test_join.py 4.7 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155
  1. from datetime import (
  2. datetime,
  3. timezone,
  4. )
  5. import numpy as np
  6. import pytest
  7. from pandas import (
  8. DatetimeIndex,
  9. Index,
  10. Timestamp,
  11. date_range,
  12. to_datetime,
  13. )
  14. import pandas._testing as tm
  15. from pandas.tseries.offsets import (
  16. BDay,
  17. BMonthEnd,
  18. )
  19. class TestJoin:
  20. def test_does_not_convert_mixed_integer(self):
  21. df = tm.makeCustomDataframe(
  22. 10,
  23. 10,
  24. data_gen_f=lambda *args, **kwargs: np.random.randn(),
  25. r_idx_type="i",
  26. c_idx_type="dt",
  27. )
  28. cols = df.columns.join(df.index, how="outer")
  29. joined = cols.join(df.columns)
  30. assert cols.dtype == np.dtype("O")
  31. assert cols.dtype == joined.dtype
  32. tm.assert_numpy_array_equal(cols.values, joined.values)
  33. def test_join_self(self, join_type):
  34. index = date_range("1/1/2000", periods=10)
  35. joined = index.join(index, how=join_type)
  36. assert index is joined
  37. def test_join_with_period_index(self, join_type):
  38. df = tm.makeCustomDataframe(
  39. 10,
  40. 10,
  41. data_gen_f=lambda *args: np.random.randint(2),
  42. c_idx_type="p",
  43. r_idx_type="dt",
  44. )
  45. s = df.iloc[:5, 0]
  46. expected = df.columns.astype("O").join(s.index, how=join_type)
  47. result = df.columns.join(s.index, how=join_type)
  48. tm.assert_index_equal(expected, result)
  49. def test_join_object_index(self):
  50. rng = date_range("1/1/2000", periods=10)
  51. idx = Index(["a", "b", "c", "d"])
  52. result = rng.join(idx, how="outer")
  53. assert isinstance(result[0], Timestamp)
  54. def test_join_utc_convert(self, join_type):
  55. rng = date_range("1/1/2011", periods=100, freq="H", tz="utc")
  56. left = rng.tz_convert("US/Eastern")
  57. right = rng.tz_convert("Europe/Berlin")
  58. result = left.join(left[:-5], how=join_type)
  59. assert isinstance(result, DatetimeIndex)
  60. assert result.tz == left.tz
  61. result = left.join(right[:-5], how=join_type)
  62. assert isinstance(result, DatetimeIndex)
  63. assert result.tz is timezone.utc
  64. def test_datetimeindex_union_join_empty(self, sort):
  65. dti = date_range(start="1/1/2001", end="2/1/2001", freq="D")
  66. empty = Index([])
  67. result = dti.union(empty, sort=sort)
  68. expected = dti.astype("O")
  69. tm.assert_index_equal(result, expected)
  70. result = dti.join(empty)
  71. assert isinstance(result, DatetimeIndex)
  72. tm.assert_index_equal(result, dti)
  73. def test_join_nonunique(self):
  74. idx1 = to_datetime(["2012-11-06 16:00:11.477563", "2012-11-06 16:00:11.477563"])
  75. idx2 = to_datetime(["2012-11-06 15:11:09.006507", "2012-11-06 15:11:09.006507"])
  76. rs = idx1.join(idx2, how="outer")
  77. assert rs.is_monotonic_increasing
  78. @pytest.mark.parametrize("freq", ["B", "C"])
  79. def test_outer_join(self, freq):
  80. # should just behave as union
  81. start, end = datetime(2009, 1, 1), datetime(2010, 1, 1)
  82. rng = date_range(start=start, end=end, freq=freq)
  83. # overlapping
  84. left = rng[:10]
  85. right = rng[5:10]
  86. the_join = left.join(right, how="outer")
  87. assert isinstance(the_join, DatetimeIndex)
  88. # non-overlapping, gap in middle
  89. left = rng[:5]
  90. right = rng[10:]
  91. the_join = left.join(right, how="outer")
  92. assert isinstance(the_join, DatetimeIndex)
  93. assert the_join.freq is None
  94. # non-overlapping, no gap
  95. left = rng[:5]
  96. right = rng[5:10]
  97. the_join = left.join(right, how="outer")
  98. assert isinstance(the_join, DatetimeIndex)
  99. # overlapping, but different offset
  100. other = date_range(start, end, freq=BMonthEnd())
  101. the_join = rng.join(other, how="outer")
  102. assert isinstance(the_join, DatetimeIndex)
  103. assert the_join.freq is None
  104. def test_naive_aware_conflicts(self):
  105. start, end = datetime(2009, 1, 1), datetime(2010, 1, 1)
  106. naive = date_range(start, end, freq=BDay(), tz=None)
  107. aware = date_range(start, end, freq=BDay(), tz="Asia/Hong_Kong")
  108. msg = "tz-naive.*tz-aware"
  109. with pytest.raises(TypeError, match=msg):
  110. naive.join(aware)
  111. with pytest.raises(TypeError, match=msg):
  112. aware.join(naive)
  113. @pytest.mark.parametrize("tz", [None, "US/Pacific"])
  114. def test_join_preserves_freq(self, tz):
  115. # GH#32157
  116. dti = date_range("2016-01-01", periods=10, tz=tz)
  117. result = dti[:5].join(dti[5:], how="outer")
  118. assert result.freq == dti.freq
  119. tm.assert_index_equal(result, dti)
  120. result = dti[:5].join(dti[6:], how="outer")
  121. assert result.freq is None
  122. expected = dti.delete(5)
  123. tm.assert_index_equal(result, expected)