import numpy as np
from numpy.random import randn
import pytest

from pandas._libs import join as libjoin

import pandas as pd
from pandas import DataFrame, Index, MultiIndex, Series, concat, merge
from pandas.tests.reshape.merge.test_merge import NGROUPS, N, get_test_data
import pandas.util.testing as tm
from pandas.util.testing import assert_frame_equal

a_ = np.array


class TestJoin:
    def setup_method(self, method):
        # aggregate multiple columns
        self.df = DataFrame(
            {
                "key1": get_test_data(),
                "key2": get_test_data(),
                "data1": np.random.randn(N),
                "data2": np.random.randn(N),
            }
        )

        # exclude a couple keys for fun
        self.df = self.df[self.df["key2"] > 1]

        self.df2 = DataFrame(
            {
                "key1": get_test_data(n=N // 5),
                "key2": get_test_data(ngroups=NGROUPS // 2, n=N // 5),
                "value": np.random.randn(N // 5),
            }
        )

        index, data = tm.getMixedTypeDict()
        self.target = DataFrame(data, index=index)

        # Join on string value
        self.source = DataFrame(
            {"MergedA": data["A"], "MergedD": data["D"]}, index=data["C"]
        )

    def test_cython_left_outer_join(self):
        left = a_([0, 1, 2, 1, 2, 0, 0, 1, 2, 3, 3], dtype=np.int64)
        right = a_([1, 1, 0, 4, 2, 2, 1], dtype=np.int64)
        max_group = 5

        ls, rs = libjoin.left_outer_join(left, right, max_group)

        exp_ls = left.argsort(kind="mergesort")
        exp_rs = right.argsort(kind="mergesort")

        exp_li = a_([0, 1, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 7, 7, 8, 8, 9, 10])
        exp_ri = a_([0, 0, 0, 1, 2, 3, 1, 2, 3, 1, 2, 3, 4, 5, 4, 5, 4, 5, -1, -1])

        exp_ls = exp_ls.take(exp_li)
        exp_ls[exp_li == -1] = -1

        exp_rs = exp_rs.take(exp_ri)
        exp_rs[exp_ri == -1] = -1

        tm.assert_numpy_array_equal(ls, exp_ls, check_dtype=False)
        tm.assert_numpy_array_equal(rs, exp_rs, check_dtype=False)

    def test_cython_right_outer_join(self):
        left = a_([0, 1, 2, 1, 2, 0, 0, 1, 2, 3, 3], dtype=np.int64)
        right = a_([1, 1, 0, 4, 2, 2, 1], dtype=np.int64)
        max_group = 5

        rs, ls = libjoin.left_outer_join(right, left, max_group)

        exp_ls = left.argsort(kind="mergesort")
        exp_rs = right.argsort(kind="mergesort")

        #            0        1        1        1
        exp_li = a_(
            [
                0,
                1,
                2,
                3,
                4,
                5,
                3,
                4,
                5,
                3,
                4,
                5,
                #            2        2        4
                6,
                7,
                8,
                6,
                7,
                8,
                -1,
            ]
        )
        exp_ri = a_([0, 0, 0, 1, 1, 1, 2, 2, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6])

        exp_ls = exp_ls.take(exp_li)
        exp_ls[exp_li == -1] = -1

        exp_rs = exp_rs.take(exp_ri)
        exp_rs[exp_ri == -1] = -1

        tm.assert_numpy_array_equal(ls, exp_ls, check_dtype=False)
        tm.assert_numpy_array_equal(rs, exp_rs, check_dtype=False)

    def test_cython_inner_join(self):
        left = a_([0, 1, 2, 1, 2, 0, 0, 1, 2, 3, 3], dtype=np.int64)
        right = a_([1, 1, 0, 4, 2, 2, 1, 4], dtype=np.int64)
        max_group = 5

        ls, rs = libjoin.inner_join(left, right, max_group)

        exp_ls = left.argsort(kind="mergesort")
        exp_rs = right.argsort(kind="mergesort")

