REGR: Fixes first_valid_index when DataFrame or Series has duplicate row index (GH21441)

doc/source/whatsnew/v0.23.2.txt

@@ -16,7 +16,7 @@ and bug fixes. We recommend that all users upgrade to this version.
 Fixed Regressions
 ~~~~~~~~~~~~~~~~~
 
--
+- Bug in both :meth:`DataFrame.first_valid_index` and :meth:`Series.first_valid_index` raised for a row index with duplicate values (:issue:`21441`)
 -
 
 .. _whatsnew_0232.performance:

pandas/core/generic.py

@@ -8969,18 +8969,21 @@ def _find_valid_index(self, how):
             is_valid = is_valid.any(1)  # reduce axis 1
 
         if how == 'first':
-            # First valid value case
-            i = is_valid.idxmax()
-            if not is_valid[i]:
-                return None
-            return i
-
-        elif how == 'last':
-            # Last valid value case
-            i = is_valid.values[::-1].argmax()
-            if not is_valid.iat[len(self) - i - 1]:
-                return None
-            return self.index[len(self) - i - 1]
+            idx = is_valid.idxmax()
+            if isinstance(is_valid[idx], ABCSeries):
+                for chk_notna in is_valid[idx]:
+                    chk_notna = False or chk_notna
+            else:
+                chk_notna = is_valid[idx]
+
+        if how == 'last':
+            idxpos = len(self) - 1 - is_valid.values[::-1].argmax()
+            chk_notna = is_valid.iat[idxpos]
+            idx = self.index[idxpos]
+
+        if not chk_notna:
+            return None
+        return idx
 
     @Appender(_shared_docs['valid_index'] % {'position': 'first',
                                              'klass': 'NDFrame'})

pandas/tests/generic/test_generic.py

@@ -612,6 +612,16 @@ def test_pct_change(self, periods, fill_method, limit, exp):
         else:
             tm.assert_series_equal(res, Series(exp))
 
+    @pytest.mark.parametrize("data,index,expected_first,expected_last", [
+        ({'A': [1, 2, 3]}, [1, 1, 2], 1, 2),
+        ({'A': [1, 2, 3]}, [1, 2, 2], 1, 2),
+        ({'A': [1, 2, 3, 4]}, ['d', 'd', 'd', 'd'], 'd', 'd')])
+    def test_valid_index(self, data, index, expected_first, expected_last):
+        # GH 21441
+        df = DataFrame(data, index=index)
+        assert expected_first == df.first_valid_index()
+        assert expected_last == df.last_valid_index()
+
 
 class TestNDFrame(object):
     # tests that don't fit elsewhere