def test_2d_multiindex_rows():
a = xarray.DataArray(
np.arange(2 * 3 * 4).reshape((2, 3, 4)),
dims=['x', 'y', 'z'],
coords={
'x': ['x0', 'x1'],
'y': ['y2', 'y0', 'y1'], # test order is not compromised
'z': ['z0', 'z2', 'z1', 'z3']
})
b = a.stack(dim_0=['y', 'z']).T
txt = ("x,,x0,x1\n"
"y,z,,\n"
"y2,z0,0,12\n"
"y2,z2,1,13\n"
"y2,z1,2,14\n"
"y2,z3,3,15\n"
"y0,z0,4,16\n"
"y0,z2,5,17\n"
"y0,z1,6,18\n"
"y0,z3,7,19\n"
"y1,z0,8,20\n"
"y1,z2,9,21\n"
"y1,z1,10,22\n"
"y1,z3,11,23\n")
buf = io.StringIO()
write_csv(b, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(a, c)
buf.seek(0)
c = read_csv(buf, unstack=False)
xarray.testing.assert_equal(b, c)
def test_2d_multiindex_both(explicit_stack):
a = xarray.DataArray(np.arange(16).reshape((2, 2, 2, 2)),
dims=['x', 'y', 'z', 'w'],
coords={
'x': ['x0', 'x1'],
'y': ['y0', 'y1'],
'z': ['z0', 'z1'],
'w': ['w0', 'w1']
})
b = a.stack(dim_0=['x', 'y']).transpose('dim_0', 'z', 'w')
c = b.stack(dim_1=['z', 'w'])
txt = ("z,,z0,z0,z1,z1\n"
"w,,w0,w1,w0,w1\n"
"x,y,,,,\n"
"x0,y0,0,1,2,3\n"
"x0,y1,4,5,6,7\n"
"x1,y0,8,9,10,11\n"
"x1,y1,12,13,14,15\n")
buf = io.StringIO()
if explicit_stack:
write_csv(c, buf)
else:
write_csv(b, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
d = read_csv(buf)
xarray.testing.assert_equal(a, d)
buf.seek(0)
d = read_csv(buf, unstack=False)
xarray.testing.assert_equal(c, d)
def test_duplicate_index_multiindex():
"""Duplicate indices are OK as long as you don't try unstacking
"""
a = xarray.DataArray(
[1, 2],
dims=['dim_0'],
coords={
'dim_0':
pandas.MultiIndex.from_tuples([(10, 10), (10, 10)],
names=['x', 'y'])
})
txt = ("x,y,\n" "10,10,1\n" "10,10,2\n")
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
b = read_csv(buf, unstack=False)
xarray.testing.assert_equal(a, b)
buf.seek(0)
with pytest.raises(ValueError) as e:
read_csv(buf, unstack=True)
assert str(e.value) == ("cannot reindex or align along dimension 'dim_0' "
"because the index has duplicate values")
def test_2d_multiindex_cols(explicit_stack):
a = xarray.DataArray(
np.arange(2 * 3 * 4).reshape((2, 3, 4)),
dims=['x', 'y', 'z'],
coords={
'x': ['x0', 'x1'],
'y': ['y2', 'y0', 'y1'], # test order is not compromised
'z': ['z0', 'z2', 'z1', 'z3']
})
b = a.stack(dim_1=['y', 'z'])
txt = ("y,y2,y2,y2,y2,y0,y0,y0,y0,y1,y1,y1,y1\n"
"z,z0,z2,z1,z3,z0,z2,z1,z3,z0,z2,z1,z3\n"
"x,,,,,,,,,,,,\n"
"x0,0,1,2,3,4,5,6,7,8,9,10,11\n"
"x1,12,13,14,15,16,17,18,19,20,21,22,23\n")
buf = io.StringIO()
# >2 dimensional arrays get their dimensions beyond the first automatically
# stacked
if explicit_stack:
write_csv(b, buf)
else:
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(a, c)
buf.seek(0)
c = read_csv(buf, unstack=False)
xarray.testing.assert_equal(b, c)
def test_ambiguous_nonindex_coords():
"""Test when non-index coords have multiple values for the matching
index coord
"""
buf = io.StringIO("x,y,z (x),\n" "0,0,0,1\n" "0,1,1,1\n")
with pytest.raises(ValueError) as e:
read_csv(buf)
assert str(e.value) == ("Non-index coord z (x) has different values for "
"the same value of its dimension x")
def test_coords_bool():
buf = io.StringIO("y,true,false,TRUE,FALSE,True,False,Y,N,y,n,YES,NO,"
"Yes,No,yes,no,T,F,T,F\n"
"x\n"
"x0" + ",1" * 20 + "\n")
a = read_csv(buf)
assert a.y.values.tolist() == [True, False] * 10
def test_nonindex_coords_with_multiindex(unstack):
buf = io.StringIO(
"x,y,z (x),\nx1,y1,z1,1\nx1,y2,z1,2\nx2,y1,z2,3\nx2,y2,z2,4\n")
a = read_csv(buf, unstack=unstack)
if unstack:
b = xarray.DataArray(
[[1, 2], [3, 4]],
dims=["x", "y"],
coords={
"x": ["x1", "x2"],
"y": ["y1", "y2"],
"z": ("x", ["z1", "z2"]),
},
)
else:
b = xarray.DataArray(
[1, 2, 3, 4],
dims=["dim_0"],
coords={
"dim_0":
pandas.MultiIndex.from_tuples(