        exp_li = a_([0, 1, 2, 3, 3, 3, 4, 4, 4, 5, 5, 5, 6, 6, 7, 7, 8, 8])
        exp_ri = a_([0, 0, 0, 1, 2, 3, 1, 2, 3, 1, 2, 3, 4, 5, 4, 5, 4, 5])

        exp_ls = exp_ls.take(exp_li)
        exp_ls[exp_li == -1] = -1

        exp_rs = exp_rs.take(exp_ri)
        exp_rs[exp_ri == -1] = -1

        tm.assert_numpy_array_equal(ls, exp_ls, check_dtype=False)
        tm.assert_numpy_array_equal(rs, exp_rs, check_dtype=False)

    def test_left_outer_join(self):
        joined_key2 = merge(self.df, self.df2, on="key2")
        _check_join(self.df, self.df2, joined_key2, ["key2"], how="left")

        joined_both = merge(self.df, self.df2)
        _check_join(self.df, self.df2, joined_both, ["key1", "key2"], how="left")

    def test_right_outer_join(self):
        joined_key2 = merge(self.df, self.df2, on="key2", how="right")
        _check_join(self.df, self.df2, joined_key2, ["key2"], how="right")

        joined_both = merge(self.df, self.df2, how="right")
        _check_join(self.df, self.df2, joined_both, ["key1", "key2"], how="right")

    def test_full_outer_join(self):
        joined_key2 = merge(self.df, self.df2, on="key2", how="outer")
        _check_join(self.df, self.df2, joined_key2, ["key2"], how="outer")

        joined_both = merge(self.df, self.df2, how="outer")
        _check_join(self.df, self.df2, joined_both, ["key1", "key2"], how="outer")

    def test_inner_join(self):
        joined_key2 = merge(self.df, self.df2, on="key2", how="inner")
        _check_join(self.df, self.df2, joined_key2, ["key2"], how="inner")

        joined_both = merge(self.df, self.df2, how="inner")
        _check_join(self.df, self.df2, joined_both, ["key1", "key2"], how="inner")

    def test_handle_overlap(self):
        joined = merge(self.df, self.df2, on="key2", suffixes=[".foo", ".bar"])

        assert "key1.foo" in joined
        assert "key1.bar" in joined

    def test_handle_overlap_arbitrary_key(self):
        joined = merge(
            self.df,
            self.df2,
            left_on="key2",
            right_on="key1",
            suffixes=[".foo", ".bar"],
        )
        assert "key1.foo" in joined
        assert "key2.bar" in joined

    def test_join_on(self):
        target = self.target
        source = self.source

        merged = target.join(source, on="C")
        tm.assert_series_equal(merged["MergedA"], target["A"], check_names=False)
        tm.assert_series_equal(merged["MergedD"], target["D"], check_names=False)

        # join with duplicates (fix regression from DataFrame/Matrix merge)
        df = DataFrame({"key": ["a", "a", "b", "b", "c"]})
        df2 = DataFrame({"value": [0, 1, 2]}, index=["a", "b", "c"])
        joined = df.join(df2, on="key")
        expected = DataFrame(
            {"key": ["a", "a", "b", "b", "c"], "value": [0, 0, 1, 1, 2]}
        )
        assert_frame_equal(joined, expected)

        # Test when some are missing
        df_a = DataFrame([[1], [2], [3]], index=["a", "b", "c"], columns=["one"])
        df_b = DataFrame([["foo"], ["bar"]], index=[1, 2], columns=["two"])
        df_c = DataFrame([[1], [2]], index=[1, 2], columns=["three"])
        joined = df_a.join(df_b, on="one")
        joined = joined.join(df_c, on="one")
        assert np.isnan(joined["two"]["c"])
        assert np.isnan(joined["three"]["c"])

