BUG: Patch handling no NA values in TextFileReader

When cleaning `na_values` during initialization of `TextFileReader`,
we return a `list` whenever we specify that `na_values` should be
empty.  However, the rest of the code expects a `set`.

Closes #15835.

Author: gfyoung <[email protected]>

Closes #15881 from gfyoung/keep-default-na-excel and squashes the following commits:

0bb6f64 [gfyoung] BUG: Patch handling no NA values in TextFileReader

Author: gfyoung

doc/source/whatsnew/v0.20.0.txt

@@ -995,6 +995,7 @@ I/O
 - Bug in ``pd.read_csv()`` for the C engine where ``usecols`` were being indexed incorrectly with ``parse_dates`` (:issue:`14792`)
 - Bug in ``pd.read_csv()`` with ``parse_dates`` when multiline headers are specified (:issue:`15376`)
 - Bug in ``pd.read_csv()`` with ``float_precision='round_trip'`` which caused a segfault when a text entry is parsed (:issue:`15140`)
+- Bug in ``pd.read_csv()`` when an index was specified and no values were specified as null values (:issue:`15835`)
 - Added checks in ``pd.read_csv()`` ensuring that values for ``nrows`` and ``chunksize`` are valid (:issue:`15767`)
 - Bug in ``pd.tools.hashing.hash_pandas_object()`` in which hashing of categoricals depended on the ordering of categories, instead of just their values. (:issue:`15143`)
 - Bug in ``.to_json()`` where ``lines=True`` and contents (keys or values) contain escaped characters (:issue:`15096`)

pandas/io/parsers.py

@@ -2890,7 +2890,7 @@ def _clean_na_values(na_values, keep_default_na=True):
         if keep_default_na:
             na_values = _NA_VALUES
         else:
-            na_values = []
+            na_values = set()
         na_fvalues = set()
     elif isinstance(na_values, dict):
         na_values = na_values.copy()  # Prevent aliasing.

pandas/tests/io/parser/na_values.py

@@ -11,7 +11,7 @@
 import pandas.io.parsers as parsers
 import pandas.util.testing as tm
 
-from pandas import DataFrame, MultiIndex
+from pandas import DataFrame, Index, MultiIndex
 from pandas.compat import StringIO, range
 
 
@@ -303,3 +303,12 @@ def test_na_values_uint64(self):
         expected = DataFrame([[str(2**63), 1], ['', 2]])
         out = self.read_csv(StringIO(data), header=None)
         tm.assert_frame_equal(out, expected)
+
+    def test_empty_na_values_no_default_with_index(self):
+        # see gh-15835
+        data = "a,1\nb,2"
+
+        expected = DataFrame({'1': [2]}, index=Index(["b"], name="a"))
+        out = self.read_csv(StringIO(data), keep_default_na=False, index_col=0)
+
+        tm.assert_frame_equal(out, expected)