def test_missing_marker_5(self):
"afunc.missing_marker_5"
import datetime
d = datetime.date(2011, 1, 1)
assert_equal(missing_marker(larry([d])), None)
def test_missing_marker_3(self):
"afunc.missing_marker_3"
assert_equal(missing_marker(larry([True])), NotImplemented)
def test_missing_marker_4(self):
"afunc.missing_marker_4"
assert_equal(missing_marker(larry(['a'])), '')
def test_missing_marker_1(self):
"afunc.missing_marker_1"
assert_equal(missing_marker(larry([1])), NotImplemented)
def test_missing_marker_2(self):
"afunc.missing_marker_2"
assert_equal(missing_marker(larry([1.0])), nan)
def test_missing_marker_4a(self):
"afunc.missing_marker_4a"
assert_equal(missing_marker(np.array(['a'])), '')
def test_missing_marker_3a(self):
"afunc.missing_marker_3a"
assert_equal(missing_marker(np.array([True])), NotImplemented)
def test_missing_marker_2a(self):
"afunc.missing_marker_2a"
assert_equal(missing_marker(np.array([1.0])), nan)
def align_raw(lar1, lar2, join='inner', cast=True):
"""
Align two larrys but return Numpy arrays and label instead of larrys.
This function is the same as la.align() except that instead of returning
two larrys, the components of the two larrys are returned (two Numpy
arrays, a label, and flags for whether the two Numpy arrays are views of
the data arrays of the corresponding input larrys).
Parameters
----------
lar1 : larry
One of the input larrys. Must have the same number of dimensions as
`lar2`.
lar2 : larry
One of the input larrys. Must have the same number of dimensions as
`lar1`.
join : {'inner', 'outer', 'left', 'right', list}, optional
The join method used to align the two larrys. The default join method
along each axis is 'inner', i.e., the intersection of the labels. If
`join` is a list of strings then the length of the list should be the
same as the number of dimensions of the two larrys. The first element
in the list is the join method for axis=0, the second element is the
join method for axis=1, and so on.
cast : bool, optional
Only float, str, and object dtypes have missing value markers (la.nan,
'', and None, respectively). Other dtypes, such as int and bool, do
not have missing value markers. If `cast` is set to True (default)
then int and bool dtypes, for example, will be cast to float if any
new rows, columns, etc are created. If cast is set to False, then a
TypeError will be raised for int and bool dtype input if the join
introduces new rows, columns, etc. An inner join will never introduce
new rows, columns, etc.
Returns
-------
x1 : ndarray
The aligned version of `lar1`.
x2 : ndarray
The aligned version of `lar2`.
label : list of lists
The label of the joined larrys.
x1isview : bool
True if x1 is a view of lar1.x; False otherwise. A view of lar1.x is
retuned if the labels of `lar1` and `lar2` are the same along all
axes; otherwise a copy is returned.
x2isview : bool
True if x2 is a view of lar2.x; False otherwise. A view of lar2.x is
retuned if the labels of `lar1` and `lar2` are the same along all
axes; otherwise a copy is returned.
See Also
--------
la.align: Align two larrys using one of five join methods.
Notes
-----
The returned Numpy arrays are views of the corresponding input larrys if
the labels of the two input larrys are the same along all axes. If the
labels are not the same along any axis then a copy is returned.
Examples
--------
Create two larrys:
>>> y1 = larry([1, 2])
>>> y2 = larry([1, 2, 3])
The default join method is an inner join:
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2)
>>> x1
array([1, 2])
>>> x2
array([1, 2])
>>> label
[[0, 1]]
>>> x1isview
False
>>> x2isview
False
An outer join adds a missing value (NaN) to lar1, therefore the the dtype
of lar1 is changed from int to float:
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2, join='outer')
>>> x1
array([ 1., 2., NaN])
>>> x2
array([1, 2, 3])
>>> label
[[0, 1, 2]]
>>> x1isview
False
>>> x2isview
False
If the labels are already aligned, then a view of the data array is
returned:
>>> lar1 = larry([1, 2])
>>> lar2 = larry([3, 4])
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2)
>>> x1isview
True
>>> x2isview
True
"""
# Check number of dimensions
ndim = lar2.ndim
if lar1.ndim != ndim:
msg = "'lar1' and 'lar2' must have the same number of dimensions."
raise ValueError, msg
# Check join type
typejoin = type(join)
if typejoin is str:
join = [join] * ndim
elif typejoin is list:
if len(join) != ndim:
msg = "Length of `join` list equal number of dimension of `lar1`."
raise ValueError, msg
else:
raise TypeError, "`join` must be a string or a list."