        # merge column not p resent
        with pytest.raises(KeyError, match="^'E'$"):
            target.join(source, on="E")

        # overlap
        source_copy = source.copy()
        source_copy["A"] = 0
        msg = (
            "You are trying to merge on float64 and object columns. If"
            " you wish to proceed you should use pd.concat"
        )
        with pytest.raises(ValueError, match=msg):
            target.join(source_copy, on="A")

    def test_join_on_fails_with_different_right_index(self):
        df = DataFrame(
            {"a": np.random.choice(["m", "f"], size=3), "b": np.random.randn(3)}
        )
        df2 = DataFrame(
            {"a": np.random.choice(["m", "f"], size=10), "b": np.random.randn(10)},
            index=tm.makeCustomIndex(10, 2),
        )
        msg = (
            r"len\(left_on\) must equal the number of levels in the index" ' of "right"'
        )
        with pytest.raises(ValueError, match=msg):
            merge(df, df2, left_on="a", right_index=True)

    def test_join_on_fails_with_different_left_index(self):
        df = DataFrame(
            {"a": np.random.choice(["m", "f"], size=3), "b": np.random.randn(3)},
            index=tm.makeCustomIndex(3, 2),
        )
        df2 = DataFrame(
            {"a": np.random.choice(["m", "f"], size=10), "b": np.random.randn(10)}
        )
        msg = (
            r"len\(right_on\) must equal the number of levels in the index" ' of "left"'
        )
        with pytest.raises(ValueError, match=msg):
            merge(df, df2, right_on="b", left_index=True)

    def test_join_on_fails_with_different_column_counts(self):
        df = DataFrame(
            {"a": np.random.choice(["m", "f"], size=3), "b": np.random.randn(3)}
        )
        df2 = DataFrame(
            {"a": np.random.choice(["m", "f"], size=10), "b": np.random.randn(10)},
            index=tm.makeCustomIndex(10, 2),
        )
        msg = r"len\(right_on\) must equal len\(left_on\)"
        with pytest.raises(ValueError, match=msg):
            merge(df, df2, right_on="a", left_on=["a", "b"])

    @pytest.mark.parametrize("wrong_type", [2, "str", None, np.array([0, 1])])
    def test_join_on_fails_with_wrong_object_type(self, wrong_type):
        # GH12081 - original issue

        # GH21220 - merging of Series and DataFrame is now allowed
        # Edited test to remove the Series object from test parameters

        df = DataFrame({"a": [1, 1]})
        msg = "Can only merge Series or DataFrame objects, a {} was passed".format(
            str(type(wrong_type))
        )
        with pytest.raises(TypeError, match=msg):
            merge(wrong_type, df, left_on="a", right_on="a")
        with pytest.raises(TypeError, match=msg):
            merge(df, wrong_type, left_on="a", right_on="a")

    def test_join_on_pass_vector(self):
        expected = self.target.join(self.source, on="C")
        del expected["C"]

        join_col = self.target.pop("C")
        result = self.target.join(self.source, on=join_col)
        assert_frame_equal(result, expected)

    def test_join_with_len0(self):
        # nothing to merge
        merged = self.target.join(self.source.reindex([]), on="C")
        for col in self.source:
            assert col in merged
            assert merged[col].isna().all()

        merged2 = self.target.join(self.source.reindex([]), on="C", how="inner")
        tm.assert_index_equal(merged2.columns, merged.columns)
        assert len(merged2) == 0

    def test_join_on_inner(self):
        df = DataFrame({"key": ["a", "a", "d", "b", "b", "c"]})
        df2 = DataFrame({"value": [0, 1]}, index=["a", "b"])

        joined = df.join(df2, on="key", how="inner")

        expected = df.join(df2, on="key")
        expected = expected[expected["value"].notna()]
        tm.assert_series_equal(joined["key"], expected["key"], check_dtype=False)
        tm.assert_series_equal(joined["value"], expected["value"], check_dtype=False)
        tm.assert_index_equal(joined.index, expected.index)

    def test_join_on_singlekey_list(self):
        df = DataFrame({"key": ["a", "a", "b", "b", "c"]})
        df2 = DataFrame({"value": [0, 1, 2]}, index=["a", "b", "c"])

