ENH: Multi-level join on multi-indexes #16162

harisbal · 2017-04-27T22:28:16Z

Allow for merging on multiple levels of multi-indexes

gfyoung · 2017-04-29T20:44:36Z

pandas/tests/reshape/test_merge.py

@@ -1136,14 +1136,14 @@ def test_join_multi_levels(self):

        def f():
            household.join(portfolio, how='inner')
-        pytest.raises(ValueError, f)
+        self.assertRaises(ValueError, f)



Do not change this. We're using the pytest framework.

gfyoung · 2017-04-29T20:45:10Z

pandas/tests/reshape/test_merge.py


        portfolio2 = portfolio.copy()
        portfolio2.index.set_names(['household_id', 'foo'])

        def f():
            portfolio2.join(portfolio, how='inner')
-        pytest.raises(ValueError, f)
+        self.assertRaises(ValueError, f)



Same as here.

gfyoung · 2017-04-29T20:45:23Z

pandas/tests/reshape/test_merge.py

+
+        def f():
+            matrix.join(distances2, how='left')
+        self.assertRaises(TypeError, f)


Same as here.

gfyoung · 2017-04-29T20:45:27Z

pandas/tests/reshape/test_merge.py

+
+        def f():
+            matrix.join(distances2, how='left')
+        self.assertRaises(ValueError, f)


Same as here.

codecov · 2017-05-15T16:10:14Z

Codecov Report

Merging #16162 into master will decrease coverage by 0.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master   #16162      +/-   ##
==========================================
- Coverage   90.87%   90.85%   -0.02%     
==========================================
  Files         162      162              
  Lines       50816    50852      +36     
==========================================
+ Hits        46178    46202      +24     
- Misses       4638     4650      +12

Flag	Coverage Δ
#multiple	`88.63% <100%> (-0.02%)`	⬇️
#single	`40.3% <0%> (-0.03%)`	⬇️

Impacted Files	Coverage Δ
pandas/core/indexes/base.py	`96.27% <100%> (+0.07%)`	⬆️
pandas/plotting/_converter.py	`63.54% <0%> (-1.82%)`	⬇️

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 075eca1...75dea48. Read the comment docs.

codecov · 2017-05-15T16:10:21Z

Codecov Report

❗ No coverage uploaded for pull request base (master@0d86742). Click here to learn what that means.
The diff coverage is 100%.

@@            Coverage Diff            @@
##             master   #16162   +/-   ##
=========================================
  Coverage          ?   91.22%           
=========================================
  Files             ?      163           
  Lines             ?    49837           
  Branches          ?        0           
=========================================
  Hits              ?    45462           
  Misses            ?     4375           
  Partials          ?        0

Flag	Coverage Δ
#multiple	`89.02% <100%> (?)`
#single	`40.3% <3.92%> (?)`

Impacted Files	Coverage Δ
pandas/core/indexes/base.py	`96.37% <100%> (ø)`
pandas/core/reshape/merge.py	`94.52% <100%> (ø)`

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 0d86742...5da2e44. Read the comment docs.

harisbal · 2017-05-17T14:51:02Z

Can anyone give some insight on the failure of Travis CI tests?

TomAugspurger · 2017-05-17T15:55:05Z

Some linting (style) issues, starting at https://travis-ci.org/pandas-dev/pandas/jobs/232494137#L1403

You can pip install flake8 and run it on the files your changing.

jreback · 2017-06-01T23:39:28Z

pandas/core/indexes/base.py

+
+            return join_index, lidx, ridx
+
+        else:


you don't need the else here

jreback · 2017-06-01T23:39:42Z

pandas/tests/reshape/test_merge.py

@@ -1195,7 +1191,7 @@ def f():
        expected = (
            DataFrame(dict(
                household_id=[1, 2, 2, 2, 2, 3, 3, 3, 3, 3, 3, 4],
-                asset_id=["nl0000301109", "nl0000289783", "gb00b03mlx29",
+                asset_id=["nl0000301109", "nl0000301109", "gb00b03mlx29",


why is this changed?