[("x1", "y1"), ("x1", "y2"), ("x2", "y1"), ("x2", "y2")],
names=["x", "y"],
),
"z": ("dim_0", ["z1", "z1", "z2", "z2"]),
},
)
xarray.testing.assert_equal(a, b)
def test_1d_multiindex():
a = xarray.DataArray(
[[1, 2], [3, 4]], dims=["r", "c"], coords={"r": [10, 20], "c": [30, 40]}
)
b = a.stack(dim_0=["r", "c"])
txt = "r,c,\n10,30,1\n10,40,2\n20,30,3\n20,40,4\n"
buf = io.StringIO()
write_csv(b, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(c, a)
buf.seek(0)
c = read_csv(buf, unstack=False)
xarray.testing.assert_equal(c, b)
def test_nonindex_coords_with_multiindex(unstack):
buf = io.StringIO("x,y,z (x),\n"
"x1,y1,z1,1\n"
"x1,y2,z1,2\n"
"x2,y1,z2,3\n"
"x2,y2,z2,4\n")
a = read_csv(buf, unstack=unstack)
if unstack:
b = xarray.DataArray([[1, 2], [3, 4]],
dims=['x', 'y'],
coords={
'x': ['x1', 'x2'],
'y': ['y1', 'y2'],
'z': ('x', ['z1', 'z2']),
})
else:
b = xarray.DataArray(
[1, 2, 3, 4],
dims=['dim_0'],
coords={
'dim_0':
pandas.MultiIndex.from_tuples([('x1', 'y1'), ('x1', 'y2'),
('x2', 'y1'), ('x2', 'y2')],
names=['x', 'y']),
'z': ('dim_0', ['z1', 'z1', 'z2', 'z2']),
})
xarray.testing.assert_equal(a, b)
def test_file_io(tmpdir, open_func, ext):
a = xarray.DataArray(1)
fname = f"{tmpdir}/test.{ext}"
write_csv(a, fname)
with open_func(fname, "rt") as fh:
assert fh.read() == "1\n"
b = read_csv(fname)
xarray.testing.assert_equal(a, b)
def test_numerical_ids2():
a = xarray.DataArray([1, 2], dims=['x'], coords={'x': ['01', '02']})
b = xarray.DataArray([1, 2], dims=['x'], coords={'x': [1, 2]})
buf = io.StringIO()
write_csv(a, buf)
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(b, c)
def test_buf_io():
a = xarray.DataArray(1)
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue() == '1\n'
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_file_io(tmpdir, open_func, ext):
a = xarray.DataArray(1)
fname = '%s/test.%s' % (tmpdir, ext)
write_csv(a, fname)
with open_func(fname, 'rt') as fh:
assert fh.read() == '1\n'
b = read_csv(fname)
xarray.testing.assert_equal(a, b)
def test_numerical_ids2():
a = xarray.DataArray([1, 2], dims=["x"], coords={"x": ["01", "02"]})
b = xarray.DataArray([1, 2], dims=["x"], coords={"x": [1, 2]})
buf = io.StringIO()
write_csv(a, buf)
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(b, c)
def test_missing_index_coord1(unstack):
"""The index coord may be missing as long as matching non-index coord(s)
spell out the dim name
"""
buf = io.StringIO("y (x),\n" "10,1\n" "20,2\n")
b = xarray.DataArray([1, 2], dims=['x'], coords={'y': ('x', [10, 20])})
a = read_csv(buf, unstack=unstack)
xarray.testing.assert_equal(a, b)
def test_1d(data, txt):
a = xarray.DataArray(data, dims=["x"], coords={"x": ["x1", "x2"]})
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_1d(data, txt):
a = xarray.DataArray(data, dims=['x'], coords={'x': ['x1', 'x2']})
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_0d(data, txt):
a = xarray.DataArray(data)
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_shape1():
# Test the edge case of an array with shape (1, )
a = xarray.DataArray([1], dims=['x'], coords={'x': ['x1']})
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == 'x,\nx1,1\n'
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_shape1():
# Test the edge case of an array with shape (1, )
a = xarray.DataArray([1], dims=["x"], coords={"x": ["x1"]})
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace("\r", "") == "x,\nx1,1\n"
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_1d_multiindex():
a = xarray.DataArray([[1, 2], [3, 4]],
dims=['r', 'c'],
coords={
'r': [10, 20],
'c': [30, 40]
})
b = a.stack(dim_0=['r', 'c'])
txt = ("r,c,\n" "10,30,1\n" "10,40,2\n" "20,30,3\n" "20,40,4\n")
buf = io.StringIO()
