From 65762428c92b944358bbe4ab21ea2c41500e2ff8 Mon Sep 17 00:00:00 2001
From: HH <jwalag87@gmail.com>
Date: Tue, 10 Sep 2019 14:09:49 +0900
Subject: [PATCH] EHN: Groupby on multiindex with missing data in group keys
 raises IndexError (#20519)

* If all index values in some level is NA, fill with NaN
---
 doc/source/whatsnew/v1.0.0.rst        |  2 +-
 pandas/core/indexes/multi.py          |  5 ++++-
 pandas/tests/groupby/test_grouping.py | 32 +++++++++++++++++++++++++++
 3 files changed, 37 insertions(+), 2 deletions(-)

diff --git a/doc/source/whatsnew/v1.0.0.rst b/doc/source/whatsnew/v1.0.0.rst
index 4decc99087a9e4..4a1e874f0c8d7d 100644
--- a/doc/source/whatsnew/v1.0.0.rst
+++ b/doc/source/whatsnew/v1.0.0.rst
@@ -174,7 +174,7 @@ Groupby/resample/rolling
 ^^^^^^^^^^^^^^^^^^^^^^^^
 
 -
--
+- ``IndexError`` would not raise if all index values in some index level is missing data (:issue:`20519`)
 - Bug in :meth:`DataFrame.groupby` not offering selection by column name when ``axis=1`` (:issue:`27614`)
 
 Reshaping
diff --git a/pandas/core/indexes/multi.py b/pandas/core/indexes/multi.py
index b614952ba1e043..528ad8e66421ca 100644
--- a/pandas/core/indexes/multi.py
+++ b/pandas/core/indexes/multi.py
@@ -1279,7 +1279,10 @@ def _get_grouper_for_level(self, mapper, level):
             # Remove unobserved levels from level_index
             level_index = level_index.take(uniques)
 
-        grouper = level_index.take(codes)
+        if len(level_index):
+            grouper = level_index.take(codes)
+        else:
+            grouper = level_index.take(codes,fill_value=True)
 
         return grouper, codes, level_index
 
diff --git a/pandas/tests/groupby/test_grouping.py b/pandas/tests/groupby/test_grouping.py
index 403f5f11ee7686..0fbd87d0f15309 100644
--- a/pandas/tests/groupby/test_grouping.py
+++ b/pandas/tests/groupby/test_grouping.py
@@ -628,6 +628,38 @@ def test_groupby_empty(self):
         # check name
         assert s.groupby(s).grouper.names == ["name"]
 
+    def test_groupby_level_index_value_all_na(self):
+        # issue 20519
+        df = pd.DataFrame([["x", np.nan, 1]], columns=["A", "B", "C"]).set_index(
+            ["A", "B"]
+        )
+        result = df.groupby(level=["A", "B"]).sum()
+        expected = DataFrame(
+            data=[],
+            index=MultiIndex(
+                levels=[["x"],[]],
+                codes=[[],[]],
+                names=["A","B"]
+            ),
+            columns=["C"]
+        )
+        tm.assert_frame_equal(result, expected, check_index_type=False, check_dtype=False)
+
+        df = pd.DataFrame(
+            [[None, None, "x", 2], [np.nan, "y", np.nan, 4]], columns=["A", "B", "C", "D"]
+        ).set_index(["A", "B", "C"])
+        result = df.groupby(level=["A", "B", "C"]).sum()
+        expected = DataFrame(
+            data=[],
+            index=MultiIndex(
+                levels=[[], ["y"], ["x"]],
+                codes=[[], [], []],
+                names=["A", "B", "C"]
+            ),
+            columns=["D"]
+        )
+        tm.assert_frame_equal(result, expected, check_index_type=False, check_dtype=False)
+
 
 # get_group
 # --------------------------------