I believe it was a mistake in the first instance. The household dataframe in this method suggests this correction

why does the existing code not fail then? before this change

This part was raising a NotImplementedError so it was never actually evaluated

jreback · 2017-06-01T23:41:04Z

pandas/tests/reshape/test_merge.py

+
+        assert_frame_equal(result, expected)
+
+    def test_join_multi_levels3(self):


where are the error cases, e.g. mistmatch on multi-levels? misnamed levels.

https://github.com/harisbal/pandas/blob/4ed307074a4c376db209b44aae5b7eef00276d26/pandas/tests/reshape/test_merge.py#L1339-L1352

jreback · 2017-06-01T23:41:35Z

pandas/core/indexes/base.py

+            other_tmp = other.droplevel(rdrop_levels)
+
+            if not (other_tmp.is_unique and self_tmp.is_unique):
+                raise TypeError(" The index resulting from the overlapping "


need testing for this

https://github.com/harisbal/pandas/blob/4ed307074a4c376db209b44aae5b7eef00276d26/pandas/tests/reshape/test_merge.py#L1323-L1337

jreback · 2017-06-01T23:42:17Z

pandas/core/indexes/base.py

-        self_is_mi = isinstance(self, MultiIndex)
-        other_is_mi = isinstance(other, MultiIndex)
+
+        def _complete_join():


this should be a separate method, too confusing here

harisbal

Changes implemented

jreback · 2017-07-01T08:26:52Z

pandas/core/indexes/base.py

@@ -3028,27 +3028,61 @@ def join(self, other, how='left', level=None, return_indexers=False,

    def _join_multi(self, other, how, return_indexers=True):
        from .multi import MultiIndex
-        self_is_mi = isinstance(self, MultiIndex)
-        other_is_mi = isinstance(other, MultiIndex)

        # figure out join names
        self_names = [n for n in self.names if n is not None]
        other_names = [n for n in other.names if n is not None]
        overlap = list(set(self_names) & set(other_names))


this should remain a set

jreback · 2017-07-01T08:27:43Z

pandas/core/indexes/base.py

-        # make the indices into mi's that match
-        if not (self_is_mi and other_is_mi):
+            if not (self_tmp.is_unique and other_tmp.is_unique):
+                raise TypeError(" The index resulting from the overlapping "


use the original error message and it should be a ValueError

jreback · 2017-07-01T08:28:40Z

pandas/core/indexes/base.py

+                join_index = self
+            elif how == 'right':
+                join_index = other
+            else:


else is unecessary

jreback · 2017-07-01T08:29:52Z

pandas/core/indexes/base.py

+
+            return join_index, lidx, ridx
+        else:
+            jl = overlap[0]


don't need the else here as you are returning

jreback · 2017-07-01T08:30:17Z

pandas/core/indexes/base.py

+        ldrop_lvls = [l for l in self_names if l not in overlap]
+        rdrop_lvls = [l for l in other_names if l not in overlap]
+
+        self_is_mi = isinstance(self, MultiIndex)


instead of doing this inline it should be a separate method, its too long

jreback · 2017-07-01T08:30:41Z

pandas/core/indexes/base.py

-        raise NotImplementedError("merging with both multi-indexes is not "
-                                  "implemented")
+    def _complete_multi_join(self, other, join_idx, lidx, ridx, dropped_lvls):
+        new_lvls = join_idx.levels


doc-string on what this is

jreback · 2017-07-01T08:31:29Z

pandas/core/indexes/base.py

+        for n in dropped_lvls:
+            if n in self.names:
+                idx = lidx
+                lvls = self.levels[self.names.index(n)].values


don't abbreviate on the variable names, use new_levels and new_labels etc

jreback · 2017-07-01T08:32:35Z

pandas/tests/reshape/test_merge.py

+        result = matrix.join(distances, how='outer')
+        assert_frame_equal(result, expected)
+
+        # Non-unique resulting index