        # corner cases
        joined = df.join(df2, on=["key"])
        expected = df.join(df2, on="key")

        assert_frame_equal(joined, expected)

    def test_join_on_series(self):
        result = self.target.join(self.source["MergedA"], on="C")
        expected = self.target.join(self.source[["MergedA"]], on="C")
        assert_frame_equal(result, expected)

    def test_join_on_series_buglet(self):
        # GH #638
        df = DataFrame({"a": [1, 1]})
        ds = Series([2], index=[1], name="b")
        result = df.join(ds, on="a")
        expected = DataFrame({"a": [1, 1], "b": [2, 2]}, index=df.index)
        tm.assert_frame_equal(result, expected)

    def test_join_index_mixed(self, join_type):
        # no overlapping blocks
        df1 = DataFrame(index=np.arange(10))
        df1["bool"] = True
        df1["string"] = "foo"

        df2 = DataFrame(index=np.arange(5, 15))
        df2["int"] = 1
        df2["float"] = 1.0

        joined = df1.join(df2, how=join_type)
        expected = _join_by_hand(df1, df2, how=join_type)
        assert_frame_equal(joined, expected)

        joined = df2.join(df1, how=join_type)
        expected = _join_by_hand(df2, df1, how=join_type)
        assert_frame_equal(joined, expected)

    def test_join_index_mixed_overlap(self):
        df1 = DataFrame(
            {"A": 1.0, "B": 2, "C": "foo", "D": True},
            index=np.arange(10),
            columns=["A", "B", "C", "D"],
        )
        assert df1["B"].dtype == np.int64
        assert df1["D"].dtype == np.bool_

        df2 = DataFrame(
            {"A": 1.0, "B": 2, "C": "foo", "D": True},
            index=np.arange(0, 10, 2),
            columns=["A", "B", "C", "D"],
        )

        # overlap
        joined = df1.join(df2, lsuffix="_one", rsuffix="_two")
        expected_columns = [
            "A_one",
            "B_one",
            "C_one",
            "D_one",
            "A_two",
            "B_two",
            "C_two",
            "D_two",
        ]
        df1.columns = expected_columns[:4]
        df2.columns = expected_columns[4:]
        expected = _join_by_hand(df1, df2)
        assert_frame_equal(joined, expected)

    def test_join_empty_bug(self):
        # generated an exception in 0.4.3
        x = DataFrame()
        x.join(DataFrame([3], index=[0], columns=["A"]), how="outer")

    def test_join_unconsolidated(self):
        # GH #331
        a = DataFrame(randn(30, 2), columns=["a", "b"])
        c = Series(randn(30))
        a["c"] = c
        d = DataFrame(randn(30, 1), columns=["q"])

        # it works!
        a.join(d)
        d.join(a)

    def test_join_multiindex(self):
        index1 = MultiIndex.from_arrays(
            [["a", "a", "a", "b", "b", "b"], [1, 2, 3, 1, 2, 3]],
            names=["first", "second"],
        )

        index2 = MultiIndex.from_arrays(
            [["b", "b", "b", "c", "c", "c"], [1, 2, 3, 1, 2, 3]],
            names=["first", "second"],
        )

        df1 = DataFrame(data=np.random.randn(6), index=index1, columns=["var X"])
        df2 = DataFrame(data=np.random.randn(6), index=index2, columns=["var Y"])

        df1 = df1.sort_index(level=0)
        df2 = df2.sort_index(level=0)

        joined = df1.join(df2, how="outer")
        ex_index = Index(index1.values).union(Index(index2.values))
        expected = df1.reindex(ex_index).join(df2.reindex(ex_index))
        expected.index.names = index1.names
        assert_frame_equal(joined, expected)
        assert joined.index.names == index1.names

        df1 = df1.sort_index(level=1)
        df2 = df2.sort_index(level=1)

        joined = df1.join(df2, how="outer").sort_index(level=0)
        ex_index = Index(index1.values).union(Index(index2.values))
        expected = df1.reindex(ex_index).join(df2.reindex(ex_index))
        expected.index.names = index1.names

        assert_frame_equal(joined, expected)
        assert joined.index.names == index1.names