# Initialize missing markers, set value later (in loop) only if needed.
# The weird initialization value ensures a user would never pick the same
undefined = 'aB!@12#E~=-'
miss1 = undefined
miss2 = undefined
# For loop initialization
label = []
x1 = lar1.x
x2 = lar2.x
label1 = lar1.label
label2 = lar2.label
x1isview = True
x2isview = True
# Loop: align one axis at a time
msg = "`fill` type not compatible with larry dtype"
for ax in range(ndim):
list1 = label1[ax]
list2 = label2[ax]
joinax = join[ax]
if joinax == 'inner':
if list1 == list2:
list3 = list(list1)
else:
list3 = list(set(list1) & (set(list2)))
list3.sort()
idx1 = listmap(list1, list3)
idx2 = listmap(list2, list3)
x1 = x1.take(idx1, ax)
x2 = x2.take(idx2, ax)
x1isview = False
x2isview = False
elif joinax == 'outer':
if list1 == list2:
list3 = list(list1)
else:
list3 = list(set(list1) | (set(list2)))
list3.sort()
idx1, idx1_miss = listmap_fill(list1, list3, fill=0)
idx2, idx2_miss = listmap_fill(list2, list3, fill=0)
x1 = x1.take(idx1, ax)
x2 = x2.take(idx2, ax)
if len(idx1_miss) > 0:
if miss1 == undefined:
miss1 = missing_marker(lar1)
if miss1 == NotImplemented:
if cast:
x1 = x1.astype(float)
miss1 = missing_marker(x1)
else:
raise TypeError, msg
index1 = [slice(None)] * ndim
index1[ax] = idx1_miss
x1[index1] = miss1
if len(idx2_miss) > 0:
if miss2 == undefined:
miss2 = missing_marker(lar2)
if miss2 == NotImplemented:
if cast:
x2 = x2.astype(float)
miss2 = missing_marker(x2)
else:
raise TypeError, msg
index2 = [slice(None)] * ndim
index2[ax] = idx2_miss
x2[index2] = miss2
x1isview = False
x2isview = False
elif joinax == 'left':
list3 = list(list1)
if list1 != list2:
idx2, idx2_miss = listmap_fill(list2, list3, fill=0)
x2 = x2.take(idx2, ax)
if len(idx2_miss) > 0:
if miss2 == undefined:
miss2 = missing_marker(lar2)
if miss2 == NotImplemented:
if miss2 is None:
miss2 = missing_marker(lar2)
if miss2 is None:
miss2 = missing_marker(lar2)
if cast:
x2 = x2.astype(float)
miss2 = missing_marker(x2)
else:
raise TypeError, msg
index2 = [slice(None)] * ndim
index2[ax] = idx2_miss
x2[index2] = miss2
x2isview = False
elif joinax == 'right':
list3 = list(list2)
if list1 != list2:
idx1, idx1_miss = listmap_fill(list1, list3, fill=0)
x1 = x1.take(idx1, ax)
if len(idx1_miss) > 0:
if miss1 == undefined:
miss1 = missing_marker(lar1)
if miss1 == NotImplemented:
if cast:
x1 = x1.astype(float)
miss1 = missing_marker(x1)
else:
raise TypeError, msg
index1 = [slice(None)] * ndim
index1[ax] = idx1_miss
x1[index1] = miss1
x1isview = False
else:
raise ValueError, 'join type not recognized'
label.append(list3)
return x1, x2, label, x1isview, x2isview
def align_raw(lar1, lar2, join='inner', cast=True):
"""
Align two larrys but return Numpy arrays and label instead of larrys.
This function is the same as la.align() except that instead of returning
two larrys, the components of the two larrys are returned (two Numpy
arrays, a label, and flags for whether the two Numpy arrays are views of
the data arrays of the corresponding input larrys).
Parameters
----------
lar1 : larry
One of the input larrys. Must have the same number of dimensions as
`lar2`.
lar2 : larry
One of the input larrys. Must have the same number of dimensions as
`lar1`.
join : {'inner', 'outer', 'left', 'right', list}, optional
The join method used to align the two larrys. The default join method
along each axis is 'inner', i.e., the intersection of the labels. If
`join` is a list of strings then the length of the list should be the
same as the number of dimensions of the two larrys. The first element
in the list is the join method for axis=0, the second element is the
join method for axis=1, and so on.
cast : bool, optional
Only float, str, and object dtypes have missing value markers (la.nan,
'', and None, respectively). Other dtypes, such as int and bool, do
not have missing value markers. If `cast` is set to True (default)
then int and bool dtypes, for example, will be cast to float if any
new rows, columns, etc are created. If cast is set to False, then a
TypeError will be raised for int and bool dtype input if the join
introduces new rows, columns, etc. An inner join will never introduce
new rows, columns, etc.
Returns
-------
x1 : ndarray
The aligned version of `lar1`.
x2 : ndarray
The aligned version of `lar2`.
label : list of lists
The label of the joined larrys.
x1isview : bool
True if x1 is a view of lar1.x; False otherwise. A view of lar1.x is
retuned if the labels of `lar1` and `lar2` are the same along all
axes; otherwise a copy is returned.
x2isview : bool
True if x2 is a view of lar2.x; False otherwise. A view of lar2.x is
retuned if the labels of `lar1` and `lar2` are the same along all
axes; otherwise a copy is returned.
See Also
--------
la.align: Align two larrys using one of five join methods.
