = Datetime64Validator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
 
 
 cpdef bint is_datetime_with_singletz_array(ndarray[object] values):
@@ -843,130 +883,104 @@ cpdef bint is_datetime_with_singletz_array(ndarray[object] values):
     return True
 
 
+cdef class TimedeltaValidator(TemporalValidator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return PyDelta_Check(value)
+
+    cdef bint is_valid_null(self, object value) except -1:
+        return is_null_timedelta64(value)
+
+
 cpdef bint is_timedelta_array(ndarray values):
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
-    for i in range(n):
-        v = values[i]
-        if is_null_timedelta64(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not PyDelta_Check(v):
-            return False
-    return null_count != n
+    cdef:
+        TimedeltaValidator validator = TimedeltaValidator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
+
+
+cdef class Timedelta64Validator(TimedeltaValidator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return util.is_timedelta64_object(value)
 
 
 cpdef bint is_timedelta64_array(ndarray values):
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
-    for i in range(n):
-        v = values[i]
-        if is_null_timedelta64(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not util.is_timedelta64_object(v):
-            return False
-    return null_count != n
+    cdef:
+        Timedelta64Validator validator = Timedelta64Validator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
+
+
+cdef class AnyTimedeltaValidator(TimedeltaValidator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_timedelta(value)
 
 
 cpdef bint is_timedelta_or_timedelta64_array(ndarray values):
     """ infer with timedeltas and/or nat/none """
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
-    for i in range(n):
-        v = values[i]
-        if is_null_timedelta64(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not is_timedelta(v):
-            return False
-    return null_count != n
+    cdef:
+        AnyTimedeltaValidator validator = AnyTimedeltaValidator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
+
+
+cdef class DateValidator(Validator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_date(value)
 
 
 cpdef bint is_date_array(ndarray[object] values, bint skipna=False):
-    cdef:
-        Py_ssize_t i, n = len(values)
-        object val
+    cdef DateValidator validator = DateValidator(len(values), skipna=skipna)
+    return validator.validate(values)
 
-    if n == 0:
-        return False
 
-    if skipna:
-        for i in range(n):
-            val = values[i]
-            if not util._checknull(val) and not is_date(val):
-                return False
-    else:
-        for i in range(n):
-            val = values[i]
-            if not is_date(val):
-                return False
-    return True
+cdef class TimeValidator(Validator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_time(value)
 
 
 cpdef bint is_time_array(ndarray[object] values, bint skipna=False):
-    cdef:
-        Py_ssize_t i, n = len(values)
-        object val
+    cdef TimeValidator validator = TimeValidator(len(values), skipna=skipna)
+    return validator.validate(values)
 
-    if n == 0:
-        return False
 
-    if skipna:
-        for i in range(n):
-            val = values[i]
-            if not util._checknull(val) and not is_time(val):
-                return False
-    else:
-        for i in range(n):
-            val = values[i]
-            if not is_time(val):
-                return False
-    return True
+cdef class PeriodValidator(TemporalValidator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_period(value)
+
+    cdef bint is_valid_null(self, object value) except -1:
+        return is_null_period(value)
 
 
 cpdef bint is_period_array(ndarray[object] values):
-    cdef Py_ssize_t i, null_count = 0, n = len(values)
-    cdef object v
-    if n == 0:
-        return False
+    cdef PeriodValidator validator = PeriodValidator(len(values), skipna=True)
+    return validator.validate(values)
 
-    # return False for all nulls
-    for i in range(n):
-        v = values[i]
-        if is_null_period(v):
-            # we are a regular null
-            if util._checknull(v):
-                null_count += 1
-        elif not is_period(v):
-            return False
-    return null_count != n
+
+cdef class IntervalValidator(Validator):
+
+    cdef bint is_value_typed(self, object value) except -1:
+        return is_interval(value)
 
 
 cpdef bint is_interval_array(ndarray[object] values):
     cdef:
-        Py_ssize_t i, n = len(values), null_count = 0
-        object v
-
-    if n == 0:
-        return False
-    for i in range(n):
-        v = values[i]
-        if util._checknull(v):
-            null_count += 1
-            continue
-        if not is_interval(v):
-            return False
-    return null_count != n
+        IntervalValidator validator = IntervalValidator(
+            len(values),
+            skipna=True,
+        )
+    return validator.validate(values)
 
 
 cdef extern from "parse_helper.h":