    def test_join_inner_multiindex(self):
        key1 = ["bar", "bar", "bar", "foo", "foo", "baz", "baz", "qux", "qux", "snap"]
        key2 = [
            "two",
            "one",
            "three",
            "one",
            "two",
            "one",
            "two",
            "two",
            "three",
            "one",
        ]

        data = np.random.randn(len(key1))
        data = DataFrame({"key1": key1, "key2": key2, "data": data})

        index = MultiIndex(
            levels=[["foo", "bar", "baz", "qux"], ["one", "two", "three"]],
            codes=[[0, 0, 0, 1, 1, 2, 2, 3, 3, 3], [0, 1, 2, 0, 1, 1, 2, 0, 1, 2]],
            names=["first", "second"],
        )
        to_join = DataFrame(
            np.random.randn(10, 3), index=index, columns=["j_one", "j_two", "j_three"]
        )

        joined = data.join(to_join, on=["key1", "key2"], how="inner")
        expected = merge(
            data,
            to_join.reset_index(),
            left_on=["key1", "key2"],
            right_on=["first", "second"],
            how="inner",
            sort=False,
        )

        expected2 = merge(
            to_join,
            data,
            right_on=["key1", "key2"],
            left_index=True,
            how="inner",
            sort=False,
        )
        assert_frame_equal(joined, expected2.reindex_like(joined))

        expected2 = merge(
            to_join,
            data,
            right_on=["key1", "key2"],
            left_index=True,
            how="inner",
            sort=False,
        )

        expected = expected.drop(["first", "second"], axis=1)
        expected.index = joined.index

        assert joined.index.is_monotonic
        assert_frame_equal(joined, expected)

        # _assert_same_contents(expected, expected2.loc[:, expected.columns])

    def test_join_hierarchical_mixed(self):
        # GH 2024
        df = DataFrame([(1, 2, 3), (4, 5, 6)], columns=["a", "b", "c"])
        new_df = df.groupby(["a"]).agg({"b": [np.mean, np.sum]})
        other_df = DataFrame([(1, 2, 3), (7, 10, 6)], columns=["a", "b", "d"])
        other_df.set_index("a", inplace=True)
        # GH 9455, 12219
        with tm.assert_produces_warning(UserWarning):
            result = merge(new_df, other_df, left_index=True, right_index=True)
        assert ("b", "mean") in result
        assert "b" in result

    def test_join_float64_float32(self):

        a = DataFrame(randn(10, 2), columns=["a", "b"], dtype=np.float64)
        b = DataFrame(randn(10, 1), columns=["c"], dtype=np.float32)
        joined = a.join(b)
        assert joined.dtypes["a"] == "float64"
        assert joined.dtypes["b"] == "float64"
        assert joined.dtypes["c"] == "float32"

        a = np.random.randint(0, 5, 100).astype("int64")
        b = np.random.random(100).astype("float64")
        c = np.random.random(100).astype("float32")
        df = DataFrame({"a": a, "b": b, "c": c})
        xpdf = DataFrame({"a": a, "b": b, "c": c})
        s = DataFrame(np.random.random(5).astype("float32"), columns=["md"])
        rs = df.merge(s, left_on="a", right_index=True)
        assert rs.dtypes["a"] == "int64"
        assert rs.dtypes["b"] == "float64"
        assert rs.dtypes["c"] == "float32"
        assert rs.dtypes["md"] == "float32"

        xp = xpdf.merge(s, left_on="a", right_index=True)
        assert_frame_equal(rs, xp)

    def test_join_many_non_unique_index(self):
        df1 = DataFrame({"a": [1, 1], "b": [1, 1], "c": [10, 20]})
        df2 = DataFrame({"a": [1, 1], "b": [1, 2], "d": [100, 200]})
        df3 = DataFrame({"a": [1, 1], "b": [1, 2], "e": [1000, 2000]})
        idf1 = df1.set_index(["a", "b"])
        idf2 = df2.set_index(["a", "b"])
        idf3 = df3.set_index(["a", "b"])

