BUG: Avoid FutureWarning by implementing equality check in Cython.

BUG: array_equivalent expects NumPy arrays as input.
DOC: Fix typo and show that the index order matters

Author: unutbu

doc/source/basics.rst

@@ -317,6 +317,16 @@ locations treated as equal.
 
    (df+df).equals(df*2)
 
+Note that the Series or DataFrame index needs to be in the same order for
+equality to be True:
+
+.. ipython:: python
+
+   df = DataFrame({'col':['foo', 0, np.nan]})
+   df2 = DataFrame({'col':[np.nan, 0, 'foo']}, index=[2,1,0])
+   df.equals(df2)
+   df.equals(df2.sort())
+
 
 Combining overlapping data sets
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

doc/source/v0.13.1.txt

@@ -121,9 +121,10 @@ API changes
 
   .. ipython:: python
 
-      df = DataFrame({'col':['foo', 0, np.nan]}).sort()
+      df = DataFrame({'col':['foo', 0, np.nan]})
       df2 = DataFrame({'col':[np.nan, 0, 'foo']}, index=[2,1,0])
-      df.equals(df)
+      df.equals(df2)
+      df.equals(df2.sort())
 
       import pandas.core.common as com
       com.array_equivalent(np.array([0, np.nan]), np.array([0, np.nan]))

doc/source/v0.15.0.txt

@@ -1110,3 +1110,4 @@ Bug Fixes
 - Regression in ``NDFrame.loc`` indexing when rows/columns were converted to Float64Index if target was an empty list/ndarray (:issue:`7774`)
 - Bug in ``Series`` that allows it to be indexed by a ``DataFrame`` which has unexpected results.  Such indexing is no longer permitted (:issue:`8444`)
 - Bug in item assignment of a ``DataFrame`` with multi-index columns where right-hand-side columns were not aligned (:issue:`7655`)
+- Suppress FutureWarning generated by NumPy when comparing object arrays containing NaN for equality (:issue:`7065`)

pandas/core/common.py

@@ -406,6 +406,7 @@ def array_equivalent(left, right, strict_nan=False):
     >>> array_equivalent(np.array([1, nan, 2]), np.array([1, 2, nan]))
     False
     """
+
     left, right = np.asarray(left), np.asarray(right)
     if left.shape != right.shape: return False
 
@@ -414,8 +415,8 @@ def array_equivalent(left, right, strict_nan=False):
 
         if not strict_nan:
             # pd.isnull considers NaN and None to be equivalent.
-            return ((left == right) | (pd.isnull(left) & pd.isnull(right))).all()
-
+            return lib.array_equivalent_object(left.ravel(), right.ravel())
+        
         for left_value, right_value in zip(left, right):
             if left_value is tslib.NaT and right_value is not tslib.NaT:
                 return False
@@ -426,7 +427,6 @@ def array_equivalent(left, right, strict_nan=False):
             else:
                 if left_value != right_value:
                     return False
-
         return True
 
     # NaNs can occur in float and complex arrays.

pandas/core/index.py

@@ -1047,7 +1047,7 @@ def equals(self, other):
         if type(other) != Index:
             return other.equals(self)
 
-        return array_equivalent(self, other)
+        return array_equivalent(_values_from_object(self), _values_from_object(other))
 
     def identical(self, other):
         """Similar to equals, but check that other comparable attributes are
@@ -2260,7 +2260,7 @@ def equals(self, other):
         #     return False
 
         try:
-            return array_equivalent(self, other)
+            return array_equivalent(_values_from_object(self), _values_from_object(other))
         except TypeError:
             # e.g. fails in numpy 1.6 with DatetimeIndex #1681
             return False
@@ -4175,7 +4175,8 @@ def equals(self, other):
             return True
 
         if not isinstance(other, MultiIndex):
-            return array_equivalent(self.values, _ensure_index(other))
+            return array_equivalent(self.values,
+                                    _values_from_object(_ensure_index(other)))
 
         if self.nlevels != other.nlevels:
             return False

pandas/lib.pyx

@@ -330,6 +330,26 @@ def list_to_object_array(list obj):
     return arr
 
 
+@cython.wraparound(False)
+@cython.boundscheck(False)
+def array_equivalent_object(ndarray left, ndarray right):
+    cdef Py_ssize_t i, n
+    cdef object lobj, robj
+
+    n = len(left)
+    for i from 0 <= i < n:
+        lobj = left[i]
+        robj = right[i]
+
+        # we are either not equal or both nan
+        # I think None == None will be true here
+        if lobj != robj:
+            if checknull(lobj) and checknull(robj):
+                continue
+            return False
+    return True
+
+
 @cython.wraparound(False)
 @cython.boundscheck(False)
 def fast_unique(ndarray[object] values):