From 68b277bf312426227a51e13562f13b172f79a86c Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 19:40:38 -0400
Subject: [PATCH 3/9] PERF: No negative index or out of bounds check

---
 pandas/_libs/src/inference.pyx | 7 +++++++
 1 file changed, 7 insertions(+)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 82c2b125badac..7ea506212868b 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -1,6 +1,7 @@
 import sys
 from decimal import Decimal
 cimport util
+cimport cython
 from tslib import NaT, get_timezone
 from datetime import datetime, timedelta
 iNaT = util.get_nat()
@@ -623,6 +624,8 @@ cdef class Validator:
         else:
             return False
 
+    @cython.wraparound(False)
+    @cython.boundscheck(False)
     cdef bint _validate(self, object[:] values):
         cdef:
             Py_ssize_t i
@@ -631,8 +634,11 @@ cdef class Validator:
         for i in range(n):
             if not self.is_valid(values[i]):
                 return False
+
         return self.finalize()
 
+    @cython.wraparound(False)
+    @cython.boundscheck(False)
     cdef bint _validate_skipna(self, object[:] values):
         cdef:
             Py_ssize_t i
@@ -641,6 +647,7 @@ cdef class Validator:
         for i in range(n):
             if not self.is_valid_skipna(values[i]):
                 return False
+
         return self.finalize()
 
     cdef bint is_valid(self, object value) except -1:

From b4552c35592b761ded460fd83fa2a8a7c7ce9c20 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 20:04:27 -0400
Subject: [PATCH 4/9] DOC: versionadded tag

---
 pandas/_libs/src/inference.pyx | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 7ea506212868b..e01973a4962a9 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -234,6 +234,8 @@ def infer_dtype(object value, bint skipna=False):
     skipna : bool, default False
         Ignore NaN values when inferring the type.
 
+        .. versionadded:: 0.21.0
+
     Returns
     -------
     string describing the common type of the input data.

From ddfca5176dd0517a9160a9d17922ab7a32786e63 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 20:13:56 -0400
Subject: [PATCH 5/9] PERF: Inline leaf classes' methods

---
 pandas/_libs/src/inference.pyx | 60 +++++++++++++++++-----------------
 1 file changed, 30 insertions(+), 30 deletions(-)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index e01973a4962a9..2b25b9029009c 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -612,7 +612,7 @@ cdef class Validator:
         self.dtype = dtype if dtype is not None else np.dtype(np.object_)
         self.skipna = skipna
 
-    cdef bint validate(self, object[:] values):
+    cdef inline bint validate(self, object[:] values):
         if not self.n:
             return False
 
@@ -628,7 +628,7 @@ cdef class Validator:
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef bint _validate(self, object[:] values):
+    cdef inline bint _validate(self, object[:] values):
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n
@@ -641,7 +641,7 @@ cdef class Validator:
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef bint _validate_skipna(self, object[:] values):
+    cdef inline bint _validate_skipna(self, object[:] values):
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n
@@ -675,10 +675,10 @@ cdef class Validator:
 
 cdef class BoolValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_bool_object(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.bool_)
 
 
@@ -694,10 +694,10 @@ cpdef bint is_bool_array(ndarray values, bint skipna=False):
 
 cdef class IntegerValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_integer_object(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.integer)
 
 
@@ -712,10 +712,10 @@ cpdef bint is_integer_array(ndarray values):
 
 cdef class IntegerFloatValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_integer_object(value) or util.is_float_object(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.integer)
 
 
@@ -730,10 +730,10 @@ cpdef bint is_integer_float_array(ndarray values):
 
 cdef class FloatValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_float_object(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.floating)
 
 
@@ -744,10 +744,10 @@ cpdef bint is_float_array(ndarray values):
 
 cdef class StringValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return PyString_Check(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return (
             PY2 and issubclass(self.dtype.type, np.string_)
         ) or not PY2 and issubclass(self.dtype.type, np.unicode_)
@@ -765,10 +765,10 @@ cpdef bint is_string_array(ndarray values, bint skipna=False):
 
 cdef class UnicodeValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return PyUnicode_Check(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.unicode_)
 