        result = idf1.join([idf2, idf3], how="outer")

        df_partially_merged = merge(df1, df2, on=["a", "b"], how="outer")
        expected = merge(df_partially_merged, df3, on=["a", "b"], how="outer")

        result = result.reset_index()
        expected = expected[result.columns]
        expected["a"] = expected.a.astype("int64")
        expected["b"] = expected.b.astype("int64")
        assert_frame_equal(result, expected)

        df1 = DataFrame({"a": [1, 1, 1], "b": [1, 1, 1], "c": [10, 20, 30]})
        df2 = DataFrame({"a": [1, 1, 1], "b": [1, 1, 2], "d": [100, 200, 300]})
        df3 = DataFrame({"a": [1, 1, 1], "b": [1, 1, 2], "e": [1000, 2000, 3000]})
        idf1 = df1.set_index(["a", "b"])
        idf2 = df2.set_index(["a", "b"])
        idf3 = df3.set_index(["a", "b"])
        result = idf1.join([idf2, idf3], how="inner")

        df_partially_merged = merge(df1, df2, on=["a", "b"], how="inner")
        expected = merge(df_partially_merged, df3, on=["a", "b"], how="inner")

        result = result.reset_index()

        assert_frame_equal(result, expected.loc[:, result.columns])

        # GH 11519
        df = DataFrame(
            {
                "A": ["foo", "bar", "foo", "bar", "foo", "bar", "foo", "foo"],
                "B": ["one", "one", "two", "three", "two", "two", "one", "three"],
                "C": np.random.randn(8),
                "D": np.random.randn(8),
            }
        )
        s = Series(
            np.repeat(np.arange(8), 2), index=np.repeat(np.arange(8), 2), name="TEST"
        )
        inner = df.join(s, how="inner")
        outer = df.join(s, how="outer")
        left = df.join(s, how="left")
        right = df.join(s, how="right")
        assert_frame_equal(inner, outer)
        assert_frame_equal(inner, left)
        assert_frame_equal(inner, right)

    def test_join_sort(self):
        left = DataFrame({"key": ["foo", "bar", "baz", "foo"], "value": [1, 2, 3, 4]})
        right = DataFrame({"value2": ["a", "b", "c"]}, index=["bar", "baz", "foo"])

        joined = left.join(right, on="key", sort=True)
        expected = DataFrame(
            {
                "key": ["bar", "baz", "foo", "foo"],
                "value": [2, 3, 1, 4],
                "value2": ["a", "b", "c", "c"],
            },
            index=[1, 2, 0, 3],
        )
        assert_frame_equal(joined, expected)

        # smoke test
        joined = left.join(right, on="key", sort=False)
        tm.assert_index_equal(joined.index, pd.Index(list(range(4))))

    def test_join_mixed_non_unique_index(self):
        # GH 12814, unorderable types in py3 with a non-unique index
        df1 = DataFrame({"a": [1, 2, 3, 4]}, index=[1, 2, 3, "a"])
        df2 = DataFrame({"b": [5, 6, 7, 8]}, index=[1, 3, 3, 4])
        result = df1.join(df2)
        expected = DataFrame(
            {"a": [1, 2, 3, 3, 4], "b": [5, np.nan, 6, 7, np.nan]},
            index=[1, 2, 3, 3, "a"],
        )
        tm.assert_frame_equal(result, expected)

        df3 = DataFrame({"a": [1, 2, 3, 4]}, index=[1, 2, 2, "a"])
        df4 = DataFrame({"b": [5, 6, 7, 8]}, index=[1, 2, 3, 4])
        result = df3.join(df4)
        expected = DataFrame(
            {"a": [1, 2, 3, 4], "b": [5, 6, 6, np.nan]}, index=[1, 2, 2, "a"]
        )
        tm.assert_frame_equal(result, expected)

