BUG: Don't parse NA-values in index when requested (#18127)

Closes gh-5239.

Author: gfyoung

doc/source/whatsnew/v0.21.1.txt

@@ -79,6 +79,7 @@ I/O
 
 - Bug in class:`~pandas.io.stata.StataReader` not converting date/time columns with display formatting addressed (:issue:`17990`). Previously columns with display formatting were normally left as ordinal numbers and not converted to datetime objects.
 - Bug in :func:`read_csv` when reading a compressed UTF-16 encoded file (:issue:`18071`)
+- Bug in :func:`read_csv` for handling null values in index columns when specifying ``na_filter=False`` (:issue:`5239`)
 - Bug in :meth:`DataFrame.to_csv` when the table had ``MultiIndex`` columns, and a list of strings was passed in for ``header`` (:issue:`5539`)
 
 Plotting

pandas/io/parsers.py

@@ -1249,6 +1249,8 @@ def __init__(self, kwds):
 
         self.na_values = kwds.get('na_values')
         self.na_fvalues = kwds.get('na_fvalues')
+        self.na_filter = kwds.get('na_filter', False)
+
         self.true_values = kwds.get('true_values')
         self.false_values = kwds.get('false_values')
         self.as_recarray = kwds.get('as_recarray', False)
@@ -1424,7 +1426,6 @@ def _make_index(self, data, alldata, columns, indexnamerow=False):
         elif not self._has_complex_date_col:
             index = self._get_simple_index(alldata, columns)
             index = self._agg_index(index)
-
         elif self._has_complex_date_col:
             if not self._name_processed:
                 (self.index_names, _,
@@ -1504,8 +1505,12 @@ def _agg_index(self, index, try_parse_dates=True):
             if try_parse_dates and self._should_parse_dates(i):
                 arr = self._date_conv(arr)
 
-            col_na_values = self.na_values
-            col_na_fvalues = self.na_fvalues
+            if self.na_filter:
+                col_na_values = self.na_values
+                col_na_fvalues = self.na_fvalues
+            else:
+                col_na_values = set()
+                col_na_fvalues = set()
 
             if isinstance(self.na_values, dict):
                 col_name = self.index_names[i]
@@ -2060,8 +2065,6 @@ def __init__(self, f, **kwds):
 
         self.names_passed = kwds['names'] or None
 
-        self.na_filter = kwds['na_filter']
-
         self.has_index_names = False
         if 'has_index_names' in kwds:
             self.has_index_names = kwds['has_index_names']

pandas/tests/io/parser/na_values.py

@@ -312,3 +312,21 @@ def test_empty_na_values_no_default_with_index(self):
         out = self.read_csv(StringIO(data), keep_default_na=False, index_col=0)
 
         tm.assert_frame_equal(out, expected)
+
+    def test_no_na_filter_on_index(self):
+        # see gh-5239
+        data = "a,b,c\n1,,3\n4,5,6"
+
+        # Don't parse NA-values in index when na_filter=False.
+        out = self.read_csv(StringIO(data), index_col=[1], na_filter=False)
+
+        expected = DataFrame({"a": [1, 4], "c": [3, 6]},
+                             index=Index(["", "5"], name="b"))
+        tm.assert_frame_equal(out, expected)
+
+        # Parse NA-values in index when na_filter=True.
+        out = self.read_csv(StringIO(data), index_col=[1], na_filter=True)
+
+        expected = DataFrame({"a": [1, 4], "c": [3, 6]},
+                             index=Index([np.nan, 5.0], name="b"))
+        tm.assert_frame_equal(out, expected)