 
@@ -784,10 +784,10 @@ cpdef bint is_unicode_array(ndarray values, bint skipna=False):
 
 cdef class BytesValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return PyBytes_Check(value)
 
-    cdef bint is_array_typed(self) except -1:
+    cdef inline bint is_array_typed(self) except -1:
         return issubclass(self.dtype.type, np.bytes_)
 
 
@@ -811,7 +811,7 @@ cdef class TemporalValidator(Validator):
         self.skipna = skipna
         self.generic_null_count = 0
 
-    cdef bint is_valid(self, object value) except -1:
+    cdef inline bint is_valid(self, object value) except -1:
         return self.is_value_typed(value) or self.is_valid_null(value)
 
     cdef bint is_value_typed(self, object value) except -1:
@@ -820,14 +820,14 @@ cdef class TemporalValidator(Validator):
     cdef bint is_valid_null(self, object value) except -1:
         raise NotImplementedError()
 
-    cdef bint is_valid_skipna(self, object value) except -1:
+    cdef inline bint is_valid_skipna(self, object value) except -1:
         cdef:
             bint is_typed_null = self.is_valid_null(value)
             bint is_generic_null = util._checknull(value)
         self.generic_null_count += is_typed_null and is_generic_null
         return self.is_value_typed(value) or is_typed_null or is_generic_null
 
-    cdef bint finalize(self):
+    cdef inline bint finalize(self):
         return self.generic_null_count != self.n
 
 
@@ -836,7 +836,7 @@ cdef class DatetimeValidator(TemporalValidator):
     cdef bint is_value_typed(self, object value) except -1:
         return is_datetime(value)
 
-    cdef bint is_valid_null(self, object value) except -1:
+    cdef inline bint is_valid_null(self, object value) except -1:
         return is_null_datetime64(value)
 
 
@@ -851,7 +851,7 @@ cpdef bint is_datetime_array(ndarray[object] values):
 
 cdef class Datetime64Validator(DatetimeValidator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_datetime64_object(value)
 
 
@@ -897,7 +897,7 @@ cdef class TimedeltaValidator(TemporalValidator):
     cdef bint is_value_typed(self, object value) except -1:
         return PyDelta_Check(value)
 
-    cdef bint is_valid_null(self, object value) except -1:
+    cdef inline bint is_valid_null(self, object value) except -1:
         return is_null_timedelta64(value)
 
 
@@ -912,7 +912,7 @@ cpdef bint is_timedelta_array(ndarray values):
 
 cdef class Timedelta64Validator(TimedeltaValidator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return util.is_timedelta64_object(value)
 
 
@@ -927,7 +927,7 @@ cpdef bint is_timedelta64_array(ndarray values):
 
 cdef class AnyTimedeltaValidator(TimedeltaValidator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return is_timedelta(value)
 
 
@@ -943,7 +943,7 @@ cpdef bint is_timedelta_or_timedelta64_array(ndarray values):
 
 cdef class DateValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return is_date(value)
 
 
@@ -954,7 +954,7 @@ cpdef bint is_date_array(ndarray[object] values, bint skipna=False):
 
 cdef class TimeValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return is_time(value)
 
 
@@ -965,10 +965,10 @@ cpdef bint is_time_array(ndarray[object] values, bint skipna=False):
 
 cdef class PeriodValidator(TemporalValidator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return is_period(value)
 
-    cdef bint is_valid_null(self, object value) except -1:
+    cdef inline bint is_valid_null(self, object value) except -1:
         return is_null_period(value)
 
 
@@ -979,7 +979,7 @@ cpdef bint is_period_array(ndarray[object] values):
 
 cdef class IntervalValidator(Validator):
 
-    cdef bint is_value_typed(self, object value) except -1:
+    cdef inline bint is_value_typed(self, object value) except -1:
         return is_interval(value)
 
 

From 234110af84d121871f43730c7e67c68097afd6dc Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Mon, 24 Jul 2017 20:16:12 -0400
Subject: [PATCH 6/9] PERF: Don't try to inline the larger methods

---
 pandas/_libs/src/inference.pyx | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 2b25b9029009c..78023bcc7f576 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -612,7 +612,7 @@ cdef class Validator:
         self.dtype = dtype if dtype is not None else np.dtype(np.object_)
         self.skipna = skipna
 