    def test_join_non_unique_period_index(self):
        # GH #16871
        index = pd.period_range("2016-01-01", periods=16, freq="M")
        df = DataFrame([i for i in range(len(index))], index=index, columns=["pnum"])
        df2 = concat([df, df])
        result = df.join(df2, how="inner", rsuffix="_df2")
        expected = DataFrame(
            np.tile(np.arange(16, dtype=np.int64).repeat(2).reshape(-1, 1), 2),
            columns=["pnum", "pnum_df2"],
            index=df2.sort_index().index,
        )
        tm.assert_frame_equal(result, expected)

    def test_mixed_type_join_with_suffix(self):
        # GH #916
        df = DataFrame(np.random.randn(20, 6), columns=["a", "b", "c", "d", "e", "f"])
        df.insert(0, "id", 0)
        df.insert(5, "dt", "foo")

        grouped = df.groupby("id")
        mn = grouped.mean()
        cn = grouped.count()

        # it works!
        mn.join(cn, rsuffix="_right")

    def test_join_many(self):
        df = DataFrame(np.random.randn(10, 6), columns=list("abcdef"))
        df_list = [df[["a", "b"]], df[["c", "d"]], df[["e", "f"]]]

        joined = df_list[0].join(df_list[1:])
        tm.assert_frame_equal(joined, df)

        df_list = [df[["a", "b"]][:-2], df[["c", "d"]][2:], df[["e", "f"]][1:9]]

        def _check_diff_index(df_list, result, exp_index):
            reindexed = [x.reindex(exp_index) for x in df_list]
            expected = reindexed[0].join(reindexed[1:])
            tm.assert_frame_equal(result, expected)

        # different join types
        joined = df_list[0].join(df_list[1:], how="outer")
        _check_diff_index(df_list, joined, df.index)

        joined = df_list[0].join(df_list[1:])
        _check_diff_index(df_list, joined, df_list[0].index)

        joined = df_list[0].join(df_list[1:], how="inner")
        _check_diff_index(df_list, joined, df.index[2:8])

        msg = "Joining multiple DataFrames only supported for joining on index"
        with pytest.raises(ValueError, match=msg):
            df_list[0].join(df_list[1:], on="a")

    def test_join_many_mixed(self):
        df = DataFrame(np.random.randn(8, 4), columns=["A", "B", "C", "D"])
        df["key"] = ["foo", "bar"] * 4
        df1 = df.loc[:, ["A", "B"]]
        df2 = df.loc[:, ["C", "D"]]
        df3 = df.loc[:, ["key"]]

        result = df1.join([df2, df3])
        assert_frame_equal(result, df)

    def test_join_dups(self):

        # joining dups
        df = concat(
            [
                DataFrame(np.random.randn(10, 4), columns=["A", "A", "B", "B"]),
                DataFrame(
                    np.random.randint(0, 10, size=20).reshape(10, 2), columns=["A", "C"]
                ),
            ],
            axis=1,
        )

        expected = concat([df, df], axis=1)
        result = df.join(df, rsuffix="_2")
        result.columns = expected.columns
        assert_frame_equal(result, expected)

        # GH 4975, invalid join on dups
        w = DataFrame(np.random.randn(4, 2), columns=["x", "y"])
        x = DataFrame(np.random.randn(4, 2), columns=["x", "y"])
        y = DataFrame(np.random.randn(4, 2), columns=["x", "y"])
        z = DataFrame(np.random.randn(4, 2), columns=["x", "y"])

        dta = x.merge(y, left_index=True, right_index=True).merge(
            z, left_index=True, right_index=True, how="outer"
        )
        dta = dta.merge(w, left_index=True, right_index=True)
        expected = concat([x, y, z, w], axis=1)
        expected.columns = ["x_x", "y_x", "x_y", "y_y", "x_x", "y_x", "x_y", "y_y"]
        assert_frame_equal(dta, expected)

    def test_join_multi_to_multi(self, join_type):
        # GH 20475
        leftindex = MultiIndex.from_product(
            [list("abc"), list("xy"), [1, 2]], names=["abc", "xy", "num"]
        )
        left = DataFrame({"v1": range(12)}, index=leftindex)