Notes
-----
The returned Numpy arrays are views of the corresponding input larrys if
the labels of the two input larrys are the same along all axes. If the
labels are not the same along any axis then a copy is returned.
Examples
--------
Create two larrys:
>>> y1 = larry([1, 2])
>>> y2 = larry([1, 2, 3])
The default join method is an inner join:
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2)
>>> x1
array([1, 2])
>>> x2
array([1, 2])
>>> label
[[0, 1]]
>>> x1isview
False
>>> x2isview
False
An outer join adds a missing value (NaN) to lar1, therefore the the dtype
of lar1 is changed from int to float:
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2, join='outer')
>>> x1
array([ 1., 2., NaN])
>>> x2
array([1, 2, 3])
>>> label
[[0, 1, 2]]
>>> x1isview
False
>>> x2isview
False
If the labels are already aligned, then a view of the data array is
returned:
>>> lar1 = larry([1, 2])
>>> lar2 = larry([3, 4])
>>> x1, x2, label, x1isview, x2isview = la.flarry._align_raw(lar1, lar2)
>>> x1isview
True
>>> x2isview
True
"""
# Check number of dimensions
ndim = lar2.ndim
if lar1.ndim != ndim:
msg = "'lar1' and 'lar2' must have the same number of dimensions."
raise ValueError, msg
# Check join type
typejoin = type(join)
if typejoin is str:
join = [join] * ndim
elif typejoin is list:
if len(join) != ndim:
msg = "Length of `join` list equal number of dimension of `lar1`."
raise ValueError, msg
else:
raise TypeError, "`join` must be a string or a list."
# Initialize missing markers, set value later (in loop) only if needed.
# The weird initialization value ensures a user would never pick the same
undefined = 'aB!@12#E~=-'
miss1 = undefined
miss2 = undefined
# For loop initialization
label = []
x1 = lar1.x
x2 = lar2.x
label1 = lar1.label
label2 = lar2.label
x1isview = True
x2isview = True
# Loop: align one axis at a time
msg = "`fill` type not compatible with larry dtype"
for ax in range(ndim):
list1 = label1[ax]
list2 = label2[ax]
joinax = join[ax]
if joinax == 'inner':
if list1 == list2:
list3 = list(list1)
else:
list3 = list(set(list1) & (set(list2)))
list3.sort()
idx1 = listmap(list1, list3)
idx2 = listmap(list2, list3)
x1 = x1.take(idx1, ax)
x2 = x2.take(idx2, ax)
x1isview = False
x2isview = False
elif joinax == 'outer':
if list1 == list2:
list3 = list(list1)
else:
list3 = list(set(list1) | (set(list2)))
list3.sort()
idx1, idx1_miss = listmap_fill(list1, list3, fill=0)
idx2, idx2_miss = listmap_fill(list2, list3, fill=0)
x1 = x1.take(idx1, ax)
x2 = x2.take(idx2, ax)
if len(idx1_miss) > 0:
if miss1 == undefined:
miss1 = missing_marker(lar1)
if miss1 == NotImplemented:
if cast:
x1 = x1.astype(float)
miss1 = missing_marker(x1)
else:
raise TypeError, msg
index1 = [slice(None)] * ndim
index1[ax] = idx1_miss
x1[index1] = miss1
if len(idx2_miss) > 0:
if miss2 == undefined:
miss2 = missing_marker(lar2)
if miss2 == NotImplemented:
if cast:
x2 = x2.astype(float)
miss2 = missing_marker(x2)
else:
raise TypeError, msg
index2 = [slice(None)] * ndim
index2[ax] = idx2_miss
x2[index2] = miss2
x1isview = False
x2isview = False
elif joinax == 'left':
list3 = list(list1)
if list1 != list2:
idx2, idx2_miss = listmap_fill(list2, list3, fill=0)
x2 = x2.take(idx2, ax)
if len(idx2_miss) > 0:
if miss2 == undefined:
miss2 = missing_marker(lar2)
if miss2 == NotImplemented:
if miss2 is None:
miss2 = missing_marker(lar2)
if miss2 is None:
miss2 = missing_marker(lar2)
if cast:
x2 = x2.astype(float)
miss2 = missing_marker(x2)
else:
raise TypeError, msg
index2 = [slice(None)] * ndim
index2[ax] = idx2_miss
x2[index2] = miss2
x2isview = False
elif joinax == 'right':
list3 = list(list2)
if list1 != list2:
idx1, idx1_miss = listmap_fill(list1, list3, fill=0)
x1 = x1.take(idx1, ax)
if len(idx1_miss) > 0:
if miss1 == undefined:
miss1 = missing_marker(lar1)
if miss1 == NotImplemented:
if cast:
x1 = x1.astype(float)
miss1 = missing_marker(x1)
else:
raise TypeError, msg
index1 = [slice(None)] * ndim
index1[ax] = idx1_miss
x1[index1] = miss1
x1isview = False
else:
raise ValueError, 'join type not recognized'
label.append(list3)
return x1, x2, label, x1isview, x2isview