separate test for non-uniques

jreback · 2017-07-01T08:33:21Z

pandas/tests/reshape/test_merge.py

+                     Destination=[1, 2, 1, 3, 1],
+                     Period=['AM', 'PM', 'IP', 'AM', 'OP'],
+                     TripPurp=['hbw', 'nhb', 'hbo', 'nhb', 'hbw'],
+                     Trips=[1987, 3647, 2470, 4296, 4444],


in fact why don't you move all of these added tests to a new TestMultiMulti class for testing (and other relevant tests if any).

jreback · 2017-07-27T00:00:02Z

pls rebase and update

pep8speaks · 2017-07-27T11:14:02Z

Hello @harisbal! Thanks for updating the PR.

In the file pandas/tests/reshape/merge/test_merge.py, following are the PEP8 issues :

Line 1433:5: E303 too many blank lines (2)

Comment last updated on March 11, 2018 at 13:40 Hours UTC

harisbal

Rebased

TomAugspurger

cc @jreback

jreback

a number of comments

jreback · 2017-09-26T12:51:46Z

pandas/core/indexes/base.py

+            other_tmp = other.droplevel(rdrop_levels)
+
+            if not (self_tmp.is_unique and other_tmp.is_unique):
+                raise ValueError("Join on level between two MultiIndex objects"


test for this?

jreback · 2017-09-26T12:53:01Z

pandas/core/indexes/base.py

+                elif how == 'right':
+                    join_index = other
+
+                levels = join_index.levels


move these 3 assignments after the .join, then you can make the 'how' an if/elif (for each of outer/left/right)

jreback · 2017-09-26T12:53:33Z

pandas/core/indexes/base.py

-                if isinstance(result, tuple):
-                    return result[0], result[2], result[1]
-            return result
+        flip_order = False


huh?

if you need to do something based on the how, then do it above

The part from jl = list(overlap)[0] and onwards refers to cases where not both indexes are multiindexes.
This is legacy code that I didn't modify. Shall I try to rewrite?

jreback · 2017-09-26T12:53:54Z

pandas/core/indexes/base.py

+        result = self._join_level(other, level, how=how,
+                                  return_indexers=return_indexers)
+
+        if flip_order:


this is non-intuitive here, have no idea what is going one

Same as above. Legacy code that I didn't modify

jreback · 2017-09-26T12:55:54Z

pandas/core/indexes/base.py

+                return result[0], result[2], result[1]
+        return result
+
+    def _complete_outer_join(self, other, join_idx, lidx, ridx,


move to this pandas/core/reshape/merge.py (below _get_join_indexers) and just import where you need

jreback · 2017-09-26T12:57:57Z

pandas/tests/reshape/test_merge.py

+        # Multi-index join tests
+        # Self join
+        matrix = (
+            pd.DataFrame(


pls parametrize or use fixtures for this.

jreback · 2017-09-26T12:58:15Z

pandas/tests/reshape/test_merge.py

+
+        assert_frame_equal(result, expected)
+
+    def test_join_multi_levels3(self):


rename to something more useful

jreback · 2017-09-26T12:59:19Z

also the docs in merging.rst need updating. pls add a small sub-section in 0.21.0 as well showcasing the new functionaility.

harisbal · 2017-09-26T13:01:08Z

Thank you @jreback, I'll implement the changes

jreback · 2017-09-26T13:05:04Z

@harisbal great. sorry it took so long on this.

harisbal · 2017-09-26T13:27:25Z

It was actually my fault. Thanks again for all the comments :)

jreback · 2017-10-01T17:52:12Z

pandas/core/reshape/merge.py

@@ -1066,6 +1066,54 @@ def _get_join_indexers(left_keys, right_keys, sort=False, how='inner',
    return join_func(lkey, rkey, count, **kwargs)