-    cdef inline bint validate(self, object[:] values):
+    cdef bint validate(self, object[:] values):
         if not self.n:
             return False
 
@@ -628,7 +628,7 @@ cdef class Validator:
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef inline bint _validate(self, object[:] values):
+    cdef bint _validate(self, object[:] values):
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n
@@ -641,7 +641,7 @@ cdef class Validator:
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef inline bint _validate_skipna(self, object[:] values):
+    cdef bint _validate_skipna(self, object[:] values):
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n

From ca2e65ba36120b2684095419f674c2c2f3922d98 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Tue, 25 Jul 2017 09:57:29 -0400
Subject: [PATCH 7/9] DOC: typo

---
 doc/source/whatsnew/v0.21.0.txt | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/doc/source/whatsnew/v0.21.0.txt b/doc/source/whatsnew/v0.21.0.txt
index d277e8d4b9434..5fd245cee7d88 100644
--- a/doc/source/whatsnew/v0.21.0.txt
+++ b/doc/source/whatsnew/v0.21.0.txt
@@ -24,8 +24,8 @@ New features
   <https://www.python.org/dev/peps/pep-0519/>`_ on most readers and writers (:issue:`13823`)
 - Added ``__fspath__`` method to :class:`~pandas.HDFStore`, :class:`~pandas.ExcelFile`,
   and :class:`~pandas.ExcelWriter` to work properly with the file system path protocol (:issue:`13823`)
-- Added ``skipna`` parameter :func:`~pandas.api.types.infer_dtype` to support
-  type inference in the presence of missing values (:issue:`17059`).
+- Added ``skipna`` parameter to :func:`~pandas.api.types.infer_dtype` to
+  support type inference in the presence of missing values (:issue:`17059`).
 
 
 .. _whatsnew_0210.enhancements.infer_objects:

From 9a428360ecfda81a9f473c49c56ea33d6ff296c1 Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Tue, 25 Jul 2017 10:32:12 -0400
Subject: [PATCH 8/9] REF: Address comments

---
 pandas/_libs/src/inference.pyx | 63 +++++++++++++++++++++++-----------
 1 file changed, 43 insertions(+), 20 deletions(-)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index 78023bcc7f576..ab19e66502848 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -232,7 +232,8 @@ def infer_dtype(object value, bint skipna=False):
     ----------
     value : scalar, list, ndarray, or pandas type
     skipna : bool, default False
-        Ignore NaN values when inferring the type.
+        Ignore NaN values when inferring the type. The default of ``False``
+        will be deprecated in a later version of pandas.
 
         .. versionadded:: 0.21.0
 
@@ -280,6 +281,9 @@ def infer_dtype(object value, bint skipna=False):
     >>> infer_dtype(['a', np.nan, 'b'], skipna=True)
     'string'
 
+    >>> infer_dtype(['a', np.nan, 'b'], skipna=False)
+    'mixed'
+
     >>> infer_dtype([b'foo', b'bar'])
     'bytes'
 
@@ -323,7 +327,8 @@ def infer_dtype(object value, bint skipna=False):
         Py_ssize_t i, n
         object val
         ndarray values
-        bint seen_pdnat = False, seen_val = False
+        bint seen_pdnat = False
+        bint seen_val = False
 
     if isinstance(value, np.ndarray):
         values = value
@@ -604,21 +609,26 @@ cdef class Validator:
 
     cdef:
         Py_ssize_t n
-        object dtype
+        np.dtype dtype
         bint skipna
 
-    def __cinit__(self, Py_ssize_t n, object dtype=None, bint skipna=False):
+    def __cinit__(
+        self,
+        Py_ssize_t n,
+        np.dtype dtype=np.dtype(np.object_),
+        bint skipna=False
+    ):
         self.n = n
-        self.dtype = dtype if dtype is not None else np.dtype(np.object_)
+        self.dtype = dtype
         self.skipna = skipna
 
-    cdef bint validate(self, object[:] values):
+    cdef bint validate(self, object[:] values) except -1:
         if not self.n:
             return False
 
         if self.is_array_typed():
             return True
-        elif self.dtype.type == np.object_:
+        elif self.dtype.type_num == NPY_OBJECT:
             if self.skipna:
                 return self._validate_skipna(values)
             else:
@@ -628,7 +638,7 @@ cdef class Validator:
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef bint _validate(self, object[:] values):
+    cdef bint _validate(self, object[:] values) except -1:
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n
@@ -637,11 +647,11 @@ cdef class Validator:
             if not self.is_valid(values[i]):
                 return False
 