        rightindex = MultiIndex.from_product(
            [list("abc"), list("xy")], names=["abc", "xy"]
        )
        right = DataFrame({"v2": [100 * i for i in range(1, 7)]}, index=rightindex)

        result = left.join(right, on=["abc", "xy"], how=join_type)
        expected = (
            left.reset_index()
            .merge(right.reset_index(), on=["abc", "xy"], how=join_type)
            .set_index(["abc", "xy", "num"])
        )
        assert_frame_equal(expected, result)

        msg = (
            r"len\(left_on\) must equal the number of levels in the index" ' of "right"'
        )
        with pytest.raises(ValueError, match=msg):
            left.join(right, on="xy", how=join_type)

        with pytest.raises(ValueError, match=msg):
            right.join(left, on=["abc", "xy"], how=join_type)

    def test_join_on_tz_aware_datetimeindex(self):
        # GH 23931, 26335
        df1 = pd.DataFrame(
            {
                "date": pd.date_range(
                    start="2018-01-01", periods=5, tz="America/Chicago"
                ),
                "vals": list("abcde"),
            }
        )

        df2 = pd.DataFrame(
            {
                "date": pd.date_range(
                    start="2018-01-03", periods=5, tz="America/Chicago"
                ),
                "vals_2": list("tuvwx"),
            }
        )
        result = df1.join(df2.set_index("date"), on="date")
        expected = df1.copy()
        expected["vals_2"] = pd.Series([np.nan] * 2 + list("tuv"), dtype=object)
        assert_frame_equal(result, expected)


def _check_join(left, right, result, join_col, how="left", lsuffix="_x", rsuffix="_y"):

    # some smoke tests
    for c in join_col:
        assert result[c].notna().all()

    left_grouped = left.groupby(join_col)
    right_grouped = right.groupby(join_col)

    for group_key, group in result.groupby(join_col):
        l_joined = _restrict_to_columns(group, left.columns, lsuffix)
        r_joined = _restrict_to_columns(group, right.columns, rsuffix)

        try:
            lgroup = left_grouped.get_group(group_key)
        except KeyError:
            if how in ("left", "inner"):
                raise AssertionError(
                    "key {group_key!s} should not have been in the join".format(
                        group_key=group_key
                    )
                )

            _assert_all_na(l_joined, left.columns, join_col)
        else:
            _assert_same_contents(l_joined, lgroup)

        try:
            rgroup = right_grouped.get_group(group_key)
        except KeyError:
            if how in ("right", "inner"):
                raise AssertionError(
                    "key {group_key!s} should not have been in the join".format(
                        group_key=group_key
                    )
                )

            _assert_all_na(r_joined, right.columns, join_col)
        else:
            _assert_same_contents(r_joined, rgroup)


def _restrict_to_columns(group, columns, suffix):
    found = [
        c for c in group.columns if c in columns or c.replace(suffix, "") in columns
    ]

    # filter
    group = group.loc[:, found]

    # get rid of suffixes, if any
    group = group.rename(columns=lambda x: x.replace(suffix, ""))

    # put in the right order...
    group = group.loc[:, columns]

    return group


def _assert_same_contents(join_chunk, source):
    NA_SENTINEL = -1234567  # drop_duplicates not so NA-friendly...

    jvalues = join_chunk.fillna(NA_SENTINEL).drop_duplicates().values
    svalues = source.fillna(NA_SENTINEL).drop_duplicates().values

    rows = {tuple(row) for row in jvalues}
    assert len(rows) == len(source)
    assert all(tuple(row) in rows for row in svalues)


def _assert_all_na(join_chunk, source_columns, join_col):
    for c in source_columns:
        if c in join_col:
            continue
        assert join_chunk[c].isna().all()


def _join_by_hand(a, b, how="left"):
    join_index = a.index.join(b.index, how=how)

    a_re = a.reindex(join_index)
    b_re = b.reindex(join_index)

    result_columns = a.columns.append(b.columns)

    for col, s in b_re.items():
        a_re[col] = s
    return a_re.reindex(columns=result_columns)