+def _complete_outer_join(self, other, join_idx, lidx, ridx, dropped_levels):


call these left/right (not self, other), add to doc-string

jreback · 2017-10-01T17:53:05Z

pandas/core/reshape/merge.py

+    Complete the index in case of outer join
+
+    Parameters
+    ----------


document like

join_index : Index the index of the join between common leves lidx : intp array left indexer ....

etc

jreback · 2017-10-01T17:55:51Z

pandas/core/indexes/base.py

+            dropped_lvls = ldrop_levels + rdrop_levels
+
+            # tmp_index is equivalent of index when how='inner'
+            tmp_index, lidx, ridx = self_tmp.join(other_tmp, how=how,


don't use tmp_index call this something else (and self_tmp)

jreback · 2017-10-01T17:56:28Z

pandas/core/indexes/base.py

+            # Append to the returned Index the non-overlapping levels
+            dropped_lvls = ldrop_levels + rdrop_levels
+
+            # tmp_index is equivalent of index when how='inner'


explicity enumerate all of the how cases

if how == 'innter': pass elif how == 'outer': # outer_join elif how == 'left' .....

jreback · 2017-10-02T00:01:52Z

doc/source/merging.rst

@@ -1120,6 +1120,13 @@ This is not Implemented via ``join`` at-the-moment, however it can be done using
          labels=['left', 'right'], vertical=False);
   plt.close('all');

+For previous versions can be done using the following.


rather I would say this is equivalent to

jreback

can you add some tests versus empty frames (matching levels but empty). test all the hows's

jreback · 2017-10-02T00:02:36Z

doc/source/merging.rst

@@ -1098,7 +1098,8 @@ This is equivalent but less verbose and more memory efficient / faster than this
 Joining with two multi-indexes
 ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

-This is not Implemented via ``join`` at-the-moment, however it can be done using the following.
+.. versionadded:: 0.21.0
+As of version 0.21.0 joining on two multi-indexes is possible:


can you add a similar example to a sub-section in whatsnew (and put a ref to here). you can use this example and show how we could do it previously (e.g. the below section), and how it will just work now.

Shall I wait to include it in 0.21.1?

going to be for next version (so you can wait to move it is ok), after we tag can merge shortly after.

jreback · 2017-10-02T00:03:22Z

pandas/core/indexes/base.py

+                                               verify_integrity=False)
+
+            return multilevel_join_index, lidx, ridx
+
        if not (self_is_mi and other_is_mi):


I think this just becomes an else here

or you can just remove the if as well (IOW we have covered both cases).

jreback · 2017-10-02T00:04:37Z

pandas/core/reshape/merge.py

+        right indexer
+    dropped_levels : str array
+        list of non-common levels
+    Returns


blank line before Returns

jreback · 2017-10-02T00:05:29Z

pandas/core/reshape/merge.py

+    """
+
+    if how == 'outer':
+        levels = join_idx.levels


maybe a few comments on what you are doing here (for the outer case), IOW why iterating over the dropped levels, etc.

jreback · 2017-10-02T00:07:52Z

pandas/tests/reshape/test_merge.py

+
+        result = left.join(right, how=how)
+        tm.assert_frame_equal(result, expected)
+


is it possible to generically also tests the equivalency (this is just a copy-paste from above, obviously have to adapt here)

result = (merge(household.reset_index(), log_return.reset_index(), on=['asset_id'], how='inner') .set_index(['household_id', 'asset_id', 't']))

jreback · 2017-10-02T00:09:21Z

pandas/tests/reshape/test_merge.py

+
+
+class TestJoinMultiMulti(object):
+    @pytest.fixture


I would simply put the fixture outside the class, right above TestJoinMultiMulti, then you can just pass them in as needed

jreback · 2017-10-02T00:09:42Z

pandas/tests/reshape/test_merge.py

+                                               ('right', expected_rightj),
+                                               ('inner', expected_innerj),
+                                               ('outer', expected_outerj)])
+    def test_join_multi_multi(self, how, expected):


e.g. you pass left, right in here

jreback · 2017-10-02T00:10:16Z

pandas/tests/reshape/test_merge.py

+        .set_index(['Origin', 'Destination', 'Period',
+                    'TripPurp', 'LinkType']))
+
+    @pytest.mark.parametrize('how, expected', [('left', expected_leftj),


these you actually DO need to evaluate (e.g. expected_left()) because these are simply calling a function and are not 'parameters' (if they are in the signature they are parameters).