write_csv(b, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(c, a)
buf.seek(0)
c = read_csv(buf, unstack=False)
xarray.testing.assert_equal(c, b)
def test_nonindex_coords(unstack):
a = xarray.DataArray(
[1, 2], dims=["x"], coords={"x": [10, 20], "y": ("x", [30, 40])}
)
txt = "x,y (x),\n10,30,1\n20,40,2\n"
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
b = read_csv(buf, unstack=unstack)
xarray.testing.assert_equal(a, b)
def test_numerical_ids1():
"""An index full of numerical IDs padded on the left with zeros
won't accidentally convert them to int as long as at least one element
can't be cast to int
"""
a = xarray.DataArray([1, 2, 3], dims=["x"], coords={"x": ["01", "02", "S1"]})
buf = io.StringIO()
write_csv(a, buf)
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_duplicate_index():
"""Duplicate indices are OK as long as you don't try unstacking"""
a = xarray.DataArray([1, 2], dims=["x"], coords={"x": [10, 10]})
txt = "x,\n10,1\n10,2\n"
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
b = read_csv(buf)
xarray.testing.assert_equal(a, b)
def test_missing_index_coord2(unstack):
"""Same as above, but with 2 non-index coords which will instigate
pandas to create a MultiIndex
"""
buf = io.StringIO("y (x),z (x),\n" "10,30,1\n" "20,40,2\n")
b = xarray.DataArray([1, 2],
dims=['x'],
coords={
'y': ('x', [10, 20]),
'z': ('x', [30, 40])
})
a = read_csv(buf, unstack=unstack)
xarray.testing.assert_equal(a, b)
def test_xarray_nocoords():
a = xarray.DataArray([[1, 2], [3, 4]], dims=['r', 'c'])
b = a.copy()
b.coords['r'] = [0, 1]
b.coords['c'] = [0, 1]
txt = ("c,0,1\n" "r,,\n" "0,1,2\n" "1,3,4\n")
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(b, c)
def test_xarray_nocoords():
a = xarray.DataArray([[1, 2], [3, 4]], dims=["r", "c"])
b = a.copy()
b.coords["r"] = [0, 1]
b.coords["c"] = [0, 1]
txt = "c,0,1\nr,,\n0,1,2\n1,3,4\n"
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(b, c)
def test_2d_multiindex_rows():
a = xarray.DataArray(
np.arange(2 * 3 * 4).reshape((2, 3, 4)),
dims=["x", "y", "z"],
coords={
"x": ["x0", "x1"],
"y": ["y2", "y0", "y1"], # test order is not compromised
"z": ["z0", "z2", "z1", "z3"],
},
)
b = a.stack(dim_0=["y", "z"]).T
txt = (
"x,,x0,x1\n"
"y,z,,\n"
"y2,z0,0,12\n"
"y2,z2,1,13\n"
"y2,z1,2,14\n"
"y2,z3,3,15\n"
"y0,z0,4,16\n"
"y0,z2,5,17\n"
"y0,z1,6,18\n"
"y0,z3,7,19\n"
"y1,z0,8,20\n"
"y1,z2,9,21\n"
"y1,z1,10,22\n"
"y1,z3,11,23\n"
)
buf = io.StringIO()
write_csv(b, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
c = read_csv(buf)
xarray.testing.assert_equal(a, c)
buf.seek(0)
c = read_csv(buf, unstack=False)
xarray.testing.assert_equal(b, c)
def test_nonindex_coords(unstack):
a = xarray.DataArray([1, 2],
dims=['x'],
coords={
'x': [10, 20],
'y': ('x', [30, 40])
})
txt = ("x,y (x),\n" "10,30,1\n" "20,40,2\n")
buf = io.StringIO()
write_csv(a, buf)
assert buf.getvalue().replace('\r', '') == txt
buf.seek(0)
b = read_csv(buf, unstack=unstack)
xarray.testing.assert_equal(a, b)
def test_2d_multiindex_both(explicit_stack):
a = xarray.DataArray(
np.arange(16).reshape((2, 2, 2, 2)),
dims=["x", "y", "z", "w"],
coords={
"x": ["x0", "x1"],
"y": ["y0", "y1"],
"z": ["z0", "z1"],
"w": ["w0", "w1"],
},
)
b = a.stack(dim_0=["x", "y"]).transpose("dim_0", "z", "w")
c = b.stack(dim_1=["z", "w"])
txt = (
"z,,z0,z0,z1,z1\n"
"w,,w0,w1,w0,w1\n"
"x,y,,,,\n"
"x0,y0,0,1,2,3\n"
"x0,y1,4,5,6,7\n"
"x1,y0,8,9,10,11\n"
"x1,y1,12,13,14,15\n"
)
buf = io.StringIO()
if explicit_stack:
write_csv(c, buf)
else:
write_csv(b, buf)
assert buf.getvalue().replace("\r", "") == txt
buf.seek(0)
d = read_csv(buf)
xarray.testing.assert_equal(a, d)
buf.seek(0)
d = read_csv(buf, unstack=False)
xarray.testing.assert_equal(c, d)