-        return self.finalize()
+        return self.finalize_validate()
 
     @cython.wraparound(False)
     @cython.boundscheck(False)
-    cdef bint _validate_skipna(self, object[:] values):
+    cdef bint _validate_skipna(self, object[:] values) except -1:
         cdef:
             Py_ssize_t i
             Py_ssize_t n = self.n
@@ -650,7 +660,7 @@ cdef class Validator:
             if not self.is_valid_skipna(values[i]):
                 return False
 
-        return self.finalize()
+        return self.finalize_validate_skipna()
 
     cdef bint is_valid(self, object value) except -1:
         return self.is_value_typed(value)
@@ -660,7 +670,9 @@ cdef class Validator:
 
     cdef bint is_value_typed(self, object value) except -1:
         raise NotImplementedError(
-            'is_value_typed must be implemented in subclasses'
+            '{} child class must define is_value_typed'.format(
+                type(self).__name__
+            )
         )
 
     cdef bint is_valid_null(self, object value) except -1:
@@ -669,7 +681,12 @@ cdef class Validator:
     cdef bint is_array_typed(self) except -1:
         return False
 
-    cdef bint finalize(self):
+    cdef inline bint finalize_validate(self):
+        return True
+
+    cdef bint finalize_validate_skipna(self):
+        # TODO(phillipc): Remove the existing validate methods and replace them
+        # with the skipna versions upon full deprecation of skipna=False
         return True
 
 
@@ -805,20 +822,26 @@ cdef class TemporalValidator(Validator):
 
     cdef Py_ssize_t generic_null_count
 
-    def __cinit__(self, Py_ssize_t n, object dtype=None, bint skipna=False):
+    def __cinit__(
+        self,
+        Py_ssize_t n,
+        np.dtype dtype=np.dtype(np.object_),
+        bint skipna=False
+    ):
         self.n = n
-        self.dtype = dtype if dtype is not None else np.dtype(np.object_)
+        self.dtype = dtype
         self.skipna = skipna
         self.generic_null_count = 0
 
     cdef inline bint is_valid(self, object value) except -1:
         return self.is_value_typed(value) or self.is_valid_null(value)
 
-    cdef bint is_value_typed(self, object value) except -1:
-        raise NotImplementedError()
-
     cdef bint is_valid_null(self, object value) except -1:
-        raise NotImplementedError()
+        raise NotImplementedError(
+            '{} child class must define is_valid_null'.format(
+                type(self).__name__
+            )
+        )
 
     cdef inline bint is_valid_skipna(self, object value) except -1:
         cdef:
@@ -827,7 +850,7 @@ cdef class TemporalValidator(Validator):
         self.generic_null_count += is_typed_null and is_generic_null
         return self.is_value_typed(value) or is_typed_null or is_generic_null
 
-    cdef inline bint finalize(self):
+    cdef inline bint finalize_validate_skipna(self):
         return self.generic_null_count != self.n
 
 

From 18839270046b35e1f0bbd836723c479d37eaecbb Mon Sep 17 00:00:00 2001
From: Phillip Cloud <cpcloud@gmail.com>
Date: Tue, 25 Jul 2017 12:04:05 -0400
Subject: [PATCH 9/9] CLN: Simplify is_array_type for strings

---
 pandas/_libs/src/inference.pyx | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

diff --git a/pandas/_libs/src/inference.pyx b/pandas/_libs/src/inference.pyx
index ab19e66502848..6b5a8f20f0067 100644
--- a/pandas/_libs/src/inference.pyx
+++ b/pandas/_libs/src/inference.pyx
@@ -765,9 +765,7 @@ cdef class StringValidator(Validator):
         return PyString_Check(value)
 
     cdef inline bint is_array_typed(self) except -1:
-        return (
-            PY2 and issubclass(self.dtype.type, np.string_)
-        ) or not PY2 and issubclass(self.dtype.type, np.unicode_)
+        return issubclass(self.dtype.type, np.str_)
 
 
 cpdef bint is_string_array(ndarray values, bint skipna=False):