jreback · 2017-10-02T00:10:55Z

pandas/tests/reshape/test_merge.py

 class TestMergeCategorical(object):

-    def test_identical(self, left):
+    @pytest.fixture


would leave these where they were

jreback · 2017-10-02T00:11:19Z

pandas/tests/reshape/test_merge.py

@@ -1439,9 +1597,11 @@ def test_identical(self, left):
                          index=['X', 'Y_x', 'Y_y'])
        assert_series_equal(result, expected)

-    def test_basic(self, left, right):


so just leave this, it is very typical, writing left = self.left() is not.

harisbal · 2017-10-19T20:51:42Z

@jreback I have implemented the changes but one test is failing and I'm struggling to understand why. Shall I upload so you can give it a look?

jreback · 2017-10-19T23:34:51Z

@harisbal what test is failing? l

jreback · 2017-11-10T21:13:24Z

can you rebase

…s-dev#19875)

…das-dev#19847)

… Series construction. (pandas-dev#19762)

…dev#19899)

…-dev#19829)

)

…68 (pandas-dev#19915)

* Added seek to buffer to fix xlwt asv failure * Added conditional to check for seek on xlrd object

Rebase

# Conflicts: # doc/source/merging.rst # doc/source/whatsnew/v0.23.0.txt # pandas/core/frame.py # pandas/core/generic.py # pandas/core/indexes/base.py # pandas/core/ops.py # pandas/core/reshape/merge.py # pandas/plotting/_misc.py # pandas/tests/reshape/merge/test_merge.py

harisbal · 2018-03-11T20:22:38Z

This pull request is dirty beyond repair. I'll squash and start a new one

harisbal changed the title ~~ENH: Multi-level merge on multi-indexes~~ ENH: Multi-level join on multi-indexes Apr 27, 2017

jreback added the Reshaping Concat, Merge/Join, Stack/Unstack, Explode label Apr 27, 2017

gfyoung reviewed Apr 29, 2017

View reviewed changes

harisbal force-pushed the multi-index-merge branch 2 times, most recently from a6ac2d3 to 4ed3070 Compare May 31, 2017 22:16

jreback requested changes Jun 1, 2017

View reviewed changes

harisbal commented Jun 22, 2017

View reviewed changes

jreback requested changes Jul 1, 2017

View reviewed changes

harisbal commented Sep 26, 2017

View reviewed changes

TomAugspurger approved these changes Sep 26, 2017

View reviewed changes

jreback requested changes Sep 26, 2017

View reviewed changes

jreback requested changes Oct 1, 2017

View reviewed changes

jreback reviewed Oct 2, 2017

View reviewed changes

jreback requested changes Oct 2, 2017

View reviewed changes

WillAyd and others added 21 commits February 28, 2018 00:04

Fixed pct_change with 'fill_method' returning NaN instead of 0 (panda…

e5981d1

…s-dev#19875)

Use pandas_datetimestruct instead of date_info (pandas-dev#19874)

a00f41a

Fix+test DTI/TDI/PI add/sub with ndarray[datetime64/timedelta64] (pan…

86dfeae

…das-dev#19847)

Fixed issue with leftover test.json file (pandas-dev#19879)

21bc4d5

ENH: ISO8601-compliant datetime string conversion in iterrows() and…

b9149b0

… Series construction. (pandas-dev#19762)

parameterize test_pct_change_periods_freq (pandas-dev#19897)

48785e6

DOC: Make API reference intro section concise (pandas-dev#19846)

6003ff6

DOC/BLD: unpin sphinx to use sphinx 1.7 (pandas-dev#19687)

769e4c2

DOC: fix numpydoc section titles in misc plotting docstrings (pandas-…

51093fc

…dev#19899)

DOC: small typo fix (pandas-dev#19901)

eed8092

cleanup order of operations kludges (pandas-dev#19895)

e3c5467

make ops.add_foo take just class (pandas-dev#19828)

d9f57a4

Test Decorators and Better Pytest Integration in 'test_excel' (pandas…

3471271

…-dev#19829)

BUG: Fix Series constructor for Categorical with index (pandas-dev#19714

5508704

)

CLN: Remove Series._from_array (pandas-dev#19893)

69f0e8b

DOC fix incorrect example in DataFrame.to_dict docstring. Close GH198…

25fc828

…68 (pandas-dev#19915)

handle NaT add/sub in one place (pandas-dev#19903)

2d10b35

ASV: Added seek to buffer to fix xlwt asv failure (pandas-dev#19926)

892dd3d

* Added seek to buffer to fix xlwt asv failure * Added conditional to check for seek on xlrd object

TST: Debug flaky plotting test (pandas-dev#19925)

dc4bf8a

Rebase

4a4f3f2

Rebase

Merge remote-tracking branch 'pandas-dev/master'

f00a2d1

harisbal force-pushed the multi-index-merge branch from f5d9c20 to 2ccbe5b Compare March 11, 2018 13:34

harisbal and others added 3 commits March 11, 2018 19:35

Rebase

593d6cb

Merge branch 'master' of https://github.com/pandas-dev/pandas

4708db0

harisbal force-pushed the multi-index-merge branch from c562204 to 5da2e44 Compare March 11, 2018 20:20

harisbal closed this Mar 11, 2018

harisbal deleted the multi-index-merge branch March 12, 2018 22:55

harisbal mentioned this pull request Mar 15, 2018

ENH: Allow for join between two multi-index dataframe instances #20356

Merged


		assert_frame_equal(result, expected)

		def test_join_multi_levels3(self):

		@@ -1066,6 +1066,54 @@ def _get_join_indexers(left_keys, right_keys, sort=False, how='inner',
		return join_func(lkey, rkey, count, **kwargs)


		def _complete_outer_join(self, other, join_idx, lidx, ridx, dropped_levels):


		result = left.join(right, how=how)
		tm.assert_frame_equal(result, expected)

ENH: Multi-level join on multi-indexes #16162

ENH: Multi-level join on multi-indexes #16162

Conversation

harisbal commented Apr 27, 2017 • edited by jreback Loading

gfyoung Apr 29, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

codecov bot commented May 15, 2017

Codecov Report

codecov bot commented May 15, 2017 • edited Loading

Codecov Report

harisbal commented May 17, 2017

TomAugspurger commented May 17, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

harisbal Jun 5, 2017 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

harisbal left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Jul 27, 2017

pep8speaks commented Jul 27, 2017 • edited Loading

Comment last updated on March 11, 2018 at 13:40 Hours UTC

harisbal left a comment

Choose a reason for hiding this comment

TomAugspurger left a comment

Choose a reason for hiding this comment

jreback left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback commented Sep 26, 2017

harisbal commented Sep 26, 2017

jreback commented Sep 26, 2017

harisbal commented Sep 26, 2017

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

jreback left a comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

harisbal commented Oct 19, 2017

harisbal commented Apr 27, 2017 •

edited by jreback

Loading

gfyoung Apr 29, 2017 •

edited

Loading

codecov bot commented May 15, 2017 •

edited

Loading

harisbal Jun 5, 2017 •

edited

Loading

pep8speaks commented Jul 27, 2017 •

edited

Loading